Merge branch 'master' into develop

.github/contributors/AlJohri.md

@@ -0,0 +1,106 @@
+# spaCy contributor agreement
+
+This spaCy Contributor Agreement (**"SCA"**) is based on the
+[Oracle Contributor Agreement](http://www.oracle.com/technetwork/oca-405177.pdf).
+The SCA applies to any contribution that you make to any product or project
+managed by us (the **"project"**), and sets out the intellectual property rights
+you grant to us in the contributed materials. The term **"us"** shall mean
+[ExplosionAI GmbH](https://explosion.ai/legal). The term
+**"you"** shall mean the person or entity identified below.
+
+If you agree to be bound by these terms, fill in the information requested
+below and include the filled-in version with your first pull request, under the
+folder [`.github/contributors/`](/.github/contributors/). The name of the file
+should be your GitHub username, with the extension `.md`. For example, the user
+example_user would create the file `.github/contributors/example_user.md`.
+
+Read this agreement carefully before signing. These terms and conditions
+constitute a binding legal agreement.
+
+## Contributor Agreement
+
+1. The term "contribution" or "contributed materials" means any source code,
+object code, patch, tool, sample, graphic, specification, manual,
+documentation, or any other material posted or submitted by you to the project.
+
+2. With respect to any worldwide copyrights, or copyright applications and
+registrations, in your contribution:
+
+    * you hereby assign to us joint ownership, and to the extent that such
+    assignment is or becomes invalid, ineffective or unenforceable, you hereby
+    grant to us a perpetual, irrevocable, non-exclusive, worldwide, no-charge,
+    royalty-free, unrestricted license to exercise all rights under those
+    copyrights. This includes, at our option, the right to sublicense these same
+    rights to third parties through multiple levels of sublicensees or other
+    licensing arrangements;
+
+    * you agree that each of us can do all things in relation to your
+    contribution as if each of us were the sole owners, and if one of us makes
+    a derivative work of your contribution, the one who makes the derivative
+    work (or has it made will be the sole owner of that derivative work;
+
+    * you agree that you will not assert any moral rights in your contribution
+    against us, our licensees or transferees;
+
+    * you agree that we may register a copyright in your contribution and
+    exercise all ownership rights associated with it; and
+
+    * you agree that neither of us has any duty to consult with, obtain the
+    consent of, pay or render an accounting to the other for any use or
+    distribution of your contribution.
+
+3. With respect to any patents you own, or that you can license without payment
+to any third party, you hereby grant to us a perpetual, irrevocable,
+non-exclusive, worldwide, no-charge, royalty-free license to:
+
+    * make, have made, use, sell, offer to sell, import, and otherwise transfer
+    your contribution in whole or in part, alone or in combination with or
+    included in any product, work or materials arising out of the project to
+    which your contribution was submitted, and
+
+    * at our option, to sublicense these same rights to third parties through
+    multiple levels of sublicensees or other licensing arrangements.
+
+4. Except as set out above, you keep all right, title, and interest in your
+contribution. The rights that you grant to us under these terms are effective
+on the date you first submitted a contribution to us, even if your submission
+took place before the date you sign these terms.
+
+5. You covenant, represent, warrant and agree that:
+
+    * Each contribution that you submit is and shall be an original work of
+    authorship and you can legally grant the rights set out in this SCA;
+
+    * to the best of your knowledge, each contribution will not violate any
+    third party's copyrights, trademarks, patents, or other intellectual
+    property rights; and
+
+    * each contribution shall be in compliance with U.S. export control laws and
+    other applicable export and import laws. You agree to notify us if you
+    become aware of any circumstance which would make any of the foregoing
+    representations inaccurate in any respect. We may publicly disclose your
+    participation in the project, including the fact that you have signed the SCA.
+
+6. This SCA is governed by the laws of the State of California and applicable
+U.S. Federal law. Any choice of law rules will not apply.
+
+7. Please place an “x” on one of the applicable statement below. Please do NOT
+mark both statements:
+
+    * [x] I am signing on behalf of myself as an individual and no other person
+    or entity, including my employer, has or will have rights with respect to my
+    contributions.
+
+    * [ ] I am signing on behalf of my employer or a legal entity and I have the
+    actual authority to contractually bind that entity.
+
+## Contributor Details
+
+| Field                          | Entry                |
+|------------------------------- | -------------------- |
+| Name                           | Al Johri             |
+| Company name (if applicable)   | N/A                  |
+| Title or role (if applicable)  | N/A                  |
+| Date                           | December 27th, 2019  |
+| GitHub username                | AlJohri              |
+| Website (optional)             | http://aljohri.com/  |

.github/contributors/Jan-711.md

@@ -0,0 +1,106 @@
+# spaCy contributor agreement
+
+This spaCy Contributor Agreement (**"SCA"**) is based on the
+[Oracle Contributor Agreement](http://www.oracle.com/technetwork/oca-405177.pdf).
+The SCA applies to any contribution that you make to any product or project
+managed by us (the **"project"**), and sets out the intellectual property rights
+you grant to us in the contributed materials. The term **"us"** shall mean
+[ExplosionAI GmbH](https://explosion.ai/legal). The term
+**"you"** shall mean the person or entity identified below.
+
+If you agree to be bound by these terms, fill in the information requested
+below and include the filled-in version with your first pull request, under the
+folder [`.github/contributors/`](/.github/contributors/). The name of the file
+should be your GitHub username, with the extension `.md`. For example, the user
+example_user would create the file `.github/contributors/example_user.md`.
+
+Read this agreement carefully before signing. These terms and conditions
+constitute a binding legal agreement.
+
+## Contributor Agreement
+
+1. The term "contribution" or "contributed materials" means any source code,
+object code, patch, tool, sample, graphic, specification, manual,
+documentation, or any other material posted or submitted by you to the project.
+
+2. With respect to any worldwide copyrights, or copyright applications and
+registrations, in your contribution:
+
+    * you hereby assign to us joint ownership, and to the extent that such
+    assignment is or becomes invalid, ineffective or unenforceable, you hereby
+    grant to us a perpetual, irrevocable, non-exclusive, worldwide, no-charge,
+    royalty-free, unrestricted license to exercise all rights under those
+    copyrights. This includes, at our option, the right to sublicense these same
+    rights to third parties through multiple levels of sublicensees or other
+    licensing arrangements;
+
+    * you agree that each of us can do all things in relation to your
+    contribution as if each of us were the sole owners, and if one of us makes
+    a derivative work of your contribution, the one who makes the derivative
+    work (or has it made will be the sole owner of that derivative work;
+
+    * you agree that you will not assert any moral rights in your contribution
+    against us, our licensees or transferees;
+
+    * you agree that we may register a copyright in your contribution and
+    exercise all ownership rights associated with it; and
+
+    * you agree that neither of us has any duty to consult with, obtain the
+    consent of, pay or render an accounting to the other for any use or
+    distribution of your contribution.
+
+3. With respect to any patents you own, or that you can license without payment
+to any third party, you hereby grant to us a perpetual, irrevocable,
+non-exclusive, worldwide, no-charge, royalty-free license to:
+
+    * make, have made, use, sell, offer to sell, import, and otherwise transfer
+    your contribution in whole or in part, alone or in combination with or
+    included in any product, work or materials arising out of the project to
+    which your contribution was submitted, and
+
+    * at our option, to sublicense these same rights to third parties through
+    multiple levels of sublicensees or other licensing arrangements.
+
+4. Except as set out above, you keep all right, title, and interest in your
+contribution. The rights that you grant to us under these terms are effective
+on the date you first submitted a contribution to us, even if your submission
+took place before the date you sign these terms.
+
+5. You covenant, represent, warrant and agree that:
+
+    * Each contribution that you submit is and shall be an original work of
+    authorship and you can legally grant the rights set out in this SCA;
+
+    * to the best of your knowledge, each contribution will not violate any
+    third party's copyrights, trademarks, patents, or other intellectual
+    property rights; and
+
+    * each contribution shall be in compliance with U.S. export control laws and
+    other applicable export and import laws. You agree to notify us if you
+    become aware of any circumstance which would make any of the foregoing
+    representations inaccurate in any respect. We may publicly disclose your
+    participation in the project, including the fact that you have signed the SCA.
+
+6. This SCA is governed by the laws of the State of California and applicable
+U.S. Federal law. Any choice of law rules will not apply.
+
+7. Please place an “x” on one of the applicable statement below. Please do NOT
+mark both statements:
+
+    * [x] I am signing on behalf of myself as an individual and no other person
+    or entity, including my employer, has or will have rights with respect to my
+    contributions.
+
+    * [ ] I am signing on behalf of my employer or a legal entity and I have the
+    actual authority to contractually bind that entity.
+
+## Contributor Details
+
+| Field                          | Entry                |
+|------------------------------- | -------------------- |
+| Name                           | Jan Jessewitsch      |
+| Company name (if applicable)   |                      |
+| Title or role (if applicable)  |                      |
+| Date                           | 16.02.2020           |
+| GitHub username                | Jan-711              |
+| Website (optional)             |                      |

.github/contributors/ceteri.md

@@ -0,0 +1,106 @@
+# spaCy contributor agreement
+
+This spaCy Contributor Agreement (**"SCA"**) is based on the
+[Oracle Contributor Agreement](http://www.oracle.com/technetwork/oca-405177.pdf).
+The SCA applies to any contribution that you make to any product or project
+managed by us (the **"project"**), and sets out the intellectual property rights
+you grant to us in the contributed materials. The term **"us"** shall mean
+[ExplosionAI GmbH](https://explosion.ai/legal). The term
+**"you"** shall mean the person or entity identified below.
+
+If you agree to be bound by these terms, fill in the information requested
+below and include the filled-in version with your first pull request, under the
+folder [`.github/contributors/`](/.github/contributors/). The name of the file
+should be your GitHub username, with the extension `.md`. For example, the user
+example_user would create the file `.github/contributors/example_user.md`.
+
+Read this agreement carefully before signing. These terms and conditions
+constitute a binding legal agreement.
+
+## Contributor Agreement
+
+1. The term "contribution" or "contributed materials" means any source code,
+object code, patch, tool, sample, graphic, specification, manual,
+documentation, or any other material posted or submitted by you to the project.
+
+2. With respect to any worldwide copyrights, or copyright applications and
+registrations, in your contribution:
+
+    * you hereby assign to us joint ownership, and to the extent that such
+    assignment is or becomes invalid, ineffective or unenforceable, you hereby
+    grant to us a perpetual, irrevocable, non-exclusive, worldwide, no-charge,
+    royalty-free, unrestricted license to exercise all rights under those
+    copyrights. This includes, at our option, the right to sublicense these same
+    rights to third parties through multiple levels of sublicensees or other
+    licensing arrangements;
+
+    * you agree that each of us can do all things in relation to your
+    contribution as if each of us were the sole owners, and if one of us makes
+    a derivative work of your contribution, the one who makes the derivative
+    work (or has it made will be the sole owner of that derivative work;
+
+    * you agree that you will not assert any moral rights in your contribution
+    against us, our licensees or transferees;
+
+    * you agree that we may register a copyright in your contribution and
+    exercise all ownership rights associated with it; and
+
+    * you agree that neither of us has any duty to consult with, obtain the
+    consent of, pay or render an accounting to the other for any use or
+    distribution of your contribution.
+
+3. With respect to any patents you own, or that you can license without payment
+to any third party, you hereby grant to us a perpetual, irrevocable,
+non-exclusive, worldwide, no-charge, royalty-free license to:
+
+    * make, have made, use, sell, offer to sell, import, and otherwise transfer
+    your contribution in whole or in part, alone or in combination with or
+    included in any product, work or materials arising out of the project to
+    which your contribution was submitted, and
+
+    * at our option, to sublicense these same rights to third parties through
+    multiple levels of sublicensees or other licensing arrangements.
+
+4. Except as set out above, you keep all right, title, and interest in your
+contribution. The rights that you grant to us under these terms are effective
+on the date you first submitted a contribution to us, even if your submission
+took place before the date you sign these terms.
+
+5. You covenant, represent, warrant and agree that:
+
+    * Each contribution that you submit is and shall be an original work of
+    authorship and you can legally grant the rights set out in this SCA;
+
+    * to the best of your knowledge, each contribution will not violate any
+    third party's copyrights, trademarks, patents, or other intellectual
+    property rights; and
+
+    * each contribution shall be in compliance with U.S. export control laws and
+    other applicable export and import laws. You agree to notify us if you
+    become aware of any circumstance which would make any of the foregoing
+    representations inaccurate in any respect. We may publicly disclose your
+    participation in the project, including the fact that you have signed the SCA.
+
+6. This SCA is governed by the laws of the State of California and applicable
+U.S. Federal law. Any choice of law rules will not apply.
+
+7. Please place an “x” on one of the applicable statement below. Please do NOT
+mark both statements:
+
+    * [ ] I am signing on behalf of myself as an individual and no other person
+    or entity, including my employer, has or will have rights with respect to my
+    contributions.
+
+    * [x] I am signing on behalf of my employer or a legal entity and I have the
+    actual authority to contractually bind that entity.
+
+## Contributor Details
+
+| Field                          | Entry                  |
+|------------------------------- | ---------------------- |
+| Name                           | Paco Nathan            |
+| Company name (if applicable)   | Derwen, Inc.           |
+| Title or role (if applicable)  | Managing Partner       |
+| Date                           | 2020-01-25             |
+| GitHub username                | ceteri                 |
+| Website (optional)             | https://derwen.ai/paco |

.github/contributors/drndos.md

@@ -0,0 +1,106 @@
+# spaCy contributor agreement
+
+This spaCy Contributor Agreement (**"SCA"**) is based on the
+[Oracle Contributor Agreement](http://www.oracle.com/technetwork/oca-405177.pdf).
+The SCA applies to any contribution that you make to any product or project
+managed by us (the **"project"**), and sets out the intellectual property rights
+you grant to us in the contributed materials. The term **"us"** shall mean
+[ExplosionAI GmbH](https://explosion.ai/legal). The term
+**"you"** shall mean the person or entity identified below.
+
+If you agree to be bound by these terms, fill in the information requested
+below and include the filled-in version with your first pull request, under the
+folder [`.github/contributors/`](/.github/contributors/). The name of the file
+should be your GitHub username, with the extension `.md`. For example, the user
+example_user would create the file `.github/contributors/example_user.md`.
+
+Read this agreement carefully before signing. These terms and conditions
+constitute a binding legal agreement.
+
+## Contributor Agreement
+
+1. The term "contribution" or "contributed materials" means any source code,
+object code, patch, tool, sample, graphic, specification, manual,
+documentation, or any other material posted or submitted by you to the project.
+
+2. With respect to any worldwide copyrights, or copyright applications and
+registrations, in your contribution:
+
+    * you hereby assign to us joint ownership, and to the extent that such
+    assignment is or becomes invalid, ineffective or unenforceable, you hereby
+    grant to us a perpetual, irrevocable, non-exclusive, worldwide, no-charge,
+    royalty-free, unrestricted license to exercise all rights under those
+    copyrights. This includes, at our option, the right to sublicense these same
+    rights to third parties through multiple levels of sublicensees or other
+    licensing arrangements;
+
+    * you agree that each of us can do all things in relation to your
+    contribution as if each of us were the sole owners, and if one of us makes
+    a derivative work of your contribution, the one who makes the derivative
+    work (or has it made will be the sole owner of that derivative work;
+
+    * you agree that you will not assert any moral rights in your contribution
+    against us, our licensees or transferees;
+
+    * you agree that we may register a copyright in your contribution and
+    exercise all ownership rights associated with it; and
+
+    * you agree that neither of us has any duty to consult with, obtain the
+    consent of, pay or render an accounting to the other for any use or
+    distribution of your contribution.
+
+3. With respect to any patents you own, or that you can license without payment
+to any third party, you hereby grant to us a perpetual, irrevocable,
+non-exclusive, worldwide, no-charge, royalty-free license to:
+
+    * make, have made, use, sell, offer to sell, import, and otherwise transfer
+    your contribution in whole or in part, alone or in combination with or
+    included in any product, work or materials arising out of the project to
+    which your contribution was submitted, and
+
+    * at our option, to sublicense these same rights to third parties through
+    multiple levels of sublicensees or other licensing arrangements.
+
+4. Except as set out above, you keep all right, title, and interest in your
+contribution. The rights that you grant to us under these terms are effective
+on the date you first submitted a contribution to us, even if your submission
+took place before the date you sign these terms.
+
+5. You covenant, represent, warrant and agree that:
+
+    * Each contribution that you submit is and shall be an original work of
+    authorship and you can legally grant the rights set out in this SCA;
+
+    * to the best of your knowledge, each contribution will not violate any
+    third party's copyrights, trademarks, patents, or other intellectual
+    property rights; and
+
+    * each contribution shall be in compliance with U.S. export control laws and
+    other applicable export and import laws. You agree to notify us if you
+    become aware of any circumstance which would make any of the foregoing
+    representations inaccurate in any respect. We may publicly disclose your
+    participation in the project, including the fact that you have signed the SCA.
+
+6. This SCA is governed by the laws of the State of California and applicable
+U.S. Federal law. Any choice of law rules will not apply.
+
+7. Please place an “x” on one of the applicable statement below. Please do NOT
+mark both statements:
+
+    * [ ] I am signing on behalf of myself as an individual and no other person
+    or entity, including my employer, has or will have rights with respect to my
+    contributions.
+
+    * [x] I am signing on behalf of my employer or a legal entity and I have the
+    actual authority to contractually bind that entity.
+
+## Contributor Details
+
+| Field                          | Entry                |
+|------------------------------- | -------------------- |
+| Name                           | Filip Bednárik       |
+| Company name (if applicable)   | Ardevop, s. r. o.    |
+| Title or role (if applicable)  | IT Consultant        |
+| Date                           | 2020-01-26           |
+| GitHub username                | drndos               |
+| Website (optional)             | https://ardevop.sk   |

