eyecite-ts 0.1.0

package/dist/data/index.d.cts

@@ -0,0 +1,116 @@
+//#region src/data/reporters.d.ts
+/**
+* Reporter database integration for citation validation
+*
+* This module provides lazy-loadable access to the reporters-db database,
+* containing 1200+ court reporters with variant forms. The library works
+* in degraded mode (pattern-based extraction only) if reporters are not loaded.
+*
+* @example
+* // Degraded mode: extraction works without reporter data
+* const citations = await extract(text)
+*
+* @example
+* // Full mode: load reporters for validation
+* await loadReporters()
+* const citations = await extract(text)  // Now with reporter validation
+*/
+/**
+* Edition entry from reporters-db
+*
+* Represents a specific edition of a reporter with start/end dates.
+*/
+interface ReporterEdition {
+  /** Start date in ISO 8601 format */
+  start: string | null;
+  /** End date in ISO 8601 format (null if ongoing) */
+  end: string | null;
+}
+/**
+* Reporter entry from reporters-db
+*
+* Represents a single court reporter with all metadata needed for
+* citation validation and enrichment.
+*
+* Note: The reporters-db structure has the actual data; this interface
+* represents it flexibly to handle all variations in the JSON.
+*/
+interface ReporterEntry {
+  /** Full reporter name (e.g., "Federal Reporter") */
+  name: string;
+  /** Citation type: state, federal, specialty, neutral, state_regional, etc. */
+  cite_type: string;
+  /** Editions keyed by abbreviation (e.g., {"F.2d": {...}, "F.3d": {...}}) */
+  editions: Record<string, ReporterEdition>;
+  /** Variant forms mapped to canonical form (e.g., {"F. 2d": "F.2d"}) */
+  variations?: Record<string, string | undefined>;
+  /** MLZ jurisdiction identifiers (optional) */
+  mlz_jurisdiction?: string[];
+  /** Publisher (optional) */
+  publisher?: string;
+  /** Notes (optional) */
+  notes?: string;
+}
+/**
+* In-memory reporter database with fast O(1) lookup
+*
+* Uses Map-based indexing for case-insensitive abbreviation lookup.
+* All variant forms are indexed to support fuzzy matching.
+*/
+interface ReportersDatabase {
+  /** Fast O(1) lookup by abbreviation (lowercase normalized keys) */
+  byAbbreviation: Map<string, ReporterEntry[]>;
+  /** All reporters (for iteration/filtering) */
+  all: ReporterEntry[];
+}
+/**
+* Load reporter database asynchronously with lazy loading
+*
+* Dynamic import prevents loading 1200+ reporters until explicitly requested.
+* Result is cached after first load for subsequent calls.
+*
+* @returns Promise resolving to indexed reporter database
+*
+* @example
+* const db = await loadReporters()
+* const reporters = db.byAbbreviation.get('f.2d')  // Fast O(1) lookup
+*/
+declare function loadReporters(): Promise<ReportersDatabase>;
+/**
+* Get cached reporter database synchronously (degraded mode support)
+*
+* Returns null if reporters not loaded yet. This enables the library to
+* work in degraded mode without reporter validation.
+*
+* @returns Cached database or null if not loaded
+*
+* @example
+* const db = getReportersSync()
+* if (db) {
+*   // Full mode: validate citations
+* } else {
+*   // Degraded mode: extract without validation
+* }
+*/
+declare function getReportersSync(): ReportersDatabase | null;
+/**
+* Find reporters by abbreviation (case-insensitive)
+*
+* Loads reporter database if not already loaded. Returns all reporters
+* matching the abbreviation (including variant forms).
