@orama/plugin-match-highlight/dist/index.cjs.map

Orama plugin for search match highlighting

{"version":3,"sources":["../src/index.ts"],"sourcesContent":["import {\n  AnyDocument,\n  AnyOrama,\n  Language,\n  RawData,\n  Result,\n  Results,\n  SearchParamsFullText,\n  TypedDocument,\n  load,\n  save,\n  search\n} from '@orama/orama'\nimport { boundedLevenshtein } from '@orama/orama/internals'\n\nexport interface Position {\n  start: number\n  length: number\n}\n\nexport type OramaWithHighlight<T extends AnyOrama> = T & {\n  data: { positions: Record<string, Record<string, Record<string, Position[]>>> }\n}\n\nexport type ResultWithPositions<ResultDocument> = Result<ResultDocument> & {\n  positions: Record<string, Record<string, Position[]>>\n}\n\nexport type SearchResultWithHighlight<ResultDocument> = Omit<Results<ResultDocument>, 'hits'> & {\n  hits: ResultWithPositions<ResultDocument>[]\n}\n\nexport type RawDataWithPositions = RawData & {\n  positions: Record<string, Record<string, Record<string, Position[]>>>\n}\n\nexport async function afterInsert<T extends AnyOrama>(orama: T, id: string): Promise<void> {\n  if (!('positions' in orama.data)) {\n    Object.assign(orama.data, { positions: {} })\n  }\n\n  await recursivePositionInsertion(\n    orama as OramaWithHighlight<T>,\n    (await orama.documentsStore.get(orama.data.docs, id))!,\n    id\n  )\n}\n\nconst wordRegEx = /[\\p{L}0-9_'-]+/gimu\n\nasync function recursivePositionInsertion<T extends AnyOrama, ResultDocument = TypedDocument<T>>(\n  orama: OramaWithHighlight<T>,\n  doc: ResultDocument,\n  id: string,\n  prefix = '',\n  schema: T['schema'] = orama.schema\n): Promise<void> {\n  orama.data.positions[id] = Object.create(null)\n  for (const key of Object.keys(doc as object) as Array<keyof ResultDocument>) {\n    const isNested = typeof doc[key] === 'object'\n    const isSchemaNested = typeof schema[key] === 'object'\n    const propName = `${prefix}${String(key)}`\n    if (isNested && key in schema && isSchemaNested) {\n      recursivePositionInsertion(orama, doc[key], id, propName + '.', schema[key])\n    }\n    if (!(typeof doc[key] === 'string' && key in schema && !isSchemaNested)) {\n      continue\n    }\n    orama.data.positions[id][propName] = Object.create(null)\n    const text = doc[key] as string\n    let regExResult: RegExpExecArray | null\n    while ((regExResult = wordRegEx.exec(text)) !== null) {\n      const word = regExResult[0].toLowerCase()\n      const key = `${orama.tokenizer.language}:${word}`\n      let token: string\n      if (orama.tokenizer.normalizationCache.has(key)) {\n        token = orama.tokenizer.normalizationCache.get(key)!