.github/contributors/iechevarria.md

@@ -0,0 +1,106 @@
+# spaCy contributor agreement
+
+This spaCy Contributor Agreement (**"SCA"**) is based on the
+[Oracle Contributor Agreement](http://www.oracle.com/technetwork/oca-405177.pdf).
+The SCA applies to any contribution that you make to any product or project
+managed by us (the **"project"**), and sets out the intellectual property rights
+you grant to us in the contributed materials. The term **"us"** shall mean
+[ExplosionAI GmbH](https://explosion.ai/legal). The term
+**"you"** shall mean the person or entity identified below.
+
+If you agree to be bound by these terms, fill in the information requested
+below and include the filled-in version with your first pull request, under the
+folder [`.github/contributors/`](/.github/contributors/). The name of the file
+should be your GitHub username, with the extension `.md`. For example, the user
+example_user would create the file `.github/contributors/example_user.md`.
+
+Read this agreement carefully before signing. These terms and conditions
+constitute a binding legal agreement.
+
+## Contributor Agreement
+
+1. The term "contribution" or "contributed materials" means any source code,
+object code, patch, tool, sample, graphic, specification, manual,
+documentation, or any other material posted or submitted by you to the project.
+
+2. With respect to any worldwide copyrights, or copyright applications and
+registrations, in your contribution:
+
+    * you hereby assign to us joint ownership, and to the extent that such
+    assignment is or becomes invalid, ineffective or unenforceable, you hereby
+    grant to us a perpetual, irrevocable, non-exclusive, worldwide, no-charge,
+    royalty-free, unrestricted license to exercise all rights under those
+    copyrights. This includes, at our option, the right to sublicense these same
+    rights to third parties through multiple levels of sublicensees or other
+    licensing arrangements;
+
+    * you agree that each of us can do all things in relation to your
+    contribution as if each of us were the sole owners, and if one of us makes
+    a derivative work of your contribution, the one who makes the derivative
+    work (or has it made will be the sole owner of that derivative work;
+
+    * you agree that you will not assert any moral rights in your contribution
+    against us, our licensees or transferees;
+
+    * you agree that we may register a copyright in your contribution and
+    exercise all ownership rights associated with it; and
+
+    * you agree that neither of us has any duty to consult with, obtain the
+    consent of, pay or render an accounting to the other for any use or
+    distribution of your contribution.
+
+3. With respect to any patents you own, or that you can license without payment
+to any third party, you hereby grant to us a perpetual, irrevocable,
+non-exclusive, worldwide, no-charge, royalty-free license to:
+
+    * make, have made, use, sell, offer to sell, import, and otherwise transfer
+    your contribution in whole or in part, alone or in combination with or
+    included in any product, work or materials arising out of the project to
+    which your contribution was submitted, and
+
+    * at our option, to sublicense these same rights to third parties through
+    multiple levels of sublicensees or other licensing arrangements.
+
+4. Except as set out above, you keep all right, title, and interest in your
+contribution. The rights that you grant to us under these terms are effective
+on the date you first submitted a contribution to us, even if your submission
+took place before the date you sign these terms.
+
+5. You covenant, represent, warrant and agree that:
+
+    * Each contribution that you submit is and shall be an original work of
+    authorship and you can legally grant the rights set out in this SCA;
+
+    * to the best of your knowledge, each contribution will not violate any
+    third party's copyrights, trademarks, patents, or other intellectual
+    property rights; and
+
+    * each contribution shall be in compliance with U.S. export control laws and
+    other applicable export and import laws. You agree to notify us if you
+    become aware of any circumstance which would make any of the foregoing
+    representations inaccurate in any respect. We may publicly disclose your
+    participation in the project, including the fact that you have signed the SCA.
+
+6. This SCA is governed by the laws of the State of California and applicable
+U.S. Federal law. Any choice of law rules will not apply.
+
+7. Please place an “x” on one of the applicable statement below. Please do NOT
+mark both statements:
+
+    * [x] I am signing on behalf of myself as an individual and no other person
+    or entity, including my employer, has or will have rights with respect to my
+    contributions.
+
+    * [ ] I am signing on behalf of my employer or a legal entity and I have the
+    actual authority to contractually bind that entity.
+
+## Contributor Details
+
+| Field                          | Entry                 |
+|------------------------------- | --------------------- |
+| Name                           | Ivan Echevarria       |
+| Company name (if applicable)   |                       |
+| Title or role (if applicable)  |                       |
+| Date                           | 2019-12-24            |
+| GitHub username                | iechevarria           |
+| Website (optional)             | https://echevarria.io |

.github/contributors/iurshina.md

@@ -0,0 +1,106 @@
+# spaCy contributor agreement
+
+This spaCy Contributor Agreement (**"SCA"**) is based on the
+[Oracle Contributor Agreement](http://www.oracle.com/technetwork/oca-405177.pdf).
+The SCA applies to any contribution that you make to any product or project
+managed by us (the **"project"**), and sets out the intellectual property rights
+you grant to us in the contributed materials. The term **"us"** shall mean
+[ExplosionAI GmbH](https://explosion.ai/legal). The term
+**"you"** shall mean the person or entity identified below.
+
+If you agree to be bound by these terms, fill in the information requested
+below and include the filled-in version with your first pull request, under the
+folder [`.github/contributors/`](/.github/contributors/). The name of the file
+should be your GitHub username, with the extension `.md`. For example, the user
+example_user would create the file `.github/contributors/example_user.md`.
+
+Read this agreement carefully before signing. These terms and conditions
+constitute a binding legal agreement.
+
+## Contributor Agreement
+
+1. The term "contribution" or "contributed materials" means any source code,
+object code, patch, tool, sample, graphic, specification, manual,
+documentation, or any other material posted or submitted by you to the project.
+
+2. With respect to any worldwide copyrights, or copyright applications and
+registrations, in your contribution:
+
+    * you hereby assign to us joint ownership, and to the extent that such
+    assignment is or becomes invalid, ineffective or unenforceable, you hereby
+    grant to us a perpetual, irrevocable, non-exclusive, worldwide, no-charge,
+    royalty-free, unrestricted license to exercise all rights under those
+    copyrights. This includes, at our option, the right to sublicense these same
+    rights to third parties through multiple levels of sublicensees or other
+    licensing arrangements;
+
+    * you agree that each of us can do all things in relation to your
+    contribution as if each of us were the sole owners, and if one of us makes
+    a derivative work of your contribution, the one who makes the derivative
+    work (or has it made will be the sole owner of that derivative work;
+
+    * you agree that you will not assert any moral rights in your contribution
+    against us, our licensees or transferees;
+
+    * you agree that we may register a copyright in your contribution and
+    exercise all ownership rights associated with it; and
+
+    * you agree that neither of us has any duty to consult with, obtain the
+    consent of, pay or render an accounting to the other for any use or
+    distribution of your contribution.
+
+3. With respect to any patents you own, or that you can license without payment
+to any third party, you hereby grant to us a perpetual, irrevocable,
+non-exclusive, worldwide, no-charge, royalty-free license to:
+
+    * make, have made, use, sell, offer to sell, import, and otherwise transfer
+    your contribution in whole or in part, alone or in combination with or
+    included in any product, work or materials arising out of the project to
+    which your contribution was submitted, and
+
+    * at our option, to sublicense these same rights to third parties through
+    multiple levels of sublicensees or other licensing arrangements.
+
+4. Except as set out above, you keep all right, title, and interest in your
+contribution. The rights that you grant to us under these terms are effective
+on the date you first submitted a contribution to us, even if your submission
+took place before the date you sign these terms.
+
+5. You covenant, represent, warrant and agree that:
+
+    * Each contribution that you submit is and shall be an original work of
+    authorship and you can legally grant the rights set out in this SCA;
+
+    * to the best of your knowledge, each contribution will not violate any
+    third party's copyrights, trademarks, patents, or other intellectual
+    property rights; and
+
+    * each contribution shall be in compliance with U.S. export control laws and
+    other applicable export and import laws. You agree to notify us if you
+    become aware of any circumstance which would make any of the foregoing
+    representations inaccurate in any respect. We may publicly disclose your
+    participation in the project, including the fact that you have signed the SCA.
+
+6. This SCA is governed by the laws of the State of California and applicable
+U.S. Federal law. Any choice of law rules will not apply.
+
+7. Please place an “x” on one of the applicable statement below. Please do NOT
+mark both statements:
+
+    * [ ] I am signing on behalf of myself as an individual and no other person
+    or entity, including my employer, has or will have rights with respect to my
+    contributions.
+
+    * [ ] I am signing on behalf of my employer or a legal entity and I have the
+    actual authority to contractually bind that entity.
+
+## Contributor Details
+
+| Field                          | Entry                |
+|------------------------------- | -------------------- |
+| Name                           | Anastasiia Iurshina  |
+| Company name (if applicable)   |                      |
+| Title or role (if applicable)  |                      |
+| Date                           | 28.12.2019           |
+| GitHub username                | iurshina             |
+| Website (optional)             |                      |

.github/contributors/onlyanegg.md

@@ -0,0 +1,106 @@
+# spaCy contributor agreement
+
+This spaCy Contributor Agreement (**"SCA"**) is based on the
+[Oracle Contributor Agreement](http://www.oracle.com/technetwork/oca-405177.pdf).
+The SCA applies to any contribution that you make to any product or project
+managed by us (the **"project"**), and sets out the intellectual property rights
+you grant to us in the contributed materials. The term **"us"** shall mean
+[ExplosionAI GmbH](https://explosion.ai/legal). The term
+**"you"** shall mean the person or entity identified below.
+
+If you agree to be bound by these terms, fill in the information requested
+below and include the filled-in version with your first pull request, under the
+folder [`.github/contributors/`](/.github/contributors/). The name of the file
+should be your GitHub username, with the extension `.md`. For example, the user
+example_user would create the file `.github/contributors/example_user.md`.
+
+Read this agreement carefully before signing. These terms and conditions
+constitute a binding legal agreement.
+
+## Contributor Agreement
+
+1.  The term "contribution" or "contributed materials" means any source code,
+    object code, patch, tool, sample, graphic, specification, manual,
+    documentation, or any other material posted or submitted by you to the project.
+
+2.  With respect to any worldwide copyrights, or copyright applications and
+    registrations, in your contribution:
+
+        * you hereby assign to us joint ownership, and to the extent that such
+        assignment is or becomes invalid, ineffective or unenforceable, you hereby
+        grant to us a perpetual, irrevocable, non-exclusive, worldwide, no-charge,
+        royalty-free, unrestricted license to exercise all rights under those
+        copyrights. This includes, at our option, the right to sublicense these same
+        rights to third parties through multiple levels of sublicensees or other
+        licensing arrangements;
+
+        * you agree that each of us can do all things in relation to your
+        contribution as if each of us were the sole owners, and if one of us makes
+        a derivative work of your contribution, the one who makes the derivative
+        work (or has it made will be the sole owner of that derivative work;
+
+        * you agree that you will not assert any moral rights in your contribution
+        against us, our licensees or transferees;
+
+        * you agree that we may register a copyright in your contribution and
+        exercise all ownership rights associated with it; and
+
+        * you agree that neither of us has any duty to consult with, obtain the
+        consent of, pay or render an accounting to the other for any use or
+        distribution of your contribution.
+
+3.  With respect to any patents you own, or that you can license without payment
+    to any third party, you hereby grant to us a perpetual, irrevocable,
+    non-exclusive, worldwide, no-charge, royalty-free license to:
+
+        * make, have made, use, sell, offer to sell, import, and otherwise transfer
+        your contribution in whole or in part, alone or in combination with or
+        included in any product, work or materials arising out of the project to
+        which your contribution was submitted, and
+
+        * at our option, to sublicense these same rights to third parties through
+        multiple levels of sublicensees or other licensing arrangements.
+
+4.  Except as set out above, you keep all right, title, and interest in your
+    contribution. The rights that you grant to us under these terms are effective
+    on the date you first submitted a contribution to us, even if your submission
+    took place before the date you sign these terms.
+
+5.  You covenant, represent, warrant and agree that:
+
+    - Each contribution that you submit is and shall be an original work of
+      authorship and you can legally grant the rights set out in this SCA;
+
+    - to the best of your knowledge, each contribution will not violate any
+      third party's copyrights, trademarks, patents, or other intellectual
+      property rights; and
+
+    - each contribution shall be in compliance with U.S. export control laws and
+      other applicable export and import laws. You agree to notify us if you
+      become aware of any circumstance which would make any of the foregoing
+      representations inaccurate in any respect. We may publicly disclose your
+      participation in the project, including the fact that you have signed the SCA.
+
+6.  This SCA is governed by the laws of the State of California and applicable
+    U.S. Federal law. Any choice of law rules will not apply.
+
+7.  Please place an “x” on one of the applicable statement below. Please do NOT
+    mark both statements:
+
+        * [x] I am signing on behalf of myself as an individual and no other person
+        or entity, including my employer, has or will have rights with respect to my
+        contributions.
+
+        * [ ] I am signing on behalf of my employer or a legal entity and I have the
+        actual authority to contractually bind that entity.
+
+## Contributor Details
+
+| Field                         | Entry            |
+| ----------------------------- | ---------------- |
+| Name                          | Tyler Couto      |
+| Company name (if applicable)  |                  |
+| Title or role (if applicable) |                  |
+| Date                          | January 29, 2020 |
+| GitHub username               | onlyanegg        |
+| Website (optional)            |                  |

MANIFEST.in

@@ -1,5 +1,5 @@
 recursive-include include *.h
-recursive-include spacy *.pyx *.pxd *.txt
+recursive-include spacy *.txt *.pyx *.pxd
 include LICENSE
 include README.md
 include bin/spacy

bin/spacy

@@ -1 +1,2 @@
+#! /bin/sh
 python -m spacy "$@"

bin/wiki_entity_linking/README.md

@@ -7,16 +7,17 @@ Run  `wikipedia_pretrain_kb.py`
   * WikiData: get `latest-all.json.bz2` from https://dumps.wikimedia.org/wikidatawiki/entities/
   * Wikipedia: get `enwiki-latest-pages-articles-multistream.xml.bz2` from https://dumps.wikimedia.org/enwiki/latest/ (or for any other language)
 * You can set the filtering parameters for KB construction:
-  * `max_per_alias`: (max) number of candidate entities in the KB per alias/synonym
+  * `max_per_alias` (`-a`): (max) number of candidate entities in the KB per alias/synonym
-  * `min_freq`: threshold of number of times an entity should occur in the corpus to be included in the KB
+  * `min_freq` (`-f`): threshold of number of times an entity should occur in the corpus to be included in the KB
-  * `min_pair`: threshold of number of times an entity+alias combination should occur in the corpus to be included in the KB
+  * `min_pair` (`-c`): threshold of number of times an entity+alias combination should occur in the corpus to be included in the KB
 * Further parameters to set:
-  * `descriptions_from_wikipedia`: whether to parse descriptions from Wikipedia (`True`) or Wikidata (`False`)
+  * `descriptions_from_wikipedia` (`-wp`): whether to parse descriptions from Wikipedia (`True`) or Wikidata (`False`)
-  * `entity_vector_length`: length of the pre-trained entity description vectors
+  * `entity_vector_length` (`-v`): length of the pre-trained entity description vectors
-  * `lang`: language for which to fetch Wikidata information (as the dump contains all languages)
+  * `lang` (`-la`): language for which to fetch Wikidata information (as the dump contains all languages)
 
 Quick testing and rerunning: 
-* When trying out the pipeline for a quick test, set `limit_prior`, `limit_train` and/or `limit_wd` to read only parts of the dumps instead of everything. 
+* When trying out the pipeline for a quick test, set `limit_prior` (`-lp`), `limit_train` (`-lt`) and/or `limit_wd` (`-lw`) to read only parts of the dumps instead of everything. 
+  * e.g. set `-lt 20000 -lp 2000 -lw 3000 -f 1`
 * If you only want to (re)run certain parts of the pipeline, just remove the corresponding files and they will be recalculated or reparsed.
 
 
@@ -24,11 +25,13 @@ Quick testing and rerunning:
 
 Run  `wikidata_train_entity_linker.py` 
 * This takes the **KB directory** produced by Step 1, and trains an **Entity Linking model**
+* Specify the output directory (`-o`) in which the final, trained model will be saved
 * You can set the learning parameters for the EL training:
-  * `epochs`: number of training iterations
+  * `epochs` (`-e`): number of training iterations
-  * `dropout`: dropout rate
+  * `dropout` (`-p`): dropout rate
-  * `lr`: learning rate
+  * `lr` (`-n`): learning rate
-  * `l2`: L2 regularization
+  * `l2` (`-r`): L2 regularization
-* Specify the number of training and dev testing entities with `train_inst` and `dev_inst` respectively
+* Specify the number of training and dev testing articles with `train_articles` (`-t`) and `dev_articles` (`-d`) respectively
+  * If not specified, the full dataset will be processed - this may take a LONG time !
 * Further parameters to set:
-  * `labels_discard`: NER label types to discard during training
+  * `labels_discard` (`-l`): NER label types to discard during training

bin/wiki_entity_linking/entity_linker_evaluation.py

@@ -1,6 +1,8 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
 import logging
 import random
-
 from tqdm import tqdm
 from collections import defaultdict
 
@@ -92,133 +94,110 @@ class BaselineResults(object):
         self.random.update_metrics(ent_label, true_entity, random_candidate)
 
 
-def measure_performance(dev_data, kb, el_pipe, baseline=True, context=True):
+def measure_performance(dev_data, kb, el_pipe, baseline=True, context=True, dev_limit=None):
-    if baseline:
+    counts = dict()
-        baseline_accuracies, counts = measure_baselines(dev_data, kb)
+    baseline_results = BaselineResults()
-        logger.info("Counts: {}".format({k: v for k, v in sorted(counts.items())}))
+    context_results = EvaluationResults()
-        logger.info(baseline_accuracies.report_performance("random"))
+    combo_results = EvaluationResults()
-        logger.info(baseline_accuracies.report_performance("prior"))
-        logger.info(baseline_accuracies.report_performance("oracle"))
 
-    if context:
+    for doc, gold in tqdm(dev_data, total=dev_limit, leave=False, desc='Processing dev data'):
-        # using only context
+        if len(doc) > 0:
-        el_pipe.cfg["incl_context"] = True
+            correct_ents = dict()
-        el_pipe.cfg["incl_prior"] = False
-        results = get_eval_results(dev_data, el_pipe)
-        logger.info(results.report_metrics("context only"))
-
-        # measuring combined accuracy (prior + context)
-        el_pipe.cfg["incl_context"] = True
-        el_pipe.cfg["incl_prior"] = True
-        results = get_eval_results(dev_data, el_pipe)
-        logger.info(results.report_metrics("context and prior"))
-
-
-def get_eval_results(data, el_pipe=None):
-    """
-    Evaluate the ent.kb_id_ annotations against the gold standard.
-    Only evaluate entities that overlap between gold and NER, to isolate the performance of the NEL.
-    If the docs in the data require further processing with an entity linker, set el_pipe.
-    """
-    docs = []
-    golds = []
-    for d, g in tqdm(data, leave=False):
-        if len(d) > 0:
-            golds.append(g)
-            if el_pipe is not None:
-                docs.append(el_pipe(d))
-            else:
-                docs.append(d)
-
-    results = EvaluationResults()
-    for doc, gold in zip(docs, golds):
-        try:
-            correct_entries_per_article = dict()
             for entity, kb_dict in gold.links.items():
                 start, end = entity
                 for gold_kb, value in kb_dict.items():
                     if value:
                         # only evaluating on positive examples
                         offset = _offset(start, end)
-                        correct_entries_per_article[offset] = gold_kb
+                        correct_ents[offset] = gold_kb
 
-            for ent in doc.ents:
+            if baseline:
-                ent_label = ent.label_
+                _add_baseline(baseline_results, counts, doc, correct_ents, kb)
-                pred_entity = ent.kb_id_
-                start = ent.start_char
-                end = ent.end_char
-                offset = _offset(start, end)
-                gold_entity = correct_entries_per_article.get(offset, None)
-                # the gold annotations are not complete so we can't evaluate missing annotations as 'wrong'
-                if gold_entity is not None:
-                    results.update_metrics(ent_label, gold_entity, pred_entity)
 
-        except Exception as e:
+            if context:
-            logging.error("Error assessing accuracy " + str(e))
+                # using only context
+                el_pipe.cfg["incl_context"] = True
+                el_pipe.cfg["incl_prior"] = False
+                _add_eval_result(context_results, doc, correct_ents, el_pipe)
 
-    return results
+                # measuring combined accuracy (prior + context)
+                el_pipe.cfg["incl_context"] = True
+                el_pipe.cfg["incl_prior"] = True
+                _add_eval_result(combo_results, doc, correct_ents, el_pipe)
+
+    if baseline:
+        logger.info("Counts: {}".format({k: v for k, v in sorted(counts.items())}))
+        logger.info(baseline_results.report_performance("random"))
+        logger.info(baseline_results.report_performance("prior"))
+        logger.info(baseline_results.report_performance("oracle"))
+
+    if context:
+        logger.info(context_results.report_metrics("context only"))
+        logger.info(combo_results.report_metrics("context and prior"))
 
