paradizelost/paperless-ngx
1
0
Fork 0
mirror of https://github.com/paperless-ngx/paperless-ngx.git synced 2025-04-25 10:49:30 -05:00
Code Issues Packages Projects Releases Wiki Activity

Adds multiprocessing to the fuzzy matching for a speedup

Browse Source
This commit is contained in:
Trenton H 2023-09-11 08:48:30 -07:00
parent ce8bf90663
commit a03a745295
3 changed files with 83 additions and 34 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
docs/administration.md
View File

@ -586,6 +586,7 @@ those which look close according to a given ratio.
document_fuzzy_match [--ratio]
```
Optional arguments:
--ratio - a number between 0 and 100, setting how similar a document must be for it to be reported.
Higher numbers mean more similarity.
| Option | Required | Default | Description |
| ----------- | -------- | ------- | ------------------------------------------------------------------------------------------------------------------------------ |
| --ratio | No | 85.0 | a number between 0 and 100, setting how similar a document must be for it to be reported. Higher numbers mean more similarity. |
| --processes | No | 4 | Number of processes to use for matching. Setting 1 disables multiple processes |

107
src/documents/management/commands/document_fuzzy_match.py
View File

@ -1,3 +1,5 @@
import dataclasses
import multiprocessing
from typing import Final
import rapidfuzz
@ -8,8 +10,35 @@ from django.core.management import CommandError
from documents.models import Document
@dataclasses.dataclass(frozen=True)
class _WorkPackage:
first_doc: Document
second_doc: Document
@dataclasses.dataclass(frozen=True)
class _WorkResult:
doc_one_pk: int
doc_two_pk: int
ratio: float
def __lt__(self, other: "_WorkResult") -> bool:
return self.doc_one_pk < other.doc_one_pk
def _process_and_match(work: _WorkPackage) -> _WorkResult:
# Normalize the string some, lower case, whitespace, etc
first_string = rapidfuzz.utils.default_process(work.first_doc.content)
second_string = rapidfuzz.utils.default_process(work.second_doc.content)
# Basic matching ratio
match = rapidfuzz.fuzz.ratio(first_string, second_string)
return _WorkResult(work.first_doc.pk, work.second_doc.pk, match)
class Command(BaseCommand):
help = "Manages the document index."
help = "Searches for documents where the content almost matches"
def add_arguments(self, parser):
parser.add_argument(
@ -18,6 +47,12 @@ class Command(BaseCommand):
type=float,
help="Ratio to consider documents a match",
)
parser.add_argument(
"--processes",
default=4,
type=int,
help="Number of processes to distribute work amongst",
)
parser.add_argument(
"--no-progress-bar",
default=False,
@ -30,8 +65,8 @@ class Command(BaseCommand):
RATIO_MAX: Final[float] = 100.0
opt_ratio = options["ratio"]
progress_bar_disable = options["no_progress_bar"]
match_pairs = set()
checked_pairs: set[tuple[int, int]] = set()
work_pkgs: list[_WorkPackage] = []
# Ratio is a float from 0.0 to 100.0
if opt_ratio < RATIO_MIN or opt_ratio > RATIO_MAX:
@ -39,42 +74,54 @@ class Command(BaseCommand):
all_docs = Document.objects.all().order_by("id")
messages = []
for first_doc in tqdm.tqdm(all_docs, disable=progress_bar_disable):
# Build work packages for processing
for first_doc in all_docs:
for second_doc in all_docs:
# doc to doc is obviously not useful
if first_doc.pk == second_doc.pk:
continue
# Skip matching which have already been matched together
# doc 1 to doc 2 is the same as doc 2 to doc 1
if (first_doc.pk, second_doc.pk) in checked_pairs or (
second_doc.pk,
first_doc.pk,
) in checked_pairs:
continue
checked_pairs.update(
[(first_doc.pk, second_doc.pk), (second_doc.pk, first_doc.pk)],
)
# Normalize the string some, lower case, whitespace, etc
first_string = rapidfuzz.utils.default_process(first_doc.content)
second_string = rapidfuzz.utils.default_process(second_doc.content)
work_pkgs.append(_WorkPackage(first_doc, second_doc))
# Basic matching ratio
match = rapidfuzz.fuzz.ratio(first_string, second_string)
# Don't spin up a pool of 1 process
if options["processes"] == 1:
results = []
for work in tqdm.tqdm(work_pkgs, disable=options["no_progress_bar"]):
results.append(_process_and_match(work))
else:
with multiprocessing.Pool(processes=options["processes"]) as pool:
results = list(
tqdm.tqdm(
pool.imap_unordered(_process_and_match, work_pkgs),
total=len(work_pkgs),
disable=options["no_progress_bar"],
),
)
if match >= opt_ratio:
# Skip matching which have already been matched together
# doc 1 to doc 2 is the same as doc 2 to doc 1
if (first_doc.pk, second_doc.pk) in match_pairs or (
second_doc.pk,
first_doc.pk,
) in match_pairs:
continue
else:
match_pairs.add((first_doc.pk, second_doc.pk))
match_pairs.add((second_doc.pk, first_doc.pk))
messages.append(
self.style.NOTICE(
f"Document {first_doc.pk} fuzzy match"
f" to {second_doc.pk} (confidence {match:.3f})",
),
)
# Check results
messages = []
for result in sorted(results):
if result.ratio >= opt_ratio:
messages.append(
self.style.NOTICE(
f"Document {result.doc_one_pk} fuzzy match"
f" to {result.doc_two_pk} (confidence {result.ratio:.3f})",
),
)
if len(messages) == 0:
messages.append(
self.style.NOTICE("No matches found"),
self.style.SUCCESS("No matches found"),
)
self.stdout.writelines(
messages,

3
src/documents/tests/test_management_fuzzy.py
View File

@ -13,6 +13,7 @@ class TestFuzzyMatchCommand(TestCase):
stderr = StringIO()
call_command(
"document_fuzzy_match",
"--no-progress-bar",
*args,
stdout=stdout,
stderr=stderr,
@ -63,7 +64,7 @@ class TestFuzzyMatchCommand(TestCase):
mime_type="application/pdf",
filename="other_test.pdf",
)
stdout, _ = self.call_command()
stdout, _ = self.call_command("--processes", "1")
self.assertEqual(stdout, "Document 1 fuzzy match to 2 (confidence 86.667)\n")
def test_with_3_matches(self):