+*
+* @param abbr - Reporter abbreviation to look up
+* @returns Promise resolving to matching reporters (empty array if none)
+*
+* @example
+* const reporters = await findReportersByAbbreviation('F.2d')
+* // [{ abbreviation: 'F.2d', name: 'Federal Reporter, Second Series', ... }]
+*
+* @example
+* const unknown = await findReportersByAbbreviation('NONEXISTENT')
+* // [] (empty array, not error)
+*/
+declare function findReportersByAbbreviation(abbr: string): Promise<ReporterEntry[]>;
+//#endregion
+export { ReporterEdition, ReporterEntry, ReportersDatabase, findReportersByAbbreviation, getReportersSync, loadReporters };
+//# sourceMappingURL=index.d.cts.map

package/dist/data/index.d.cts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.cts","names":[],"sources":["../../src/data/reporters.ts"],"mappings":";;AAsBA;;;;;AAgBA;;;;;;;;;;;;;;;UAhBiB,eAAA;;EAEf,KAAA;;EAEA,GAAA;AAAA;;;;;;;;;;UAYe,aAAA;;EAEf,IAAA;;EAEA,SAAA;EA2CF;EAzCE,QAAA,EAAU,MAAA,SAAe,eAAA;;EAEzB,UAAA,GAAa,MAAA;EAuCgC;EArC7C,gBAAA;EAwGc;EAtGd,SAAA;EAsGkC;EApGlC,KAAA;AAAA;;;;;;;UASe,iBAAA;EAkHN;EAhHT,cAAA,EAAgB,GAAA,SAAY,aAAA;;EAE5B,GAAA,EAAK,aAAA;AAAA;;;;;;;;;;;;;iBAoBe,aAAA,CAAA,GAAiB,OAAA,CAAQ,iBAAA;;;;;;;;;;;;;;;;;iBAmE/B,gBAAA,CAAA,GAAoB,iBAAA;;;;;;;;;;;;;;;;;;iBAqBd,2BAAA,CACpB,IAAA,WACC,OAAA,CAAQ,aAAA"}

package/dist/data/index.d.mts

@@ -0,0 +1,116 @@
+//#region src/data/reporters.d.ts
+/**
+* Reporter database integration for citation validation
+*
+* This module provides lazy-loadable access to the reporters-db database,
+* containing 1200+ court reporters with variant forms. The library works
+* in degraded mode (pattern-based extraction only) if reporters are not loaded.
+*
+* @example
+* // Degraded mode: extraction works without reporter data
+* const citations = await extract(text)
+*
+* @example
+* // Full mode: load reporters for validation
+* await loadReporters()
+* const citations = await extract(text)  // Now with reporter validation
+*/
+/**
+* Edition entry from reporters-db
+*
+* Represents a specific edition of a reporter with start/end dates.
+*/
+interface ReporterEdition {
+  /** Start date in ISO 8601 format */
+  start: string | null;
+  /** End date in ISO 8601 format (null if ongoing) */
+  end: string | null;
+}
+/**
+* Reporter entry from reporters-db
+*
+* Represents a single court reporter with all metadata needed for
+* citation validation and enrichment.
+*
+* Note: The reporters-db structure has the actual data; this interface
+* represents it flexibly to handle all variations in the JSON.
+*/
+interface ReporterEntry {
+  /** Full reporter name (e.g., "Federal Reporter") */
+  name: string;
+  /** Citation type: state, federal, specialty, neutral, state_regional, etc. */
+  cite_type: string;
+  /** Editions keyed by abbreviation (e.g., {"F.2d": {...}, "F.3d": {...}}) */
+  editions: Record<string, ReporterEdition>;
+  /** Variant forms mapped to canonical form (e.g., {"F. 2d": "F.2d"}) */
+  variations?: Record<string, string | undefined>;
+  /** MLZ jurisdiction identifiers (optional) */
+  mlz_jurisdiction?: string[];
+  /** Publisher (optional) */
+  publisher?: string;
+  /** Notes (optional) */
+  notes?: string;
+}
+/**
+* In-memory reporter database with fast O(1) lookup
+*
+* Uses Map-based indexing for case-insensitive abbreviation lookup.
+* All variant forms are indexed to support fuzzy matching.
+*/
+interface ReportersDatabase {
+  /** Fast O(1) lookup by abbreviation (lowercase normalized keys) */
+  byAbbreviation: Map<string, ReporterEntry[]>;
+  /** All reporters (for iteration/filtering) */
+  all: ReporterEntry[];
+}
+/**
+* Load reporter database asynchronously with lazy loading
+*
+* Dynamic import prevents loading 1200+ reporters until explicitly requested.
+* Result is cached after first load for subsequent calls.
+*
+* @returns Promise resolving to indexed reporter database
+*
+* @example
+* const db = await loadReporters()
+* const reporters = db.byAbbreviation.get('f.2d')  // Fast O(1) lookup
+*/
+declare function loadReporters(): Promise<ReportersDatabase>;
+/**
+* Get cached reporter database synchronously (degraded mode support)
+*
+* Returns null if reporters not loaded yet. This enables the library to
+* work in degraded mode without reporter validation.
+*
+* @returns Cached database or null if not loaded
+*
+* @example
+* const db = getReportersSync()
+* if (db) {
+*   // Full mode: validate citations
+* } else {
+*   // Degraded mode: extract without validation
+* }
+*/
+declare function getReportersSync(): ReportersDatabase | null;
+/**
+* Find reporters by abbreviation (case-insensitive)
+*
+* Loads reporter database if not already loaded. Returns all reporters
+* matching the abbreviation (including variant forms).
+*
+* @param abbr - Reporter abbreviation to look up
+* @returns Promise resolving to matching reporters (empty array if none)
+*
+* @example
+* const reporters = await findReportersByAbbreviation('F.2d')
+* // [{ abbreviation: 'F.2d', name: 'Federal Reporter, Second Series', ... }]
+*
+* @example
+* const unknown = await findReportersByAbbreviation('NONEXISTENT')
+* // [] (empty array, not error)
+*/
+declare function findReportersByAbbreviation(abbr: string): Promise<ReporterEntry[]>;
+//#endregion
+export { ReporterEdition, ReporterEntry, ReportersDatabase, findReportersByAbbreviation, getReportersSync, loadReporters };
+//# sourceMappingURL=index.d.mts.map

package/dist/data/index.d.mts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.mts","names":[],"sources":["../../src/data/reporters.ts"],"mappings":";;AAsBA;;;;;AAgBA;;;;;;;;;;;;;;;UAhBiB,eAAA;;EAEf,KAAA;;EAEA,GAAA;AAAA;;;;;;;;;;UAYe,aAAA;;EAEf,IAAA;;EAEA,SAAA;EA2CF;EAzCE,QAAA,EAAU,MAAA,SAAe,eAAA;;EAEzB,UAAA,GAAa,MAAA;EAuCgC;EArC7C,gBAAA;EAwGc;EAtGd,SAAA;EAsGkC;EApGlC,KAAA;AAAA;;;;;;;UASe,iBAAA;EAkHN;EAhHT,cAAA,EAAgB,GAAA,SAAY,aAAA;;EAE5B,GAAA,EAAK,aAAA;AAAA;;;;;;;;;;;;;iBAoBe,aAAA,CAAA,GAAiB,OAAA,CAAQ,iBAAA;;;;;;;;;;;;;;;;;iBAmE/B,gBAAA,CAAA,GAAoB,iBAAA;;;;;;;;;;;;;;;;;;iBAqBd,2BAAA,CACpB,IAAA,WACC,OAAA,CAAQ,aAAA"}

package/dist/data/index.mjs

@@ -0,0 +1,2 @@
+let e=null;async function t(){if(e)return e;let t=await import(`../../data/reporters.json`,{assert:{type:`json`}}),n=new Map,r=[],i=t.default||t;for(let[e,t]of Object.entries(i))for(let e of t){r.push(e);for(let t of Object.keys(e.editions)){let r=t.toLowerCase();n.has(r)||n.set(r,[]),n.get(r).push(e)}for(let[t,r]of Object.entries(e.variations||{})){let r=t.toLowerCase();n.has(r)||n.set(r,[]),n.get(r).push(e)}}return e={byAbbreviation:n,all:r},e}function n(){return e}async function r(e){return(await t()).byAbbreviation.get(e.toLowerCase())??[]}export{r as findReportersByAbbreviation,n as getReportersSync,t as loadReporters};
+//# sourceMappingURL=index.mjs.map