 
-def measure_baselines(data, kb):
+def _add_eval_result(results, doc, correct_ents, el_pipe):
     """
-    Measure 3 performance baselines: random selection, prior probabilities, and 'oracle' prediction for upper bound.
+    Evaluate the ent.kb_id_ annotations against the gold standard.
     Only evaluate entities that overlap between gold and NER, to isolate the performance of the NEL.
-    Also return a dictionary of counts by entity label.
     """
-    counts_d = dict()
+    try:
-
+        doc = el_pipe(doc)
-    baseline_results = BaselineResults()
-
-    docs = [d for d, g in data if len(d) > 0]
-    golds = [g for d, g in data if len(d) > 0]
-
-    for doc, gold in zip(docs, golds):
-        correct_entries_per_article = dict()
-        for entity, kb_dict in gold.links.items():
-            start, end = entity
-            for gold_kb, value in kb_dict.items():
-                # only evaluating on positive examples
-                if value:
-                    offset = _offset(start, end)
-                    correct_entries_per_article[offset] = gold_kb
-
         for ent in doc.ents:
             ent_label = ent.label_
             start = ent.start_char
             end = ent.end_char
             offset = _offset(start, end)
-            gold_entity = correct_entries_per_article.get(offset, None)
+            gold_entity = correct_ents.get(offset, None)
-
             # the gold annotations are not complete so we can't evaluate missing annotations as 'wrong'
             if gold_entity is not None:
-                candidates = kb.get_candidates(ent.text)
+                pred_entity = ent.kb_id_
-                oracle_candidate = ""
+                results.update_metrics(ent_label, gold_entity, pred_entity)
-                prior_candidate = ""
-                random_candidate = ""
-                if candidates:
-                    scores = []
 
-                    for c in candidates:
+    except Exception as e:
-                        scores.append(c.prior_prob)
+        logging.error("Error assessing accuracy " + str(e))
-                        if c.entity_ == gold_entity:
-                            oracle_candidate = c.entity_
 
-                    best_index = scores.index(max(scores))
-                    prior_candidate = candidates[best_index].entity_
-                    random_candidate = random.choice(candidates).entity_
 
-                current_count = counts_d.get(ent_label, 0)
+def _add_baseline(baseline_results, counts, doc, correct_ents, kb):
-                counts_d[ent_label] = current_count+1
+    """
+    Measure 3 performance baselines: random selection, prior probabilities, and 'oracle' prediction for upper bound.
+    Only evaluate entities that overlap between gold and NER, to isolate the performance of the NEL.
+    """
+    for ent in doc.ents:
+        ent_label = ent.label_
+        start = ent.start_char
+        end = ent.end_char
+        offset = _offset(start, end)
+        gold_entity = correct_ents.get(offset, None)
 
-                baseline_results.update_baselines(
+        # the gold annotations are not complete so we can't evaluate missing annotations as 'wrong'
-                    gold_entity,
+        if gold_entity is not None:
-                    ent_label,
+            candidates = kb.get_candidates(ent.text)
-                    random_candidate,
+            oracle_candidate = ""
-                    prior_candidate,
+            prior_candidate = ""
-                    oracle_candidate,
+            random_candidate = ""
-                )
+            if candidates:
+                scores = []
 
-    return baseline_results, counts_d
+                for c in candidates:
+                    scores.append(c.prior_prob)
+                    if c.entity_ == gold_entity:
+                        oracle_candidate = c.entity_
+
+                best_index = scores.index(max(scores))
+                prior_candidate = candidates[best_index].entity_
+                random_candidate = random.choice(candidates).entity_
+
+            current_count = counts.get(ent_label, 0)
+            counts[ent_label] = current_count+1
+
+            baseline_results.update_baselines(
+                gold_entity,
+                ent_label,
+                random_candidate,
+                prior_candidate,
+                oracle_candidate,
+            )
 
 
 def _offset(start, end):

bin/wiki_entity_linking/wikidata_pretrain_kb.py

@@ -40,7 +40,7 @@ logger = logging.getLogger(__name__)
     loc_prior_prob=("Location to file with prior probabilities", "option", "p", Path),
     loc_entity_defs=("Location to file with entity definitions", "option", "d", Path),
     loc_entity_desc=("Location to file with entity descriptions", "option", "s", Path),
-    descr_from_wp=("Flag for using wp descriptions not wd", "flag", "wp"),
+    descr_from_wp=("Flag for using descriptions from WP instead of WD (default False)", "flag", "wp"),
     limit_prior=("Threshold to limit lines read from WP for prior probabilities", "option", "lp", int),
     limit_train=("Threshold to limit lines read from WP for training set", "option", "lt", int),
     limit_wd=("Threshold to limit lines read from WD", "option", "lw", int),

bin/wiki_entity_linking/wikidata_train_entity_linker.py

@@ -1,5 +1,5 @@
 # coding: utf-8
-"""Script to take a previously created Knowledge Base and train an entity linking
+"""Script that takes a previously created Knowledge Base and trains an entity linking
 pipeline. The provided KB directory should hold the kb, the original nlp object and
 its vocab used to create the KB, and a few auxiliary files such as the entity definitions,
 as created by the script `wikidata_create_kb`.
@@ -14,9 +14,16 @@ import logging
 import spacy
 from pathlib import Path
 import plac
+from tqdm import tqdm
 
 from bin.wiki_entity_linking import wikipedia_processor
-from bin.wiki_entity_linking import TRAINING_DATA_FILE, KB_MODEL_DIR, KB_FILE, LOG_FORMAT, OUTPUT_MODEL_DIR
+from bin.wiki_entity_linking import (
+    TRAINING_DATA_FILE,
+    KB_MODEL_DIR,
+    KB_FILE,
+    LOG_FORMAT,
+    OUTPUT_MODEL_DIR,
+)
 from bin.wiki_entity_linking.entity_linker_evaluation import measure_performance
 from bin.wiki_entity_linking.kb_creator import read_kb
 
@@ -33,8 +40,8 @@ logger = logging.getLogger(__name__)
     dropout=("Dropout to prevent overfitting (default 0.5)", "option", "p", float),
     lr=("Learning rate (default 0.005)", "option", "n", float),
     l2=("L2 regularization", "option", "r", float),
-    train_inst=("# training instances (default 90% of all)", "option", "t", int),
+    train_articles=("# training articles (default 90% of all)", "option", "t", int),
-    dev_inst=("# test instances (default 10% of all)", "option", "d", int),
+    dev_articles=("# dev test articles (default 10% of all)", "option", "d", int),
     labels_discard=("NER labels to discard (default None)", "option", "l", str),
 )
 def main(
@@ -45,10 +52,15 @@ def main(
     dropout=0.5,
     lr=0.005,
     l2=1e-6,
-    train_inst=None,
+    train_articles=None,
-    dev_inst=None,
+    dev_articles=None,
-    labels_discard=None
+    labels_discard=None,
 ):
+    if not output_dir:
+        logger.warning(
+            "No output dir specified so no results will be written, are you sure about this ?"
+        )
+
     logger.info("Creating Entity Linker with Wikipedia and WikiData")
 
     output_dir = Path(output_dir) if output_dir else dir_kb
@@ -64,47 +76,57 @@ def main(
     # STEP 1 : load the NLP object
     logger.info("STEP 1a: Loading model from {}".format(nlp_dir))
     nlp = spacy.load(nlp_dir)
-    logger.info("STEP 1b: Loading KB from {}".format(kb_path))
+    logger.info(
-    kb = read_kb(nlp, kb_path)
+        "Original NLP pipeline has following pipeline components: {}".format(
+            nlp.pipe_names
+        )
+    )
 
     # check that there is a NER component in the pipeline
     if "ner" not in nlp.pipe_names:
         raise ValueError("The `nlp` object should have a pretrained `ner` component.")
 
-    # STEP 2: read the training dataset previously created from WP
+    logger.info("STEP 1b: Loading KB from {}".format(kb_path))
-    logger.info("STEP 2: Reading training dataset from {}".format(training_path))
+    kb = read_kb(nlp, kb_path)
 
+    # STEP 2: read the training dataset previously created from WP
+    logger.info("STEP 2: Reading training & dev dataset from {}".format(training_path))
+    train_indices, dev_indices = wikipedia_processor.read_training_indices(
+        training_path
+    )
+    logger.info(
+        "Training set has {} articles, limit set to roughly {} articles per epoch".format(
+            len(train_indices), train_articles if train_articles else "all"
+        )
+    )
+    logger.info(
+        "Dev set has {} articles, limit set to rougly {} articles for evaluation".format(
+            len(dev_indices), dev_articles if dev_articles else "all"
+        )
+    )
+    if dev_articles:
+        dev_indices = dev_indices[0:dev_articles]
+
+    # STEP 3: create and train an entity linking pipe
+    logger.info(
+        "STEP 3: Creating and training an Entity Linking pipe for {} epochs".format(
+            epochs
+        )
+    )
     if labels_discard:
         labels_discard = [x.strip() for x in labels_discard.split(",")]
-        logger.info("Discarding {} NER types: {}".format(len(labels_discard), labels_discard))
+        logger.info(
+            "Discarding {} NER types: {}".format(len(labels_discard), labels_discard)
+        )
     else:
         labels_discard = []
 
-    train_data = wikipedia_processor.read_training(
-        nlp=nlp,
-        entity_file_path=training_path,
-        dev=False,
-        limit=train_inst,
-        kb=kb,
-        labels_discard=labels_discard
-    )
-
-    # for testing, get all pos instances (independently of KB)
-    dev_data = wikipedia_processor.read_training(
-        nlp=nlp,
-        entity_file_path=training_path,
-        dev=True,
-        limit=dev_inst,
-        kb=None,
-        labels_discard=labels_discard
-    )
-
-    # STEP 3: create and train an entity linking pipe
-    logger.info("STEP 3: Creating and training an Entity Linking pipe")
-
     el_pipe = nlp.create_pipe(
-        name="entity_linker", config={"pretrained_vectors": nlp.vocab.vectors,
+        name="entity_linker",
-                                      "labels_discard": labels_discard}
+        config={
+            "pretrained_vectors": nlp.vocab.vectors,
+            "labels_discard": labels_discard,
+        },
     )
     el_pipe.set_kb(kb)
     nlp.add_pipe(el_pipe, last=True)
@@ -115,78 +137,96 @@ def main(
         optimizer.learn_rate = lr
         optimizer.L2 = l2
 
-    logger.info("Training on {} articles".format(len(train_data)))
-    logger.info("Dev testing on {} articles".format(len(dev_data)))
-
-    # baseline performance on dev data
     logger.info("Dev Baseline Accuracies:")
-    measure_performance(dev_data, kb, el_pipe, baseline=True, context=False)
+    dev_data = wikipedia_processor.read_el_docs_golds(
+        nlp=nlp,
+        entity_file_path=training_path,
+        dev=True,
+        line_ids=dev_indices,
+        kb=kb,
+        labels_discard=labels_discard,
+    )
+
+    measure_performance(
+        dev_data, kb, el_pipe, baseline=True, context=False, dev_limit=len(dev_indices)
+    )
 
     for itn in range(epochs):
-        random.shuffle(train_data)
+        random.shuffle(train_indices)
         losses = {}
-        batches = minibatch(train_data, size=compounding(4.0, 128.0, 1.001))
+        batches = minibatch(train_indices, size=compounding(8.0, 128.0, 1.001))
         batchnr = 0
+        articles_processed = 0
 
-        with nlp.disable_pipes(*other_pipes):
+        # we either process the whole training file, or just a part each epoch
+        bar_total = len(train_indices)
+        if train_articles:
+            bar_total = train_articles
+
+        with tqdm(total=bar_total, leave=False, desc=f"Epoch {itn}") as pbar:
             for batch in batches:
-                try:
+                if not train_articles or articles_processed < train_articles:
-                    nlp.update(
+                    with nlp.disable_pipes("entity_linker"):
-                        examples=batch,
+                        train_batch = wikipedia_processor.read_el_docs_golds(
-                        sgd=optimizer,
+                            nlp=nlp,
-                        drop=dropout,
+                            entity_file_path=training_path,
-                        losses=losses,
+                            dev=False,
-                    )
+                            line_ids=batch,
-                    batchnr += 1
+                            kb=kb,
-                except Exception as e:
+                            labels_discard=labels_discard,
-                    logger.error("Error updating batch:" + str(e))
+                        )
+                        docs, golds = zip(*train_batch)
+                    try:
+                        with nlp.disable_pipes(*other_pipes):
+                            nlp.update(
+                                docs=docs,
+                                golds=golds,
+                                sgd=optimizer,
+                                drop=dropout,
+                                losses=losses,
+                            )
+                            batchnr += 1
+                            articles_processed += len(docs)
+                            pbar.update(len(docs))
+                    except Exception as e:
+                        logger.error("Error updating batch:" + str(e))
         if batchnr > 0:
-            logging.info("Epoch {}, train loss {}".format(itn, round(losses["entity_linker"] / batchnr, 2)))
+            logging.info(
-            measure_performance(dev_data, kb, el_pipe, baseline=False, context=True)
+                "Epoch {} trained on {} articles, train loss {}".format(
-
+                    itn, articles_processed, round(losses["entity_linker"] / batchnr, 2)
-    # STEP 4: measure the performance of our trained pipe on an independent dev set
+                )
-    logger.info("STEP 4: Final performance measurement of Entity Linking pipe")
+            )
-    measure_performance(dev_data, kb, el_pipe)
+            # re-read the dev_data (data is returned as a generator)
-
+            dev_data = wikipedia_processor.read_el_docs_golds(
-    # STEP 5: apply the EL pipe on a toy example
+                nlp=nlp,
-    logger.info("STEP 5: Applying Entity Linking to toy example")
+                entity_file_path=training_path,
-    run_el_toy_example(nlp=nlp)
+                dev=True,
+                line_ids=dev_indices,
+                kb=kb,
+                labels_discard=labels_discard,
+            )
+            measure_performance(
+                dev_data,
+                kb,
+                el_pipe,
+                baseline=False,
+                context=True,
+                dev_limit=len(dev_indices),
+            )
 
     if output_dir:
-        # STEP 6: write the NLP pipeline (now including an EL model) to file
+        # STEP 4: write the NLP pipeline (now including an EL model) to file
-        logger.info("STEP 6: Writing trained NLP to {}".format(nlp_output_dir))
+        logger.info(
+            "Final NLP pipeline has following pipeline components: {}".format(
+                nlp.pipe_names
+            )
+        )
+        logger.info("STEP 4: Writing trained NLP to {}".format(nlp_output_dir))
         nlp.to_disk(nlp_output_dir)
 
         logger.info("Done!")
 
 
-def check_kb(kb):
-    for mention in ("Bush", "Douglas Adams", "Homer", "Brazil", "China"):
-        candidates = kb.get_candidates(mention)
-
-        logger.info("generating candidates for " + mention + " :")
-        for c in candidates:
-            logger.info(" ".join[
-                str(c.prior_prob),
-                c.alias_,
-                "-->",
-                c.entity_ + " (freq=" + str(c.entity_freq) + ")"
-            ])
-
-
-def run_el_toy_example(nlp):
-    text = (
-        "In The Hitchhiker's Guide to the Galaxy, written by Douglas Adams, "
-        "Douglas reminds us to always bring our towel, even in China or Brazil. "
-        "The main character in Doug's novel is the man Arthur Dent, "
-        "but Dougledydoug doesn't write about George Washington or Homer Simpson."
-    )
-    doc = nlp(text)
-    logger.info(text)
-    for ent in doc.ents:
-        logger.info(" ".join(["ent", ent.text, ent.label_, ent.kb_id_]))
-
-
 if __name__ == "__main__":
     logging.basicConfig(level=logging.INFO, format=LOG_FORMAT)
     plac.call(main)

bin/wiki_entity_linking/wikipedia_processor.py

@@ -6,9 +6,6 @@ import bz2
 import logging
 import random
 import json
-from tqdm import tqdm
-
-from functools import partial
 
 from spacy.gold import GoldParse
 from bin.wiki_entity_linking import wiki_io as io
@@ -454,25 +451,40 @@ def _write_training_entities(outputfile, article_id, clean_text, entities):
     outputfile.write(line)
 
 
-def read_training(nlp, entity_file_path, dev, limit, kb, labels_discard=None):
+def read_training_indices(entity_file_path):
-    """ This method provides training examples that correspond to the entity annotations found by the nlp object.
+    """ This method creates two lists of indices into the training file: one with indices for the
+     training examples, and one for the dev examples."""
+    train_indices = []
+    dev_indices = []
+
+    with entity_file_path.open("r", encoding="utf8") as file:
+        for i, line in enumerate(file):
+            example = json.loads(line)
+            article_id = example["article_id"]
+            clean_text = example["clean_text"]
+
+            if is_valid_article(clean_text):
+                if is_dev(article_id):
+                    dev_indices.append(i)
+                else:
+                    train_indices.append(i)
+
+    return train_indices, dev_indices
+
+
+def read_el_docs_golds(nlp, entity_file_path, dev, line_ids, kb, labels_discard=None):
+    """ This method provides training/dev examples that correspond to the entity annotations found by the nlp object.
      For training, it will include both positive and negative examples by using the candidate generator from the kb.
      For testing (kb=None), it will include all positive examples only."""
     if not labels_discard:
         labels_discard = []
 
-    data = []
+    texts = []
-    num_entities = 0
+    entities_list = []
-    get_gold_parse = partial(
-        _get_gold_parse, dev=dev, kb=kb, labels_discard=labels_discard
-    )
 
-    logger.info(
-        "Reading {} data with limit {}".format("dev" if dev else "train", limit)
-    )
     with entity_file_path.open("r", encoding="utf8") as file:
-        with tqdm(total=limit, leave=False) as pbar:
+        for i, line in enumerate(file):
-            for i, line in enumerate(file):
+            if i in line_ids:
                 example = json.loads(line)
                 article_id = example["article_id"]
                 clean_text = example["clean_text"]
@@ -481,16 +493,15 @@ def read_training(nlp, entity_file_path, dev, limit, kb, labels_discard=None):
                 if dev != is_dev(article_id) or not is_valid_article(clean_text):
                     continue
 
-                doc = nlp(clean_text)
+                texts.append(clean_text)
-                gold = get_gold_parse(doc, entities)
+                entities_list.append(entities)
-                if gold and len(gold.links) > 0:
+
-                    data.append((doc, gold))
+    docs = nlp.pipe(texts, batch_size=50)
-                    num_entities += len(gold.links)
+
-                    pbar.update(len(gold.links))
+    for doc, entities in zip(docs, entities_list):
-                if limit and num_entities >= limit:
+        gold = _get_gold_parse(doc, entities, dev=dev, kb=kb, labels_discard=labels_discard)
-                    break
+        if gold and len(gold.links) > 0:
-    logger.info("Read {} entities in {} articles".format(num_entities, len(data)))
+            yield doc, gold
-    return data
 
 
 def _get_gold_parse(doc, entities, dev, kb, labels_discard):

examples/streamlit_spacy.py

@@ -26,12 +26,12 @@ DEFAULT_TEXT = "Mark Zuckerberg is the CEO of Facebook."
 HTML_WRAPPER = """<div style="overflow-x: auto; border: 1px solid #e6e9ef; border-radius: 0.25rem; padding: 1rem; margin-bottom: 2.5rem">{}</div>"""
 
 
-@st.cache(ignore_hash=True)
+@st.cache(allow_output_mutation=True)
 def load_model(name):
     return spacy.load(name)
 
 
-@st.cache(ignore_hash=True)
+@st.cache(allow_output_mutation=True)
 def process_text(model_name, text):
     nlp = load_model(model_name)
     return nlp(text)
@@ -79,7 +79,9 @@ if "ner" in nlp.pipe_names:
     st.header("Named Entities")
     st.sidebar.header("Named Entities")
     label_set = nlp.get_pipe("ner").labels
-    labels = st.sidebar.multiselect("Entity labels", label_set, label_set)
+    labels = st.sidebar.multiselect(
+        "Entity labels", options=label_set, default=list(label_set)
+    )
     html = displacy.render(doc, style="ent", options={"ents": labels})
     # Newlines seem to mess with the rendering
     html = html.replace("\n", " ")

examples/training/pretrain_kb.py

@@ -32,27 +32,24 @@ DESC_WIDTH = 64  # dimension of output entity vectors
 
 
 @plac.annotations(
-    vocab_path=("Path to the vocab for the kb", "option", "v", Path),
+    model=("Model name, should have pretrained word embeddings", "positional", None, str),
-    model=("Model name, should have pretrained word embeddings", "option", "m", str),
     output_dir=("Optional output directory", "option", "o", Path),
     n_iter=("Number of training iterations", "option", "n", int),
 )
-def main(vocab_path=None, model=None, output_dir=None, n_iter=50):
+def main(model=None, output_dir=None, n_iter=50):
     """Load the model, create the KB and pretrain the entity encodings.
-    Either an nlp model or a vocab is needed to provide access to pretrained word embeddings.
     If an output_dir is provided, the KB will be stored there in a file 'kb'.
-    When providing an nlp model, the updated vocab will also be written to a directory in the output_dir."""
+    The updated vocab will also be written to a directory in the output_dir."""
-    if model is None and vocab_path is None:
-        raise ValueError("Either the `nlp` model or the `vocab` should be specified.")
 
