X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/f46044c66663049e286c20ee015db99d47d9dd8a..540fd68c40df72763aee5d75598675c45cfa9aba:/youtube_dl/postprocessor/metadatafromtitle.py

diff --git a/youtube_dl/postprocessor/metadatafromtitle.py b/youtube_dl/postprocessor/metadatafromtitle.py
new file mode 100644
index 0000000..a56077f
--- /dev/null
+++ b/youtube_dl/postprocessor/metadatafromtitle.py
@@ -0,0 +1,47 @@
+from __future__ import unicode_literals
+
+import re
+
+from .common import PostProcessor
+from ..utils import PostProcessingError
+
+
+class MetadataFromTitlePPError(PostProcessingError):
+    pass
+
+
+class MetadataFromTitlePP(PostProcessor):
+    def __init__(self, downloader, titleformat):
+        super(MetadataFromTitlePP, self).__init__(downloader)
+        self._titleformat = titleformat
+        self._titleregex = self.format_to_regex(titleformat)
+
+    def format_to_regex(self, fmt):
+        """
+        Converts a string like
+           '%(title)s - %(artist)s'
+        to a regex like
+           '(?P<title>.+)\ \-\ (?P<artist>.+)'
+        """
+        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<' + match.group(1) + '>.+)'
+            lastpos = match.end()
+        if lastpos < len(fmt):
+            regex += re.escape(fmt[lastpos:len(fmt)])
+        return regex
+
+    def run(self, info):
+        title = info['title']
+        match = re.match(self._titleregex, title)
+        if match is None:
+            raise MetadataFromTitlePPError('Could not interpret title of video as "%s"' % self._titleformat)
+        for attribute, value in match.groupdict().items():
+            value = match.group(attribute)
+            info[attribute] = value
+            self._downloader.to_screen('[fromtitle] parsed ' + attribute + ': ' + value)
+
+        return [], info