package/dist/data/index.mjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.mjs","names":[],"sources":["../../src/data/reporters.ts"],"sourcesContent":["/**\n * Reporter database integration for citation validation\n *\n * This module provides lazy-loadable access to the reporters-db database,\n * containing 1200+ court reporters with variant forms. The library works\n * in degraded mode (pattern-based extraction only) if reporters are not loaded.\n *\n * @example\n * // Degraded mode: extraction works without reporter data\n * const citations = await extract(text)\n *\n * @example\n * // Full mode: load reporters for validation\n * await loadReporters()\n * const citations = await extract(text)  // Now with reporter validation\n */\n\n/**\n * Edition entry from reporters-db\n *\n * Represents a specific edition of a reporter with start/end dates.\n */\nexport interface ReporterEdition {\n  /** Start date in ISO 8601 format */\n  start: string | null\n  /** End date in ISO 8601 format (null if ongoing) */\n  end: string | null\n}\n\n/**\n * Reporter entry from reporters-db\n *\n * Represents a single court reporter with all metadata needed for\n * citation validation and enrichment.\n *\n * Note: The reporters-db structure has the actual data; this interface\n * represents it flexibly to handle all variations in the JSON.\n */\nexport interface ReporterEntry {\n  /** Full reporter name (e.g., \"Federal Reporter\") */\n  name: string\n  /** Citation type: state, federal, specialty, neutral, state_regional, etc. */\n  cite_type: string\n  /** Editions keyed by abbreviation (e.g., {\"F.2d\": {...}, \"F.3d\": {...}}) */\n  editions: Record<string, ReporterEdition>\n  /** Variant forms mapped to canonical form (e.g., {\"F. 2d\": \"F.2d\"}) */\n  variations?: Record<string, string | undefined>\n  /** MLZ jurisdiction identifiers (optional) */\n  mlz_jurisdiction?: string[]\n  /** Publisher (optional) */\n  publisher?: string\n  /** Notes (optional) */\n  notes?: string\n}\n\n/**\n * In-memory reporter database with fast O(1) lookup\n *\n * Uses Map-based indexing for case-insensitive abbreviation lookup.\n * All variant forms are indexed to support fuzzy matching.\n */\nexport interface ReportersDatabase {\n  /** Fast O(1) lookup by abbreviation (lowercase normalized keys) */\n  byAbbreviation: Map<string, ReporterEntry[]>\n  /** All reporters (for iteration/filtering) */\n  all: ReporterEntry[]\n}\n\n/**\n * Cached database instance (null until loadReporters() called)\n */\nlet cached: ReportersDatabase | null = null\n\n/**\n * Load reporter database asynchronously with lazy loading\n *\n * Dynamic import prevents loading 1200+ reporters until explicitly requested.\n * Result is cached after first load for subsequent calls.\n *\n * @returns Promise resolving to indexed reporter database\n *\n * @example\n * const db = await loadReporters()\n * const reporters = db.byAbbreviation.get('f.2d')  // Fast O(1) lookup\n */\nexport async function loadReporters(): Promise<ReportersDatabase> {\n  if (cached) return cached\n\n  // Dynamic import prevents loading until requested (keeps core bundle small)\n  const data = await import(\"../../data/reporters.json\", {\n    assert: { type: \"json\" },\n  })\n\n  const byAbbreviation = new Map<string, ReporterEntry[]>()\n  const all: ReporterEntry[] = []\n\n  // reporters.json structure: { \"A.