Add option `--replace-in-metadata`

3 files changed, 123 insertions, 76 deletions

diff --git a/yt_dlp/postprocessor/__init__.py b/yt_dlp/postprocessor/__init__.py
index 98cbe8665..b1a6917d7 100644
--- a/yt_dlp/postprocessor/__init__.py
+++ b/yt_dlp/postprocessor/__init__.py
@@ -20,8 +20,11 @@ from .ffmpeg import (
 )
 from .xattrpp import XAttrMetadataPP
 from .execafterdownload import ExecAfterDownloadPP
-from .metadatafromfield import MetadataFromFieldPP
-from .metadatafromfield import MetadataFromTitlePP
+from .metadataparser import (
+    MetadataFromFieldPP,
+    MetadataFromTitlePP,
+    MetadataParserPP,
+)
 from .movefilesafterdownload import MoveFilesAfterDownloadPP
 from .sponskrub import SponSkrubPP
 
@@ -48,6 +51,7 @@ __all__ = [
     'FFmpegThumbnailsConvertorPP',
     'FFmpegVideoConvertorPP',
     'FFmpegVideoRemuxerPP',
+    'MetadataParserPP',
     'MetadataFromFieldPP',
     'MetadataFromTitlePP',
     'MoveFilesAfterDownloadPP',
diff --git a/yt_dlp/postprocessor/metadatafromfield.py b/yt_dlp/postprocessor/metadatafromfield.py
deleted file mode 100644
index 002794765..000000000
--- a/yt_dlp/postprocessor/metadatafromfield.py
+++ /dev/null
@@ -1,74 +0,0 @@
-from __future__ import unicode_literals
-
-import re
-
-from .common import PostProcessor
-from ..compat import compat_str
-
-
-class MetadataFromFieldPP(PostProcessor):
-    regex = r'(?P<in>.*?)(?<!\\):(?P<out>.+)$'
-
-    def __init__(self, downloader, formats):
-        PostProcessor.__init__(self, downloader)
-        assert isinstance(formats, (list, tuple))
-        self._data = []
-        for f in formats:
-            assert isinstance(f, compat_str)
-            match = re.match(self.regex, f)
-            assert match is not None
-            inp = match.group('in').replace('\\:', ':')
-            self._data.append({
-                'in': inp,
-                'out': match.group('out'),
-                'tmpl': self.field_to_template(inp),
-                'regex': self.format_to_regex(match.group('out')),
-            })
-
-    @staticmethod
-    def field_to_template(tmpl):
-        if re.match(r'[a-zA-Z_]+$', tmpl):
-            return '%%(%s)s' % tmpl
-        return tmpl
-
-    @staticmethod
-    def format_to_regex(fmt):
-        r"""
-        Converts a string like
-           '%(title)s - %(artist)s'
-        to a regex like
-           '(?P<title>.+)\ \-\ (?P<artist>.+)'
-        """
-        if not re.search(r'%\(\w+\)s', fmt):
-            return fmt
-        lastpos = 0
-        regex = ''
-        # replace %(..)s with regex group and escape other string parts
-        for match in re.finditer(r'%\((\w+)\)s', fmt):
-            regex += re.escape(fmt[lastpos:match.start()])
-            regex += r'(?P<%s>.+)' % match.group(1)
-            lastpos = match.end()
-        if lastpos < len(fmt):
-            regex += re.escape(fmt[lastpos:])
-        return regex
-
-    def run(self, info):
-        for dictn in self._data:
-            tmpl, tmpl_dict = self._downloader.prepare_outtmpl(dictn['tmpl'], info)
-            data_to_parse = self._downloader.escape_outtmpl(tmpl) % tmpl_dict
-            self.write_debug('Searching for r"%s" in %s' % (dictn['regex'], dictn['tmpl']))
-            match = re.search(dictn['regex'], data_to_parse)
-            if match is None:
-                self.report_warning('Could not interpret video %s as "%s"' % (dictn['in'], dictn['out']))
-                continue
-            for attribute, value in match.groupdict().items():
-                info[attribute] = value
-                self.to_screen('parsed %s from "%s": %s' % (attribute, dictn['tmpl'], value if value is not None else 'NA'))
-        return [], info
-
-
-class MetadataFromTitlePP(MetadataFromFieldPP):  # for backward compatibility
-    def __init__(self, downloader, titleformat):
-        super(MetadataFromTitlePP, self).__init__(downloader, ['%%(title)s:%s' % titleformat])
-        self._titleformat = titleformat
-        self._titleregex = self._data[0]['regex']
diff --git a/yt_dlp/postprocessor/metadataparser.py b/yt_dlp/postprocessor/metadataparser.py
new file mode 100644
index 000000000..4d3c0e0ed
--- /dev/null
+++ b/yt_dlp/postprocessor/metadataparser.py
@@ -0,0 +1,117 @@
+import re
+
