yt_dlp/postprocessor/metadataparser.py

   1 import re
   2
   3 from .common import PostProcessor
   4 from ..utils import Namespace, filter_dict, function_with_repr
   5
   6
   7 class MetadataParserPP(PostProcessor):
   8     def __init__(self, downloader, actions):
   9         super().__init__(downloader)
  10         self._actions = []
  11         for f in actions:
  12             action, *args = f
  13             assert action in self.Actions
  14             self._actions.append(action(self, *args))
  15
  16     @classmethod
  17     def validate_action(cls, action, *data):
  18         """Each action can be:
  19                 (Actions.INTERPRET, from, to) OR
  20                 (Actions.REPLACE, field, search, replace)
  21         """
  22         if action not in cls.Actions:
  23             raise ValueError(f'{action!r} is not a valid action')
  24         action(cls, *data)  # So this can raise error to validate
  25
  26     @staticmethod
  27     def field_to_template(tmpl):
  28         if re.match(r'[a-zA-Z_]+$', tmpl):
  29             return f'%({tmpl})s'
  30
  31         from ..YoutubeDL import YoutubeDL
  32         err = YoutubeDL.validate_outtmpl(tmpl)
  33         if err:
  34             raise err
  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     @function_with_repr
  64     def interpretter(self, inp, out):
  65         def f(info):
  66             data_to_parse = self._downloader.evaluate_outtmpl(template, info)
  67             self.write_debug(f'Searching for {out_re.pattern!r} in {template!r}')
  68             match = out_re.search(data_to_parse)
  69             if match is None:
  70                 self.to_screen(f'Could not interpret {inp!r} as {out!r}')
  71                 return
  72             for attribute, value in filter_dict(match.groupdict()).items():
  73                 info[attribute] = value
  74                 self.to_screen(f'Parsed {attribute} from {template!r}: {value!r}')
  75
  76         template = self.field_to_template(inp)
  77         out_re = re.compile(self.format_to_regex(out))
  78         return f
  79
  80     @function_with_repr
  81     def replacer(self, field, search, replace):
  82         def f(info):
  83             val = info.get(field)
  84             if val is None:
  85                 self.to_screen(f'Video does not have a {field}')
  86                 return
  87             elif not isinstance(val, str):
  88                 self.report_warning(f'Cannot replace in field {field} since it is a {type(val).__name__}')
  89                 return
  90             self.write_debug(f'Replacing all {search!r} in {field} with {replace!r}')
  91             info[field], n = search_re.subn(replace, val)
  92             if n:
  93                 self.to_screen(f'Changed {field} to: {info[field]}')
  94             else:
  95                 self.to_screen(f'Did not find {search!r} in {field}')
  96
  97         search_re = re.compile(search)
  98         return f
  99
 100     Actions = Namespace(INTERPRET=interpretter, REPLACE=replacer)
 101
 102
 103 class MetadataFromFieldPP(MetadataParserPP):
 104     @classmethod
 105     def to_action(cls, f):
 106         match = re.match(r'(?s)(?P<in>.*?)(?<!\\):(?P<out>.+)$', f)
 107         if match is None:
 108             raise ValueError(f'it should be FROM:TO, not {f!r}')
 109         return (
 110             cls.Actions.INTERPRET,
 111             match.group('in').replace('\\:', ':'),
 112             match.group('out'),
 113         )
 114
 115     def __init__(self, downloader, formats):
 116         super().__init__(downloader, [self.to_action(f) for f in formats])
 117
 118
 119 # Deprecated
 120 class MetadataFromTitlePP(MetadataParserPP):
 121     def __init__(self, downloader, titleformat):
 122         super().__init__(downloader, [(self.Actions.INTERPRET, 'title', titleformat)])
 123         self.deprecation_warning(
 124             'yt_dlp.postprocessor.MetadataFromTitlePP is deprecated '
 125             'and may be removed in a future version. Use yt_dlp.postprocessor.MetadataFromFieldPP instead')