mirror of
https://github.com/bellingcat/auto-archiver.git
synced 2026-06-08 03:18:28 +03:00
Merge branch 'dev' into fix/improve-deleted-post-detection
This commit is contained in:
1238
poetry.lock
generated
1238
poetry.lock
generated
File diff suppressed because it is too large
Load Diff
1
src/auto_archiver/modules/antibot_extractor_enricher/captcha_services/.gitignore
vendored
Normal file
1
src/auto_archiver/modules/antibot_extractor_enricher/captcha_services/.gitignore
vendored
Normal file
@@ -0,0 +1 @@
|
||||
*.py
|
||||
@@ -3,6 +3,13 @@
|
||||
"type": ["enricher"],
|
||||
"requires_setup": True,
|
||||
"dependencies": {"python": ["loguru"], "bin": ["exiftool"]},
|
||||
"configs": {
|
||||
"look_for_keys": {
|
||||
"default": [],
|
||||
"help": "list of lowercased metadata keys that will be included in the enriched metadata. Special keys: 'author', 'datetimes', 'location' to include related metadata fields. The default empty list `[]` means all metadata will be included.",
|
||||
"type": "list",
|
||||
},
|
||||
},
|
||||
"description": """
|
||||
Extracts metadata information from files using ExifTool.
|
||||
|
||||
|
||||
@@ -16,6 +16,8 @@ class MetadataEnricher(Enricher):
|
||||
|
||||
for i, m in enumerate(to_enrich.media):
|
||||
if len(md := self.get_metadata(m.filename)):
|
||||
if self.look_for_keys != []:
|
||||
md = self.select_metadata(md, self.look_for_keys)
|
||||
to_enrich.media[i].set("metadata", md)
|
||||
|
||||
def get_metadata(self, filename: str) -> dict:
|
||||
@@ -23,7 +25,6 @@ class MetadataEnricher(Enricher):
|
||||
# Run ExifTool command to extract metadata from the file
|
||||
cmd = ["exiftool", filename]
|
||||
result = subprocess.run(cmd, capture_output=True, text=True)
|
||||
|
||||
# Process the output to extract individual metadata fields
|
||||
metadata = {}
|
||||
for line in result.stdout.splitlines():
|
||||
@@ -35,3 +36,33 @@ class MetadataEnricher(Enricher):
|
||||
except Exception as e:
|
||||
logger.error(f"Error occurred: {e}: {traceback.format_exc()}")
|
||||
return {}
|
||||
|
||||
def select_metadata(self, all_md, requested_metadata_keys):
|
||||
"""
|
||||
coordinates the selection of metadata from the general exiftool output to the user-specified grocery list
|
||||
"""
|
||||
# defining the batches of metadata that get pulled for special terms
|
||||
author_key_terms = ["author", "producer", "creator"]
|
||||
datetime_key_terms = ["date", "time"]
|
||||
location_key_terms = ["gps", "latitude", "longitude"]
|
||||
|
||||
specified_md = {}
|
||||
for md_key in all_md.keys():
|
||||
md_key_lower = md_key.lower()
|
||||
# checking for special baskets within the grocery list of requested metadata
|
||||
if ("author" in requested_metadata_keys) and any(
|
||||
term in md_key_lower and len(all_md[md_key]) for term in author_key_terms
|
||||
):
|
||||
specified_md[md_key] = all_md[md_key]
|
||||
if ("datetime" in requested_metadata_keys) and any(
|
||||
term in md_key_lower and len(all_md[md_key]) for term in datetime_key_terms
|
||||
):
|
||||
specified_md[md_key] = all_md[md_key]
|
||||
if ("location" in requested_metadata_keys) and any(
|
||||
term in md_key_lower and len(all_md[md_key]) for term in location_key_terms
|
||||
):
|
||||
specified_md[md_key] = all_md[md_key]
|
||||