-    if model is not None:
+    nlp = spacy.load(model)  # load existing spaCy model
-        nlp = spacy.load(model)  # load existing spaCy model
+    print("Loaded model '%s'" % model)
-        print("Loaded model '%s'" % model)
+
-    else:
+    # check the length of the nlp vectors
-        vocab = Vocab().from_disk(vocab_path)
+    if "vectors" not in nlp.meta or not nlp.vocab.vectors.size:
-        # create blank Language class with specified vocab
+        raise ValueError(
-        nlp = spacy.blank("en", vocab=vocab)
+            "The `nlp` object should have access to pretrained word vectors, "
-        print("Created blank 'en' model with vocab from '%s'" % vocab_path)
+            " cf. https://spacy.io/usage/models#languages."
+        )
 
     kb = KnowledgeBase(vocab=nlp.vocab)
 
@@ -103,11 +100,9 @@ def main(vocab_path=None, model=None, output_dir=None, n_iter=50):
         print()
         print("Saved KB to", kb_path)
 
-        # only storing the vocab if we weren't already reading it from file
+        vocab_path = output_dir / "vocab"
-        if not vocab_path:
+        kb.vocab.to_disk(vocab_path)
-            vocab_path = output_dir / "vocab"
+        print("Saved vocab to", vocab_path)
-            kb.vocab.to_disk(vocab_path)
-            print("Saved vocab to", vocab_path)
 
         print()
 

examples/training/pretrain_textcat.py

@@ -131,7 +131,8 @@ def train_textcat(nlp, n_texts, n_iter=10):
     train_data = list(zip(train_texts, [{"cats": cats} for cats in train_cats]))
 
     # get names of other pipes to disable them during training
-    other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "textcat"]
+    pipe_exceptions = ["textcat", "trf_wordpiecer", "trf_tok2vec"]
+    other_pipes = [pipe for pipe in nlp.pipe_names if pipe not in pipe_exceptions]
     with nlp.disable_pipes(*other_pipes):  # only train textcat
         optimizer = nlp.begin_training()
         textcat.model.tok2vec.from_bytes(tok2vec_weights)

examples/training/rehearsal.py

@@ -63,7 +63,8 @@ def main(model_name, unlabelled_loc):
     optimizer.b2 = 0.0
 
     # get names of other pipes to disable them during training
-    other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "ner"]
+    pipe_exceptions = ["ner", "trf_wordpiecer", "trf_tok2vec"]
+    other_pipes = [pipe for pipe in nlp.pipe_names if pipe not in pipe_exceptions]
     sizes = compounding(1.0, 4.0, 1.001)
     with nlp.disable_pipes(*other_pipes):
         for itn in range(n_iter):

examples/training/train_entity_linker.py

@@ -113,7 +113,8 @@ def main(kb_path, vocab_path=None, output_dir=None, n_iter=50):
         TRAIN_DOCS.append((doc, annotation_clean))
 
     # get names of other pipes to disable them during training
-    other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "entity_linker"]
+    pipe_exceptions = ["entity_linker", "trf_wordpiecer", "trf_tok2vec"]
+    other_pipes = [pipe for pipe in nlp.pipe_names if pipe not in pipe_exceptions]
     with nlp.disable_pipes(*other_pipes):  # only train entity linker
         # reset and initialize the weights randomly
         optimizer = nlp.begin_training()

examples/training/train_intent_parser.py

@@ -124,7 +124,8 @@ def main(model=None, output_dir=None, n_iter=15):
         for dep in annotations.get("deps", []):
             parser.add_label(dep)
 
-    other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "parser"]
+    pipe_exceptions = ["parser", "trf_wordpiecer", "trf_tok2vec"]
+    other_pipes = [pipe for pipe in nlp.pipe_names if pipe not in pipe_exceptions]
     with nlp.disable_pipes(*other_pipes):  # only train parser
         optimizer = nlp.begin_training()
         for itn in range(n_iter):

examples/training/train_ner.py

@@ -55,7 +55,8 @@ def main(model=None, output_dir=None, n_iter=100):
             ner.add_label(ent[2])
 
     # get names of other pipes to disable them during training
-    other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "ner"]
+    pipe_exceptions = ["ner", "trf_wordpiecer", "trf_tok2vec"]
+    other_pipes = [pipe for pipe in nlp.pipe_names if pipe not in pipe_exceptions]
     with nlp.disable_pipes(*other_pipes):  # only train NER
         # reset and initialize the weights randomly – but only if we're
         # training a new model

examples/training/train_new_entity_type.py

@@ -95,7 +95,8 @@ def main(model=None, new_model_name="animal", output_dir=None, n_iter=30):
         optimizer = nlp.resume_training()
     move_names = list(ner.move_names)
     # get names of other pipes to disable them during training
-    other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "ner"]
+    pipe_exceptions = ["ner", "trf_wordpiecer", "trf_tok2vec"]
+    other_pipes = [pipe for pipe in nlp.pipe_names if pipe not in pipe_exceptions]
     with nlp.disable_pipes(*other_pipes):  # only train NER
         sizes = compounding(1.0, 4.0, 1.001)
         # batch up the examples using spaCy's minibatch

examples/training/train_parser.py

@@ -65,7 +65,8 @@ def main(model=None, output_dir=None, n_iter=15):
             parser.add_label(dep)
 
     # get names of other pipes to disable them during training
-    other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "parser"]
+    pipe_exceptions = ["parser", "trf_wordpiecer", "trf_tok2vec"]
+    other_pipes = [pipe for pipe in nlp.pipe_names if pipe not in pipe_exceptions]
     with nlp.disable_pipes(*other_pipes):  # only train parser
         optimizer = nlp.begin_training()
         for itn in range(n_iter):

examples/training/train_textcat.py

@@ -68,7 +68,8 @@ def main(model=None, output_dir=None, n_iter=20, n_texts=2000, init_tok2vec=None
     train_data = list(zip(train_texts, [{"cats": cats} for cats in train_cats]))
 
     # get names of other pipes to disable them during training
-    other_pipes = [pipe for pipe in nlp.pipe_names if pipe != "textcat"]
+    pipe_exceptions = ["textcat", "trf_wordpiecer", "trf_tok2vec"]
+    other_pipes = [pipe for pipe in nlp.pipe_names if pipe not in pipe_exceptions]
     with nlp.disable_pipes(*other_pipes):  # only train textcat
         optimizer = nlp.begin_training()
         if init_tok2vec is not None:

setup.cfg

@@ -49,6 +49,7 @@ install_requires =
     catalogue>=0.0.7,<1.1.0
     ml_datasets
     # Third-party dependencies
+    tqdm>=4.38.0,<5.0.0
     setuptools
     numpy>=1.15.0
     plac>=0.9.6,<1.2.0

spacy/attrs.pxd

@@ -92,3 +92,4 @@ cdef enum attr_id_t:
     LANG
     ENT_KB_ID = symbols.ENT_KB_ID
     MORPH
+    ENT_ID = symbols.ENT_ID

spacy/attrs.pyx

@@ -81,6 +81,7 @@ IDS = {
     "DEP": DEP,
     "ENT_IOB": ENT_IOB,
     "ENT_TYPE": ENT_TYPE,
+    "ENT_ID": ENT_ID,
     "ENT_KB_ID": ENT_KB_ID,
     "HEAD": HEAD,
     "SENT_START": SENT_START,

spacy/cli/converters/conllu2json.py

@@ -9,8 +9,14 @@ from wasabi import Printer
 
 
 def conllu2json(
-    input_data, n_sents=10, append_morphology=False, lang=None, ner_map=None,
+    input_data,
-    merge_subtokens=False, no_print=False, **_
+    n_sents=10,
+    append_morphology=False,
+    lang=None,
+    ner_map=None,
+    merge_subtokens=False,
+    no_print=False,
+    **_
 ):
     """
     Convert conllu files into JSON format for use with train cli.
@@ -26,9 +32,13 @@ def conllu2json(
     docs = []
     raw = ""
     sentences = []
-    conll_data = read_conllx(input_data, append_morphology=append_morphology,
+    conll_data = read_conllx(
-                             ner_tag_pattern=MISC_NER_PATTERN, ner_map=ner_map,
+        input_data,
-                             merge_subtokens=merge_subtokens)
+        append_morphology=append_morphology,
+        ner_tag_pattern=MISC_NER_PATTERN,
+        ner_map=ner_map,
+        merge_subtokens=merge_subtokens,
+    )
     has_ner_tags = has_ner(input_data, ner_tag_pattern=MISC_NER_PATTERN)
     for i, example in enumerate(conll_data):
         raw += example.text
@@ -72,20 +82,28 @@ def has_ner(input_data, ner_tag_pattern):
                     return False
 
 
-def read_conllx(input_data, append_morphology=False, merge_subtokens=False,
+def read_conllx(
-            ner_tag_pattern="", ner_map=None):
+    input_data,
+    append_morphology=False,
+    merge_subtokens=False,
+    ner_tag_pattern="",
+    ner_map=None,
+):
     """ Yield examples, one for each sentence """
-    vocab = Language.Defaults.create_vocab() # need vocab to make a minimal Doc
+    vocab = Language.Defaults.create_vocab()  # need vocab to make a minimal Doc
-    i = 0
     for sent in input_data.strip().split("\n\n"):
         lines = sent.strip().split("\n")
         if lines:
             while lines[0].startswith("#"):
                 lines.pop(0)
-            example = example_from_conllu_sentence(vocab, lines,
+            example = example_from_conllu_sentence(
-                    ner_tag_pattern, merge_subtokens=merge_subtokens,
+                vocab,
-                    append_morphology=append_morphology,
+                lines,
-                    ner_map=ner_map)
+                ner_tag_pattern,
+                merge_subtokens=merge_subtokens,
+                append_morphology=append_morphology,
+                ner_map=ner_map,
+            )
             yield example
 
 
@@ -157,8 +175,14 @@ def create_json_doc(raw, sentences, id_):
     return doc
 
 
-def example_from_conllu_sentence(vocab, lines, ner_tag_pattern,
+def example_from_conllu_sentence(
-        merge_subtokens=False, append_morphology=False, ner_map=None):
+    vocab,
+    lines,
+    ner_tag_pattern,
+    merge_subtokens=False,
+    append_morphology=False,
+    ner_map=None,
+):
     """Create an Example from the lines for one CoNLL-U sentence, merging
     subtokens and appending morphology to tags if required.
 
@@ -182,7 +206,6 @@ def example_from_conllu_sentence(vocab, lines, ner_tag_pattern,
     in_subtok = False
     for i in range(len(lines)):
         line = lines[i]
-        subtok_lines = []
         parts = line.split("\t")
         id_, word, lemma, pos, tag, morph, head, dep, _1, misc = parts
         if "." in id_:
@@ -266,9 +289,17 @@ def example_from_conllu_sentence(vocab, lines, ner_tag_pattern,
         if space:
             raw += " "
     example = Example(doc=raw)
-    example.set_token_annotation(ids=ids, words=words, tags=tags, pos=pos,
+    example.set_token_annotation(
-                                 morphs=morphs, lemmas=lemmas, heads=heads,
+        ids=ids,
-                                 deps=deps, entities=ents)
+        words=words,
+        tags=tags,
+        pos=pos,
+        morphs=morphs,
+        lemmas=lemmas,
+        heads=heads,
+        deps=deps,
+        entities=ents,
+    )
     return example
 
 
@@ -280,7 +311,7 @@ def merge_conllu_subtokens(lines, doc):
         id_, word, lemma, pos, tag, morph, head, dep, _1, misc = parts
         if "-" in id_:
             subtok_start, subtok_end = id_.split("-")
-            subtok_span = doc[int(subtok_start) - 1:int(subtok_end)]
+            subtok_span = doc[int(subtok_start) - 1 : int(subtok_end)]
             subtok_spans.append(subtok_span)
             # create merged tag, morph, and lemma values
             tags = []
@@ -292,7 +323,7 @@ def merge_conllu_subtokens(lines, doc):
                 if token._.merged_morph:
                     for feature in token._.merged_morph.split("|"):
                         field, values = feature.split("=", 1)
-                        if not field in morphs:
+                        if field not in morphs:
                             morphs[field] = set()
                         for value in values.split(","):
                             morphs[field].add(value)
@@ -306,7 +337,9 @@ def merge_conllu_subtokens(lines, doc):
                 token._.merged_lemma = " ".join(lemmas)
                 token.tag_ = "_".join(tags)
                 token._.merged_morph = "|".join(sorted(morphs.values()))
-                token._.merged_spaceafter = True if subtok_span[-1].whitespace_ else False
+                token._.merged_spaceafter = (
+                    True if subtok_span[-1].whitespace_ else False
+                )
 
     with doc.retokenize() as retokenizer:
         for span in subtok_spans:

spacy/cli/debug_data.py

@@ -166,6 +166,7 @@ def debug_data(
         has_low_data_warning = False
         has_no_neg_warning = False
         has_ws_ents_error = False
+        has_punct_ents_warning = False
 
         msg.divider("Named Entity Recognition")
         msg.info(
@@ -190,6 +191,14 @@ def debug_data(
             msg.fail(f"{gold_train_data['ws_ents']} invalid whitespace entity spans")
             has_ws_ents_error = True
 
+        if gold_train_data["punct_ents"]:
+            msg.warn(
+                "{} entity span(s) with punctuation".format(
+                    gold_train_data["punct_ents"]
+                )
+            )
+            has_punct_ents_warning = True
+
         for label in new_labels:
             if label_counts[label] <= NEW_LABEL_THRESHOLD:
                 msg.warn(
@@ -209,6 +218,8 @@ def debug_data(
             msg.good("Examples without occurrences available for all labels")
         if not has_ws_ents_error:
             msg.good("No entities consisting of or starting/ending with whitespace")
+        if not has_punct_ents_warning:
+            msg.good("No entities consisting of or starting/ending with punctuation")
 
         if has_low_data_warning:
             msg.text(
@@ -229,6 +240,12 @@ def debug_data(
                 "with whitespace characters are considered invalid."
             )
 
+        if has_punct_ents_warning:
+            msg.text(
+                "Entity spans consisting of or starting/ending "
+                "with punctuation can not be trained with a noise level > 0."
+            )
+
     if "textcat" in pipeline:
         msg.divider("Text Classification")
         labels = [label for label in gold_train_data["cats"]]
@@ -446,6 +463,7 @@ def _compile_gold(examples, pipeline):
         "words": Counter(),
         "roots": Counter(),
         "ws_ents": 0,
+        "punct_ents": 0,
         "n_words": 0,
         "n_misaligned_words": 0,
         "n_sents": 0,
@@ -469,6 +487,16 @@ def _compile_gold(examples, pipeline):
                 if label.startswith(("B-", "U-", "L-")) and doc[i].is_space:
                     # "Illegal" whitespace entity
                     data["ws_ents"] += 1
+                if label.startswith(("B-", "U-", "L-")) and doc[i].text in [
+                    ".",
+                    "'",
+                    "!",
+                    "?",
+                    ",",
+                ]:
+                    # punctuation entity: could be replaced by whitespace when training with noise,
+                    # so add a warning to alert the user to this unexpected side effect.
+                    data["punct_ents"] += 1
                 if label.startswith(("B-", "U-")):
                     combined_label = label.split("-")[1]
                     data["ner"][combined_label] += 1

spacy/cli/pretrain.py

@@ -28,7 +28,7 @@ def pretrain(
     vectors_model: ("Name or path to spaCy model with vectors to learn from", "positional", None, str),
     output_dir: ("Directory to write models to on each epoch", "positional", None, str),
     width: ("Width of CNN layers", "option", "cw", int) = 96,
-    depth: ("Depth of CNN layers", "option", "cd", int) = 4,
+    conv_depth: ("Depth of CNN layers", "option", "cd", int) = 4,
     bilstm_depth: ("Depth of BiLSTM layers (requires PyTorch)", "option", "lstm", int) = 0,
     cnn_pieces: ("Maxout size for CNN layers. 1 for Mish", "option", "cP", int) = 3,
     sa_depth: ("Depth of self-attention layers", "option", "sa", int) = 0,
@@ -77,9 +77,15 @@ def pretrain(
     msg.info("Using GPU" if has_gpu else "Not using GPU")
 
     output_dir = Path(output_dir)
+    if output_dir.exists() and [p for p in output_dir.iterdir()]:
+        msg.warn(
+            "Output directory is not empty",
+            "It is better to use an empty directory or refer to a new output path, "
+            "then the new directory will be created for you.",
+        )
     if not output_dir.exists():
         output_dir.mkdir()
-        msg.good("Created output directory")
+        msg.good("Created output directory: {}".format(output_dir))
     srsly.write_json(output_dir / "config.json", config)
     msg.good("Saved settings to config.json")
 
