feat: pfad-metadata-parser mit semester/fach/typ

app/ingest/metadata.py

@@ -0,0 +1,44 @@
+import re
+from dataclasses import dataclass
+from pathlib import PurePosixPath
+
+
+SEMESTER_RE = re.compile(r"^\d+\.Semester$")
+
+
+@dataclass(frozen=True)
+class PathMetadata:
+    semester: str
+    fach: str
+    typ: str | None
+
+
+def parse_path(file_path: str, ingest_root: str) -> PathMetadata | None:
+    """Parse a Nextcloud file path into structured metadata.
+
+    Returns None when the path is outside the ingest root or does not match
+    the expected `<root>/<N>.Semester/<Fach>/[<typ>/...]/<file>` pattern.
+    """
+    norm_path = file_path.lstrip("/")
+    norm_root = ingest_root.strip("/")
+
+    if not norm_path.startswith(norm_root + "/"):
+        return None
+
+    relative = norm_path[len(norm_root) + 1:]
+    parts = PurePosixPath(relative).parts
+
+    # Need at least: semester / fach / file.ext  → 3 parts
+    if len(parts) < 3:
+        return None
+
+    semester, fach = parts[0], parts[1]
+
+    if not SEMESTER_RE.match(semester):
+        return None
+
+    # parts[-1] is the filename. Anything between fach and filename is "deeper".
+    # The first deeper segment becomes `typ`. None if file lives directly in fach.
+    typ = parts[2] if len(parts) > 3 else None
+
+    return PathMetadata(semester=semester, fach=fach, typ=typ)