remove pdf_pagecount()
This commit is contained in:
bovenliggende
07b3479029
commit
e006a95673
7
addindex
7
addindex
@ -41,11 +41,8 @@ def strip_irrelevant(content):
|
||||
result = re.sub(' +', ' ', result)
|
||||
return result;
|
||||
|
||||
def pdf_pagecount(path):
|
||||
cmd = "pdfinfo " + path + " | grep Pages | awk '{print $2}'"
|
||||
stdout,stderr = subprocess.Popen(cmd,shell=True,stdout=subprocess.PIPE, stderr=subprocess.PIPE).communicate()
|
||||
result = int(stdout)
|
||||
return result
|
||||
|
||||
|
||||
def process_pdf(path):
|
||||
result = list()
|
||||
args=["pdftotext", path, "-"]
|
||||
|
Laden…
Verwijs in nieuw issue
Block a user