mirror of
https://github.com/paperless-ngx/paperless-ngx.git
synced 2026-11-16 01:22:21 +00:00
Compare commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
c91a09fe77 | ||
|
|
8f029f30a3 |
No files matched your search
@@ -3,7 +3,6 @@ from __future__ import annotations
|
||||
import logging
|
||||
import re
|
||||
from fnmatch import fnmatch
|
||||
from fnmatch import translate as fnmatch_translate
|
||||
from typing import TYPE_CHECKING
|
||||
|
||||
from rest_framework import serializers
|
||||
@@ -548,6 +547,48 @@ def existing_document_matches_workflow(
|
||||
return (True, None)
|
||||
|
||||
|
||||
def _glob_to_portable_regex(pattern: str) -> str:
|
||||
"""
|
||||
Translate an fnmatch-style glob into an anchored regex which is accepted by
|
||||
Python, PostgreSQL (ARE), and MariaDB (PCRE).
|
||||
|
||||
fnmatch.translate() can't be used, as its output (`(?s:...)\\Z`) uses syntax
|
||||
PostgreSQL rejects. Supports `*`, `?`, `[seq]` and `[!seq]`.
|
||||
"""
|
||||
out: list[str] = []
|
||||
i = 0
|
||||
n = len(pattern)
|
||||
while i < n:
|
||||
c = pattern[i]
|
||||
i += 1
|
||||
if c == "*":
|
||||
out.append(".*")
|
||||
elif c == "?":
|
||||
out.append(".")
|
||||
elif c == "[":
|
||||
j = i
|
||||
if j < n and pattern[j] == "!":
|
||||
j += 1
|
||||
if j < n and pattern[j] == "]":
|
||||
j += 1
|
||||
while j < n and pattern[j] != "]":
|
||||
j += 1
|
||||
if j >= n:
|
||||
# No closing bracket, treat as a literal
|
||||
out.append(r"\[")
|
||||
else:
|
||||
body = pattern[i:j]
|
||||
i = j + 1
|
||||
negate = body.startswith("!")
|
||||
if negate:
|
||||
body = body[1:]
|
||||
body = "".join(f"\\{ch}" if ch in "\\[^" else ch for ch in body)
|
||||
out.append(f"[{'^' if negate else ''}{body}]")
|
||||
else:
|
||||
out.append(re.escape(c))
|
||||
return f"^{''.join(out)}$"
|
||||
|
||||
|
||||
def prefilter_documents_by_workflowtrigger(
|
||||
documents: QuerySet[Document],
|
||||
trigger: WorkflowTrigger,
|
||||
@@ -628,8 +669,9 @@ def prefilter_documents_by_workflowtrigger(
|
||||
documents = documents.annotate(**annotations).filter(custom_field_q)
|
||||
|
||||
if trigger.filter_filename:
|
||||
regex = fnmatch_translate(trigger.filter_filename).lstrip("^").rstrip("$")
|
||||
documents = documents.filter(original_filename__iregex=regex)
|
||||
documents = documents.filter(
|
||||
original_filename__iregex=_glob_to_portable_regex(trigger.filter_filename),
|
||||
)
|
||||
|
||||
return documents
|
||||
|
||||
|
||||
@@ -0,0 +1,62 @@
|
||||
import fnmatch
|
||||
import re
|
||||
|
||||
import pytest
|
||||
|
||||
from documents.matching import _glob_to_portable_regex
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
("pattern", "expected"),
|
||||
[
|
||||
pytest.param("", r"^$", id="empty"),
|
||||
pytest.param("abc", r"^abc$", id="literal"),
|
||||
pytest.param("*", r"^.*$", id="star"),
|
||||
pytest.param("?", r"^.$", id="question"),
|
||||
pytest.param("*.pdf", r"^.*\.pdf$", id="star-escapes-dot"),
|
||||
pytest.param("[abc]", r"^[abc]$", id="class"),
|
||||
pytest.param("[a-z]", r"^[a-z]$", id="class-range"),
|
||||
pytest.param("[!abc]", r"^[^abc]$", id="negated-class"),
|
||||
pytest.param("[]a]", r"^[]a]$", id="leading-close-bracket-is-member"),
|
||||
pytest.param("[!]a]", r"^[^]a]$", id="negated-leading-close-bracket"),
|
||||
pytest.param("[a^b]", r"^[a\^b]$", id="caret-escaped-in-class"),
|
||||
pytest.param("[a[b]", r"^[a\[b]$", id="open-bracket-escaped-in-class"),
|
||||
pytest.param("[a\\b]", r"^[a\\b]$", id="backslash-escaped-in-class"),
|
||||
pytest.param("[abc", r"^\[abc$", id="unclosed-class-literal"),
|
||||
pytest.param("[", r"^\[$", id="lone-open-bracket"),
|
||||
pytest.param("[!", r"^\[!$", id="unclosed-negation-literal"),
|
||||
pytest.param("[]", r"^\[\]$", id="unclosed-with-only-close-bracket"),
|
||||
pytest.param("a[0-9]?*", r"^a[0-9]..*$", id="mixed"),
|
||||
],
|
||||
)
|
||||
def test_translation(pattern: str, expected: str) -> None:
|
||||
"""
|
||||
GIVEN:
|
||||
- A glob pattern
|
||||
WHEN:
|
||||
- It is translated to a portable regex
|
||||
THEN:
|
||||
- The exact anchored regex is produced
|
||||
"""
|
||||
assert _glob_to_portable_regex(pattern) == expected
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"pattern",
|
||||
["*", "?", "*.pdf", "a?c", "[abc]x", "[!abc]x", "[]a]", "[!]a]", "[a-c]*", "[a"],
|
||||
)
|
||||
@pytest.mark.parametrize(
|
||||
"name",
|
||||
["", "a", "abc", "x", "]", "]x", "file.pdf", "ax", "bx", "dx", "[a"],
|
||||
)
|
||||
def test_agrees_with_fnmatch(pattern: str, name: str) -> None:
|
||||
"""
|
||||
GIVEN:
|
||||
- A glob pattern and a name
|
||||
WHEN:
|
||||
- Matched via the portable regex and via stdlib fnmatch
|
||||
THEN:
|
||||
- Both agree
|
||||
"""
|
||||
regex = _glob_to_portable_regex(pattern)
|
||||
assert bool(re.search(regex, name)) == fnmatch.fnmatchcase(name, pattern)
|
||||
Reference in new issue
Block a user