\": [...], \"F.2d\": [...], ... }\n  const reportersData = (data.default || data) as Record<\n    string,\n    ReporterEntry[]\n  >\n\n  // Build fast lookup index with lowercase normalization\n  for (const [canonicalAbbr, reporters] of Object.entries(reportersData)) {\n    for (const reporter of reporters) {\n      all.push(reporter)\n\n      // Index by all edition abbreviations\n      for (const editionAbbr of Object.keys(reporter.editions)) {\n        const key = editionAbbr.toLowerCase()\n        if (!byAbbreviation.has(key)) {\n          byAbbreviation.set(key, [])\n        }\n        byAbbreviation.get(key)!.push(reporter)\n      }\n\n      // Index all variations for fuzzy matching\n      for (const [variant, canonical] of Object.entries(\n        reporter.variations || {},\n      )) {\n        const variantKey = variant.toLowerCase()\n        if (!byAbbreviation.has(variantKey)) {\n          byAbbreviation.set(variantKey, [])\n        }\n        byAbbreviation.get(variantKey)!.push(reporter)\n      }\n    }\n  }\n\n  cached = {\n    byAbbreviation,\n    all,\n  }\n  return cached\n}\n\n/**\n * Get cached reporter database synchronously (degraded mode support)\n *\n * Returns null if reporters not loaded yet. This enables the library to\n * work in degraded mode without reporter validation.\n *\n * @returns Cached database or null if not loaded\n *\n * @example\n * const db = getReportersSync()\n * if (db) {\n *   // Full mode: validate citations\n * } else {\n *   // Degraded mode: extract without validation\n * }\n */\nexport function getReportersSync(): ReportersDatabase | null {\n  return cached\n}\n\n/**\n * Find reporters by abbreviation (case-insensitive)\n *\n * Loads reporter database if not already loaded. Returns all reporters\n * matching the abbreviation (including variant forms).\n *\n * @param abbr - Reporter abbreviation to look up\n * @returns Promise resolving to matching reporters (empty array if none)\n *\n * @example\n * const reporters = await findReportersByAbbreviation('F.2d')\n * // [{ abbreviation: 'F.2d', name: 'Federal Reporter, Second Series', ... }]\n *\n * @example\n * const unknown = await findReportersByAbbreviation('NONEXISTENT')\n * // [] (empty array, not error)\n */\nexport async function findReportersByAbbreviation(\n  abbr: string,\n): Promise<ReporterEntry[]> {\n  const db = await loadReporters()\n  return db.byAbbreviation.get(abbr.toLowerCase()) ?? []\n}\n"],"mappings":"AAuEA,IAAI,EAAmC,KAcvC,eAAsB,GAA4C,CAChE,GAAI,EAAQ,OAAO,EAGnB,IAAM,EAAO,MAAM,OAAO,4BAA6B,CACrD,OAAQ,CAAE,KAAM,OAAQ,CACzB,EAEK,EAAiB,IAAI,IACrB,EAAuB,EAAE,CAGzB,EAAiB,EAAK,SAAW,EAMvC,IAAK,GAAM,CAAC,EAAe,KAAc,OAAO,QAAQ,EAAc,CACpE,IAAK,IAAM,KAAY,EAAW,CAChC,EAAI,KAAK,EAAS,CAGlB,IAAK,IAAM,KAAe,OAAO,KAAK,EAAS,SAAS,CAAE,CACxD,IAAM,EAAM,EAAY,aAAa,CAChC,EAAe,IAAI,EAAI,EAC1B,EAAe,IAAI,EAAK,EAAE,CAAC,CAE7B,EAAe,IAAI,EAAI,CAAE,KAAK,EAAS,CAIzC,IAAK,GAAM,CAAC,EAAS,KAAc,OAAO,QACxC,EAAS,YAAc,EAAE,CAC1B,CAAE,CACD,IAAM,EAAa,EAAQ,aAAa,CACnC,EAAe,IAAI,EAAW,EACjC,EAAe,IAAI,EAAY,EAAE,CAAC,CAEpC,EAAe,IAAI,EAAW,CAAE,KAAK,EAAS,EASpD,MAJA,GAAS,CACP,iBACA,MACD,CACM,EAmBT,SAAgB,GAA6C,CAC3D,OAAO,EAoBT,eAAsB,EACpB,EAC0B,CAE1B,OADW,MAAM,GAAe,EACtB,eAAe,IAAI,EAAK,aAAa,CAAC,EAAI,EAAE"}

package/dist/index.cjs

@@ -0,0 +1,2 @@
