yt_dlp/postprocessor/metadataparser.py

   1 import re
   2
   3 from enum import Enum
   4
   5 from .common import PostProcessor
   6
   7
   8 class MetadataParserPP(PostProcessor):
   9     class Actions(Enum):
  10         INTERPRET = 'interpretter'
  11         REPLACE = 'replacer'
  12
  13     def __init__(self, downloader, actions):
  14         PostProcessor.__init__(self, downloader)
  15         self._actions = []
  16         for f in actions:
  17             action = f[0]
  18             assert isinstance(action, self.Actions)
  19             self._actions.append(getattr(self, action.value)(*f[1:]))
  20
  21     @classmethod
  22     def validate_action(cls, action, *data):
  23         ''' Each action can be:
  24                 (Actions.INTERPRET, from, to) OR
  25                 (Actions.REPLACE, field, search, replace)
  26         '''
  27         if not isinstance(action, cls.Actions):
  28             raise ValueError(f'{action!r} is not a valid action')
  29         getattr(cls, action.value)(cls, *data)
  30
  31     @staticmethod
  32     def field_to_template(tmpl):
  33         if re.match(r'[a-zA-Z_]+$', tmpl):
  34             return f'%({tmpl})s'
  35         return tmpl
  36
  37     @staticmethod
  38     def format_to_regex(fmt):
  39         r"""
  40         Converts a string like
  41            '%(title)s - %(artist)s'
  42         to a regex like
  43            '(?P<title>.+)\ \-\ (?P<artist>.+)'
  44         """
  45         if not re.search(r'%\(\w+\)s', fmt):
  46             return fmt
  47         lastpos = 0
  48         regex = ''
  49         # replace %(..)s with regex group and escape other string parts
  50         for match in re.finditer(r'%\((\w+)\)s', fmt):
  51             regex += re.escape(fmt[lastpos:match.start()])
  52             regex += rf'(?P<{match.group(1)}>.+)'
  53             lastpos = match.end()
  54         if lastpos < len(fmt):
  55             regex += re.escape(fmt[lastpos:])
  56         return regex
  57
  58     def run(self, info):
  59         for f in self._actions:
  60             f(info)
  61         return [], info
  62
  63     def interpretter(self, inp, out):
  64         def f(info):
  65             data_to_parse = self._downloader.evaluate_outtmpl(template, info)
  66             self.write_debug(f'Searching for {out_re.pattern!r} in {template!r}')
  67             match = out_re.search(data_to_parse)
  68             if match is None:
  69                 self.report_warning(f'Could not interpret {inp!r} as {out!r}')
  70                 return
  71             for attribute, value in match.groupdict().items():
  72                 info[attribute] = value
  73                 self.to_screen('Parsed %s from %r: %r' % (attribute, template, value if value is not None else 'NA'))
  74
  75         template = self.field_to_template(inp)
  76         out_re = re.compile(self.format_to_regex(out))
  77         return f
  78
  79     def replacer(self, field, search, replace):
  80         def f(info):
  81             val = info.get(field)
  82             if val is None:
  83                 self.report_warning(f'Video does not have a {field}')
  84                 return
  85             elif not isinstance(val, str):
  86                 self.report_warning(f'Cannot replace in field {field} since it is a {type(val).__name__}')
  87                 return
  88             self.write_debug(f'Replacing all {search!r} in {field} with {replace!r}')
  89             info[field], n = search_re.subn(replace, val)
  90             if n:
  91                 self.to_screen(f'Changed {field} to: {info[field]}')
  92             else:
  93                 self.to_screen(f'Did not find {search!r} in {field}')
  94
  95         search_re = re.compile(search)
  96         return f
  97
  98
  99 class MetadataFromFieldPP(MetadataParserPP):
 100     @classmethod
 101     def to_action(cls, f):
 102         match = re.match(r'(?s)(?P<in>.*?)(?<!\\):(?P<out>.+)$', f)
 103         if match is None:
 104             raise ValueError(f'it should be FROM:TO, not {f!r}')
 105         return (
 106             cls.Actions.INTERPRET,
 107             match.group('in').replace('\\:', ':'),
 108             match.group('out'),
 109         )
 110
 111     def __init__(self, downloader, formats):
 112         super().__init__(downloader, [self.to_action(f) for f in formats])
 113
 114
 115 # Deprecated
 116 class MetadataFromTitlePP(MetadataParserPP):
 117     def __init__(self, downloader, titleformat):
 118         super().__init__(downloader, [(self.Actions.INTERPRET, 'title', titleformat)])
 119         self.deprecation_warning(
 120             'yt_dlp.postprocessor.MetadataFromTitlePP is deprecated '
 121             'and may be removed in a future version. Use yt_dlp.postprocessor.MetadataFromFieldPP instead')