@@ -107,7 +113,7 @@ def pretrain(
         Tok2Vec(
             width,
             embed_rows,
-            conv_depth=depth,
+            conv_depth=conv_depth,
             pretrained_vectors=pretrained_vectors,
             bilstm_depth=bilstm_depth,  # Requires PyTorch. Experimental.
             subword_features=not use_chars,  # Set to False for Chinese etc

spacy/cli/train.py

@@ -10,6 +10,7 @@ import contextlib
 import random
 
 from ..util import create_default_optimizer
+from ..util import use_gpu as set_gpu
 from ..attrs import PROB, IS_OOV, CLUSTER, LANG
 from ..gold import GoldCorpus
 from .. import util
@@ -26,6 +27,14 @@ def train(
     base_model: ("Name of model to update (optional)", "option", "b", str) = None,
     pipeline: ("Comma-separated names of pipeline components", "option", "p", str) = "tagger,parser,ner",
     vectors: ("Model to load vectors from", "option", "v", str) = None,
+    replace_components: ("Replace components from base model", "flag", "R", bool) = False,
+    width: ("Width of CNN layers of Tok2Vec component", "option", "cw", int) = 96,
+    conv_depth: ("Depth of CNN layers of Tok2Vec component", "option", "cd", int) = 4,
+    cnn_window: ("Window size for CNN layers of Tok2Vec component", "option", "cW", int) = 1,
+    cnn_pieces: ("Maxout size for CNN layers of Tok2Vec component. 1 for Mish", "option", "cP", int) = 3,
+    use_chars: ("Whether to use character-based embedding of Tok2Vec component", "flag", "chr", bool) = False,
+    bilstm_depth: ("Depth of BiLSTM layers of Tok2Vec component (requires PyTorch)", "option", "lstm", int) = 0,
+    embed_rows: ("Number of embedding rows of Tok2Vec component", "option", "er", int) = 2000,
     n_iter: ("Number of iterations", "option", "n", int) = 30,
     n_early_stopping: ("Maximum number of training epochs without dev accuracy improvement", "option", "ne", int) = None,
     n_examples: ("Number of examples", "option", "ns", int) = 0,
@@ -80,6 +89,7 @@ def train(
         )
     if not output_path.exists():
         output_path.mkdir()
+        msg.good("Created output directory: {}".format(output_path))
 
     tag_map = {}
     if tag_map_path is not None:
@@ -113,6 +123,21 @@ def train(
     # training starts from a blank model, intitalize the language class.
     pipeline = [p.strip() for p in pipeline.split(",")]
     msg.text(f"Training pipeline: {pipeline}")
+    disabled_pipes = None
+    pipes_added = False
+    msg.text("Training pipeline: {}".format(pipeline))
+    if use_gpu >= 0:
+        activated_gpu = None
+        try:
+            activated_gpu = set_gpu(use_gpu)
+        except Exception as e:
+            msg.warn("Exception: {}".format(e))
+        if activated_gpu is not None:
+            msg.text("Using GPU: {}".format(use_gpu))
+        else:
+            msg.warn("Unable to activate GPU: {}".format(use_gpu))
+            msg.text("Using CPU only")
+            use_gpu = -1
     if base_model:
         msg.text(f"Starting with base model '{base_model}'")
         nlp = util.load_model(base_model)
@@ -122,20 +147,24 @@ def train(
                 f"specified as `lang` argument ('{lang}') ",
                 exits=1,
             )
-        nlp.disable_pipes([p for p in nlp.pipe_names if p not in pipeline])
         for pipe in pipeline:
+            pipe_cfg = {}
+            if pipe == "parser":
+                pipe_cfg = {"learn_tokens": learn_tokens}
+            elif pipe == "textcat":
+                pipe_cfg = {
+                    "exclusive_classes": not textcat_multilabel,
+                    "architecture": textcat_arch,
+                    "positive_label": textcat_positive_label,
+                }
             if pipe not in nlp.pipe_names:
-                if pipe == "parser":
+                msg.text("Adding component to base model '{}'".format(pipe))
-                    pipe_cfg = {"learn_tokens": learn_tokens}
-                elif pipe == "textcat":
-                    pipe_cfg = {
-                        "exclusive_classes": not textcat_multilabel,
-                        "architecture": textcat_arch,
-                        "positive_label": textcat_positive_label,
-                    }
-                else:
-                    pipe_cfg = {}
                 nlp.add_pipe(nlp.create_pipe(pipe, config=pipe_cfg))
+                pipes_added = True
+            elif replace_components:
+                msg.text("Replacing component from base model '{}'".format(pipe))
+                nlp.replace_pipe(pipe, nlp.create_pipe(pipe, config=pipe_cfg))
+                pipes_added = True
             else:
                 if pipe == "textcat":
                     textcat_cfg = nlp.get_pipe("textcat").cfg
@@ -144,11 +173,6 @@ def train(
                         "architecture": textcat_cfg["architecture"],
                         "positive_label": textcat_cfg["positive_label"],
                     }
-                    pipe_cfg = {
-                        "exclusive_classes": not textcat_multilabel,
-                        "architecture": textcat_arch,
-                        "positive_label": textcat_positive_label,
-                    }
                     if base_cfg != pipe_cfg:
                         msg.fail(
                             f"The base textcat model configuration does"
@@ -156,6 +180,10 @@ def train(
                             f"Existing cfg: {base_cfg}, provided cfg: {pipe_cfg}",
                             exits=1,
                         )
+                msg.text("Extending component from base model '{}'".format(pipe))
+        disabled_pipes = nlp.disable_pipes(
+            [p for p in nlp.pipe_names if p not in pipeline]
+        )
     else:
         msg.text(f"Starting with blank model '{lang}'")
         lang_cls = util.get_lang_class(lang)
@@ -198,13 +226,20 @@ def train(
     corpus = GoldCorpus(train_path, dev_path, limit=n_examples)
     n_train_words = corpus.count_train()
 
-    if base_model:
+    if base_model and not pipes_added:
         # Start with an existing model, use default optimizer
         optimizer = create_default_optimizer()
     else:
         # Start with a blank model, call begin_training
-        optimizer = nlp.begin_training(lambda: corpus.train_examples, device=use_gpu)
+        cfg = {"device": use_gpu}
-
+        cfg["conv_depth"] = conv_depth
+        cfg["token_vector_width"] = width
+        cfg["bilstm_depth"] = bilstm_depth
+        cfg["cnn_maxout_pieces"] = cnn_pieces
+        cfg["embed_size"] = embed_rows
+        cfg["conv_window"] = cnn_window
+        cfg["subword_features"] = not use_chars
+        optimizer = nlp.begin_training(lambda: corpus.train_tuples, **cfg)
     nlp._optimizer = None
 
     # Load in pretrained weights
@@ -214,7 +249,7 @@ def train(
 
     # Verify textcat config
     if "textcat" in pipeline:
-        textcat_labels = nlp.get_pipe("textcat").cfg["labels"]
+        textcat_labels = nlp.get_pipe("textcat").cfg.get("labels", [])
         if textcat_positive_label and textcat_positive_label not in textcat_labels:
             msg.fail(
                 f"The textcat_positive_label (tpl) '{textcat_positive_label}' "
@@ -327,12 +362,22 @@ def train(
                 for batch in util.minibatch_by_words(train_data, size=batch_sizes):
                     if not batch:
                         continue
-                    nlp.update(
+                    docs, golds = zip(*batch)
-                        batch,
+                    try:
-                        sgd=optimizer,
+                        nlp.update(
-                        drop=next(dropout_rates),
+                            docs,
-                        losses=losses,
+                            golds,
-                    )
+                            sgd=optimizer,
+                            drop=next(dropout_rates),
+                            losses=losses,
+                        )
+                    except ValueError as e:
+                        msg.warn("Error during training")
+                        if init_tok2vec:
+                            msg.warn(
+                                "Did you provide the same parameters during 'train' as during 'pretrain'?"
+                            )
+                        msg.fail("Original error message: {}".format(e), exits=1)
                     if raw_text:
                         # If raw text is available, perform 'rehearsal' updates,
                         # which use unlabelled data to reduce overfitting.
@@ -396,11 +441,16 @@ def train(
                             "cpu": cpu_wps,
                             "gpu": gpu_wps,
                         }
-                        meta["accuracy"] = scorer.scores
+                        meta.setdefault("accuracy", {})
+                        for component in nlp.pipe_names:
+                            for metric in _get_metrics(component):
+                                meta["accuracy"][metric] = scorer.scores[metric]
                     else:
                         meta.setdefault("beam_accuracy", {})
                         meta.setdefault("beam_speed", {})
-                        meta["beam_accuracy"][beam_width] = scorer.scores
+                        for component in nlp.pipe_names:
+                            for metric in _get_metrics(component):
+                                meta["beam_accuracy"][metric] = scorer.scores[metric]
                         meta["beam_speed"][beam_width] = {
                             "nwords": nwords,
                             "cpu": cpu_wps,
@@ -453,13 +503,23 @@ def train(
                             f"Best score = {best_score}; Final iteration score = {current_score}"
                         )
                         break
+    except Exception as e:
+        msg.warn(
+            "Aborting and saving the final best model. Encountered exception: {}".format(
+                e
+            )
+        )
     finally:
+        best_pipes = nlp.pipe_names
+        if disabled_pipes:
+            disabled_pipes.restore()
         with nlp.use_params(optimizer.averages):
             final_model_path = output_path / "model-final"
             nlp.to_disk(final_model_path)
+            final_meta = srsly.read_json(output_path / "model-final" / "meta.json")
         msg.good("Saved model to output directory", final_model_path)
         with msg.loading("Creating best model..."):
-            best_model_path = _collate_best_model(meta, output_path, nlp.pipe_names)
+            best_model_path = _collate_best_model(final_meta, output_path, best_pipes)
         msg.good("Created best model", best_model_path)
 
 
@@ -519,15 +579,14 @@ def _load_pretrained_tok2vec(nlp, loc):
 
 def _collate_best_model(meta, output_path, components):
     bests = {}
+    meta.setdefault("accuracy", {})
     for component in components:
         bests[component] = _find_best(output_path, component)
     best_dest = output_path / "model-best"
     shutil.copytree(str(output_path / "model-final"), str(best_dest))
     for component, best_component_src in bests.items():
         shutil.rmtree(str(best_dest / component))
-        shutil.copytree(
+        shutil.copytree(str(best_component_src / component), str(best_dest / component))
-            str(best_component_src / component), str(best_dest / component)
-        )
         accs = srsly.read_json(best_component_src / "accuracy.json")
         for metric in _get_metrics(component):
             meta["accuracy"][metric] = accs[metric]
@@ -550,13 +609,15 @@ def _find_best(experiment_dir, component):
 
 def _get_metrics(component):
     if component == "parser":
-        return ("las", "uas", "token_acc", "sent_f")
+        return ("las", "uas", "las_per_type", "token_acc", "sent_f")
     elif component == "tagger":
         return ("tags_acc",)
     elif component == "ner":
-        return ("ents_f", "ents_p", "ents_r")
+        return ("ents_f", "ents_p", "ents_r", "enty_per_type")
     elif component == "sentrec":
         return ("sent_f", "sent_p", "sent_r")
+    elif component == "textcat":
+        return ("textcat_score",)
     return ("token_acc",)
 
 
@@ -568,8 +629,12 @@ def _configure_training_output(pipeline, use_gpu, has_beam_widths):
             row_head.extend(["Tag Loss ", " Tag %  "])
             output_stats.extend(["tag_loss", "tags_acc"])
         elif pipe == "parser":
-            row_head.extend(["Dep Loss ", " UAS  ", " LAS  ", "Sent P", "Sent R", "Sent F"])
+            row_head.extend(
-            output_stats.extend(["dep_loss", "uas", "las", "sent_p", "sent_r", "sent_f"])
+                ["Dep Loss ", " UAS  ", " LAS  ", "Sent P", "Sent R", "Sent F"]
+            )
+            output_stats.extend(
+                ["dep_loss", "uas", "las", "sent_p", "sent_r", "sent_f"]
+            )
         elif pipe == "ner":
             row_head.extend(["NER Loss ", "NER P ", "NER R ", "NER F "])
             output_stats.extend(["ner_loss", "ents_p", "ents_r", "ents_f"])

spacy/displacy/__init__.py

@@ -51,9 +51,10 @@ def render(
         html = RENDER_WRAPPER(html)
     if jupyter or (jupyter is None and is_in_jupyter()):
         # return HTML rendered by IPython display()
+        # See #4840 for details on span wrapper to disable mathjax
         from IPython.core.display import display, HTML
 
-        return display(HTML(html))
+        return display(HTML('<span class="tex2jax_ignore">{}</span>'.format(html)))
     return html
 
 

spacy/errors.py

@@ -75,10 +75,9 @@ class Warnings(object):
     W015 = ("As of v2.1.0, the use of keyword arguments to exclude fields from "
             "being serialized or deserialized is deprecated. Please use the "
             "`exclude` argument instead. For example: exclude=['{arg}'].")
-    W016 = ("The keyword argument `n_threads` on the is now deprecated, as "
+    W016 = ("The keyword argument `n_threads` is now deprecated. As of v2.2.2, "
-            "the v2.x models cannot release the global interpreter lock. "
+            "the argument `n_process` controls parallel inference via "
-            "Future versions may introduce a `n_process` argument for "
+            "multiprocessing.")
-            "parallel inference via multiprocessing.")
     W017 = ("Alias '{alias}' already exists in the Knowledge Base.")
     W018 = ("Entity '{entity}' already exists in the Knowledge Base - "
             "ignoring the duplicate entry.")
@@ -170,7 +169,8 @@ class Errors(object):
             "and satisfies the correct annotations specified in the GoldParse. "
             "For example, are all labels added to the model? If you're "
             "training a named entity recognizer, also make sure that none of "
-            "your annotated entity spans have leading or trailing whitespace. "
+            "your annotated entity spans have leading or trailing whitespace "
+            "or punctuation. "
             "You can also use the experimental `debug-data` command to "
             "validate your JSON-formatted training data. For details, run:\n"
             "python -m spacy debug-data --help")

spacy/gold.pyx

@@ -991,6 +991,11 @@ cdef class GoldParse:
         self.cats = {} if cats is None else dict(cats)
         self.links = {} if links is None else dict(links)
 
+        # orig_annot is used as an iterator in `nlp.evalate` even if self.length == 0,
+        # so set a empty list to avoid error.
+        # if self.lenght > 0, this is modified latter.
+        self.orig_annot = []
+
         # avoid allocating memory if the doc does not contain any tokens
         if self.length > 0:
             if not words:

spacy/lang/char_classes.py

@@ -14,6 +14,17 @@ _tamil = r"\u0B80-\u0BFF"
 
 _telugu = r"\u0C00-\u0C7F"
 
+# from the final table in: https://en.wikipedia.org/wiki/CJK_Unified_Ideographs
+_cjk = (
+    r"\u4E00-\u62FF\u6300-\u77FF\u7800-\u8CFF\u8D00-\u9FFF\u3400-\u4DBF"
+    r"\U00020000-\U000215FF\U00021600-\U000230FF\U00023100-\U000245FF"
+    r"\U00024600-\U000260FF\U00026100-\U000275FF\U00027600-\U000290FF"
+    r"\U00029100-\U0002A6DF\U0002A700-\U0002B73F\U0002B740-\U0002B81F"
+    r"\U0002B820-\U0002CEAF\U0002CEB0-\U0002EBEF\u2E80-\u2EFF\u2F00-\u2FDF"
+    r"\u2FF0-\u2FFF\u3000-\u303F\u31C0-\u31EF\u3200-\u32FF\u3300-\u33FF"
+    r"\uF900-\uFAFF\uFE30-\uFE4F\U0001F200-\U0001F2FF\U0002F800-\U0002FA1F"
+)
+
 # Latin standard
 _latin_u_standard = r"A-Z"
 _latin_l_standard = r"a-z"
@@ -212,6 +223,7 @@ _uncased = (
     + _tamil
     + _telugu
     + _hangul
+    + _cjk
 )
 
 ALPHA = group_chars(LATIN + _russian + _tatar + _greek + _ukrainian + _uncased)

spacy/lang/de/stop_words.py

@@ -19,14 +19,14 @@ dort drei drin dritte dritten dritter drittes du durch durchaus dürfen dürft
 durfte durften
 
 eben ebenso ehrlich eigen eigene eigenen eigener eigenes ein einander eine
-einem einen einer eines einigeeinigen einiger einiges einmal einmaleins elf en
+einem einen einer eines einige einigen einiger einiges einmal einmaleins elf en
 ende endlich entweder er erst erste ersten erster erstes es etwa etwas euch
 
 früher fünf fünfte fünften fünfter fünftes für
 
 gab ganz ganze ganzen ganzer ganzes gar gedurft gegen gegenüber gehabt gehen
 geht gekannt gekonnt gemacht gemocht gemusst genug gerade gern gesagt geschweige
-gewesen gewollt geworden gibt ging gleich gott gross groß grosse große grossen
+gewesen gewollt geworden gibt ging gleich gross groß grosse große grossen
 großen grosser großer grosses großes gut gute guter gutes
 
 habe haben habt hast hat hatte hätte hatten hätten heisst heißt her heute hier
@@ -44,9 +44,8 @@ kleines kommen kommt können könnt konnte könnte konnten kurz
 lang lange leicht leider lieber los
 
 machen macht machte mag magst man manche manchem manchen mancher manches mehr
-mein meine meinem meinen meiner meines mensch menschen mich mir mit mittel
+mein meine meinem meinen meiner meines mich mir mit mittel mochte möchte mochten 
-mochte möchte mochten mögen möglich mögt morgen muss muß müssen musst müsst
+mögen möglich mögt morgen muss muß müssen musst müsst musste mussten
-musste mussten
 
 na nach nachdem nahm natürlich neben nein neue neuen neun neunte neunten neunter
 neuntes nicht nichts nie niemand niemandem niemanden noch nun nur

spacy/lang/el/__init__.py

@@ -1,5 +1,5 @@
 from .tokenizer_exceptions import TOKENIZER_EXCEPTIONS
-from .tag_map_general import TAG_MAP
+from ..tag_map import TAG_MAP
 from .stop_words import STOP_WORDS
 from .lex_attrs import LEX_ATTRS
 from .lemmatizer import GreekLemmatizer

spacy/lang/el/tag_map_general.py

@@ -1,24 +0,0 @@
-from ...symbols import POS, ADV, NOUN, ADP, PRON, SCONJ, PROPN, DET, SYM, INTJ
-from ...symbols import PUNCT, NUM, AUX, X, ADJ, VERB, PART, SPACE, CCONJ
-
-
-TAG_MAP = {
-    "ADJ": {POS: ADJ},
-    "ADV": {POS: ADV},
-    "INTJ": {POS: INTJ},
-    "NOUN": {POS: NOUN},
-    "PROPN": {POS: PROPN},
-    "VERB": {POS: VERB},
-    "ADP": {POS: ADP},
-    "CCONJ": {POS: CCONJ},
-    "SCONJ": {POS: SCONJ},
-    "PART": {POS: PART},
-    "PUNCT": {POS: PUNCT},
-    "SYM": {POS: SYM},
-    "NUM": {POS: NUM},
-    "PRON": {POS: PRON},
-    "AUX": {POS: AUX},
-    "SPACE": {POS: SPACE},
-    "DET": {POS: DET},
-    "X": {POS: X},
-}

spacy/lang/fi/punctuation.py

@@ -1,9 +1,10 @@
-from ..char_classes import LIST_ELLIPSES, LIST_ICONS
+from ..char_classes import LIST_ELLIPSES, LIST_ICONS, LIST_HYPHENS
 from ..char_classes import CONCAT_QUOTES, ALPHA, ALPHA_LOWER, ALPHA_UPPER
 from ..punctuation import TOKENIZER_SUFFIXES
 
 
 _quotes = CONCAT_QUOTES.replace("'", "")
+DASHES = "|".join(x for x in LIST_HYPHENS if x != "-")
 
 _infixes = (
     LIST_ELLIPSES
@@ -11,11 +12,9 @@ _infixes = (
     + [
         r"(?<=[{al}])\.(?=[{au}])".format(al=ALPHA_LOWER, au=ALPHA_UPPER),
         r"(?<=[{a}])[,!?](?=[{a}])".format(a=ALPHA),
-        r"(?<=[{a}])[:<>=](?=[{a}])".format(a=ALPHA),
-        r"(?<=[{a}]),(?=[{a}])".format(a=ALPHA),
         r"(?<=[{a}])([{q}\)\]\(\[])(?=[{a}])".format(a=ALPHA, q=_quotes),
-        r"(?<=[{a}])--(?=[{a}])".format(a=ALPHA),
+        r"(?<=[{a}])(?:{d})(?=[{a}])".format(a=ALPHA, d=DASHES),
-        r"(?<=[{a}0-9])[:<>=/](?=[{a}])".format(a=ALPHA),
+        r"(?<=[{a}0-9])[<>=/](?=[{a}])".format(a=ALPHA),
     ]
 )
 

spacy/lang/fi/tokenizer_exceptions.py

@@ -28,6 +28,9 @@ for exc_data in [
     {ORTH: "myöh.", LEMMA: "myöhempi"},
     {ORTH: "n.", LEMMA: "noin"},
     {ORTH: "nimim.", LEMMA: "nimimerkki"},
+    {ORTH: "n:o", LEMMA: "numero"},
+    {ORTH: "N:o", LEMMA: "numero"},
+    {ORTH: "nro", LEMMA: "numero"},
     {ORTH: "ns.", LEMMA: "niin sanottu"},
     {ORTH: "nyk.", LEMMA: "nykyinen"},
     {ORTH: "oik.", LEMMA: "oikealla"},

spacy/lang/sk/__init__.py

@@ -1,11 +1,16 @@
 from .stop_words import STOP_WORDS
+from .tag_map import TAG_MAP
+from .lex_attrs import LEX_ATTRS
+
 from ...language import Language
 from ...attrs import LANG
 
 
 class SlovakDefaults(Language.Defaults):
     lex_attr_getters = dict(Language.Defaults.lex_attr_getters)
+    lex_attr_getters.update(LEX_ATTRS)
     lex_attr_getters[LANG] = lambda text: "sk"
+    tag_map = TAG_MAP
     stop_words = STOP_WORDS
 