+Object.defineProperty(exports,Symbol.toStringTag,{value:`Module`});function e(e){return e.replace(/<[^>]+>/g,``)}function t(e){return e.replace(/[\t\n\r]+/g,` `).replace(/ {2,}/g,` `)}function n(e){return e.normalize(`NFKC`)}function r(e){return e.replace(/[\u201C\u201D]/g,`"`).replace(/[\u2018\u2019]/g,`'`)}function i(i,o=[e,t,n,r]){let s=i,c=new Map,l=new Map;for(let e=0;e<=i.length;e++)c.set(e,e),l.set(e,e);for(let e of o){let t=s,n=e(s);if(t!==n){let{newCleanToOriginal:e,newOriginalToClean:r}=a(t,n,c,l);c=e,l=r,s=n}}return{cleaned:s,transformationMap:{cleanToOriginal:c,originalToClean:l},warnings:[]}}function a(e,t,n,r){let i=new Map,a=new Map,o=0,s=0;for(;o<=e.length||s<=t.length;){if(o>=e.length&&s>=t.length){let e=n.get(o)??o;i.set(s,e),a.set(e,s);break}if(o>=e.length){let e=n.get(o)??o;i.set(s,e),s++;continue}if(s>=t.length){let e=n.get(o)??o;a.set(e,s),o++;continue}if(e[o]===t[s]){let e=n.get(o)??o;i.set(s,e),a.set(e,s),o++,s++}else{let r=!1;for(let i=1;i<=20&&!(o+i>=e.length);i++)if(e[o+i]===t[s]){for(let e=0;e<i;e++){let t=n.get(o+e)??o+e;a.set(t,s)}o+=i,r=!0;break}if(r)continue;for(let a=1;a<=20&&!(s+a>=t.length);a++)if(e[o]===t[s+a]){let e=n.get(o)??o;for(let t=0;t<a;t++)i.set(s+t,e);s+=a,r=!0;break}if(r)continue;let c=n.get(o)??o;i.set(s,c),a.set(c,s),o++,s++}}return{newCleanToOriginal:i,newOriginalToClean:a}}const o=[{id:`federal-reporter`,regex:/\b(\d+)\s+(F\.|F\.2d|F\.3d|F\.\s?Supp\.|F\.\s?Supp\.\s?2d|F\.\s?Supp\.\s?3d)\s+(\d+)\b/g,description:`Federal Reporter (F., F.2d, F.3d, F.Supp., etc.)`,type:`case`},{id:`supreme-court`,regex:/\b(\d+)\s+(U\.S\.|S\.\s?Ct\.|L\.\s?Ed\.(?:\s?2d)?)\s+(\d+)\b/g,description:`U.S. Supreme Court reporters`,type:`case`},{id:`state-reporter`,regex:/\b(\d+)\s+([A-Z][A-Za-z\.]+(?:\s?2d|\s?3d)?)\s+(\d+)\b/g,description:`State reporters (broad pattern, validated against reporters-db in Phase 3)`,type:`case`}],s=[{id:`usc`,regex:/\b(\d+)\s+U\.S\.C\.?\s+§+\s*(\d+)\b/g,description:`U.S. Code citations (e.g., "42 U.S.C. § 1983")`,type:`statute`},{id:`state-code`,regex:/\b([A-Z][a-z]+\.?\s+[A-Za-z\.]+\s+Code)\s+§\s*(\d+)\b/g,description:`State code citations (broad pattern, e.g., "Cal. Penal Code § 187")`,type:`statute`}],c=[{id:`law-review`,regex:/\b(\d+)\s+([A-Z][A-Za-z\.\s]+)\s+(\d+)\b/g,description:`Law review citations (e.g., "120 Harv. L. Rev. 500"), validated against journals-db in Phase 3`,type:`journal`}],l=[{id:`westlaw`,regex:/\b(\d{4})\s+WL\s+(\d+)\b/g,description:`WestLaw citations (e.g., "2021 WL 123456")`,type:`neutral`},{id:`lexis`,regex:/\b(\d{4})\s+U\.S\.\s+LEXIS\s+(\d+)\b/g,description:`LexisNexis citations (e.g., "2021 U.S. LEXIS 5000")`,type:`neutral`},{id:`public-law`,regex:/\bPub\.\s?L\.\s?No\.\s?(\d+-\d+)\b/g,description:`Public Law citations (e.g., "Pub. L. No. 117-58")`,type:`publicLaw`},{id:`federal-register`,regex:/\b(\d+)\s+Fed\.\s?Reg\.\s+(\d+)\b/g,description:`Federal Register citations (e.g., "86 Fed. Reg. 12345")`,type:`federalRegister`}],u=[{id:`id`,regex:/\b[Ii]d\.(?:\s+at\s+(\d+))?/g,description:`Id. citations (e.g., "Id." or "Id. at 253")`,type:`case`},{id:`ibid`,regex:/\b[Ii]bid\.(?:\s+at\s+(\d+))?/g,description:`Ibid. citations (e.g., "Ibid." or "Ibid. at 125")`,type:`case`},{id:`supra`,regex:/\b([A-Z][a-zA-Z]+(?:(?:\s+v\.?\s+|\s+)[A-Z][a-zA-Z]+)*),?\s+supra(?:,?\s+at\s+(\d+))?/g,description:`Supra citations (e.g., "Smith, supra" or "Smith, supra, at 460")`,type:`case`},{id:`shortFormCase`,regex:/\b(\d+)\s+([A-Z][A-Za-z.