1
0
mirror of https://github.com/janeczku/calibre-web synced 2024-10-31 15:16:20 +00:00

Bugfix for comicapi

This commit is contained in:
Ozzieisaacs 2019-08-03 14:09:54 +02:00
parent 6261981656
commit e9fb5d9f25

View File

@ -41,11 +41,13 @@ def extractCover(tmp_file_name, original_file_extension):
if use_comic_meta:
archive = ComicArchive(tmp_file_name)
cover_data = None
ext = os.path.splitext(archive.getPageName(0))
if len(ext) > 1:
extension = ext[1].lower()
if extension == '.jpg' or extension == '.jpeg':
cover_data = archive.getPage(0)
for index, name in enumerate(archive.getPageNameList()):
ext = os.path.splitext(name)
if len(ext) > 1:
extension = ext[1].lower()
if extension == '.jpg' or extension == '.jpeg':
cover_data = archive.getPage(index)
break
else:
if original_file_extension.upper() == '.CBZ':
cf = zipfile.ZipFile(tmp_file_name)
@ -53,7 +55,7 @@ def extractCover(tmp_file_name, original_file_extension):
ext = os.path.splitext(name)
if len(ext) > 1:
extension = ext[1].lower()
if extension == '.jpg':
if extension == '.jpg' or extension == '.jpeg':
cover_data = cf.read(name)
break
elif original_file_extension.upper() == '.CBT':
@ -62,7 +64,7 @@ def extractCover(tmp_file_name, original_file_extension):
ext = os.path.splitext(name)
if len(ext) > 1:
extension = ext[1].lower()
if extension == '.jpg':
if extension == '.jpg' or extension == '.jpeg':
cover_data = cf.extractfile(name).read()
break
prefix = os.path.dirname(tmp_file_name)
@ -87,16 +89,17 @@ def get_comic_info(tmp_file_path, original_file_name, original_file_extension):
else:
style = None
if style is not None:
loadedMetadata = archive.readMetadata(style)
# if style is not None:
loadedMetadata = archive.readMetadata(style)
lang = loadedMetadata.language
if len(lang) == 2:
loadedMetadata.language = isoLanguages.get(part1=lang).name
elif len(lang) == 3:
loadedMetadata.language = isoLanguages.get(part3=lang).name
else:
loadedMetadata.language = ""
lang = loadedMetadata.language
if lang:
if len(lang) == 2:
loadedMetadata.language = isoLanguages.get(part1=lang).name
elif len(lang) == 3:
loadedMetadata.language = isoLanguages.get(part3=lang).name
else:
loadedMetadata.language = ""
return BookMeta(
file_path=tmp_file_path,