 

spacy/lang/sk/examples.py

@@ -0,0 +1,27 @@
+# coding: utf8
+from __future__ import unicode_literals
+
+
+"""
+Example sentences to test spaCy and its language models.
+
+>>> from spacy.lang.sk.examples import sentences
+>>> docs = nlp.pipe(sentences)
+"""
+
+
+sentences = [
+    "Ardevop, s.r.o. je malá startup firma na území SR.",
+    "Samojazdiace autá presúvajú poistnú zodpovednosť na výrobcov automobilov.",
+    "Košice sú na východe.",
+    "Bratislava je hlavné mesto Slovenskej republiky.",
+    "Kde si?",
+    "Kto je prezidentom Francúzska?",
+    "Aké je hlavné mesto Slovenska?",
+    "Kedy sa narodil Andrej Kiska?",
+    "Včera som dostal 100€ na ruku.",
+    "Dnes je nedeľa 26.1.2020.",
+    "Narodil sa 15.4.1998 v Ružomberku.",
+    "Niekto mi povedal, že 500 eur je veľa peňazí.",
+    "Podaj mi ruku!",
+]

spacy/lang/sk/lex_attrs.py

@@ -0,0 +1,62 @@
+# coding: utf8
+from __future__ import unicode_literals
+
+from ...attrs import LIKE_NUM
+
+_num_words = [
+    "nula",
+    "jeden",
+    "dva",
+    "tri",
+    "štyri",
+    "päť",
+    "šesť",
+    "sedem",
+    "osem",
+    "deväť",
+    "desať",
+    "jedenásť",
+    "dvanásť",
+    "trinásť",
+    "štrnásť",
+    "pätnásť",
+    "šestnásť",
+    "sedemnásť",
+    "osemnásť",
+    "devätnásť",
+    "dvadsať",
+    "tridsať",
+    "štyridsať",
+    "päťdesiat",
+    "šesťdesiat",
+    "sedemdesiat",
+    "osemdesiat",
+    "deväťdesiat",
+    "sto",
+    "tisíc",
+    "milión",
+    "miliarda",
+    "bilión",
+    "biliarda",
+    "trilión",
+    "triliarda",
+    "kvadrilión",
+]
+
+
+def like_num(text):
+    if text.startswith(("+", "-", "±", "~")):
+        text = text[1:]
+    text = text.replace(",", "").replace(".", "")
+    if text.isdigit():
+        return True
+    if text.count("/") == 1:
+        num, denom = text.split("/")
+        if num.isdigit() and denom.isdigit():
+            return True
+    if text.lower() in _num_words:
+        return True
+    return False
+
+
+LEX_ATTRS = {LIKE_NUM: like_num}

spacy/lang/sk/stop_words.py

@@ -1,5 +1,5 @@
 
-# Source: https://github.com/stopwords-iso/stopwords-sk
+# Source: https://github.com/Ardevop-sk/stopwords-sk
 
 STOP_WORDS = set(
     """
@@ -7,17 +7,41 @@ a
 aby
 aj
 ak
+akej
+akejže
 ako
+akom
+akomže
+akou
+akouže
+akože
+aká
+akáže
+aké
+akého
+akéhože
+akému
+akémuže
+akéže
+akú
+akúže
 aký
+akých
+akýchže
+akým
+akými
+akýmiže
+akýmže
+akýže
 ale
 alebo
-and
 ani
 asi
 avšak
 až
 ba
 bez
+bezo
 bol
 bola
 boli
@@ -28,23 +52,32 @@ budeme
 budete
 budeš
 budú
-buï
 buď
 by
 byť
 cez
+cezo
 dnes
 do
 ešte
-for
 ho
 hoci
 i
 iba
 ich
 im
+inej
+inom
+iná
 iné
+iného
+inému
+iní
+inú
 iný
+iných
+iným
+inými
 ja
 je
 jeho
@@ -53,80 +86,185 @@ jemu
 ju
 k
 kam
+kamže
+každou
 každá
 každé
+každého
+každému
 každí
+každú
 každý
+každých
+každým
+každými
 kde
-kedže
+kej
-keï
+kejže
 keď
+keďže
+kie
+kieho
+kiehože
+kiemu
+kiemuže
+kieže
+koho
+kom
+komu
+kou
+kouže
 kto
+ktorej
 ktorou
 ktorá
 ktoré
 ktorí
+ktorú
 ktorý
+ktorých
+ktorým
+ktorými
 ku
+ká
+káže
+ké
+kéže
+kú
+kúže
+ký
+kýho
+kýhože
+kým
+kýmu
+kýmuže
+kýže
 lebo
+leda
+ledaže
 len
 ma
+majú
+mal
+mala
+mali
 mať
 medzi
-menej
 mi
-mna
 mne
 mnou
 moja
 moje
+mojej
+mojich
+mojim
+mojimi
+mojou
+moju
+možno
 mu
+musia
 musieť
+musí
+musím
+musíme
+musíte
+musíš
 my
 má
+mám
+máme
 máte
-mòa
+máš
 môcť
 môj
+môjho
 môže
+môžem
+môžeme
+môžete
+môžeš
+môžu
+mňa
 na
 nad
+nado
+najmä
 nami
+naša
+naše
+našej
 naši
+našich
+našim
+našimi
+našou
+ne
 nech
 neho
 nej
+nejakej
+nejakom
+nejakou
+nejaká
+nejaké
+nejakého
+nejakému
+nejakú
+nejaký
+nejakých
+nejakým
+nejakými
 nemu
 než
 nich
 nie
+niektorej
+niektorom
+niektorou
+niektorá
+niektoré
+niektorého
+niektorému
+niektorú
 niektorý
+niektorých
+niektorým
+niektorými
 nielen
+niečo
 nim
+nimi
 nič
+ničoho
+ničom
+ničomu
+ničím
 no
-nová
-nové
-noví
-nový
 nám
 nás
 náš
+nášho
 ním
 o
 od
 odo
-of
 on
 ona
 oni
 ono
 ony
+oň
+oňho
 po
 pod
+podo
 podľa
 pokiaľ
+popod
+popri
 potom
+poza
 pre
 pred
 predo
@@ -134,42 +272,56 @@ preto
 pretože
 prečo
 pri
-prvá
-prvé
-prví
-prvý
 práve
-pýta
 s
 sa
 seba
+sebe
+sebou
 sem
 si
 sme
 so
 som
-späť
 ste
 svoj
+svoja
 svoje
+svojho
 svojich
+svojim
+svojimi
+svojou
+svoju
 svojím
-svojími
 sú
 ta
 tak
+takej
+takejto
+taká
+takáto
+také
+takého
+takéhoto
+takému
+takémuto
+takéto
+takí
+takú
+takúto
 taký
+takýto
 takže
 tam
-te
 teba
 tebe
 tebou
 teda
 tej
+tejto
 ten
 tento
-the
 ti
 tie
 tieto
@@ -177,52 +329,97 @@ tiež
 to
 toho
 tohoto
+tohto
 tom
 tomto
 tomu
 tomuto
 toto
 tou
+touto
 tu
 tvoj
-tvojími
+tvoja
+tvoje
+tvojej
+tvojho
+tvoji
+tvojich
+tvojim
+tvojimi
+tvojím
 ty
 tá
 táto
+tí
+títo
 tú
 túto
+tých
 tým
+tými
 týmto
-tě
+u
 už
 v
 vami
+vaša
 vaše
-veï
+vašej
+vaši
+vašich
+vašim
+vaším
+veď
 viac
 vo
 vy
 vám
 vás
 váš
+vášho
 však
+všetci
+všetka
+všetko
+všetky
 všetok
 z
 za
+začo
+začože
 zo
-a
 áno
-èi
+čej
-èo
-èí
-òom
-òou
-òu
 či
+čia
+čie
+čieho
+čiemu
+čiu
 čo
+čoho
+čom
+čomu
+čou
+čože
+čí
+čím
+čími
 ďalšia
 ďalšie
+ďalšieho
+ďalšiemu
+ďalšiu
+ďalšom
+ďalšou
 ďalší
+ďalších
+ďalším
+ďalšími
+ňom
+ňou
+ňu
 že
 """.split()
 )

spacy/lang/tokenizer_exceptions.py

@@ -1,13 +1,17 @@
 import re
 
+from .char_classes import ALPHA_LOWER
 from ..symbols import ORTH, POS, TAG, LEMMA, SPACE
 
 
 # URL validation regex courtesy of: https://mathiasbynens.be/demo/url-regex
-# A few minor mods to this regex to account for use cases represented in test_urls
+# and https://gist.github.com/dperini/729294 (Diego Perini, MIT License)
+# A few mods to this regex to account for use cases represented in test_urls
 URL_PATTERN = (
+    # fmt: off
     r"^"
-    # protocol identifier (see: https://www.iana.org/assignments/uri-schemes/uri-schemes.xhtml)
+    # protocol identifier (mods: make optional and expand schemes)
+    # (see: https://www.iana.org/assignments/uri-schemes/uri-schemes.xhtml)
     r"(?:(?:[\w\+\-\.]{2,})://)?"
     # mailto:user or user:pass authentication
     r"(?:\S+(?::\S*)?@)?"
@@ -28,18 +32,27 @@ URL_PATTERN = (
     r"(?:\.(?:1?\d{1,2}|2[0-4]\d|25[0-5])){2}"
     r"(?:\.(?:[1-9]\d?|1\d\d|2[0-4]\d|25[0-4]))"
     r"|"
-    # host name
+    # host & domain names
-    r"(?:(?:[a-z0-9\-]*)?[a-z0-9]+)"
+    # mods: match is case-sensitive, so include [A-Z]
-    # domain name
+      "(?:"
-    r"(?:\.(?:[a-z0-9])(?:[a-z0-9\-])*[a-z0-9])?"
+        "(?:"
+          "[A-Za-z0-9\u00a1-\uffff]"
+          "[A-Za-z0-9\u00a1-\uffff_-]{0,62}"
+        ")?"
+        "[A-Za-z0-9\u00a1-\uffff]\."
+      ")+"
     # TLD identifier
-    r"(?:\.(?:[a-z]{2,}))"
+    # mods: use ALPHA_LOWER instead of a wider range so that this doesn't match
+    # strings like "lower.Upper", which can be split on "." by infixes in some
+    # languages
+    r"(?:[" + ALPHA_LOWER + "]{2,63})"
     r")"
     # port number
     r"(?::\d{2,5})?"
     # resource path
     r"(?:[/?#]\S*)?"
     r"$"
+    # fmt: on
 ).strip()
 
 TOKEN_MATCH = re.compile(URL_PATTERN, re.UNICODE).match

spacy/language.py

@@ -4,7 +4,6 @@ import weakref
 import functools
 from contextlib import contextmanager
 from copy import copy, deepcopy
-from thinc.model import Model
 from thinc.backends import get_current_ops
 import srsly
 import multiprocessing as mp
@@ -481,7 +480,6 @@ class Language(object):
             component_cfg.setdefault(name, {})
             component_cfg[name].setdefault("drop", drop)
             component_cfg[name].setdefault("set_annotations", False)
-        grads = {}
         for name, proc in self.pipeline:
             if not hasattr(proc, "update"):
                 continue
@@ -581,7 +579,8 @@ class Language(object):
                 self.vocab.vectors.data = ops.asarray(self.vocab.vectors.data)
         link_vectors_to_models(self.vocab)
         if self.vocab.vectors.data.shape[1]:
-            cfg["pretrained_vectors"] = self.vocab.vectors
+            cfg["pretrained_vectors"] = self.vocab.vectors.name
+            cfg["pretrained_dims"] = self.vocab.vectors.data.shape[1]
         if sgd is None:
             sgd = create_default_optimizer()
         self._optimizer = sgd
@@ -746,7 +745,7 @@ class Language(object):
 
         pipes = (
             []
-        )  # contains functools.partial objects so that easily create multiprocess worker.
+        )  # contains functools.partial objects to easily create multiprocess worker.
         for name, proc in self.pipeline:
             if name in disable:
                 continue
@@ -803,7 +802,7 @@ class Language(object):
         texts, raw_texts = itertools.tee(texts)
         # for sending texts to worker
         texts_q = [mp.Queue() for _ in range(n_process)]
-        # for receiving byte encoded docs from worker
+        # for receiving byte-encoded docs from worker
         bytedocs_recv_ch, bytedocs_send_ch = zip(
             *[mp.Pipe(False) for _ in range(n_process)]
         )
@@ -813,7 +812,7 @@ class Language(object):
         # This is necessary to properly handle infinite length of texts.
         # (In this case, all data cannot be sent to the workers at once)
         sender = _Sender(batch_texts, texts_q, chunk_size=n_process)
-        # send twice so that make process busy
+        # send twice to make process busy
         sender.send()
         sender.send()
 
@@ -825,7 +824,7 @@ class Language(object):
             proc.start()
 
         # Cycle channels not to break the order of docs.
-        # The received object is batch of byte encoded docs, so flatten them with chain.from_iterable.
+        # The received object is a batch of byte-encoded docs, so flatten them with chain.from_iterable.
         byte_docs = chain.from_iterable(recv.recv() for recv in cycle(bytedocs_recv_ch))
         docs = (Doc(self.vocab).from_bytes(byte_doc) for byte_doc in byte_docs)
         try:

spacy/pipeline/entityruler.py

@@ -4,7 +4,7 @@ import srsly
 from ..language import component
 from ..errors import Errors
 from ..util import ensure_path, to_disk, from_disk
-from ..tokens import Span
+from ..tokens import Doc, Span
 from ..matcher import Matcher, PhraseMatcher
 
 DEFAULT_ENT_ID_SEP = "||"
@@ -125,20 +125,31 @@ class EntityRuler(object):
 
         DOCS: https://spacy.io/api/entityruler#labels
         """
-        all_labels = set(self.token_patterns.keys())
+        keys = set(self.token_patterns.keys())
-        all_labels.update(self.phrase_patterns.keys())
+        keys.update(self.phrase_patterns.keys())
+        all_labels = set()
+
+        for l in keys:
+            if self.ent_id_sep in l:
+                label, _ = self._split_label(l)
+                all_labels.add(label)
+            else:
+                all_labels.add(l)
         return tuple(all_labels)
 
     @property
     def ent_ids(self):
-        """All entity ids present in the match patterns `id` properties.
+        """All entity ids present in the match patterns `id` properties
 
         RETURNS (set): The string entity ids.
 
         DOCS: https://spacy.io/api/entityruler#ent_ids
         """
+        keys = set(self.token_patterns.keys())
+        keys.update(self.phrase_patterns.keys())
         all_ent_ids = set()
-        for l in self.labels:
+
+        for l in keys:
             if self.ent_id_sep in l:
                 _, ent_id = self._split_label(l)
                 all_ent_ids.add(ent_id)
@@ -147,6 +158,7 @@ class EntityRuler(object):
     @property
     def patterns(self):
         """Get all patterns that were added to the entity ruler.
+
         RETURNS (list): The original patterns, one dictionary per pattern.
 
         DOCS: https://spacy.io/api/entityruler#patterns
@@ -179,6 +191,7 @@ class EntityRuler(object):
 
         DOCS: https://spacy.io/api/entityruler#add_patterns
         """
+
         # disable the nlp components after this one in case they hadn't been initialized / deserialised yet
         try:
             current_index = self.nlp.pipe_names.index(self.name)
@@ -188,7 +201,31 @@ class EntityRuler(object):
         except ValueError:
             subsequent_pipes = []
         with self.nlp.disable_pipes(subsequent_pipes):
+            token_patterns = []
+            phrase_pattern_labels = []
+            phrase_pattern_texts = []
+            phrase_pattern_ids = []
+
             for entry in patterns:
+                if isinstance(entry["pattern"], str):
+                    phrase_pattern_labels.append(entry["label"])
+                    phrase_pattern_texts.append(entry["pattern"])
+                    phrase_pattern_ids.append(entry.get("id"))
+                elif isinstance(entry["pattern"], list):
+                    token_patterns.append(entry)
+
+            phrase_patterns = []
+            for label, pattern, ent_id in zip(
+                phrase_pattern_labels,
+                self.nlp.pipe(phrase_pattern_texts),
+                phrase_pattern_ids,
+            ):
+                phrase_pattern = {"label": label, "pattern": pattern, "id": ent_id}
+                if ent_id:
+                    phrase_pattern["id"] = ent_id
+                phrase_patterns.append(phrase_pattern)
+
+            for entry in token_patterns + phrase_patterns:
                 label = entry["label"]
                 if "id" in entry:
                     ent_label = label
@@ -197,8 +234,8 @@ class EntityRuler(object):
                     self._ent_ids[key] = (ent_label, entry["id"])
 
                 pattern = entry["pattern"]
-                if isinstance(pattern, str):
+                if isinstance(pattern, Doc):
-                    self.phrase_patterns[label].append(self.nlp(pattern))
+                    self.phrase_patterns[label].append(pattern)
                 elif isinstance(pattern, list):
                     self.token_patterns[label].append(pattern)
                 else:
@@ -211,6 +248,8 @@ class EntityRuler(object):
     def _split_label(self, label):
         """Split Entity label into ent_label and ent_id if it contains self.ent_id_sep
 
+        label (str): The value of label in a pattern entry
+
         RETURNS (tuple): ent_label, ent_id
         """
         if self.ent_id_sep in label:
@@ -224,6 +263,9 @@ class EntityRuler(object):
     def _create_label(self, label, ent_id):
         """Join Entity label with ent_id if the pattern has an `id` attribute
 
+        label (str): The label to set for ent.label_
+        ent_id (str): The label
+
         RETURNS (str): The ent_label joined with configured `ent_id_sep`
         """
         if isinstance(ent_id, str):
@@ -235,6 +277,7 @@ class EntityRuler(object):
 
         patterns_bytes (bytes): The bytestring to load.
         **kwargs: Other config paramters, mostly for consistency.
+
         RETURNS (EntityRuler): The loaded entity ruler.
 
         DOCS: https://spacy.io/api/entityruler#from_bytes
@@ -274,6 +317,7 @@ class EntityRuler(object):
 
         path (unicode / Path): The JSONL file to load.
         **kwargs: Other config paramters, mostly for consistency.
+
         RETURNS (EntityRuler): The loaded entity ruler.
 
