fixed #60, UnicodeEncodeError: 'charmap' codec can't encode character

r0oth3x49 · Dec 28, 2017 · 20c9084 · 20c9084
1 parent c35e66f
commit 20c9084
Show file tree

Hide file tree

Showing 2 changed files with 71 additions and 15 deletions.
diff --git a/udemy/_compat.py b/udemy/_compat.py
@@ -76,6 +76,7 @@
             "re",
             "os",
             "sys",
+            "pyver",
             "compat_str",
             "login_popup",
             "compat_urllib",

diff --git a/udemy/_utils.py b/udemy/_utils.py
@@ -1,13 +1,14 @@
 #!/usr/bin/python
 
-
+from pprint import pprint
 from . import __author__
 from . import __version__
 from ._compat import (
                         os,
                         re,
                         sys,
                         json,
+                        pyver,
                         NO_DEFAULT,
                         compat_HTMLParser,
                         )
@@ -184,10 +185,58 @@ def fix_kv(m):
         '''.format(comment=COMMENT_RE, skip=SKIP_RE), fix_kv, code)
 
 class WEBVTT2SRT:
-    def _fix_subtitles(self, content):
+
+    def _write_srtcontent(self, filename, _srtfilename, _srtcontent):
+        retVal = {}
+        if pyver == 3:
+            with open(_srtfilename, 'w', encoding='utf-8') as sub:
+                try:
+                    sub.write('{}'.format(_srtcontent))
+                except Exception as e:
+                    retVal = {'status' : 'False', 'msg' : 'Python3 Exception : {}'.format(e)}
+                else:
+                    retVal = {'status' : 'True', 'msg' : 'srt content written successfully'}
+                    try:
+                        os.unlink(filename)
+                    except Exception as e:
+                        pass
+            sub.close()
+        else:
+            with open(_srtfilename, 'w') as sub:
+                try:
+                    sub.write('{}'.format(_srtcontent))
+                except Exception as e:
+                    retVal = {'status' : 'False', 'msg' : 'Python2 Exception : {}'.format(e)}
+                else:
+                    retVal = {'status' : 'True', 'msg' : 'srt content written successfully'}
+                    try:
+                        os.unlink(filename)
+                    except Exception as e:
+                        pass
+            sub.close()
+
+        return retVal
+
+    def _get_index(self, content, flag=True):
+        if flag:
+            i = 0
+            for line in content:
+                if '-->' in line:
+                    index = i
+                    break
+                i += 1
+            return index
+        if not flag:
+            try:
+                index       =   content.index('1\r\n') if pyver == 2 else content.index(b'1\r\n')
+            except Exception as e:
+                index       =   2
+            return index
+
+    def _fix_subtitles(self, content, index):
         _container = ''
-        for line in content[2:]:
-            if sys.version_info[:2] >= (3, 0):
+        for line in content[index:]:
+            if pyver == 3:
                 _container += line.decode('utf-8', 'ignore')
             else:
                 _container += line
@@ -226,15 +275,21 @@ def convert(self, filename=None):
             else:
                 content     =   [line for line in (l.decode('utf-8', 'ignore').strip() for l in f_in) if line]
                 f_in.close()
+                try:
+                    check       =   content.index('1') or content.index('1\r\n')
+                except:
+                    check       =   0
                 if len(content) > 4:
                     if content[0] == 'WEBVTT' or content[0].endswith('WEBVTT') or 'WEBVTT' in content[0]:
-                        if content[1] == '1':
+                        if content[check] == '1':
                             f           = open(filename, 'rb')
                             content     = f.readlines()
                             f.close()
-                            _srtcontent = self._fix_subtitles(content)
+                            index       = self._get_index(content, flag=False)
+                            _srtcontent = self._fix_subtitles(content, index)
                         else:
-                            for line in content[1:]:
+                            index   =   self._get_index(content)
+                            for line in content[index:]:
                                 if '-->' in line:
                                     _start, _end  = line.split(' --> ')
                                     _stcode       = _start.split(':')
@@ -249,14 +304,14 @@ def convert(self, filename=None):
                                         _srtcontent += '{}\r\n{} --> {}\r\n{}\r\n\r\n'.format(_seqcounter, _appeartime, _disappertime, _textcontainer)
 
                         if _srtcontent:
-                            with open(_srtfilename, 'w') as sub:
-                                sub.write('{}'.format(_srtcontent))
-                            sub.close()
-                            try:
-                                os.unlink(filename)
-                            except Exception as e:
-                                pass
-                            _flag = {'status' : 'True', 'msg' : 'successfully generated subtitle in srt ...'}
+                            retVal = self._write_srtcontent(filename, _srtfilename, _srtcontent)
+                            if isinstance(retVal, dict) and len(retVal) > 0:
+                                status = retVal.get('status')
+                                msg    = retVal.get('msg')
+                                if status == 'True':
+                                    _flag = {'status' : 'True', 'msg' : 'successfully generated subtitle in srt ...'}
+                                else:
+                                    _flag = {'status' : 'False', 'msg' : '{}'.format(msg)}
                 else:
                     _flag = {'status' : 'False', 'msg' : 'subtitle file seems to be empty skipping conversion from WEBVTT to SRT ..'}
         return _flag