Snipets tooling update

- update README
- generate-md.py: ignore trailing newlines when complaining about diffs
- all langs: add more dependencies to templates
- all langs: support @hide-start and @hide-end for multi-line hiding
- go: also allow one-line imports
- java: allow declarations other than `main` method
- python, typescript: implement `unshorten` for implicit imports and
  client initialization
- python: add mypy config as `datasets` lacks type py.typed
- python: hide `# mypy:` annotations
- typescript: add tsconfig that allows top-level `await`s
- typescript: depend on @types/node to fix `import ... from "crypto"`

automation/snippets/lib/languages/python.py

@@ -1,7 +1,19 @@
+import re
 import subprocess
 from pathlib import Path
 
-from .base import CompileResult, Language, copy_template, trim_commonpath
+from .base import (
+    CompileResult,
+    Language,
+    copy_template,
+    generic_shorten,
+    trim_commonpath,
+)
+
+RE_IMPORTS = re.compile(
+    r"^from qdrant_client import (.*)$",
+    re.MULTILINE,
+)
 
 
 class LanguagePython(Language):
@@ -28,7 +40,57 @@ class LanguagePython(Language):
                 target_path,
             ]
 
-        p = subprocess.run(["uv", "--project=templates/python", "run", "mypy", tmpdir])
+        p = subprocess.run(
+            [
+                "uv",
+                "--project=templates/python",
+                "run",
+                "mypy",
+                "--config-file=templates/python/pyproject.toml",
+                tmpdir,
+            ]
+        )
         result.has_issues = p.returncode != 0
 
         return result
+
+    @classmethod
+    def shorten(cls, contents: str) -> str:
+        lines = [
+            line
+            for line in contents.splitlines(keepends=True)
+            if not line.lstrip().startswith("# mypy:")
+        ]
+        return generic_shorten("".join(lines))
+
+    @classmethod
+    def unshorten(cls, contents: str) -> str:
+        if "client." in contents and "QdrantClient" not in contents:
+            contents = (
+                'client = QdrantClient(url="http://localhost:6333")  # @hide\n\n'
+                + contents.lstrip()
+            )
+
+        has_imports = set()
+        if m := RE_IMPORTS.search(contents):
+            has_imports = set(i.strip() for i in m[1].split(","))
+
+        need_imports = set()
+        if "QdrantClient" in contents:
+            need_imports.add("QdrantClient")
+        if "models." in contents:
+            need_imports.add("models")
+
+        if need_imports - has_imports:
+            if m is not None:
+                contents = RE_IMPORTS.sub(
+                    f"from qdrant_client import {', '.join(sorted(need_imports | has_imports))}",
+                    contents,
+                )
+            else:
+                contents = (
+                    f"from qdrant_client import {', '.join(sorted(need_imports))}  # @hide\n\n"
+                    + contents.lstrip()
+                )
+
+        return contents