makiolaj/rag-ingestor
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 2 Wiki Activity

refactor: klarere typ-extraktion, fullmatch, root-prefix-test

Browse Source
This commit is contained in:
Jean-Luc Makiola
2026-05-04 22:09:41 +02:00
parent 8d15f02187
commit 0224581587
2 changed files with 12 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
app/ingest/metadata.py
View File

@@ -18,6 +18,10 @@ def parse_path(file_path: str, ingest_root: str) -> PathMetadata | None:
Returns None when the path is outside the ingest root or does not match Returns None when the path is outside the ingest root or does not match
the expected `<root>/<N>.Semester/<Fach>/[<typ>/...]/<file>` pattern. the expected `<root>/<N>.Semester/<Fach>/[<typ>/...]/<file>` pattern.
Caller must pass a file path. Directory paths (with or without trailing
slash) produce undefined results — `PurePosixPath` strips trailing slashes,
so a directory ending in `/foo/` is indistinguishable from a file `/foo`.
""" """
norm_path = file_path.lstrip("/") norm_path = file_path.lstrip("/")
norm_root = ingest_root.strip("/") norm_root = ingest_root.strip("/")
@@ -28,17 +32,16 @@ def parse_path(file_path: str, ingest_root: str) -> PathMetadata | None:
relative = norm_path[len(norm_root) + 1:] relative = norm_path[len(norm_root) + 1:]
parts = PurePosixPath(relative).parts parts = PurePosixPath(relative).parts
# Need at least: semester / fach / file.ext → 3 parts # Layout: [semester, fach, *inner, filename] — minimum 3 parts
if len(parts) < 3: if len(parts) < 3:
return None return None
semester, fach = parts[0], parts[1] semester, fach = parts[0], parts[1]
if not SEMESTER_RE.match(semester): if not SEMESTER_RE.fullmatch(semester):
return None return None
# parts[-1] is the filename. Anything between fach and filename is "deeper". inner = parts[2:-1] # everything between fach and filename
# The first deeper segment becomes `typ`. None if file lives directly in fach. typ = inner[0] if inner else None
typ = parts[2] if len(parts) > 3 else None
return PathMetadata(semester=semester, fach=fach, typ=typ) return PathMetadata(semester=semester, fach=fach, typ=typ)

4
tests/test_metadata.py
View File

@@ -44,3 +44,7 @@ def test_parse_path_no_fach_returns_none():
def test_parse_path_with_leading_slash_normalizes(): def test_parse_path_with_leading_slash_normalizes():
md = parse_path("/Documents/THB/Studium/2.Semester/Databases/x.pdf", ROOT) md = parse_path("/Documents/THB/Studium/2.Semester/Databases/x.pdf", ROOT)
assert md == PathMetadata(semester="2.Semester", fach="Databases", typ=None) assert md == PathMetadata(semester="2.Semester", fach="Databases", typ=None)
def test_parse_path_root_prefix_collision_returns_none():
assert parse_path("Documents/THB/StudiumExam/2.Semester/Foo/x.pdf", ROOT) is None