Mercurial > hg > svcore
diff transform/TransformFactory.cpp @ 1842:19fa7bf208d8
Ensure search results that match the whole "phrase" get higher scores than any others
author | Chris Cannam |
---|---|
date | Fri, 17 Apr 2020 17:45:15 +0100 |
parents | 5f8fbbde08ff |
children | 5b1b03c1d8d4 |
line wrap: on
line diff
--- a/transform/TransformFactory.cpp Tue Apr 14 10:12:07 2020 +0100 +++ b/transform/TransformFactory.cpp Fri Apr 17 17:45:15 2020 +0100 @@ -1110,11 +1110,36 @@ { populateTransforms(); + SearchResults results = searchUnadjusted(keywords); + if (keywords.size() > 1) { - // Additional score for all keywords in a row - keywords.push_back(keywords.join(" ")); + + // If there are any hits for all keywords in a row, put them + // in (replacing previous hits for the same transforms) but + // ensure they score more than any of the others + + int maxScore = 0; + for (auto r: results) { + if (r.second.score > maxScore) { + maxScore = r.second.score; + } + } + + QStringList oneBigKeyword; + oneBigKeyword << keywords.join(" "); + SearchResults oneBigKeywordResults = searchUnadjusted(oneBigKeyword); + for (auto r: oneBigKeywordResults) { + results[r.first] = r.second; + results[r.first].score += maxScore; + } } + return results; +} + +TransformFactory::SearchResults +TransformFactory::searchUnadjusted(QStringList keywords) +{ SearchResults results; TextMatcher matcher; @@ -1177,6 +1202,26 @@ if (match.score > 0) results[i->first] = match; } +#ifdef DEBUG_TRANSFORM_FACTORY + SVCERR << "TransformFactory::search: keywords are: " << keywords.join(", ") + << endl; + int n = int(results.size()), i = 1; + SVCERR << "TransformFactory::search: results (" << n << "):" << endl; + + for (const auto &r: results) { + QStringList frags; + for (const auto &f: r.second.fragments) { + frags << QString("{\"%1\": \"%2\"}").arg(f.first).arg(f.second); + } + SVCERR << "[" << i << "/" << n << "] id " << r.first + << ": score " << r.second.score + << ", key " << r.second.key << ", fragments " + << frags.join(";") << endl; + ++i; + } + SVCERR << endl; +#endif + return results; }