\s]+?(?:\d[a-z])?)\s+at\s+(\d+)\b/g,description:`Short-form case citations (e.g., "500 F.2d at 125")`,type:`case`}];function d(e,t=[...o,...s,...c,...l,...u]){let n=[];for(let r of t)try{let t=e.matchAll(r.regex);for(let e of t)n.push({text:e[0],span:{cleanStart:e.index,cleanEnd:e.index+e[0].length},type:r.type,patternId:r.id})}catch(e){console.warn(`Pattern ${r.id} threw error, skipping:`,e instanceof Error?e.message:String(e));continue}return n.sort((e,t)=>e.span.cleanStart-t.span.cleanStart),n}function f(e,t){let{text:n,span:r}=e,i=/^(\d+)\s+([A-Za-z0-9.\s]+)\s+(\d+)/.exec(n);if(!i)throw Error(`Failed to parse case citation: ${n}`);let a=Number.parseInt(i[1],10),o=i[2].trim(),s=Number.parseInt(i[3],10),c=/,\s*(\d+)/.exec(n),l=c?Number.parseInt(c[1],10):void 0,u=/\((?:[^)]*\s)?(\d{4})\)/.exec(n),d=u?Number.parseInt(u[1],10):void 0,f=/\(([^)]*[A-Za-z][^)]*)\)/.exec(n),p=f?f[1].trim():void 0,m=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,h=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd,g=.5;return`F.,F.2d,F.3d,F.4th,U.S.,S. Ct.,L. Ed.,P.,P.2d,P.3d,A.,A.2d,A.3d,N.E.,N.E.2d,N.E.3d,N.W.,N.W.2d,S.E.,S.E.2d,S.W.,S.W.2d,S.W.3d,So.,So. 2d,So. 3d`.split(`,`).some(e=>o.includes(e))&&(g+=.3),d!==void 0&&d<=new Date().getFullYear()&&(g+=.2),g=Math.min(g,1),{type:`case`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:m,originalEnd:h},confidence:g,matchedText:n,processTimeMs:0,patternsChecked:1,volume:a,reporter:o,page:s,pincite:l,court:p,year:d}}function p(e,t){let{text:n,span:r}=e,i=/^(?:(\d+)\s+)?([A-Za-z.\s]+?)\s*§\s*(\d+[A-Za-z0-9\-]*)/.exec(n);if(!i)throw Error(`Failed to parse statute citation: ${n}`);let a=i[1]?Number.parseInt(i[1],10):void 0,o=i[2].trim(),s=i[3],c=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,l=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd,u=.5;return[`U.S.C.`,`C.F.R.`,`Cal. Civ. Code`,`Cal. Penal Code`,`N.Y. Civ. Prac. L. & R.`,`Tex. Civ. Prac. & Rem. Code`].some(e=>o.includes(e))&&(u+=.3),u=Math.min(u,1),{type:`statute`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:c,originalEnd:l},confidence:u,matchedText:n,processTimeMs:0,patternsChecked:1,title:a,code:o,section:s}}function m(e,t){let{text:n,span:r}=e,i=/^(\d+)\s+([A-Za-z.\s]+?)\s+(\d+)/.exec(n);if(!i)throw Error(`Failed to parse journal citation: ${n}`);let a=Number.parseInt(i[1],10),o=i[2].trim(),s=Number.parseInt(i[3],10),c=/,\s*(\d+)/.exec(n),l=c?Number.parseInt(c[1],10):void 0,u=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,d=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd;return{type:`journal`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:u,originalEnd:d},confidence:.6,matchedText:n,processTimeMs:0,patternsChecked:1,volume:a,journal:o,abbreviation:o,page:s,pincite:l}}function h(e,t){let{text:n,span:r}=e,i=/^(\d{4})\s+(WL|LEXIS|U\.S\.\s+LEXIS)\s+(\d+)/.exec(n);if(!i)throw Error(`Failed to parse neutral citation: ${n}`);let a=Number.parseInt(i[1],10),o=i[2],s=i[3],c=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,l=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd;return{type:`neutral`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:c,originalEnd:l},confidence:1,matchedText:n,processTimeMs:0,patternsChecked:1,year:a,court:o,documentNumber:s}}function g(e,t){let{text:n,span:r}=e,i=/Pub\.