+from enum import Enum
+
+from .common import PostProcessor
+
+
+class MetadataParserPP(PostProcessor):
+    class Actions(Enum):
+        INTERPRET = 'interpretter'
+        REPLACE = 'replacer'
+
+    def __init__(self, downloader, actions):
+        PostProcessor.__init__(self, downloader)
+        self._actions = []
+        for f in actions:
+            action = f[0]
+            assert isinstance(action, self.Actions)
+            self._actions.append(getattr(self, action._value_)(*f[1:]))
+
+    @classmethod
+    def validate_action(cls, action, *data):
+        ''' Each action can be:
+                (Actions.INTERPRET, from, to) OR
+                (Actions.REPLACE, field, search, replace)
+        '''
+        if not isinstance(action, cls.Actions):
+            raise ValueError(f'{action!r} is not a valid action')
+        getattr(cls, action._value_)(cls, *data)
+
+    @staticmethod
+    def field_to_template(tmpl):
+        if re.match(r'[a-zA-Z_]+$', tmpl):
+            return f'%({tmpl})s'
+        return tmpl
+
+    @staticmethod
+    def format_to_regex(fmt):
+        r"""
+        Converts a string like
+           '%(title)s - %(artist)s'
+        to a regex like
+           '(?P<title>.+)\ \-\ (?P<artist>.+)'
+        """
+        if not re.search(r'%\(\w+\)s', fmt):
+            return fmt
+        lastpos = 0
+        regex = ''
+        # replace %(..)s with regex group and escape other string parts
+        for match in re.finditer(r'%\((\w+)\)s', fmt):
+            regex += re.escape(fmt[lastpos:match.start()])
+            regex += rf'(?P<{match.group(1)}>.+)'
+            lastpos = match.end()
+        if lastpos < len(fmt):
+            regex += re.escape(fmt[lastpos:])
+        return regex
+
+    def run(self, info):
+        for f in self._actions:
+            f(info)
+        return [], info
+
+    def interpretter(self, inp, out):
+        def f(info):
+            outtmpl, tmpl_dict = self._downloader.prepare_outtmpl(template, info)
+            data_to_parse = self._downloader.escape_outtmpl(outtmpl) % tmpl_dict
+            self.write_debug(f'Searching for r{out_re.pattern!r} in {template!r}')
+            match = out_re.search(data_to_parse)
+            if match is None:
+                self.report_warning('Could not interpret {inp!r} as {out!r}')
+                return
+            for attribute, value in match.groupdict().items():
+                info[attribute] = value
+                self.to_screen('Parsed %s from %r: %r' % (attribute, template, value if value is not None else 'NA'))
+
+        template = self.field_to_template(inp)
+        out_re = re.compile(self.format_to_regex(out))
+        return f
+
+    def replacer(self, field, search, replace):
+        def f(info):
+            val = info.get(field)
+            if val is None:
+                self.report_warning(f'Video does not have a {field}')
+                return
+            elif not isinstance(val, str):
+                self.report_warning(f'Cannot replace in field {field} since it is a {type(val).__name__}')
+                return
+            self.write_debug(f'Replacing all r{search!r} in {field} with {replace!r}')
+            info[field], n = search_re.subn(replace, val)
+            if n:
+                self.to_screen(f'Changed {field} to: {info[field]}')
+            else:
+                self.to_screen(f'Did not find r{search!r} in {field}')
+
+        search_re = re.compile(search)
+        return f
+
+
+class MetadataFromFieldPP(MetadataParserPP):
+    @classmethod
+    def to_action(cls, f):
+        match = re.match(r'(?P<in>.*?)(?<!\\):(?P<out>.+)$', f)
+        if match is None:
+            raise ValueError(f'it should be FROM:TO, not {f!r}')
+        return (
+            cls.Actions.INTERPRET,
+            match.group('in').replace('\\:', ':'),
+            match.group('out'))
+
+    def __init__(self, downloader, formats):
+        MetadataParserPP.__init__(self, downloader, [self.to_action(f) for f in formats])
+
+
+class MetadataFromTitlePP(MetadataParserPP):  # for backward compatibility
+    def __init__(self, downloader, titleformat):
+        MetadataParserPP.__init__(self, downloader, [(self.Actions.INTERPRET, 'title', titleformat)])