bellingcat/auto-archiver
1
0
Fork 0
mirror of https://github.com/bellingcat/auto-archiver.git synced 2026-06-08 03:18:28 +03:00
Code Issues Packages Projects Releases Wiki Activity

refactorws PR to obey standard code approach

Browse Source
This commit is contained in:
msramalho
2026-01-08 14:30:26 +00:00
parent c1f312d42a
commit 53dc9904ce
4 changed files with 22 additions and 22 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/auto_archiver/core/orchestrator.py
View File

@@ -274,14 +274,6 @@ Here's how that would look: \n\nsteps:\n extractors:\n - [your_extractor_name_
default=False,
)
parser.add_argument(
"--metadata",
dest="requested_metadata",
help="An array of specific metadata fields to select from the collected content.",
default=[],
nargs="?",
)
def add_individual_module_args(
self, modules: list[LazyBaseModule] = None, parser: argparse.ArgumentParser = None
) -> None:

7
src/auto_archiver/modules/metadata_enricher/__manifest__.py
View File

@@ -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.

19
src/auto_archiver/modules/metadata_enricher/metadata_enricher.py
View File

@@ -11,15 +11,14 @@ class MetadataEnricher(Enricher):
Extracts metadata information from files using exiftool.
"""
def enrich(self, to_enrich: Metadata, md_grocery_list=["author", "datetimes", "location"]) -> None:
def enrich(self, to_enrich: Metadata) -> None:
logger.debug("Extracting EXIF metadata")
for i, m in enumerate(to_enrich.media):
if len(md := self.get_metadata(m.filename)):
# feature flag has this currently turned on
specified_md = self.select_metadata(md, md_grocery_list)
to_enrich.media[i].set("metadata", specified_md)
# to_enrich.media[i].set("metadata", md)
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:
try:
@@ -38,7 +37,7 @@ class MetadataEnricher(Enricher):
logger.error(f"Error occurred: {e}: {traceback.format_exc()}")
return {}
def select_metadata(self, all_md, md_grocery_list):
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
"""
@@ -51,19 +50,19 @@ class MetadataEnricher(Enricher):
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 md_grocery_list) and any(
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 md_grocery_list) and any(
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 md_grocery_list) and any(
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 md_grocery_list or md_key in md_grocery_list and len(all_md[md_key]):
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