         DOCS: https://spacy.io/api/entityruler#from_disk

spacy/pipeline/pipes.pyx

@@ -1632,7 +1632,7 @@ class EntityLinker(Pipe):
         for i, doc in enumerate(docs):
             if len(doc) > 0:
                 # Looping through each sentence and each entity
-                # This may go wrong if there are entities across sentences - because they might not get a KB ID
+                # This may go wrong if there are entities across sentences - which shouldn't happen normally.
                 for sent in doc.sents:
                     sent_doc = sent.as_doc()
                     # currently, the context is the same for each entity in a sentence (should be refined)
@@ -1829,7 +1829,7 @@ class Sentencizer(Pipe):
                     yield ex
             else:
                 yield from docs
-    
+
     def predict(self, docs):
         """Apply the pipeline's model to a batch of docs, without
         modifying them.
@@ -1840,20 +1840,21 @@ class Sentencizer(Pipe):
             return guesses
         guesses = []
         for doc in docs:
-            start = 0
-            seen_period = False
             doc_guesses = [False] * len(doc)
-            doc_guesses[0] = True
+            if len(doc) > 0:
-            for i, token in enumerate(doc):
+                start = 0
-                is_in_punct_chars = token.text in self.punct_chars
+                seen_period = False
-                if seen_period and not token.is_punct and not is_in_punct_chars:
+                doc_guesses[0] = True
+                for i, token in enumerate(doc):
+                    is_in_punct_chars = token.text in self.punct_chars
+                    if seen_period and not token.is_punct and not is_in_punct_chars:
+                        doc_guesses[start] = True
+                        start = token.i
+                        seen_period = False
+                    elif is_in_punct_chars:
+                        seen_period = True
+                if start < len(doc):
                     doc_guesses[start] = True
-                    start = token.i
-                    seen_period = False
-                elif is_in_punct_chars:
-                    seen_period = True
-            if start < len(doc):
-                doc_guesses[start] = True
             guesses.append(doc_guesses)
         return guesses
 

spacy/symbols.pxd

@@ -463,3 +463,4 @@ cdef enum symbol_t:
 
     ENT_KB_ID
     MORPH
+    ENT_ID

spacy/symbols.pyx

@@ -82,6 +82,7 @@ IDS = {
     "DEP": DEP,
     "ENT_IOB": ENT_IOB,
     "ENT_TYPE": ENT_TYPE,
+    "ENT_ID": ENT_ID,
     "ENT_KB_ID": ENT_KB_ID,
     "HEAD": HEAD,
     "SENT_START": SENT_START,

spacy/syntax/nn_parser.pyx

@@ -57,7 +57,7 @@ cdef class Parser:
         subword_features = util.env_opt('subword_features',
                             cfg.get('subword_features', True))
         conv_depth = util.env_opt('conv_depth', cfg.get('conv_depth', 4))
-        window_size = util.env_opt('window_size', cfg.get('window_size', 1))
+        conv_window = util.env_opt('conv_window', cfg.get('conv_window', 1))
         t2v_pieces = util.env_opt('cnn_maxout_pieces', cfg.get('cnn_maxout_pieces', 3))
         bilstm_depth = util.env_opt('bilstm_depth', cfg.get('bilstm_depth', 0))
         self_attn_depth = util.env_opt('self_attn_depth', cfg.get('self_attn_depth', 0))

spacy/tests/doc/test_doc_api.py

@@ -4,7 +4,7 @@ import numpy
 from spacy.tokens import Doc, Span
 from spacy.vocab import Vocab
 from spacy.errors import ModelsWarning
-from spacy.attrs import ENT_TYPE, ENT_IOB
+from spacy.attrs import ENT_TYPE, ENT_IOB, SENT_START, HEAD, DEP
 
 from ..util import get_doc
 
@@ -271,6 +271,39 @@ def test_doc_is_nered(en_vocab):
     assert new_doc.is_nered
 
 
+def test_doc_from_array_sent_starts(en_vocab):
+    words = ["I", "live", "in", "New", "York", ".", "I", "like", "cats", "."]
+    heads = [0, 0, 0, 0, 0, 0, 6, 6, 6, 6]
+    deps = ["ROOT", "dep", "dep", "dep", "dep", "dep", "ROOT", "dep", "dep", "dep", "dep"]
+    doc = Doc(en_vocab, words=words)
+    for i, (dep, head) in enumerate(zip(deps, heads)):
+        doc[i].dep_ = dep
+        doc[i].head = doc[head]
+        if head == i:
+            doc[i].is_sent_start = True
+    doc.is_parsed
+
+    attrs = [SENT_START, HEAD]
+    arr = doc.to_array(attrs)
+    new_doc = Doc(en_vocab, words=words)
+    with pytest.raises(ValueError):
+        new_doc.from_array(attrs, arr)
+
+    attrs = [SENT_START, DEP]
+    arr = doc.to_array(attrs)
+    new_doc = Doc(en_vocab, words=words)
+    new_doc.from_array(attrs, arr)
+    assert [t.is_sent_start for t in doc] == [t.is_sent_start for t in new_doc]
+    assert not new_doc.is_parsed
+
+    attrs = [HEAD, DEP]
+    arr = doc.to_array(attrs)
+    new_doc = Doc(en_vocab, words=words)
+    new_doc.from_array(attrs, arr)
+    assert [t.is_sent_start for t in doc] == [t.is_sent_start for t in new_doc]
+    assert new_doc.is_parsed
+
+
 def test_doc_lang(en_vocab):
     doc = Doc(en_vocab, words=["Hello", "world"])
     assert doc.lang_ == "en"

spacy/tests/doc/test_span.py

@@ -276,3 +276,12 @@ def test_filter_spans(doc):
     assert len(filtered[1]) == 5
     assert filtered[0].start == 1 and filtered[0].end == 4
     assert filtered[1].start == 5 and filtered[1].end == 10
+
+
+def test_span_eq_hash(doc, doc_not_parsed):
+    assert doc[0:2] == doc[0:2]
+    assert doc[0:2] != doc[1:3]
+    assert doc[0:2] != doc_not_parsed[0:2]
+    assert hash(doc[0:2]) == hash(doc[0:2])
+    assert hash(doc[0:2]) != hash(doc[1:3])
+    assert hash(doc[0:2]) != hash(doc_not_parsed[0:2])

spacy/tests/lang/fi/test_tokenizer.py

@@ -16,6 +16,21 @@ HYPHENATED_TESTS = [
     )
 ]
 
+ABBREVIATION_INFLECTION_TESTS = [
+    (
+        "VTT:ssa ennen v:ta 2010 suoritetut mittaukset",
+        ["VTT:ssa", "ennen", "v:ta", "2010", "suoritetut", "mittaukset"]
+    ),
+    (
+        "ALV:n osuus on 24 %.",
+        ["ALV:n", "osuus", "on", "24", "%", "."]
+    ),
+    (
+        "Hiihtäjä oli kilpailun 14:s.",
+        ["Hiihtäjä", "oli", "kilpailun", "14:s", "."]
+    )
+]
+
 
 @pytest.mark.parametrize("text,expected_tokens", ABBREVIATION_TESTS)
 def test_fi_tokenizer_abbreviations(fi_tokenizer, text, expected_tokens):
@@ -29,3 +44,10 @@ def test_fi_tokenizer_hyphenated_words(fi_tokenizer, text, expected_tokens):
     tokens = fi_tokenizer(text)
     token_list = [token.text for token in tokens if not token.is_space]
     assert expected_tokens == token_list
+
+
+@pytest.mark.parametrize("text,expected_tokens", ABBREVIATION_INFLECTION_TESTS)
+def test_fi_tokenizer_abbreviation_inflections(fi_tokenizer, text, expected_tokens):
+    tokens = fi_tokenizer(text)
+    token_list = [token.text for token in tokens if not token.is_space]
+    assert expected_tokens == token_list

spacy/tests/lang/hu/test_tokenizer.py

@@ -293,9 +293,8 @@ WIKI_TESTS = [
     ("cérium(IV)-oxid", ["cérium", "(", "IV", ")", "-oxid"]),
 ]
 
-TESTCASES = (
+EXTRA_TESTS = (
-    DEFAULT_TESTS
+    DOT_TESTS
-    + DOT_TESTS
     + QUOTE_TESTS
     + NUMBER_TESTS
     + HYPHEN_TESTS
@@ -303,8 +302,16 @@ TESTCASES = (
     + TYPO_TESTS
 )
 
+# normal: default tests + 10% of extra tests
+TESTS = DEFAULT_TESTS
+TESTS.extend([x for i, x in enumerate(EXTRA_TESTS) if i % 10 == 0])
 
-@pytest.mark.parametrize("text,expected_tokens", TESTCASES)
+# slow: remaining 90% of extra tests
+SLOW_TESTS = [x for i, x in enumerate(EXTRA_TESTS) if i % 10 != 0]
+TESTS.extend([pytest.param(x[0], x[1], marks=pytest.mark.slow()) if not isinstance(x[0], tuple) else x for x in SLOW_TESTS])
+
+
+@pytest.mark.parametrize("text,expected_tokens", TESTS)
 def test_hu_tokenizer_handles_testcases(hu_tokenizer, text, expected_tokens):
     tokens = hu_tokenizer(text)
     token_list = [token.text for token in tokens if not token.is_space]

spacy/tests/lang/tt/test_tokenizer.py

@@ -41,15 +41,15 @@ TYPOS_IN_PUNC_TESTS = [
 
 LONG_TEXTS_TESTS = [
     (
-        "Иң борынгы кешеләр суыклар һәм салкын кышлар булмый торган җылы"
+        "Иң борынгы кешеләр суыклар һәм салкын кышлар булмый торган җылы "
-        "якларда яшәгәннәр, шуңа күрә аларга кием кирәк булмаган.Йөз"
+        "якларда яшәгәннәр, шуңа күрә аларга кием кирәк булмаган.Йөз "
-        "меңнәрчә еллар үткән, борынгы кешеләр акрынлап Европа һәм Азиянең"
+        "меңнәрчә еллар үткән, борынгы кешеләр акрынлап Европа һәм Азиянең "
-        "салкын илләрендә дә яши башлаганнар. Алар кырыс һәм салкын"
+        "салкын илләрендә дә яши башлаганнар. Алар кырыс һәм салкын "
         "кышлардан саклану өчен кием-салым уйлап тапканнар - итәк.",
-        "Иң борынгы кешеләр суыклар һәм салкын кышлар булмый торган җылы"
+        "Иң борынгы кешеләр суыклар һәм салкын кышлар булмый торган җылы "
-        "якларда яшәгәннәр , шуңа күрә аларга кием кирәк булмаган . Йөз"
+        "якларда яшәгәннәр , шуңа күрә аларга кием кирәк булмаган . Йөз "
-        "меңнәрчә еллар үткән , борынгы кешеләр акрынлап Европа һәм Азиянең"
+        "меңнәрчә еллар үткән , борынгы кешеләр акрынлап Европа һәм Азиянең "
-        "салкын илләрендә дә яши башлаганнар . Алар кырыс һәм салкын"
+        "салкын илләрендә дә яши башлаганнар . Алар кырыс һәм салкын "
         "кышлардан саклану өчен кием-салым уйлап тапканнар - итәк .".split(),
     )
 ]

spacy/tests/pipeline/test_entity_ruler.py

@@ -18,6 +18,7 @@ def patterns():
         {"label": "HELLO", "pattern": [{"ORTH": "HELLO"}]},
         {"label": "COMPLEX", "pattern": [{"ORTH": "foo", "OP": "*"}]},
         {"label": "TECH_ORG", "pattern": "Apple", "id": "a1"},
+        {"label": "TECH_ORG", "pattern": "Microsoft", "id": "a2"},
     ]
 
 
@@ -144,3 +145,14 @@ def test_entity_ruler_validate(nlp):
     # invalid pattern raises error with validate
     with pytest.raises(MatchPatternError):
         validated_ruler.add_patterns([invalid_pattern])
+
+
+def test_entity_ruler_properties(nlp, patterns):
+    ruler = EntityRuler(nlp, patterns=patterns, overwrite_ents=True)
+    assert sorted(ruler.labels) == sorted([
+        "HELLO",
+        "BYE",
+        "COMPLEX",
+        "TECH_ORG"
+    ])
+    assert sorted(ruler.ent_ids) == ["a1", "a2"]

spacy/tests/pipeline/test_sentencizer.py

@@ -32,6 +32,22 @@ def test_sentencizer_pipe():
         assert len(list(doc.sents)) == 2
 
 
+def test_sentencizer_empty_docs():
+    one_empty_text = [""]
+    many_empty_texts = ["", "", ""]
+    some_empty_texts = ["hi", "", "This is a test. Here are two sentences.", ""]
+    nlp = English()
+    nlp.add_pipe(nlp.create_pipe("sentencizer"))
+    for texts in [one_empty_text, many_empty_texts, some_empty_texts]:
+        for doc in nlp.pipe(texts):
+            assert doc.is_sentenced
+            sent_starts = [t.is_sent_start for t in doc]
+            if len(doc) == 0:
+                assert sent_starts == []
+            else:
+                assert len(sent_starts) > 0
+
+
 @pytest.mark.parametrize(
     "words,sent_starts,n_sents",
     [

spacy/tests/regression/test_issue4665.py

@@ -0,0 +1,31 @@
+from spacy.cli.converters.conllu2json import conllu2json
+
+input_data = """
+1	[	_	PUNCT	-LRB-	_	_	punct	_	_
+2	This	_	DET	DT	_	_	det	_	_
+3	killing	_	NOUN	NN	_	_	nsubj	_	_
+4	of	_	ADP	IN	_	_	case	_	_
+5	a	_	DET	DT	_	_	det	_	_
+6	respected	_	ADJ	JJ	_	_	amod	_	_
+7	cleric	_	NOUN	NN	_	_	nmod	_	_
+8	will	_	AUX	MD	_	_	aux	_	_
+9	be	_	AUX	VB	_	_	aux	_	_
+10	causing	_	VERB	VBG	_	_	root	_	_
+11	us	_	PRON	PRP	_	_	iobj	_	_
+12	trouble	_	NOUN	NN	_	_	dobj	_	_
+13	for	_	ADP	IN	_	_	case	_	_
+14	years	_	NOUN	NNS	_	_	nmod	_	_
+15	to	_	PART	TO	_	_	mark	_	_
+16	come	_	VERB	VB	_	_	acl	_	_
+17	.	_	PUNCT	.	_	_	punct	_	_
+18	]	_	PUNCT	-RRB-	_	_	punct	_	_
+"""
+
+
+def test_issue4665():
+    """
+    conllu2json should not raise an exception if the HEAD column contains an
+    underscore
+    """
+
+    conllu2json(input_data)

spacy/tests/regression/test_issue4849.py

@@ -0,0 +1,36 @@
+# coding: utf8
+from __future__ import unicode_literals
+
+from spacy.lang.en import English
+from spacy.pipeline import EntityRuler
+
+
+def test_issue4849():
+    nlp = English()
+
+    ruler = EntityRuler(
+        nlp, patterns=[
+            {"label": "PERSON", "pattern": 'joe biden', "id": 'joe-biden'},
+            {"label": "PERSON", "pattern": 'bernie sanders', "id": 'bernie-sanders'},
+        ],
+        phrase_matcher_attr="LOWER"
+    )
+
+    nlp.add_pipe(ruler)
+
+    text = """
+    The left is starting to take aim at Democratic front-runner Joe Biden.
+    Sen. Bernie Sanders joined in her criticism: "There is no 'middle ground' when it comes to climate policy."
+    """
+
+    # USING 1 PROCESS
+    count_ents = 0
+    for doc in nlp.pipe([text], n_process=1):
+        count_ents += len([ent for ent in doc.ents if ent.ent_id > 0])
+    assert(count_ents == 2)
+
+    # USING 2 PROCESSES
+    count_ents = 0
+    for doc in nlp.pipe([text], n_process=2):
+        count_ents += len([ent for ent in doc.ents if ent.ent_id > 0])
+    assert (count_ents == 2)

spacy/tests/regression/test_issue4924.py

@@ -0,0 +1,16 @@
+# coding: utf8
+from __future__ import unicode_literals
+
+import pytest
+
+import spacy
+
+
+@pytest.fixture
+def nlp():
+    return spacy.blank("en")
+
+
+def test_evaluate(nlp):
+    docs_golds = [("", {})]
+    nlp.evaluate(docs_golds)

spacy/tests/serialize/test_serialize_extension_attrs.py

@@ -1,5 +1,5 @@
 import pytest
-from spacy.tokens import Doc
+from spacy.tokens import Doc, Token
 from spacy.vocab import Vocab
 
 
@@ -10,6 +10,10 @@ def doc_w_attrs(en_tokenizer):
     Doc.set_extension("_test_method", method=lambda doc, arg: f"{len(doc.text)}{arg}")
     doc = en_tokenizer("This is a test.")
     doc._._test_attr = "test"
+
+    Token.set_extension("_test_token", default="t0")
+    doc[1]._._test_token = "t1"
+
     return doc
 
 
@@ -20,3 +24,6 @@ def test_serialize_ext_attrs_from_bytes(doc_w_attrs):
     assert doc._._test_attr == "test"
     assert doc._._test_prop == len(doc.text)
     assert doc._._test_method("test") == f"{len(doc.text)}test"
+    assert doc[0]._._test_token == "t0"
+    assert doc[1]._._test_token == "t1"
+    assert doc[2]._._test_token == "t0"

spacy/tests/tokenizer/test_urls.py

@@ -19,6 +19,7 @@ URLS_FULL = URLS_BASIC + [
 # URL SHOULD_MATCH and SHOULD_NOT_MATCH patterns courtesy of https://mathiasbynens.be/demo/url-regex
 URLS_SHOULD_MATCH = [
     "http://foo.com/blah_blah",
+    "http://BlahBlah.com/Blah_Blah",
     "http://foo.com/blah_blah/",
     "http://www.example.com/wpstyle/?p=364",
     "https://www.example.com/foo/?bar=baz&inga=42&quux",
@@ -56,14 +57,17 @@ URLS_SHOULD_MATCH = [
     ),
     "http://foo.com/blah_blah_(wikipedia)",
     "http://foo.com/blah_blah_(wikipedia)_(again)",
-    pytest.param("http://⌘.ws", marks=pytest.mark.xfail()),
+    "http://www.foo.co.uk",
-    pytest.param("http://⌘.ws/", marks=pytest.mark.xfail()),
+    "http://www.foo.co.uk/",
-    pytest.param("http://☺.damowmow.com/", marks=pytest.mark.xfail()),
+    "http://www.foo.co.uk/blah/blah",
-    pytest.param("http://✪df.ws/123", marks=pytest.mark.xfail()),
+    "http://⌘.ws",
-    pytest.param("http://➡.ws/䨹", marks=pytest.mark.xfail()),
+    "http://⌘.ws/",
-    pytest.param("http://مثال.إختبار", marks=pytest.mark.xfail()),
+    "http://☺.damowmow.com/",
-    pytest.param("http://例子.测试", marks=pytest.mark.xfail()),
+    "http://✪df.ws/123",
-    pytest.param("http://उदाहरण.परीक्षा", marks=pytest.mark.xfail()),
+    "http://➡.ws/䨹",
+    "http://مثال.إختبار",
+    "http://例子.测试",
+    "http://उदाहरण.परीक्षा",
 ]
 
 URLS_SHOULD_NOT_MATCH = [

spacy/tests/util.py

@@ -91,7 +91,11 @@ def assert_docs_equal(doc1, doc2):
 
     assert [t.ent_type for t in doc1] == [t.ent_type for t in doc2]
     assert [t.ent_iob for t in doc1] == [t.ent_iob for t in doc2]
-    assert [ent for ent in doc1.ents] == [ent for ent in doc2.ents]
+    for ent1, ent2 in zip(doc1.ents, doc2.ents):
+        assert ent1.start == ent2.start
+        assert ent1.end == ent2.end
+        assert ent1.label == ent2.label
+        assert ent1.kb_id == ent2.kb_id
 
 
 def assert_packed_msg_equal(b1, b2):

spacy/tokens/doc.pyx

@@ -19,7 +19,7 @@ from ..lexeme cimport Lexeme, EMPTY_LEXEME
 from ..typedefs cimport attr_t, flags_t
 from ..attrs cimport ID, ORTH, NORM, LOWER, SHAPE, PREFIX, SUFFIX, CLUSTER
 from ..attrs cimport LENGTH, POS, LEMMA, TAG, DEP, HEAD, SPACY, ENT_IOB
-from ..attrs cimport ENT_TYPE, ENT_KB_ID, SENT_START, attr_id_t
+from ..attrs cimport ENT_TYPE, ENT_ID, ENT_KB_ID, SENT_START, attr_id_t
 from ..parts_of_speech cimport CCONJ, PUNCT, NOUN, univ_pos_t
 
 from ..attrs import intify_attrs, IDS
@@ -65,6 +65,8 @@ cdef attr_t get_token_attr(const TokenC* token, attr_id_t feat_name) nogil:
         return token.ent_iob
     elif feat_name == ENT_TYPE:
         return token.ent_type
+    elif feat_name == ENT_ID:
+        return token.ent_id
     elif feat_name == ENT_KB_ID:
         return token.ent_kb_id
     else:
@@ -807,7 +809,7 @@ cdef class Doc:
                 if attr_ids[j] != TAG:
                     Token.set_struct_attr(token, attr_ids[j], array[i, j])
         # Set flags
-        self.is_parsed = bool(self.is_parsed or HEAD in attrs or DEP in attrs)
+        self.is_parsed = bool(self.is_parsed or HEAD in attrs)
         self.is_tagged = bool(self.is_tagged or TAG in attrs or POS in attrs)
         # If document is parsed, set children
         if self.is_parsed:
@@ -864,7 +866,7 @@ cdef class Doc:
 
