Changeset 492

Timestamp:

02/17/09 10:28:37 (16 years ago)

Author:

Nicholas Riley

Message:

Handle BRs and use short/long descriptions.

File:

• trunk/Update Podcasts/update_podcasts.py (modified) (4 diffs)

trunk/Update Podcasts/update_podcasts.py

@@ -34 +34 @@
 RE_TOO_MANY_CR = re.compile(r'\s*\n\s*\n\s*\n+', re.U)
 RE_TAG = re.compile(r'<[^>]*>')
+RE_BR = re.compile(r'<[Bb][Rr][^>]*>\s*', re.U)
 RE_WHITESPACE = re.compile(r'\s+', re.U)
 RE_LEADING_WHITESPACE = re.compile(r'^\s+(.*)', re.U)
@@ -46 +47 @@
         s = RE_MATCHED_P.sub(r'\1\n\n', s)
         s = RE_OPEN_P.sub(r'\n\n', s)
+        s = RE_BR.sub(r'\n', s)
         s = RE_TAG.sub('', s)
         s = RE_TOO_MANY_CR.sub(r'\n\n', s)
@@ -51 +53 @@
     s = RE_TRAILING_WHITESPACE.sub(r'\1', s)
     return s
+
+def words(s):
+    return set(RE_WHITESPACE.split(s))
 
 def podcasts_to_lyrics(iTunes):
@@ -70 +75 @@
         else:
             longdesc = html_to_text(longdesc)
+            if desc != k.missing_value:
+                desc = html_to_text(desc)
+                if (desc not in longdesc and
+                    len(words(desc) - words(longdesc)) > 1):
+                    longdesc = '%s\n\n%s' % (desc, longdesc)
 
         if lyrics == longdesc: