mirror of
https://github.com/explosion/spaCy.git
synced 2024-09-21 19:39:13 +03:00
2457318b7a
* updating rules and vocabulary for French lemmatization of verbs * updating the file with French auxiliary verb * updating rules and vocabulary for French lemmatization of verbs * adding contributor agreement for amperinet * adding rules for words with inclusive parentheses wrongly tokenized
370 lines
9.6 KiB
Python
370 lines
9.6 KiB
Python
# coding: utf8
|
|
from __future__ import unicode_literals
|
|
|
|
|
|
AUXILIARY_VERBS_IRREG = {
|
|
"été": ("être",),
|
|
"suis": ("être",),
|
|
"es": ("être",),
|
|
"est": ("être",),
|
|
"sommes": ("être",),
|
|
"êtes": ("être",),
|
|
"sont": ("être",),
|
|
"étais": ("être",),
|
|
"étais": ("être",),
|
|
"était": ("être",),
|
|
"étions": ("être",),
|
|
"étiez": ("être",),
|
|
"étaient": ("être",),
|
|
"fus": ("être",),
|
|
"fut": ("être",),
|
|
"fûmes": ("être",),
|
|
"fûtes": ("être",),
|
|
"furent": ("être",),
|
|
"serai": ("être",),
|
|
"seras": ("être",),
|
|
"sera": ("être",),
|
|
"serons": ("être",),
|
|
"serez": ("être",),
|
|
"seront": ("être",),
|
|
"serais": ("être",),
|
|
"serait": ("être",),
|
|
"serions": ("être",),
|
|
"seriez": ("être",),
|
|
"seraient": ("être",),
|
|
"sois": ("être",),
|
|
"soit": ("être",),
|
|
"soyons": ("être",),
|
|
"soyez": ("être",),
|
|
"soient": ("être",),
|
|
"fusse": ("être",),
|
|
"fusses": ("être",),
|
|
"fût": ("être",),
|
|
"fussions": ("être",),
|
|
"fussiez": ("être",),
|
|
"fussent": ("être",),
|
|
"étant": ("être",),
|
|
"ai": ("avoir",),
|
|
"as": ("avoir",),
|
|
"a": ("avoir",),
|
|
"avons": ("avoir",),
|
|
"avez": ("avoir",),
|
|
"ont": ("avoir",),
|
|
"avais": ("avoir",),
|
|
"avait": ("avoir",),
|
|
"avions": ("avoir",),
|
|
"aviez": ("avoir",),
|
|
"avaient": ("avoir",),
|
|
"eus": ("avoir",),
|
|
"eut": ("avoir",),
|
|
"eûmes": ("avoir",),
|
|
"eûtes": ("avoir",),
|
|
"eurent": ("avoir",),
|
|
"aurai": ("avoir",),
|
|
"auras": ("avoir",),
|
|
"aura": ("avoir",),
|
|
"aurons": ("avoir",),
|
|
"aurez": ("avoir",),
|
|
"auront": ("avoir",),
|
|
"aurais": ("avoir",),
|
|
"aurait": ("avoir",),
|
|
"aurions": ("avoir",),
|
|
"auriez": ("avoir",),
|
|
"auraient": ("avoir",),
|
|
"aie": ("avoir",),
|
|
"aies": ("avoir",),
|
|
"ait": ("avoir",),
|
|
"ayons": ("avoir",),
|
|
"ayez": ("avoir",),
|
|
"aient": ("avoir",),
|
|
"eusse": ("avoir",),
|
|
"eusses": ("avoir",),
|
|
"eût": ("avoir",),
|
|
"eussions": ("avoir",),
|
|
"eussiez": ("avoir",),
|
|
"eussent": ("avoir",),
|
|
"ayant": ("avoir",),
|
|
"eu": ("avoir",),
