manage styles in the output

.gitignore

@@ -0,0 +1,6 @@
+/venv
+/.idea
+*.vtt
+*.srt
+/build/
+/vtt_to_srt.egg-info

README.md

@@ -1,23 +1,54 @@
-[![asciicast](https://asciinema.org/a/234035.svg)](https://asciinema.org/a/234035)
+# vtt-to-srt subtitle converter
 
-# Python 3
-## Packages
+A simple python module to convert vtt files to srt files **_with colours_**!
 
-- pysrt
-- webvtt-py
+## ❓ Why this fork?
+
+Some streaming platforms use vtt subtitles with colours. (ie. Auvio)
+
+- [upstream](https://github.com/lbrayner/vtt-to-srt) doesn't convert vtt to srt subtitles when there are colours.  
+see [issue #2](https://github.com/lbrayner/vtt-to-srt/issues/2) 
+
+
+- other tools like `ffmpeg` don't work either  
+`ffmpeg -i captions.vtt captions.srt`    
+see ticket [subtitles: vtt conversion to srt creates an empty output](https://trac.ffmpeg.org/ticket/9609)
+
+
+- the library used is `webvtt-py`
+it can't convert vtt files with colours to srt files  
+see [pull request #39](https://github.com/glut23/webvtt-py/pull/39)
+
+## 💾 Installation
 
 ```
-pip3 install pysrt webvtt-py
+pip3 install .
 ```
 
-# Synopsis
+## 🖥️ Command line interface
 
 ```
-./vtt-to-srt FILE...
+usage: vtt_to_srt [-h] [-s] [file [file ...]]
+
+vtt_to_srt  is a command line tool to convert vtt subtitles to srt files
+
+positional arguments:
+  file        a file. The command accepts zero, one or more files as arguments. For each .vtt, a .srt will be
+              generated in the same folder. Any other extension is ignored.
+
+optional arguments:
+  -h, --help  show this help message and exit
+  -s          strip all tags in output srt
 ```
 
-# Description
+## 📙 Usage
+
+The command accepts zero, one or more files as arguments.  
+For each _.vtt_, a _.srt_ will be generated in the same folder.  
+Any other extension is ignored.  
+If you specify no file, it will convert all _.vtt_ files in the directory.
+
+## Other projects
 
-The command accepts one or more files as arguments.
-For each _.vtt_, a _.srt_ will be generated in the same folder.
-Any other extension is ignored.
+`vtt_to_srt` is used in a docker image for youtube-dl with vtt to srt conversion
+https://github.com/darodi/docker-youtube-dl-vtt-to-srt

requirements.txt

@@ -0,0 +1,3 @@
+pysrt
+webvtt-py
+argparse

setup.py

@@ -0,0 +1,21 @@
+from setuptools import setup, find_packages
+
+setup(
+    name='vtt_to_srt',
+    version='0.1',
+    url='',
+    license='LGPLv2',
+    python_requires='>=3.4',
+    author='lbrayner',
+    author_email='[email protected]',
+    description='vtt_to_srt',
+    py_modules=['vtt_to_srt'],
+    install_requires=[
+        "pysrt",
+        "webvtt-py",
+        "argparse"
+    ],
+    entry_points={
+        'console_scripts': ['vtt_to_srt=vtt_to_srt:main']
+    }
+)

vtt_to_srt.py

@@ -0,0 +1,85 @@
+#!/usr/bin/env python3
+import argparse
+import html
+import os
+import re
+import sys
+
+from pysrt.srtitem import SubRipItem, SubRipTime
+from webvtt import WebVTT
+
+
+def replace_colors(raw_text, colours_arg, tag_name):
+    result = raw_text
+    for k, v in colours_arg.items():
+        regex_string = "<" + tag_name + "(?:\\..*?)?\\." + str(k) + "(?:\\..*?)?>(.*?)</" + tag_name + ">"
+        if re.search(regex_string, result) is not None:
+            result = re.sub(regex_string, lambda x: replace_color(x, tag_name, v), result)
+    return result
+
+
+def replace_color(x, tag_name, v):
+    return ("" if tag_name == "c" else ("<" + tag_name + ">")) \
+           + "<font color=\"" + v + "\">" \
+           + html.unescape(x.group(1)) \
+           + "</font>" \
+           + ("" if tag_name == "c" else ("</" + tag_name + ">"))
+
+
+COLOURS_PATTERN = re.compile(r'::cue\(\.([^)]+)\)\s*{.*?color:(.*?);.*?}')
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description='vtt_to_srt is a command line tool to convert vtt subtitles to srt files')
+    parser.add_argument('file', nargs='*',
+                        help='a file. The command accepts zero, one or more files as arguments.\n'
+                             'For each .vtt, a .srt will be generated in the same folder.\n'
+                             'Any other extension is ignored.')
+    parser.add_argument('-s', dest='strip', action='store_true', help='strip all tags in output srt')
+    args = parser.parse_args()
+
+    if len(args.file) == 0:
+        for file in os.listdir():
+            if file.endswith(".vtt"):
+                args.file.append(file)
+
+    for file in args.file:
+        index = 0
+
+        file_name, file_extension = os.path.splitext(file)
+
+        if not file_extension.lower() == ".vtt":
+            sys.stderr.write("Skipping %s.\n" % file)
+            continue
+
+        srt = open(file_name + ".srt", "w", encoding='utf-8')
+
+        read = WebVTT().read(file)
+
+        colours = dict()
+        if args.strip is False:
+            for style in read.styles:
+                colours_found = COLOURS_PATTERN.findall(style.text)
+                colours_classes = list(map(lambda x: x[0], colours_found))
+                colours_values = list(map(lambda x: x[1].replace(" ", ""), colours_found))
+                colours = dict(zip(colours_classes, colours_values))
+
+        for caption in read.captions:
+            index += 1
+            start = SubRipTime(0, 0, caption.start_in_seconds)
+            end = SubRipTime(0, 0, caption.end_in_seconds)
+            caption_text = caption.raw_text
+            no_tag_found = True
+            if args.strip is False:
+                for tag in ['c', 'i', 'b', 'u']:
+                    if re.search("<" + tag + "\\..*?>.*?</" + tag + ">", caption_text) is not None:
+                        caption_text = replace_colors(caption_text, colours, tag)
+                        no_tag_found = False
+            if no_tag_found:
+                caption_text = html.unescape(caption.text)
+            srt.write(SubRipItem(index, start, end, caption_text).__str__() + "\n")
+
+
+if __name__ == "__main__":
+    main()