1
0
mirror of https://github.com/mitb-archive/filebot synced 2024-12-23 16:28:51 -05:00

* fix the "law.and.order.svu" special-fuckin-corner-case

This commit is contained in:
Reinhard Pointner 2013-12-15 18:35:41 +00:00
parent 5289227d2b
commit cdc4a94daa
2 changed files with 57 additions and 14 deletions

View File

@ -199,6 +199,7 @@ def addSeriesAlias = { from, to ->
// additional custom mappings // additional custom mappings
addSeriesAlias('Law & Order: Special Victims Unit', 'Law and Order SVU') addSeriesAlias('Law & Order: Special Victims Unit', 'Law and Order SVU')
addSeriesAlias('Law & Order: Special Victims Unit', 'Law & Order SVU')
addSeriesAlias('Battlestar Galactica (2003)', 'BSG') addSeriesAlias('Battlestar Galactica (2003)', 'BSG')
addSeriesAlias('CSI: Crime Scene Investigation', 'CSI') addSeriesAlias('CSI: Crime Scene Investigation', 'CSI')
addSeriesAlias('M*A*S*H', 'MASH') addSeriesAlias('M*A*S*H', 'MASH')
@ -208,10 +209,11 @@ addSeriesAlias('World Series of Poker', 'WSOP')
thetvdb_index = thetvdb_index.findResults{ [it[0] as Integer, it[1].replaceAll(/\s+/, ' ').trim()] }.findAll{ !(it[1] =~ /(?i:duplicate)/ || it[1] =~ /\d{6,}/ || it[1].startsWith('*') || it[1].endsWith('*') || it[1].length() < 2) } thetvdb_index = thetvdb_index.findResults{ [it[0] as Integer, it[1].replaceAll(/\s+/, ' ').trim()] }.findAll{ !(it[1] =~ /(?i:duplicate)/ || it[1] =~ /\d{6,}/ || it[1].startsWith('*') || it[1].endsWith('*') || it[1].length() < 2) }
thetvdb_index = thetvdb_index.sort({a,b -> a[0] <=> b[0]} as Comparator) thetvdb_index = thetvdb_index.sort({ a, b -> a[0] <=> b[0] } as Comparator)
// join and sort // join and sort
def thetvdb_txt = thetvdb_index.groupBy{ it[0] }.findResults{ k, v -> ([k.pad(6)] + v*.getAt(1).unique{it.toLowerCase()}).join('\t') } def thetvdb_txt = thetvdb_index.groupBy{ it[0] }.findResults{ k, v -> ([k.pad(6)] + v*.getAt(1).unique{ it.toLowerCase() }).join('\t') }
pack(thetvdb_out, thetvdb_txt) pack(thetvdb_out, thetvdb_txt)
println "TheTVDB Index: " + thetvdb_txt.size() println "TheTVDB Index: " + thetvdb_txt.size()

View File

@ -236,22 +236,46 @@ public enum EpisodeMetrics implements SimilarityMetric {
@Override @Override
public float getSimilarity(Object o1, Object o2) { public float getSimilarity(Object o1, Object o2) {
String[] f1 = getNormalizedEffectiveIdentifiers(o1);
String[] f2 = getNormalizedEffectiveIdentifiers(o2);
// match all fields and average similarity
float max = 0;
for (String s1 : f1) {
for (String s2 : f2) {
max = max(super.getSimilarity(s1, s2), max);
}
}
// normalize absolute similarity to similarity rank (4 ranks in total), // normalize absolute similarity to similarity rank (4 ranks in total),
// so we are less likely to fall for false positives in this pass, and move on to the next one // so we are less likely to fall for false positives in this pass, and move on to the next one
return (float) (floor(super.getSimilarity(o1, o2) * 4) / 4); return (float) (floor(max * 4) / 4);
} }
@Override @Override
protected String normalize(Object object) { protected String normalize(Object object) {
if (object instanceof Episode) { return object.toString();
object = removeTrailingBrackets(((Episode) object).getSeriesName()); }
} else if (object instanceof Movie) {
object = ((Movie) object).getName(); protected String[] getNormalizedEffectiveIdentifiers(Object object) {
} else if (object instanceof File) { List<?> identifiers = getEffectiveIdentifiers(object);
object = getNameWithoutExtension(getRelativePathTail((File) object, 3).getPath()); String[] names = new String[identifiers.size()];
for (int i = 0; i < names.length; i++) {
names[i] = normalizeObject(identifiers.get(i));
} }
// simplify file name, if possible return names;
return normalizeObject(object); }
protected List<?> getEffectiveIdentifiers(Object object) {
if (object instanceof Episode) {
return ((Episode) object).getSeries().getEffectiveNames();
} else if (object instanceof Movie) {
return ((Movie) object).getEffectiveNames();
} else if (object instanceof File) {
return listPathTail((File) object, 3, true);
}
return singletonList(object);
} }
}), }),
@ -300,9 +324,16 @@ public enum EpisodeMetrics implements SimilarityMetric {
} }
} else if (object instanceof File) { } else if (object instanceof File) {
object = ((File) object).getName(); // try to narrow down on series name object = ((File) object).getName(); // try to narrow down on series name
String sn = seriesNameMatcher.matchByEpisodeIdentifier(object.toString());
if (sn != null) { try {
object = sn; object = resolveSeriesDirectMapping((String) object);
} catch (IOException e) {
Logger.getLogger(EpisodeMetrics.class.getName()).log(Level.WARNING, e.getMessage());
}
String snm = seriesNameMatcher.matchByEpisodeIdentifier((String) object);
if (snm != null) {
object = snm;
} }
} }
@ -318,6 +349,16 @@ public enum EpisodeMetrics implements SimilarityMetric {
// simplify file name, if possible // simplify file name, if possible
return normalizeObject(object); return normalizeObject(object);
} }
protected String resolveSeriesDirectMapping(String input) throws IOException {
for (Pattern it : releaseInfo.getSeriesDirectMappings().keySet()) {
Matcher m = it.matcher(input);
if (m.find()) {
return m.replaceAll(releaseInfo.getSeriesDirectMappings().get(it));
}
}
return input;
}
}), }),
// Match by generic name similarity (absolute) // Match by generic name similarity (absolute)