|
|
"eue": ("avoir",),
|
|
"eues": ("avoir",),
|
|
"devaient": ("devoir",),
|
|
"devais": ("devoir",),
|
|
"devait": ("devoir",),
|
|
"devant": ("devoir",),
|
|
"devez": ("devoir",),
|
|
"deviez": ("devoir",),
|
|
"devions": ("devoir",),
|
|
"devons": ("devoir",),
|
|
"devra": ("devoir",),
|
|
"devrai": ("devoir",),
|
|
"devraient": ("devoir",),
|
|
"devrais": ("devoir",),
|
|
"devrait": ("devoir",),
|
|
"devras": ("devoir",),
|
|
"devrez": ("devoir",),
|
|
"devriez": ("devoir",),
|
|
"devrions": ("devoir",),
|
|
"devrons": ("devoir",),
|
|
"devront": ("devoir",),
|
|
"dois": ("devoir",),
|
|
"doit": ("devoir",),
|
|
"doive": ("devoir",),
|
|
"doivent": ("devoir",),
|
|
"doives": ("devoir",),
|
|
"dû": ("devoir",),
|
|
"due": ("devoir",),
|
|
"dues": ("devoir",),
|
|
"dûmes": ("devoir",),
|
|
"durent": ("devoir",),
|
|
"dus": ("devoir",),
|
|
"dûs": ("devoir",),
|
|
"dusse": ("devoir",),
|
|
"dussent": ("devoir",),
|
|
"dusses": ("devoir",),
|
|
"dussiez": ("devoir",),
|
|
"dussions": ("devoir",),
|
|
"dut": ("devoir",),
|
|
"dût": ("devoir",),
|
|
"dûtes": ("devoir",),
|
|
"peut": ("pouvoir",),
|
|
"peuvent": ("pouvoir",),
|
|
"peux": ("pouvoir",),
|
|
"pourraient": ("pouvoir",),
|
|
"pourrai": ("pouvoir",),
|
|
"pourrais": ("pouvoir",),
|
|
"pourrait": ("pouvoir",),
|
|
"pourra": ("pouvoir",),
|
|
"pourras": ("pouvoir",),
|
|
"pourrez": ("pouvoir",),
|
|
"pourriez": ("pouvoir",),
|
|
"pourrions": ("pouvoir",),
|
|
"pourrons": ("pouvoir",),
|
|
"pourront": ("pouvoir",),
|
|
"pouvaient": ("pouvoir",),
|
|
"pouvais": ("pouvoir",),
|
|
"pouvait": ("pouvoir",),
|
|
"pouvez": ("pouvoir",),
|
|
"pouviez": ("pouvoir",),
|
|
"pouvions": ("pouvoir",),
|
|
"pouvons": ("pouvoir",),
|
|
"pûmes": ("pouvoir",),
|
|
"pu": ("pouvoir",),
|
|
"purent": ("pouvoir",),
|
|
"pus": ("pouvoir",),
|
|
"pûtes": ("pouvoir",),
|
|
"put": ("pouvoir",),
|
|
"pouvant": ("pouvoir",),
|
|
"puisse": ("pouvoir",),
|
|
"puissions": ("pouvoir",),
|
|
"puissiez": ("pouvoir",),
|
|
"puissent": ("pouvoir",),
|
|
"pusse": ("pouvoir",),
|
|
"pusses": ("pouvoir",),
|
|
"pussions": ("pouvoir",),
|
|
"pussiez": ("pouvoir",),
|
|
"pussent": ("pouvoir",),
|
|
"faisaient": ("faire",),
|
|
"faisais": ("faire",),
|
|
"faisait": ("faire",),
|
|
"faisant": ("faire",),
|
|
"fais": ("faire",),
|
|
"faisiez": ("faire",),
|
|
"faisions": ("faire",),
|
|
"faisons": ("faire",),
|
|
"faite": ("faire",),
|
|
"faites": ("faire",),
|
|
"fait": ("faire",),
|
|
"faits": ("faire",),
|
|
"fasse": ("faire",),
|
|
"fassent": ("faire",),
|
|
"fasses": ("faire",),