         DOCS: https://spacy.io/api/doc#to_bytes
         """
-        array_head = [LENGTH, SPACY, LEMMA, ENT_IOB, ENT_TYPE]  # TODO: ENT_KB_ID ?
+        array_head = [LENGTH, SPACY, LEMMA, ENT_IOB, ENT_TYPE, ENT_ID]  # TODO: ENT_KB_ID ?
         if self.is_tagged:
             array_head.extend([TAG, POS])
         # If doc parsed add head and dep attribute
@@ -990,9 +992,9 @@ cdef class Doc:
          order, and no span intersection is allowed.
 
         spans (Span[]): Spans to merge, in document order, with all span
-            intersections empty. Cannot be emty.
+            intersections empty. Cannot be empty.
         attributes (Dictionary[]): Attributes to assign to the merged tokens. By default,
-            must be the same lenghth as spans, emty dictionaries are allowed.
+            must be the same length as spans, empty dictionaries are allowed.
             attributes are inherited from the syntactic root of the span.
         RETURNS (Token): The first newly merged token.
         """

spacy/tokens/span.pyx

@@ -124,22 +124,27 @@ cdef class Span:
                 return False
             else:
                 return True
-        # Eq
+        # <
         if op == 0:
             return self.start_char < other.start_char
+        # <=
         elif op == 1:
             return self.start_char <= other.start_char
+        # ==
         elif op == 2:
-            return self.start_char == other.start_char and self.end_char == other.end_char
+            return (self.doc, self.start_char, self.end_char, self.label, self.kb_id) == (other.doc, other.start_char, other.end_char, other.label, other.kb_id)
+        # !=
         elif op == 3:
-            return self.start_char != other.start_char or self.end_char != other.end_char
+            return (self.doc, self.start_char, self.end_char, self.label, self.kb_id) != (other.doc, other.start_char, other.end_char, other.label, other.kb_id)
+        # >
         elif op == 4:
             return self.start_char > other.start_char
+        # >=
         elif op == 5:
             return self.start_char >= other.start_char
 
     def __hash__(self):
-        return hash((self.doc, self.label, self.start_char, self.end_char))
+        return hash((self.doc, self.start_char, self.end_char, self.label, self.kb_id))
 
     def __len__(self):
         """Get the number of tokens in the span.
@@ -207,7 +212,7 @@ cdef class Span:
         words = [t.text for t in self]
         spaces = [bool(t.whitespace_) for t in self]
         cdef Doc doc = Doc(self.doc.vocab, words=words, spaces=spaces)
-        array_head = [LENGTH, SPACY, LEMMA, ENT_IOB, ENT_TYPE, ENT_KB_ID]
+        array_head = [LENGTH, SPACY, LEMMA, ENT_IOB, ENT_TYPE, ENT_ID, ENT_KB_ID]
         if self.doc.is_tagged:
             array_head.append(TAG)
         # If doc parsed add head and dep attribute

spacy/tokens/token.pxd

@@ -55,6 +55,8 @@ cdef class Token:
             return token.ent_iob
         elif feat_name == ENT_TYPE:
             return token.ent_type
+        elif feat_name == ENT_ID:
+            return token.ent_id
         elif feat_name == ENT_KB_ID:
             return token.ent_kb_id
         elif feat_name == SENT_START:
@@ -85,6 +87,8 @@ cdef class Token:
             token.ent_iob = value
         elif feat_name == ENT_TYPE:
             token.ent_type = value
+        elif feat_name == ENT_ID:
+            token.ent_id = value
         elif feat_name == ENT_KB_ID:
             token.ent_kb_id = value
         elif feat_name == SENT_START:

spacy/vectors.pyx

@@ -278,7 +278,11 @@ cdef class Vectors:
 
         DOCS: https://spacy.io/api/vectors#add
         """
-        key = get_string_id(key)
+        # use int for all keys and rows in key2row for more efficient access
+        # and serialization
+        key = int(get_string_id(key))
+        if row is not None:
+            row = int(row)
         if row is None and key in self.key2row:
             row = self.key2row[key]
         elif row is None:

website/docs/api/cli.md

@@ -372,7 +372,7 @@ $ python -m spacy train [lang] [output_path] [train_path] [dev_path]
 | `--n-iter`, `-n`                                                | option        | Number of iterations (default: `30`).                                                                                                                             |
 | `--n-early-stopping`, `-ne`                                     | option        | Maximum number of training epochs without dev accuracy improvement.                                                                                               |
 | `--n-examples`, `-ns`                                           | option        | Number of examples to use (defaults to `0` for all examples).                                                                                                     |
-| `--use-gpu`, `-g`                                               | option        | Whether to use GPU. Can be either `0`, `1` or `-1`.                                                                                                               |
+| `--use-gpu`, `-g`                                               | option        | GPU ID or `-1` for CPU only (default: `-1`).                                                                                                                      |
 | `--version`, `-V`                                               | option        | Model version. Will be written out to the model's `meta.json` after training.                                                                                     |
 | `--meta-path`, `-m` <Tag variant="new">2</Tag>                  | option        | Optional path to model [`meta.json`](/usage/training#models-generating). All relevant properties like `lang`, `pipeline` and `spacy_version` will be overwritten. |
 | `--init-tok2vec`, `-t2v` <Tag variant="new">2.1</Tag>           | option        | Path to pretrained weights for the token-to-vector parts of the models. See `spacy pretrain`. Experimental.                                                       |

website/docs/api/language.md

@@ -77,9 +77,9 @@ more efficient than processing texts one-by-one.
 Early versions of spaCy used simple statistical models that could be efficiently
 multi-threaded, as we were able to entirely release Python's global interpreter
 lock. The multi-threading was controlled using the `n_threads` keyword argument
-to the `.pipe` method. This keyword argument is now deprecated as of v2.1.0.
+to the `.pipe` method. This keyword argument is now deprecated as of v2.1.0. A
-Future versions may introduce a `n_process` argument for parallel inference via
+new keyword argument, `n_process`, was introduced to control parallel inference
-multiprocessing.
+via multiprocessing in v2.2.2.
 
 </Infobox>
 
@@ -98,6 +98,7 @@ multiprocessing.
 | `batch_size`                                 | int   | The number of texts to buffer.                                                                                                                             |
 | `disable`                                    | list  | Names of pipeline components to [disable](/usage/processing-pipelines#disabling).                                                                          |
 | `component_cfg` <Tag variant="new">2.1</Tag> | dict  | Config parameters for specific pipeline components, keyed by component name.                                                                               |
+| `n_process` <Tag variant="new">2.2.2</Tag>   | int   | Number of processors to use, only supported in Python 3. Defaults to `1`.                                                                                  |
 | **YIELDS**                                   | `Doc` | Documents in the order of the original text.                                                                                                               |
 
 ## Language.update {#update tag="method"}

website/docs/api/phrasematcher.md

@@ -38,7 +38,7 @@ be shown.
 | Name                                    | Type            | Description                                                                                 |
 | --------------------------------------- | --------------- | ------------------------------------------------------------------------------------------- |
 | `vocab`                                 | `Vocab`         | The vocabulary object, which must be shared with the documents the matcher will operate on. |
-| `max_length`                            | int             | Deprecated argument -  the `PhraseMatcher` does not have a phrase length limit anymore.     |
+| `max_length`                            | int             | Deprecated argument - the `PhraseMatcher` does not have a phrase length limit anymore.      |
 | `attr` <Tag variant="new">2.1</Tag>     | int / unicode   | The token attribute to match on. Defaults to `ORTH`, i.e. the verbatim token text.          |
 | `validate` <Tag variant="new">2.1</Tag> | bool            | Validate patterns added to the matcher.                                                     |
 | **RETURNS**                             | `PhraseMatcher` | The newly constructed object.                                                               |
@@ -70,6 +70,18 @@ Find all token sequences matching the supplied patterns on the `Doc`.
 | `doc`       | `Doc` | The document to match over.                                                                                                                                              |
 | **RETURNS** | list  | A list of `(match_id, start, end)` tuples, describing the matches. A match tuple describes a span `doc[start:end]`. The `match_id` is the ID of the added match pattern. |
 
+<Infobox title="Note on retrieving the string representation of the match_id" variant="warning">
+
+Because spaCy stores all strings as integers, the `match_id` you get back will
+be an integer, too – but you can always get the string representation by looking
+it up in the vocabulary's `StringStore`, i.e. `nlp.vocab.strings`:
+
+```python
+match_id_string = nlp.vocab.strings[match_id]
+```
+
+</Infobox>
+
 ## PhraseMatcher.pipe {#pipe tag="method"}
 
 Match a stream of documents, yielding them in turn.

website/docs/usage/facts-figures.md

@@ -5,8 +5,6 @@ next: /usage/spacy-101
 menu:
   - ['Feature Comparison', 'comparison']
   - ['Benchmarks', 'benchmarks']
-  - ['Powered by spaCy', 'powered-by']
-  - ['Other Libraries', 'other-libraries']
 ---
 
 ## Feature comparison {#comparison}

website/docs/usage/index.md

@@ -135,9 +135,8 @@ interface for GPU arrays.
 spaCy can be installed on GPU by specifying `spacy[cuda]`, `spacy[cuda90]`,
 `spacy[cuda91]`, `spacy[cuda92]` or `spacy[cuda100]`. If you know your cuda
 version, using the more explicit specifier allows cupy to be installed via
-wheel, saving some compilation time. The specifiers should install two
+wheel, saving some compilation time. The specifiers should install
-libraries: [`cupy`](https://cupy.chainer.org) and
+[`cupy`](https://cupy.chainer.org).
-[`thinc_gpu_ops`](https://github.com/explosion/thinc_gpu_ops).
 
 ```bash
 $ pip install -U spacy[cuda92]

website/docs/usage/linguistic-features.md

@@ -327,7 +327,7 @@ displaCy in our [online demo](https://explosion.ai/demos/displacy)..
 ### Disabling the parser {#disabling}
 
 In the [default models](/models), the parser is loaded and enabled as part of
-the [standard processing pipeline](/usage/processing-pipelin). If you don't need
+the [standard processing pipeline](/usage/processing-pipelines). If you don't need
 any of the syntactic information, you should disable the parser. Disabling the
 parser will make spaCy load and run much faster. If you want to load the parser,
 but need to disable it for specific documents, you can also control its use on

website/docs/usage/rule-based-matching.md

@@ -9,7 +9,7 @@ menu:
 ---
 
 Compared to using regular expressions on raw text, spaCy's rule-based matcher
-engines and components not only let you find you the words and phrases you're
+engines and components not only let you find the words and phrases you're
 looking for – they also give you access to the tokens within the document and
 their relationships. This means you can easily access and analyze the
 surrounding tokens, merge spans into single tokens or add entries to the named
@@ -1096,6 +1096,33 @@ with the patterns. When you load the model back in, all pipeline components will
 be restored and deserialized – including the entity ruler. This lets you ship
 powerful model packages with binary weights _and_ rules included!
 
+### Using a large number of phrase patterns {#entityruler-large-phrase-patterns new="2.2.4"}
+
+When using a large amount of **phrase patterns** (roughly > 10000) it's useful to understand how the `add_patterns` function of the EntityRuler works. For each **phrase pattern**,
+the EntityRuler calls the nlp object to construct a doc object. This happens in case you try
+to add the EntityRuler at the end of an existing pipeline with, for example, a POS tagger and want to 
+extract matches based on the pattern's POS signature.
+
+In this case you would pass a config value of `phrase_matcher_attr="POS"` for the EntityRuler.
+
+Running the full language pipeline across every pattern in a large list scales linearly and can therefore take a long time on large amounts of phrase patterns.
+
+As of spaCy 2.2.4 the `add_patterns` function has been refactored to use nlp.pipe on all phrase patterns resulting in about a 10x-20x speed up with 5,000-100,000 phrase patterns respectively. 
+
+Even with this speedup (but especially if you're using an older version) the `add_patterns` function can still take a long time.
+
+An easy workaround to make this function run faster is disabling the other language pipes
+while adding the phrase patterns.
+
+```python
+entityruler = EntityRuler(nlp)
+patterns = [{"label": "TEST", "pattern": str(i)} for i in range(100000)]
+
+other_pipes = [p for p in nlp.pipe_names if p != "tagger"]
+with nlp.disable_pipes(*disable_pipes):
+    entityruler.add_patterns(patterns)
+```
+
 ## Combining models and rules {#models-rules}
 
 You can combine statistical and rule-based components in a variety of ways.

website/docs/usage/vectors-similarity.md

@@ -229,10 +229,10 @@ For more details on **adding hooks** and **overwriting** the built-in `Doc`,
 If you're using a GPU, it's much more efficient to keep the word vectors on the
 device. You can do that by setting the [`Vectors.data`](/api/vectors#attributes)
 attribute to a `cupy.ndarray` object if you're using spaCy or
-[Chainer]("https://chainer.org"), or a `torch.Tensor` object if you're using
+[Chainer](https://chainer.org), or a `torch.Tensor` object if you're using
-[PyTorch]("http://pytorch.org"). The `data` object just needs to support
+[PyTorch](http://pytorch.org). The `data` object just needs to support
 `__iter__` and `__getitem__`, so if you're using another library such as
-[TensorFlow]("https://www.tensorflow.org"), you could also create a wrapper for
+[TensorFlow](https://www.tensorflow.org), you could also create a wrapper for
 your vectors data.
 
 ```python

website/meta/universe.json

@@ -999,6 +999,17 @@
             "author": "Graphbrain",
             "category": ["standalone"]
         },
+        {
+            "type": "education",
+            "id": "nostarch-nlp-python",
+            "title": "Natural Language Processing Using Python",
+            "slogan": "No Starch Press, 2020",
+            "description": "Natural Language Processing Using Python is an introduction to natural language processing (NLP), the task of converting human language into data that a computer can process. The book uses spaCy, a leading Python library for NLP, to guide readers through common NLP tasks related to generating and understanding human language with code. It addresses problems like understanding a user's intent, continuing a conversation with a human, and maintaining the state of a conversation.",
+            "cover": "https://nostarch.com/sites/default/files/styles/uc_product_full/public/NaturalLanguageProcessing_final_v01.jpg",
+            "url": "https://nostarch.com/NLPPython",
+            "author": "Yuli Vasiliev",
+            "category": ["books"]
+        },                     
         {
             "type": "education",
             "id": "oreilly-python-ds",
@@ -1509,28 +1520,30 @@
         {
             "id": "spacy-conll",
             "title": "spacy_conll",
-            "slogan": "Parse text with spaCy and print the output in CoNLL-U format",
+            "slogan": "Parse text with spaCy and gets its output in CoNLL-U format",
-            "description": "This module allows you to parse a text to CoNLL-U format. You can use it as a command line tool, or embed it in your own scripts.",
+            "description": "This module allows you to parse a text to CoNLL-U format. It contains a pipeline component for spaCy that adds CoNLL-U properties to a Doc and its sentences. It can also be used as a command-line tool.",
             "code_example": [
-                "from spacy_conll import Spacy2ConllParser",
+                "import spacy",
-                "spacyconll = Spacy2ConllParser()",
+                "from spacy_conll import ConllFormatter",
                 "",
-                "# `parse` returns a generator of the parsed sentences",
+                "nlp = spacy.load('en')",
-                "for parsed_sent in spacyconll.parse(input_str='I like cookies.\nWhat about you?\nI don't like 'em!'):",
+                "conllformatter = ConllFormatter(nlp)",
-                "    do_something_(parsed_sent)",
+                "nlp.add_pipe(conllformatter, after='parser')",
-                "",
+                "doc = nlp('I like cookies. Do you?')",
-                "# `parseprint` prints output to stdout (default) or a file (use `output_file` parameter)",
+                "conll = doc._.conll",
-                "# This method is called when using the command line",
+                "print(doc._.conll_str_headers)",             
-                "spacyconll.parseprint(input_str='I like cookies.')"
+                "print(doc._.conll_str)"
             ],
             "code_language": "python",
             "author": "Bram Vanroy",
             "author_links": {
-                "github": "BramVanroy",
+                "github": "BramVanroy",                
+                "twitter": "BramVanroy",
                 "website": "https://bramvanroy.be"
             },
             "github": "BramVanroy/spacy_conll",
-            "category": ["standalone"]
+            "category": ["standalone", "pipeline"],
+            "tags": ["linguistics", "computational linguistics", "conll"]
         },
         {
             "id": "spacy-langdetect",
@@ -1837,6 +1850,20 @@
                 "github": "microsoft"
             }
         },
+        {
+            "id": "presidio-research",
+            "title": "Presidio Research",
+            "slogan": "Toolbox for developing and evaluating PII detectors, NER models for PII and generating fake PII data",
+            "description": "This package features data-science related tasks for developing new recognizers for Microsoft Presidio. It is used for the evaluation of the entire system, as well as for evaluating specific PII recognizers or PII detection models. Anyone interested in evaluating an existing Microsoft Presidio instance, a specific PII recognizer or to develop new models or logic for detecting PII could leverage the preexisting work in this package. Additionally, anyone interested in generating new data based on previous datasets (e.g. to increase the coverage of entity values) for Named Entity Recognition models could leverage the data generator contained in this package.",
+            "url": "https://aka.ms/presidio-research",
+            "github": "microsoft/presidio-research",
+            "category": ["standalone"],
+            "thumb": "https://avatars0.githubusercontent.com/u/6154722",
+            "author": "Microsoft",
+            "author_links": {
+                "github": "microsoft"
+            }
+        },
         {
             "id": "python-sentence-boundary-disambiguation",
             "title": "pySBD - python Sentence Boundary Disambiguation",
@@ -1901,6 +1928,43 @@
                 "twitter": "PatadiaYash",
                 "github": "yash1994"
             }
+        },
+        {
+            "id": "spacy-pytextrank",
+            "title": "PyTextRank",
+            "slogan": "Py impl of TextRank for lightweight phrase extraction",
+            "description": "An implementation of TextRank in Python for use in spaCy pipelines which provides fast, effective phrase extraction from texts, along with extractive summarization. The graph algorithm works independent of a specific natural language and does not require domain knowledge. See (Mihalcea 2004) https://web.eecs.umich.edu/~mihalcea/papers/mihalcea.emnlp04.pdf",
+            "github": "DerwenAI/pytextrank",
+            "pip": "pytextrank",
+            "code_example": [
+                "import spacy",
+                "import pytextrank",
+                "",
+                "nlp = spacy.load('en_core_web_sm')",
+                "",
+                "tr = pytextrank.TextRank()",
+                "nlp.add_pipe(tr.PipelineComponent, name='textrank', last=True)",
+                "",
+                "text = 'Compatibility of systems of linear constraints over the set of natural numbers. Criteria of compatibility of a system of linear Diophantine equations, strict inequations, and nonstrict inequations are considered.'",
+                "doc = nlp(text)",
+                "",
+                "# examine the top-ranked phrases in the document",
+                "for p in doc._.phrases:",
+                "    print('{:.4f} {:5d}  {}'.format(p.rank, p.count, p.text))",
+                "    print(p.chunks)"
+            ],
+            "code_language": "python",
+            "url": "https://github.com/DerwenAI/pytextrank/wiki",
+            "thumb": "https://memegenerator.net/img/instances/66942896.jpg",
+            "image": "https://memegenerator.net/img/instances/66942896.jpg",
+            "author": "Paco Nathan",
+            "author_links": {
+                "twitter": "pacoid",
+                "github": "ceteri",
+                "website": "https://derwen.ai/paco"
+            },
+            "category": ["pipeline"],
+            "tags": ["phrase extraction", "ner", "summarization", "graph algorithms", "textrank"]
         }
     ],