fix warc path in snapshot_icons
This commit is contained in:
parent
48eddadc9b
commit
988a10a9f6
1 changed files with 9 additions and 6 deletions
|
@ -1,11 +1,11 @@
|
||||||
__package__ = 'archivebox.index'
|
__package__ = 'archivebox.index'
|
||||||
|
|
||||||
from datetime import datetime
|
|
||||||
from typing import List, Optional, Iterator, Mapping
|
|
||||||
from pathlib import Path
|
from pathlib import Path
|
||||||
|
from datetime import datetime
|
||||||
|
from collections import defaultdict
|
||||||
|
from typing import List, Optional, Iterator, Mapping
|
||||||
|
|
||||||
from django.utils.html import format_html, mark_safe
|
from django.utils.html import format_html, mark_safe
|
||||||
from collections import defaultdict
|
|
||||||
|
|
||||||
from .schema import Link
|
from .schema import Link
|
||||||
from ..system import atomic_write
|
from ..system import atomic_write
|
||||||
|
@ -116,10 +116,9 @@ def render_django_template(template: str, context: Mapping[str, str]) -> str:
|
||||||
|
|
||||||
def snapshot_icons(snapshot) -> str:
|
def snapshot_icons(snapshot) -> str:
|
||||||
from core.models import EXTRACTORS
|
from core.models import EXTRACTORS
|
||||||
|
|
||||||
# start = datetime.now()
|
# start = datetime.now()
|
||||||
|
|
||||||
archive_results = snapshot.archiveresult_set.filter(status="succeeded")
|
archive_results = snapshot.archiveresult_set.filter(status="succeeded", output__isnull=False)
|
||||||
link = snapshot.as_link()
|
link = snapshot.as_link()
|
||||||
path = link.archive_path
|
path = link.archive_path
|
||||||
canon = link.canonical_outputs()
|
canon = link.canonical_outputs()
|
||||||
|
@ -166,7 +165,7 @@ def snapshot_icons(snapshot) -> str:
|
||||||
exists = extractor_outputs[extractor] and extractor_outputs[extractor].status == 'succeeded' and extractor_outputs[extractor].output
|
exists = extractor_outputs[extractor] and extractor_outputs[extractor].status == 'succeeded' and extractor_outputs[extractor].output
|
||||||
# get from filesystem (slower but more accurate)
|
# get from filesystem (slower but more accurate)
|
||||||
# exists = list((Path(path) / canon["warc_path"]).glob("*.warc.gz"))
|
# exists = list((Path(path) / canon["warc_path"]).glob("*.warc.gz"))
|
||||||
output += format_html(output_template, 'warc/', canon["warc_path"], str(bool(exists)), "warc", icons.get("warc", "?"))
|
output += format_html(output_template, path, canon["warc_path"], str(bool(exists)), "warc", icons.get("warc", "?"))
|
||||||
|
|
||||||
if extractor == "archive_org":
|
if extractor == "archive_org":
|
||||||
# The check for archive_org is different, so it has to be handled separately
|
# The check for archive_org is different, so it has to be handled separately
|
||||||
|
@ -183,3 +182,7 @@ def snapshot_icons(snapshot) -> str:
|
||||||
# end = datetime.now()
|
# end = datetime.now()
|
||||||
# print(((end - start).total_seconds()*1000) // 1, 'ms')
|
# print(((end - start).total_seconds()*1000) // 1, 'ms')
|
||||||
return result
|
return result
|
||||||
|
|
||||||
|
# return cache.get_or_set(cache_key, calc_snapshot_icons)
|
||||||
|
|
||||||
|
|
||||||
|
|
Loading…
Reference in a new issue