sys.exit(0)
sys.stdout.write("%s/*.pdf --> %s/*.swf\n\n" % (dirin, dirout))
try:
fdone = open(os.path.join(dirout, "done.txt"), 'a+')
ferr = open(os.path.join(dirout, "error.txt"), 'a')
except IOError as e:
sys.stderr.write("parser done file error: %s\n" % e)
else:
completed = set()
fdone.seek(0)
for line in fdone:
completed.add(line.rstrip())
pdfs = scandir(dirin)
for pdf in pdfs:
if pdf.name in completed:
sys.stdout.write("skipping: %s\n" % pdf.name)
continue
sys.stdout.write("%s\n" % pdf.name)
try:
parsed = FrankenParser(pdf.path)
except Exception as e:
try:
ferr.write("%s:%s\n" % (pdf.name, str(e)))
except Exception:
sys.exit(0)
sys.stdout.write("%s/*.pdf --> %s/*.swf\n\n" % (dirin, dirout))
try:
fdone = open(os.path.join(dirout, "done.txt"), 'a+')
ferr = open(os.path.join(dirout, "error.txt"), 'a')
except IOError as e:
sys.stderr.write("parser done file error: %s\n" % e)
else:
completed = set()
fdone.seek(0)
for line in fdone:
completed.add(line.rstrip())
pdfs = scandir(dirin)
for pdf in pdfs:
if pdf.name in completed:
sys.stdout.write("skipping: %s\n" % pdf.name)
continue
sys.stdout.write("%s\n" % pdf.name)
try:
parsed = FrankenParser(pdf.path)
except Exception as e:
try:
ferr.write("%s:%s\n" % (pdf.name, str(e)))
except Exception:
with LOCK:
sys.stdout.write(msg)
sys.stdout.flush()
if __name__ == '__main__':
pdfs = []
args = ParserFactory().new_parser().parse()
num_procs = multiprocessing.cpu_count() / 2 - 3
num_procs = num_procs if num_procs > 0 else 1
print('Running on %d processes' % num_procs)
if os.path.isdir(args.pdf_in):
dir_name = os.path.join(args.pdf_in, '*')
print('Examining directory %s' % dir_name)
pdfs = scandir(args.pdf_in)
elif os.path.exists(args.pdf_in):
print('Analyzing file: %s' % args.pdf_in)
fin = ''
try:
fin = open(args.pdf_in, 'r')
except IOError as e:
print("%s" % e)
sys.exit(0)
else:
pdfs = [line.rstrip() for line in fin.readlines()]
fin.close()
print('Found %d jobs in file' % len(pdfs))
else:
print('Unable to find PDF file/directory: %s' % args.pdf_in)
sys.exit(1)
with LOCK:
sys.stdout.write(msg)
sys.stdout.flush()
if __name__ == '__main__':
pdfs = []
args = ParserFactory().new_parser().parse()
num_procs = multiprocessing.cpu_count() / 2 - 3
num_procs = num_procs if num_procs > 0 else 1
print('Running on %d processes' % num_procs)
if os.path.isdir(args.pdf_in):
dir_name = os.path.join(args.pdf_in, '*')
print('Examining directory %s' % dir_name)
pdfs = scandir(args.pdf_in)
elif os.path.exists(args.pdf_in):
print('Analyzing file: %s' % args.pdf_in)
fin = ''
try:
fin = open(args.pdf_in, 'r')
except IOError as e:
print("%s" % e)
sys.exit(0)
else:
pdfs = [line.rstrip() for line in fin.readlines()]
fin.close()
print('Found %d jobs in file' % len(pdfs))
else:
print('Unable to find PDF file/directory: %s' % args.pdf_in)
sys.exit(1)
