From a2708ec9641045bfa972d4f201fd3ef100aa45b7 Mon Sep 17 00:00:00 2001 From: sbuser Date: Fri, 5 Aug 2011 18:59:56 -0500 Subject: [PATCH] Attempt to weed out substring results in searcher. --- headphones/searcher.py | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) diff --git a/headphones/searcher.py b/headphones/searcher.py index 89a2fc93..ee1aa2eb 100644 --- a/headphones/searcher.py +++ b/headphones/searcher.py @@ -75,10 +75,12 @@ def searchNZB(albumid=None, new=False): dic = {'...':'', ' & ':' ', ' = ': ' ', '?':'', '$':'s', ' + ':' ', '"':'', ',':''} cleanartistalbum = helpers.latinToAscii(helpers.replace_all(albums[0]+' '+albums[1], dic)) + cleanartist = helpers.latinToAscii(helpers.replace_all(albums[0], dic)) # FLAC usually doesn't have a year for some reason so I'll leave it out: term = re.sub('[\.\-\/]', ' ', '%s' % (cleanartistalbum)).encode('utf-8') altterm = re.sub('[\.\-\/]', ' ', '%s %s' % (cleanartistalbum, year)).encode('utf-8') + artistterm = re.sub('[\.\-\/]', ' ', '%s' % (cleanartist)).encode('utf-8') # Only use the year if the term could return a bunch of different albums, i.e. self-titled albums if albums[0] in albums[1] or len(albums[0]) < 4 or len(albums[1]) < 4: @@ -331,9 +333,22 @@ def searchNZB(albumid=None, new=False): else: logger.info('No results found from NEWZBIN for %s' % term) + + #attempt to verify that this isn't a substring result + #when looking for "Avant - Avant" we don't want "Avantasia" + #this should be less of an issue when it isn't a self-titled album so we'll only check vs artist + #for reasons I can't explain this still fails on "Lavanttaler" - maybe you have a better regex? + if resultlist: + for result in resultlist: + if re.search('\w' + re.escape(artistterm), result[0], re.IGNORECASE): + logger.info("Removed from results: " + result[0] + " (substring result).") + resultlist.remove(result) + elif re.search(re.escape(artistterm) + '\w', result[0], re.IGNORECASE): + logger.info("Removed from results: " + result[0] + " (substring result).") + resultlist.remove(result) if len(resultlist): - + if headphones.PREFERRED_QUALITY == 2 and headphones.PREFERRED_BITRATE: logger.debug('Target bitrate: %s kbps' % headphones.PREFERRED_BITRATE)