# if the metadata value is requested directly
|
||||
if md_key_lower in requested_metadata_keys or md_key in requested_metadata_keys and len(all_md[md_key]):
|
||||
specified_md[md_key] = all_md[md_key]
|
||||
return specified_md
|
||||
|
||||
@@ -56,6 +56,19 @@ def test_enrich_sets_metadata(enricher, mocker):
|
||||
assert metadata.media == [media1, media2]
|
||||
|
||||
|
||||
def test_enrich_no_metadata_selection(enricher, mocker):
|
||||
media1 = mocker.Mock(filename="img1.jpg")
|
||||
media2 = mocker.Mock(filename="img2.jpg")
|
||||
metadata = mocker.Mock()
|
||||
metadata.media = [media1, media2]
|
||||
enricher.get_metadata = lambda f: {"key": "value"} if f == "img1.jpg" else {}
|
||||
enricher.look_for_keys = ["no-key"]
|
||||
enricher.enrich(metadata)
|
||||
media1.set.assert_called_once_with("metadata", {})
|
||||
media2.set.assert_not_called()
|
||||
assert metadata.media == [media1, media2]
|
||||
|
||||
|
||||
def test_enrich_empty_media(enricher, mocker):
|
||||
metadata = mocker.Mock()
|
||||
metadata.media = []
|
||||
@@ -71,7 +84,9 @@ def test_get_metadata_error_handling(enricher, mocker):
|
||||
assert "Error occurred: " in mock_log.call_args[0][0]
|
||||
|
||||
|
||||
def test_metadata_pickle(enricher, unpickle, mocker):
|
||||
# TODO depends on the expected functionality
|
||||
"""
|
||||
def test_default_metadata_pickle(enricher, unpickle, mocker):
|
||||
mock_run = mocker.patch("subprocess.run")
|
||||
# Uses pickled values
|
||||
mock_run.return_value = unpickle("metadata_enricher_exif.pickle")
|
||||
@@ -79,6 +94,39 @@ def test_metadata_pickle(enricher, unpickle, mocker):
|
||||
expected = unpickle("metadata_enricher_ytshort_expected.pickle")
|
||||
enricher.enrich(metadata)
|
||||
expected_media = expected.media
|
||||
print(expected_media)
|
||||
actual_media = metadata.media
|
||||
|
||||
assert len(expected_media) == len(actual_media)
|
||||
assert actual_media[0].properties.get("metadata") == expected_media[0].properties.get("metadata")
|
||||
"""
|
||||
|
||||
|
||||
def test_metadata_pickle_megapixel(enricher, unpickle, mocker):
|
||||
mock_run = mocker.patch("subprocess.run")
|
||||
mock_run.return_value = unpickle("metadata_enricher_exif.pickle")
|
||||
metadata = unpickle("metadata_enricher_ytshort_input.pickle")
|
||||
|
||||
enricher.look_for_keys = ["megapixels"]
|
||||
enricher.enrich(metadata)
|
||||
actual_media = metadata.media
|
||||
|
||||
assert actual_media[0].properties.get("metadata") == {"Megapixels": "0.922"}
|
||||
|
||||
|
||||
def test_metadata_specify_datetime_and_metapixels(enricher, unpickle, mocker):
|
||||
mock_run = mocker.patch("subprocess.run")
|
||||
mock_run.return_value = unpickle("metadata_enricher_exif.pickle")
|
||||
metadata = unpickle("metadata_enricher_ytshort_input.pickle")
|
||||
|
||||
enricher.look_for_keys = ["datetime", "megapixels", "image height"]
|
||||
enricher.enrich(metadata)
|
||||
actual_media = metadata.media
|
||||
|
||||
assert actual_media[0].properties.get("metadata") == {
|
||||
"File Modification Date/Time": "2025:02:18 19:42:50+00:00",
|
||||
"File Access Date/Time": "2025:02:18 19:42:50+00:00",
|
||||
"File Inode Change Date/Time": "2025:02:18 19:42:50+00:00",
|
||||
"Megapixels": "0.922",
|
||||
"Image Height": "720",
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user