|
|
"fassiez": ("faire",),
|
|
"fassions": ("faire",),
|
|
"fera": ("faire",),
|
|
"feraient": ("faire",),
|
|
"ferai": ("faire",),
|
|
"ferais": ("faire",),
|
|
"ferait": ("faire",),
|
|
"feras": ("faire",),
|
|
"ferez": ("faire",),
|
|
"feriez": ("faire",),
|
|
"ferions": ("faire",),
|
|
"ferons": ("faire",),
|
|
"feront": ("faire",),
|
|
"fîmes": ("faire",),
|
|
"firent": ("faire",),
|
|
"fis": ("faire",),
|
|
"fisse": ("faire",),
|
|
"fissent": ("faire",),
|
|
"fisses": ("faire",),
|
|
"fissiez": ("faire",),
|
|
"fissions": ("faire",),
|
|
"fîtes": ("faire",),
|
|
"fit": ("faire",),
|
|
"fît": ("faire",),
|
|
"font": ("faire",),
|
|
"veuillent": ("vouloir",),
|
|
"veuilles": ("vouloir",),
|
|
"veuille": ("vouloir",),
|
|
"veuillez": ("vouloir",),
|
|
"veuillons": ("vouloir",),
|
|
"veulent": ("vouloir",),
|
|
"veut": ("vouloir",),
|
|
"veux": ("vouloir",),
|
|
"voudraient": ("vouloir",),
|
|
"voudrais": ("vouloir",),
|
|
"voudrait": ("vouloir",),
|
|
"voudrai": ("vouloir",),
|
|
"voudras": ("vouloir",),
|
|
"voudra": ("vouloir",),
|
|
"voudrez": ("vouloir",),
|
|
"voudriez": ("vouloir",),
|
|
"voudrions": ("vouloir",),
|
|
"voudrons": ("vouloir",),
|
|
"voudront": ("vouloir",),
|
|
"voulaient": ("vouloir",),
|
|
"voulais": ("vouloir",),
|
|
"voulait": ("vouloir",),
|
|
"voulant": ("vouloir",),
|
|
"voulez": ("vouloir",),
|
|
"vouliez": ("vouloir",),
|
|
"voulions": ("vouloir",),
|
|
"voulons": ("vouloir",),
|
|
"voulues": ("vouloir",),
|
|
"voulue": ("vouloir",),
|
|
"voulûmes": ("vouloir",),
|
|
"voulurent": ("vouloir",),
|
|
"voulussent": ("vouloir",),
|
|
"voulusses": ("vouloir",),
|
|
"voulusse": ("vouloir",),
|
|
"voulussiez": ("vouloir",),
|
|
"voulussions": ("vouloir",),
|
|
"voulus": ("vouloir",),
|
|
"voulûtes": ("vouloir",),
|
|
"voulut": ("vouloir",),
|
|
"voulût": ("vouloir",),
|
|
"voulu": ("vouloir",),
|
|
"sachant": ("savoir",),
|
|
"sachent": ("savoir",),
|
|
"sache": ("savoir",),
|
|
"saches": ("savoir",),
|
|
"sachez": ("savoir",),
|
|
"sachiez": ("savoir",),
|
|
"sachions": ("savoir",),
|
|
"sachons": ("savoir",),
|
|
"sais": ("savoir",),
|
|
"sait": ("savoir",),
|
|
"sauraient": ("savoir",),
|
|
"saurai": ("savoir",),
|
|
"saurais": ("savoir",),
|
|
"saurait": ("savoir",),
|
|
"saura": ("savoir",),
|
|
"sauras": ("savoir",),
|
|
"saurez": ("savoir",),
|
|
"sauriez": ("savoir",),
|
|
"saurions": ("savoir",),
|
|
"saurons": ("savoir",),
|
|
"sauront": ("savoir",),
|
|
"savaient": ("savoir",),
|
|
"savais": ("savoir",),
|
|
"savait": ("savoir",),
|
|
"savent": ("savoir",),
|
|
