diff options
Diffstat (limited to 'src/static/js/search-worker.js')
-rw-r--r-- | src/static/js/search-worker.js | 178 |
1 files changed, 106 insertions, 72 deletions
diff --git a/src/static/js/search-worker.js b/src/static/js/search-worker.js index 1b4684ad..c3002b18 100644 --- a/src/static/js/search-worker.js +++ b/src/static/js/search-worker.js @@ -371,58 +371,75 @@ function postActionResult(id, status, value) { } function performSearchAction({query, options}) { - const {generic, ...otherIndexes} = indexes; + const {queriedKind} = processTerms(query); + const genericResults = queryGenericIndex(query, options); + const verbatimResults = queryVerbatimIndex(query, options); - const genericResults = - queryGenericIndex(generic, query, options); + const verbatimIDs = + new Set(verbatimResults?.map(result => result.id)); - const otherResults = - withEntries(otherIndexes, entries => entries - .map(([indexName, index]) => [ - indexName, - index.search(query, options), - ])); + const commonResults = + (verbatimResults && genericResults + ? genericResults + .filter(({id}) => verbatimIDs.has(id)) + : verbatimResults ?? genericResults); return { - generic: genericResults, - ...otherResults, + results: commonResults, + queriedKind, }; } -function queryGenericIndex(index, query, options) { - const interestingFieldCombinations = [ - ['primaryName', 'parentName', 'groups'], - ['primaryName', 'parentName'], - ['primaryName', 'groups', 'contributors'], - ['primaryName', 'groups', 'artTags'], - ['primaryName', 'groups'], - ['primaryName', 'contributors'], - ['primaryName', 'artTags'], - ['parentName', 'groups', 'artTags'], - ['parentName', 'artTags'], - ['groups', 'contributors'], - ['groups', 'artTags'], - - // This prevents just matching *everything* tagged "john" if you - // only search "john", but it actually supports matching more than - // *two* tags at once: "john rose lowas" works! This is thanks to - // flexsearch matching multiple field values in a single query. - ['artTags', 'artTags'], - - ['contributors', 'parentName'], - ['contributors', 'groups'], - ['primaryName', 'contributors'], - ['primaryName'], - ]; +const interestingFieldCombinations = [ + ['primaryName', 'parentName', 'groups'], + ['primaryName', 'parentName'], + ['primaryName', 'groups', 'contributors'], + ['primaryName', 'groups', 'artTags'], + ['primaryName', 'groups'], + ['primaryName', 'contributors'], + ['primaryName', 'artTags'], + ['parentName', 'groups', 'artTags'], + ['parentName', 'artTags'], + ['groups', 'contributors'], + ['groups', 'artTags'], + + // This prevents just matching *everything* tagged "john" if you + // only search "john", but it actually supports matching more than + // *two* tags at once: "john rose lowas" works! This is thanks to + // flexsearch matching multiple field values in a single query. + ['artTags', 'artTags'], + + ['contributors', 'parentName'], + ['contributors', 'groups'], + ['primaryName', 'contributors'], + ['primaryName'], +]; + +function queryGenericIndex(query, options) { + return queryIndex({ + indexKey: 'generic', + termsKey: 'genericTerms', + }, query, options); +} + +function queryVerbatimIndex(query, options) { + return queryIndex({ + indexKey: 'verbatim', + termsKey: 'verbatimTerms', + }, query, options); +} +function queryIndex({termsKey, indexKey}, query, options) { const interestingFields = unique(interestingFieldCombinations.flat()); - const {genericTerms, queriedKind} = + const {[termsKey]: terms, queriedKind} = processTerms(query); + if (empty(terms)) return null; + const particles = - particulate(genericTerms); + particulate(terms); const