Merge pull request #5488 from explosion/feature/better-model-compat

Better model compatibility and validation

requirements.txt

@@ -13,6 +13,7 @@ numpy>=1.15.0
 requests>=2.13.0,<3.0.0
 plac>=0.9.6,<1.2.0
 tqdm>=4.38.0,<5.0.0
+importlib_metadata>=0.20; python_version < "3.8"
 # Optional dependencies
 jsonschema>=2.6.0,<3.1.0
 pydantic>=1.3.0,<2.0.0

setup.cfg

@@ -56,6 +56,7 @@ install_requires =
     requests>=2.13.0,<3.0.0
     pydantic>=1.3.0,<2.0.0
     tqdm>=4.38.0,<5.0.0
+    importlib_metadata>=0.20; python_version < "3.8"
 
 [options.extras_require]
 lookups =

spacy/cli/info.py

@@ -48,7 +48,9 @@ def info(
         "Location": str(Path(__file__).parent.parent),
         "Platform": platform.platform(),
         "Python version": platform.python_version(),
-        "Models": ", ".join(model["name"] for model in all_models.values()),
+        "Models": ", ".join(
+            f"{m['name']} ({m['version']})" for m in all_models.values()
+        ),
     }
     if not silent:
         title = "Info about spaCy"

spacy/cli/package.py

@@ -83,14 +83,14 @@ def generate_meta(model_path, existing_meta, msg):
         ("lang", "Model language", meta.get("lang", "en")),
         ("name", "Model name", meta.get("name", "model")),
         ("version", "Model version", meta.get("version", "0.0.0")),
-        ("spacy_version", "Required spaCy version", f">={about.__version__},<3.0.0"),
         ("description", "Model description", meta.get("description", False)),
         ("author", "Author", meta.get("author", False)),
         ("email", "Author email", meta.get("email", False)),
         ("url", "Author website", meta.get("url", False)),
-        ("license", "License", meta.get("license", "CC BY-SA 3.0")),
+        ("license", "License", meta.get("license", "MIT")),
     ]
     nlp = util.load_model_from_path(Path(model_path))
+    meta["spacy_version"] = about.__version__
     meta["pipeline"] = nlp.pipe_names
     meta["vectors"] = {
         "width": nlp.vocab.vectors_length,
@@ -168,6 +168,7 @@ def setup_package():
         package_data={model_name: list_files(model_dir)},
         install_requires=list_requirements(meta),
         zip_safe=False,
+        entry_points={'spacy_models': ['{m} = {m}'.format(m=model_name)]}
     )
 
 

spacy/cli/train.py

@@ -467,7 +467,7 @@ def train(
                     # Update model meta.json
                     meta["lang"] = nlp.lang
                     meta["pipeline"] = nlp.pipe_names
-                    meta["spacy_version"] = f">={about.__version__}"
+                    meta["spacy_version"] = about.__version__
                     if beam_width == 1:
                         meta["speed"] = {
                             "nwords": nwords,

spacy/cli/validate.py

@@ -4,6 +4,8 @@ import requests
 from wasabi import msg
 
 from .. import about
+from ..util import get_package_version, get_installed_models, split_version
+from ..util import get_package_path, get_model_meta, is_compatible_model
 
 
 def validate():
@@ -25,7 +27,7 @@ def validate():
     msg.info(f"spaCy installation: {spacy_dir}")
 
     if model_pkgs:
-        header = ("NAME", "VERSION", "")
+        header = ("NAME", "SPACY", "VERSION", "")
         rows = []
         for name, data in model_pkgs.items():
             if data["compat"]:
@@ -34,7 +36,7 @@ def validate():
             else:
                 version = msg.text(data["version"], color="red", no_print=True)
                 comp = f"--> {compat.get(data['name'], ['n/a'])[0]}"
-            rows.append((data["name"], version, comp))
+            rows.append((data["name"], data["spacy"], version, comp))
         msg.table(rows, header=header)
     else:
         msg.text("No models found in your current environment.", exits=0)
@@ -44,8 +46,9 @@ def validate():
         cmd = "python -m spacy download {}"
         print("\n".join([cmd.format(pkg) for pkg in update_models]) + "\n")
     if na_models:
-        msg.warn(
+        msg.info(
-            f"The following models are not available for spaCy v{about.__version__}:",
+            f"The following models are custom spaCy models or not "
+            f"available for spaCy v{about.__version__}:",
             ", ".join(na_models),
         )
     if incompat_models:
@@ -53,8 +56,6 @@ def validate():
 
