remove blank lines, update version number (#29)

* Remove useless blank lines in some vtt files. 1- add a new function to remove any unwanted blank lines in vtt file 2- also update test files by removing useless empty lines from them. * Backporting the vtt_to_srt script from python 3 to python 2 and test files as will. - backporting vtt_to_srt2 script to python 2.7.18 - backporting test files to run with pytest 4.6.11 * remove uesless blank lines update setup file with new version number moving version 2 code to it's new repository --------- Co-authored-by: karimkoko <karimkoko0@yahoo.com>

Author: KarimReefat

setup.py

@@ -5,7 +5,7 @@
  long_description = fh.read()
 
 setuptools.setup(name='vtt_to_srt3',
- version='0.2.0.1',
+ version='0.2.0.3',
  author="Jeison Cardoso",
  author_email="j@jsonzilla.com",
  maintainer="Jeison Cardoso",

tests/valid_output_idd.srt

@@ -1,4 +1,3 @@
-
 1
 00:00:08,393 --> 00:00:10,437
 ♪ ♪
@@ -8,7 +7,6 @@
 Narrator: <i>blaba</i>
 2
 
-
 3
 00:00:15,024 --> 00:00:15,817
 <i>bla</i>
@@ -30,15 +28,12 @@ ah</i>
 00:00:25,994 --> 00:00:28,371
 <i>blaba</i>
 
-
 8
 00:00:28,455 --> 00:00:32,125
 <i>blaba
 blaba</i>
 
-
 9
 00:00:32,208 --> 00:00:34,002
 <i>blaba</i>
 1010
-

tests/valid_output_iso-8859-2.srt

@@ -1,4 +1,3 @@
-
 1
 00:00:01,000 --> 00:00:04,000
 - Trinken Sie niemals flüssigen Stickstoff.
@@ -7,4 +6,3 @@
 00:00:05,000 --> 00:00:09,000
 - Es wird Ihren Magen perforieren.
 - Du könntest sterben.
-

tests/valid_output_utf8.srt

@@ -1,4 +1,3 @@
-
 1
 00:00:01,000 --> 00:00:04,000
 - Never drink liquid nitrogen.
@@ -7,4 +6,3 @@
 00:00:05,000 --> 00:00:09,000
 - It will perforate your stomach.
 - You could die.
-

vtt_to_srt/vtt_to_srt.py

@@ -63,6 +63,7 @@ def convert_content(self, contents: str) -> str:
  replacement = re.sub(
  r"::[\-\w]+\([\-.\w\d]+\)[ ]*{[.,:;\(\) \-\w\d]+\n }\n", "", replacement)
  replacement = re.sub(r"Style:\n##\n", "", replacement)
+ replacement = self.remove_blank_lines(replacement)
  replacement = self.remove_simple_identifiers(replacement)
  replacement = self.add_sequence_numbers(replacement)
 
@@ -89,7 +90,36 @@ def add_sequence_numbers(self, contents: str) -> str:
  counter += 1
  out += line + '\n'
  return out
-
+ 
+ def remove_blank_lines(self, contents: str) -> str:
+ # Remove useless blank lines from the vtt file 
+ lines = contents.split('\n')
+ lines = [x for x in lines if x != '']
+ lines.append('')
+ out = []
+ num = 0
+ while num < len(lines) :
+ if re.match(r"^\d+$", lines[num]) and self.has_timestamp(lines[num + 1]):
+ if num == 0 :
+ pass
+ else:
+ out.append('')
+ out.append(lines[num])
+ out.append(lines[num + 1])
+ num += 2
+ elif self.has_timestamp(lines[num]): 
+ if num == 0 :
+ pass
+ else :
+ out.append('')
+ out.append(lines[num])
+ num += 1
+ else:
+ out.append(lines[num])
+ num += 1
+ out.pop()
+ return '\n'.join(out)
+ 
  def remove_simple_identifiers(self, contents: str) -> str:
  """Remove simple identifiers of vtt file
 
@@ -102,7 +132,7 @@ def remove_simple_identifiers(self, contents: str) -> str:
  if re.match(r"^\d+$", lines[i - 1]):
  out.pop()
  out.append(line)
- return '\n'.join(out)
+ return '\n'.join(out) 
 
  def write_file(self, filename: str, data, encoding_format: str = "utf-8"):
  """Create a file with some data