0x2620/openmedialibrary
1
0
Fork 0
Code Issues 131 Pull requests Releases Wiki Activity

add meta.extract_text

Browse source
This commit is contained in:
j 2016-01-19 20:48:25 +05:30
commit f43fc6a172
2 changed files with 30 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

14
oml/media/__init__.py
View file

@ -123,3 +123,17 @@ def metadata(f, from_=None):
data[key] = [data[key]] if data[key] else []
return data
def extract_text(path):
ext = path.split('.')[-1]
text = ''
try:
if ext in ('epub', 'kepub'):
text = epub.extract_text(path)
elif ext == 'pdf':
text = pdf.extract_text(path)
elif ext == 'txt':
text = txt.extract_text(path)
except:
logger.debug('failed to extract text from %s', path, exc_info=1)
text = ''
return text