 
 def get_model_pkgs():
-    import pkg_resources
-
     with msg.loading("Loading compatibility table..."):
         r = requests.get(about.__compatibility__)
         if r.status_code != 200:
@@ -66,20 +67,30 @@ def get_model_pkgs():
     msg.good("Loaded compatibility table")
     compat = r.json()["spacy"]
     all_models = set()
+    installed_models = get_installed_models()
     for spacy_v, models in dict(compat).items():
         all_models.update(models.keys())
         for model, model_vs in models.items():
             compat[spacy_v][model] = [reformat_version(v) for v in model_vs]
     pkgs = {}
-    for pkg_name, pkg_data in pkg_resources.working_set.by_key.items():
+    for pkg_name in installed_models:
         package = pkg_name.replace("-", "_")
-        if package in all_models:
+        version = get_package_version(pkg_name)
-            version = pkg_data.version
+        if package in compat:
-            pkgs[pkg_name] = {
+            is_compat = version in compat[package]
-                "name": package,
+            v_maj, v_min = split_version(about.__version__)
-                "version": version,
+            spacy_version = f"{v_maj}.{v_min}"
-                "compat": package in compat and version in compat[package],
+        else:
-            }
+            model_path = get_package_path(package)
+            model_meta = get_model_meta(model_path)
+            is_compat = is_compatible_model(model_meta)
+            spacy_version = model_meta.get("spacy_version", "n/a")
+        pkgs[pkg_name] = {
+            "name": package,
+            "version": version,
+            "spacy": spacy_version,
+            "compat": is_compat,
+        }
     return pkgs, compat
 
 

spacy/language.py

@@ -196,7 +196,7 @@ class Language(object):
             self._meta.setdefault("lang", self.lang)
         self._meta.setdefault("name", "model")
         self._meta.setdefault("version", "0.0.0")
-        self._meta.setdefault("spacy_version", f">={about.__version__}")
+        self._meta.setdefault("spacy_version", about.__version__)
         self._meta.setdefault("description", "")
         self._meta.setdefault("author", "")
         self._meta.setdefault("email", "")

spacy/util.py

@@ -21,9 +21,16 @@ try:
 except ImportError:
     cupy = None
 
+try:  # Python 3.8
+    import importlib.metadata as importlib_metadata
+except ImportError:
+    import importlib_metadata
+
 from .symbols import ORTH
 from .compat import cupy, CudaStream
 from .errors import Errors, Warnings
+from . import about
+
 
 _PRINT_ENV = False
 
@@ -35,6 +42,10 @@ class registry(thinc.registry):
     factories = catalogue.create("spacy", "factories", entry_points=True)
     displacy_colors = catalogue.create("spacy", "displacy_colors", entry_points=True)
     assets = catalogue.create("spacy", "assets", entry_points=True)
+    # This is mostly used to get a list of all installed models in the current
+    # environment. spaCy models packaged with `spacy package` will "advertise"
+    # themselves via entry points.
+    models = catalogue.create("spacy", "models", entry_points=True)
 
 
 def set_env_log(value):
@@ -204,6 +215,56 @@ def load_model_from_init_py(init_file, **overrides):
     return load_model_from_path(data_path, meta, **overrides)
 
 
+def get_installed_models():
+    """List all model packages currently installed in the environment.
+
+    RETURNS (list): The string names of the models.
+    """
+    return list(registry.models.get_all().keys())
+
+
+def get_package_version(name):
+    """Get the version of an installed package. Typically used to get model
+    package versions.
+
+    name (unicode): The name of the installed Python package.
+    RETURNS (unicode / None): The version or None if package not installed.
+    """
+    try:
+        return importlib_metadata.version(name)
+    except importlib_metadata.PackageNotFoundError:
+        return None
+
+
+def split_version(version):
+    """RETURNS (tuple): Two integers, the major and minor spaCy version."""
+    pieces = version.split(".", 3)
+    return int(pieces[0]), int(pieces[1])
+
+
+def is_compatible_model(meta):
+    """Check if a model is compatible with the current version of spaCy, based
+    on its meta.json. We compare the version of spaCy the model was created with
+    with the current version. If the minor version is different, it's considered
+    incompatible.
+
+    meta (dict): The model's meta.
+    RETURNS (bool / None): Whether the model is compatible with the current
+        spaCy or None if we don't have enough info.
+    """
+    cur_v = about.__version__
+    pkg_v = meta.get("spacy_version")
+    if not pkg_v or not isinstance(pkg_v, str):
+        return None
+    # Handle spacy_version values like >=x,<y, just in case
+    pkg_v = re.sub(r"[^0-9.]", "", pkg_v.split(",")[0])
+    cur_major, cur_minor = split_version(cur_v)
+    pkg_major, pkg_minor = split_version(pkg_v)
+    if cur_major != pkg_major or cur_minor != pkg_minor:
+        return False
+    return True
+
+
 def load_config(path, create_objects=False):
     """Load a Thinc-formatted config file, optionally filling in objects where
     the config references registry entries. See "Thinc config files" for details.