"savez": ("savoir",),
|
|
"saviez": ("savoir",),
|
|
"savions": ("savoir",),
|
|
"savons": ("savoir",),
|
|
"sue": ("savoir",),
|
|
"sues": ("savoir",),
|
|
"sûmes": ("savoir",),
|
|
"surent": ("savoir",),
|
|
"su": ("savoir",),
|
|
"sus": ("savoir",),
|
|
"sussent": ("savoir",),
|
|
"susse": ("savoir",),
|
|
"susses": ("savoir",),
|
|
"sussiez": ("savoir",),
|
|
"sussions": ("savoir",),
|
|
"sûtes": ("savoir",),
|
|
"sut": ("savoir",),
|
|
"sût": ("savoir",),
|
|
"venaient": ("venir",),
|
|
"venais": ("venir",),
|
|
"venait": ("venir",),
|
|
"venant": ("venir",),
|
|
"venez": ("venir",),
|
|
"veniez": ("venir",),
|
|
"venions": ("venir",),
|
|
"venons": ("venir",),
|
|
"venues": ("venir",),
|
|
"venue": ("venir",),
|
|
"venus": ("venir",),
|
|
"venu": ("venir",),
|
|
"viendraient": ("venir",),
|
|
"viendrais": ("venir",),
|
|
"viendrait": ("venir",),
|
|
"viendrai": ("venir",),
|
|
"viendras": ("venir",),
|
|
"viendra": ("venir",),
|
|
"viendrez": ("venir",),
|
|
"viendriez": ("venir",),
|
|
"viendrions": ("venir",),
|
|
"viendrons": ("venir",),
|
|
"viendront": ("venir",),
|
|
"viennent": ("venir",),
|
|
"viennes": ("venir",),
|
|
"vienne": ("venir",),
|
|
"viens": ("venir",),
|
|
"vient": ("venir",),
|
|
"vînmes": ("venir",),
|
|
"vinrent": ("venir",),
|
|
"vinssent": ("venir",),
|
|
"vinsses": ("venir",),
|
|
"vinsse": ("venir",),
|
|
"vinssiez": ("venir",),
|
|
"vinssions": ("venir",),
|
|
"vins": ("venir",),
|
|
"vîntes": ("venir",),
|
|
"vint": ("venir",),
|
|
"vînt": ("venir",),
|
|
"aille": ("aller",),
|
|
"aillent": ("aller",),
|
|
"ailles": ("aller",),
|
|
"alla": ("aller",),
|
|
"allai": ("aller",),
|
|
"allaient": ("aller",),
|
|
"allais": ("aller",),
|
|
"allait": ("aller",),
|
|
"allâmes": ("aller",),
|
|
"allant": ("aller",),
|
|
"allas": ("aller",),
|
|
"allasse": ("aller",),
|
|
"allassent": ("aller",),
|
|
"allasses": ("aller",),
|
|
"allassiez": ("aller",),
|
|
"allassions": ("aller",),
|
|
"allât": ("aller",),
|
|
"allâtes": ("aller",),
|
|
"allé": ("aller",),
|
|
"allée": ("aller",),
|
|
"allées": ("aller",),
|
|
"allèrent": ("aller",),
|
|
"allés": ("aller",),
|
|
"allez": ("aller",),
|
|
"allons": ("aller",),
|
|
"ira": ("aller",),
|
|
"irai": ("aller",),
|
|
"iraient": ("aller",),
|
|
"irais": ("aller",),
|
|
"irait": ("aller",),
|
|
"iras": ("aller",),
|
|
"irez": ("aller",),
|
|
"iriez": ("aller",),
|
|
"irions": ("aller",),
|
|
"irons": ("aller",),
|
|
"iront": ("aller",),
|
|
"va": ("aller",),
|
|
"vais": ("aller",),
|
|
"vas": ("aller",),
|
|
"vont": ("aller",)
|
|
}
|