Merge pull request #60 from martindurant/hf

Add minimal hf (read license)

Author: martindurant

docs/source/api.rst

@@ -42,6 +42,7 @@ User Classes
     proj.documentation.MDBook
     proj.documentation.RTD
     proj.git.GitRepo
+    proj.hf.HuggingFaceRepo
     proj.ide.JetbrainsIDE
     proj.ide.NvidiaAIWorkbench
     proj.ide.VSCode
@@ -77,6 +78,7 @@ User Classes
 .. autoclass:: projspec.proj.documentation.MDBook
 .. autoclass:: projspec.proj.documentation.RTD
 .. autoclass:: projspec.proj.git.GitRepo
+.. autoclass:: projspec.proj.hf.HuggingFaceRepo
 .. autoclass:: projspec.proj.ide.JetbrainsIDE
 .. autoclass:: projspec.proj.ide.NvidiaAIWorkbench
 .. autoclass:: projspec.proj.ide.VSCode

pyproject.toml

@@ -28,7 +28,7 @@ dependencies = [
     "toml",
     "fsspec",
     "click",
-    "jinja2"
+    "jinja2",
 ]
 
 [project.optional-dependencies]

src/projspec/proj/__init__.py

@@ -6,6 +6,7 @@
 from projspec.proj.documentation import RTD, MDBook
 from projspec.proj.git import GitRepo
 from projspec.proj.golang import Golang
+from projspec.proj.hf import HuggingFaceRepo
 from projspec.proj.ide import JetbrainsIDE, NvidiaAIWorkbench, VSCode
 from projspec.proj.node import JLabExtension, Node, Yarn
 from projspec.proj.pixi import Pixi
@@ -25,6 +26,7 @@
     "CondaProject",
     "Golang",
     "GitRepo",
+    "HuggingFaceRepo",
     "JetbrainsIDE",
     "JLabExtension",
     "Marimo",

src/projspec/proj/hf.py

@@ -18,12 +18,47 @@ def match(self) -> bool:
         return self.proj.fs.exists(readme)
 
     def parse(self) -> None:
-        # for now, we just stash the metadata declaration
+        from projspec.content.metadata import DescriptiveMetadata, License
         import yaml
 
         readme = f"{self.proj.url}/README.md"
 
-        with self.proj.fs.open(readme) as f:
+        with self.proj.fs.open(readme, "rt") as f:
             txt = f.read()
         meta = txt.split("---\n")[1]
-        self.meta = yaml.safe_load(StringIO(meta))
+        meta = yaml.safe_load(StringIO(meta))
+        if "licence" in meta:
+            self.contents["license"] = License(
+                proj=self.proj,
+                shortname=meta["licence"],
+                fullname=meta.get("license_name"),
+                url=meta.get("license_link"),
+                artifacts=set(),
+            )
+        self.contents["desciptive_metadata"] = DescriptiveMetadata(
+            proj=self.proj,
+            meta={
+                k: meta[k] for k in ["language", "library_name", "tags"] if k in meta
+            },
+            artifacts=set(),
+        )
+
+    @staticmethod
+    def _create(path: str) -> None:
+        with open(f"{path}/README.md", "w") as f:
+            f.write(
+                """---
+license: other
+license_name: coqui-public-model-license
+license_link: https://coqui.ai/cpml
+library_name: flair
+tags:
+- flair
+base_model: HuggingFaceH4/zephyr-7b-beta
+new_version: 0.1
+datasets:
+- stanfordnlp/imdb
+- HuggingFaceFW/fineweb
+---
+"""
+            )

tests/test_roundtrips.py

@@ -21,6 +21,7 @@
         "dash",
         "panel",
         "Golang",
+        "HuggingFaceRepo",
     ],
 )
 def test_compliant(tmpdir, cls_name):