\n      } else {\n        ;[token] = orama.tokenizer.tokenize(word)\n        orama.tokenizer.normalizationCache.set(key, token)\n      }\n      if (!Array.isArray(orama.data.positions[id][propName][token])) {\n        orama.data.positions[id][propName][token] = []\n      }\n      const start = regExResult.index\n      const length = regExResult[0].length\n      orama.data.positions[id][propName][token].push({ start, length })\n    }\n  }\n}\n\nexport async function searchWithHighlight<T extends AnyOrama, ResultDocument = TypedDocument<T>>(\n  orama: T,\n  params: SearchParamsFullText<T, ResultDocument>,\n  language?: Language\n): Promise<SearchResultWithHighlight<ResultDocument>> {\n  const result = await search(orama, params, language)\n  const queryTokens: string[] = orama.tokenizer.tokenize(params.term ?? '', language)\n\n  const hitsWithPosition: ResultWithPositions<ResultDocument>[] = []\n  for (const hit of result.hits) {\n    const hitPositions = Object.entries<any>((orama as OramaWithHighlight<T>).data.positions[hit.id])\n\n    const hits: AnyDocument[] = []\n    for (const [propName, tokens] of hitPositions) {\n      const matchWithSearchTokens: [string, unknown][] = []\n\n      const tokenEntries = Object.entries(tokens)\n      for (const tokenEntry of tokenEntries) {\n        const [token] = tokenEntry\n\n        for (const queryToken of queryTokens) {\n          if (params.tolerance) {\n            const distance = boundedLevenshtein(token, queryToken, params.tolerance)\n            if (distance.isBounded) {\n              matchWithSearchTokens.push(tokenEntry)\n              break\n            }\n          } else if (token.startsWith(queryToken)) {\n            matchWithSearchTokens.push(tokenEntry)\n            break\n          }\n        }\n      }\n      hits.push([propName, Object.fromEntries(matchWithSearchTokens)])\n    }\n\n    hitsWithPosition.push(Object.assign(hit, { positions: Object.fromEntries(hits) }))\n  }\n\n  result.hits = hitsWithPosition\n\n  return result as SearchResultWithHighlight<ResultDocument>\n}\n\nexport function saveWithHighlight<T extends AnyOrama>(orama: T): RawDataWithPositions {\n  const data = save(orama)\n\n  return {\n    ...data,\n    positions: (orama as OramaWithHighlight<T>).data.positions\n  }\n}\n\nexport function loadWithHighlight<T extends AnyOrama>(orama: T, raw: RawDataWithPositions): void {\n  load(orama, raw)\n  ;(orama as OramaWithHighlight<T>).data.positions = raw.positions\n}\n"],"mappings":"yaAAA,IAAAA,EAAA,GAAAC,EAAAD,EAAA,iBAAAE,EAAA,sBAAAC,EAAA,sBAAAC,EAAA,wBAAAC,IAAA,eAAAC,EAAAN,GAAA,IAAAO,EAYO,wBACPC,EAAmC,kCAuBnC,eAAsBN,EAAgCO,EAAUC,EAA2B,CACnF,cAAeD,EAAM,MACzB,OAAO,OAAOA,EAAM,KAAM,CAAE,UAAW,CAAC,CAAE,CAAC,EAG7C,MAAME,EACJF,EACC,MAAMA,EAAM,eAAe,IAAIA,EAAM,KAAK,KAAMC,CAAE,EACnDA,CACF,CACF,CAEA,IAAME,EAAY,qBAElB,eAAeD,EACbF,EACAI,EACAH,EACAI,EAAS,GACTC,EAAsBN,EAAM,OACb,CACfA,EAAM,KAAK,UAAUC,CAAE,EAAI,OAAO,OAAO,IAAI,EAC7C,QAAWM,KAAO,OAAO,KAAKH,CAAa,EAAkC,CAC3E,IAAMI,EAAW,OAAOJ,EAAIG,CAAG,GAAM,SAC/BE,EAAiB,OAAOH,EAAOC,CAAG,GAAM,SACxCG,EAAW,GAAGL,CAAM,GAAG,OAAOE,CAAG,CAAC,GAIxC,GAHIC,GAAYD,KAAOD,GAAUG,GAC/BP,EAA2BF,EAAOI,EAAIG,CAAG,EAAGN,EAAIS,EAAW,IAAKJ,EAAOC,CAAG,CAAC,EAEzE,EAAE,OAAOH,EAAIG,CAAG,GAAM,UAAYA,KAAOD,GAAU,CAACG,GACtD,SAEFT,EAAM,KAAK,UAAUC,CAAE,EAAES,CAAQ,EAAI,OAAO,OAAO,IAAI,EACvD,IAAMC,EAAOP,EAAIG,CAAG,EAChBK,EACJ,MAAQA,EAAcT,EAAU,KAAKQ,CAAI,KAAO,MAAM,CACpD,IAAME,EAAOD,EAAY,CAAC,EAAE,YAAY,EAClCL,EAAM,GAAGP,EAAM,UAAU,QAAQ,IAAIa,CAAI,GAC3CC,EACAd,EAAM,UAAU,mBAAmB,IAAIO,CAAG,EAC5CO,EAAQd,EAAM,UAAU,mBAAmB,IAAIO,CAAG,GAEjD,CAACO,CAAK,EAAId,EAAM,UAAU,SAASa,CAAI,EACxCb,EAAM,UAAU,mBAAmB,IAAIO,EAAKO,CAAK,GAE9C,MAAM,QAAQd,EAAM,KAAK,UAAUC,CAAE,EAAES,CAAQ,EAAEI,CAAK,CAAC,IAC1Dd,EAAM,KAAK,UAAUC,CAAE,EAAES,CAAQ,EAAEI,CAAK,EAAI,CAAC,GAE/C,IAAMC,EAAQH,EAAY,MACpBI,EAASJ,EAAY,CAAC,EAAE,OAC9BZ,EAAM,KAAK,UAAUC,CAAE,EAAES,CAAQ,EAAEI,CAAK,EAAE,KAAK,CAAE,MAAAC,EAAO,OAAAC,CAAO,CAAC,CAClE,CACF,CACF,CAEA,eAAsBpB,EACpBI,EACAiB,EACAC,EACoD,CACpD,IAAMC,EAAS,QAAM,UAAOnB,EAAOiB,EAAQC,CAAQ,EAC7CE,EAAwBpB,EAAM,UAAU,SAASiB,EAAO,MAAQ,GAAIC,CAAQ,EAE5EG,EAA0D,CAAC,EACjE,QAAWC,KAAOH,EAAO,KAAM,CAC7B,IAAMI,EAAe,OAAO,QAAcvB,EAAgC,KAAK,UAAUsB,EAAI,EAAE,CAAC,EAE1FE,EAAsB,CAAC,EAC7B,OAAW,CAACd,EAAUe,CAAM,IAAKF,EAAc,CAC7C,IAAMG,EAA6C,CAAC,EAE9CC,EAAe,OAAO,QAAQF,CAAM,EAC1C,QAAWG,KAAcD,EAAc,CACrC,GAAM,CAACb,CAAK,EAAIc,EAEhB,QAAWC,KAAcT,EACvB,GAAIH,EAAO,WAET,MADiB,sBAAmBH,EAAOe,EAAYZ,EAAO,SAAS,EAC1D,UAAW,CACtBS,EAAsB,KAAKE,CAAU,EACrC,KACF,UACSd,EAAM,WAAWe,CAAU,EAAG,CACvCH,EAAsB,KAAKE,CAAU,EACrC,KACF,CAEJ,CACAJ,EAAK,KAAK,CAACd,EAAU,OAAO,YAAYgB,CAAqB,CAAC,CAAC,CACjE,CAEAL,EAAiB,KAAK,OAAO,OAAOC,EAAK,CAAE,UAAW,OAAO,YAAYE,CAAI,CAAE,CAAC,CAAC,CACnF,CAEA,OAAAL,EAAO,KAAOE,EAEPF,CACT,CAEO,SAASxB,EAAsCK,EAAgC,CAGpF,MAAO,CACL,MAHW,QAAKA,CAAK,EAIrB,UAAYA,EAAgC,KAAK,SACnD,CACF,CAEO,SAASN,EAAsCM,EAAU8B,EAAiC,IAC/F,QAAK9B,EAAO8B,CAAG,EACb9B,EAAgC,KAAK,UAAY8B,EAAI,SACzD","names":["src_exports","__export","afterInsert","loadWithHighlight","saveWithHighlight","searchWithHighlight","__toCommonJS","import_orama","import_internals","orama","id","recursivePositionInsertion","wordRegEx","doc","prefix","schema","key","isNested","isSchemaNested","propName","text","regExResult","word","token","start","length","params","language","result","queryTokens","hitsWithPosition","hit","hitPositions","hits","tokens","matchWithSearchTokens","tokenEntries","tokenEntry","queryToken","raw"]}