groupedParticles = groupArray(particles, ({length}) => length); @@ -437,7 +454,7 @@ function queryGenericIndex(index, query, options) { query: values, })); - const boilerplate = queryBoilerplate(index); + const boilerplate = queryBoilerplate(indexes[indexKey]); const particleResults = Object.fromEntries( @@ -459,62 +476,73 @@ function queryGenericIndex(index, query, options) { ])), ])); - const results = new Set(); + let matchedResults = new Set(); for (const interestingFieldCombination of interestingFieldCombinations) { for (const query of queriesBy(interestingFieldCombination)) { - const idToMatchingFieldsMap = new Map(); - for (const {field, query: fieldQuery} of query) { - for (const id of particleResults[field][fieldQuery]) { - if (idToMatchingFieldsMap.has(id)) { - idToMatchingFieldsMap.get(id).push(field); - } else { - idToMatchingFieldsMap.set(id, [field]); - } - } - } + const [firstQueryFieldLine, ...restQueryFieldLines] = query; const commonAcrossFields = - Array.from(idToMatchingFieldsMap.entries()) - .filter(([id, matchingFields]) => - matchingFields.length === interestingFieldCombination.length) - .map(([id]) => id); + new Set( + particleResults + [firstQueryFieldLine.field] + [firstQueryFieldLine.query]); + + for (const currQueryFieldLine of restQueryFieldLines) { + const tossResults = new Set(commonAcrossFields); + + const keepResults = + particleResults + [currQueryFieldLine.field] + [currQueryFieldLine.query]; + + for (const result of keepResults) { + tossResults.delete(result); + } + + for (const result of tossResults) { + commonAcrossFields.delete(result); + } + } for (const result of commonAcrossFields) { - results.add(result); + matchedResults.add(result); } } } - const constituted = - boilerplate.constitute(results); + matchedResults = Array.from(matchedResults); - const constitutedAndFiltered = - constituted - .filter(({id}) => - (queriedKind - ? id.split(':')[0] === queriedKind - : true)); + const filteredResults = + (queriedKind + ? matchedResults.filter(id => id.split(':')[0] === queriedKind) + : matchedResults); - return constitutedAndFiltered; + const constitutedResults = + boilerplate.constitute(filteredResults); + + return constitutedResults; } function processTerms(query) { const kindTermSpec = [ - {kind: 'album', terms: ['album']}, - {kind: 'artist', terms: ['artist']}, - {kind: 'flash', terms: ['flash']}, - {kind: 'group', terms: ['group']}, - {kind: 'tag', terms: ['art tag', 'tag']}, - {kind: 'track', terms: ['track']}, + {kind: 'album', terms: ['album', 'albums']}, + {kind: 'artist', terms: ['artist', 'artists']}, + {kind: 'flash', terms: ['flash', 'flashes']}, + {kind: 'group', terms: ['group', 'groups']}, + {kind: 'tag', terms: ['art tag', 'art tags', 'tag', 'tags']}, + {kind: 'track', terms: ['track', 'tracks']}, ]; const genericTerms = []; + const verbatimTerms = []; let queriedKind = null; const termRegexp = new RegExp( - String.raw`(?<kind>${kindTermSpec.flatMap(spec => spec.terms).join('|')})` + + String.raw`(?<kind>(?<=^|\s)(?:${kindTermSpec.flatMap(spec => spec.terms).join('|')})(?=$|\s))` + + String.raw`|(?<=^|\s)(?<quote>["'])(?<regularVerbatim>.+?)\k<quote>(?=$|\s)` + + String.raw`|(?<=^|\s)[“”‘’](?<curlyVerbatim>.+?)[“”‘’](?=$|\s)` + String.raw`|[^\s\-]+`, 'gi'); @@ -530,10 +558,16 @@ function processTerms(query) { continue; } + const verbatim = groups.regularVerbatim || groups.curlyVerbatim; + if (verbatim) { + verbatimTerms.push(verbatim); + continue; + } + genericTerms.push(match[0]); } - return {genericTerms, queriedKind}; + return {genericTerms, verbatimTerms, queriedKind}; } function particulate(terms) { |