\s?L\.(?:\s?No\.)?\s?(\d+)-(\d+)/.exec(n);if(!i)throw Error(`Failed to parse public law citation: ${n}`);let a=Number.parseInt(i[1],10),o=Number.parseInt(i[2],10),s=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,c=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd;return{type:`publicLaw`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:s,originalEnd:c},confidence:.9,matchedText:n,processTimeMs:0,patternsChecked:1,congress:a,lawNumber:o}}function _(e,t){let{text:n,span:r}=e,i=/^(\d+)\s+Fed\.\s?Reg\.\s+(\d+)/.exec(n);if(!i)throw Error(`Failed to parse Federal Register citation: ${n}`);let a=Number.parseInt(i[1],10),o=Number.parseInt(i[2],10),s=/\((?:.*?\s)?(\d{4})\)/.exec(n),c=s?Number.parseInt(s[1],10):void 0,l=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,u=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd;return{type:`federalRegister`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:l,originalEnd:u},confidence:.9,matchedText:n,processTimeMs:0,patternsChecked:1,volume:a,page:o,year:c}}function v(e,t){let{text:n,span:r}=e,i=/[Ii](?:d|bid)\.(?:\s+at\s+(\d+))?/.exec(n);if(!i)throw Error(`Failed to parse Id. citation: ${n}`);let a=i[1]?Number.parseInt(i[1],10):void 0,o=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,s=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd;return{type:`id`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:o,originalEnd:s},confidence:1,matchedText:n,processTimeMs:0,patternsChecked:1,pincite:a}}function y(e,t){let{text:n,span:r}=e,i=/\b([A-Z][a-zA-Z]+(?:(?:\s+v\.?\s+|\s+)[A-Z][a-zA-Z]+)*),?\s+supra(?:,?\s+at\s+(\d+))?/.exec(n);if(!i)throw Error(`Failed to parse supra citation: ${n}`);let a=i[1],o=i[2]?Number.parseInt(i[2],10):void 0,s=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,c=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd;return{type:`supra`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:s,originalEnd:c},confidence:.9,matchedText:n,processTimeMs:0,patternsChecked:1,partyName:a,pincite:o}}function b(e,t){let{text:n,span:r}=e,i=/(\d+)\s+([A-Z][A-Za-z.\s]+?(?:\d[a-z])?)\s+at\s+(\d+)/.exec(n);if(!i)throw Error(`Failed to parse short-form case citation: ${n}`);let a=Number.parseInt(i[1],10),o=i[2].trim(),s=Number.parseInt(i[3],10),c=t.cleanToOriginal.get(r.cleanStart)??r.cleanStart,l=t.cleanToOriginal.get(r.cleanEnd)??r.cleanEnd;return{type:`shortFormCase`,text:n,span:{cleanStart:r.cleanStart,cleanEnd:r.cleanEnd,originalStart:c,originalEnd:l},confidence:.7,matchedText:n,processTimeMs:0,patternsChecked:1,volume:a,reporter:o,pincite:s}}function x(e,t,n=/\n\n+/g){let r=new Map,i=[0],a;for(;(a=n.exec(e))!==null;)i.push(a.index+a[0].length);i.push(e.length);for(let e=0;e<t.length;e++){let n=t[e].span.originalStart,a=0;for(let e=0;e<i.length-1;e++)if(n>=i[e]&&n<i[e+1]){a=e;break}r.set(e,a)}return r}function S(e,t,n,r){if(r===`none`)return!0;let i=n.get(e),a=n.get(t);return i===void 0||a===void 0?!0:i===a}function C(e,t){if(e.length===0)return t.length;if(t.length===0)return e.length;let n=Array.from({length:e.length+1},()=>Array(t.length+1).fill(0));for(let t=0;t<=e.length;t++)n[t][0]=t;for(let e=0;e<=t.length;e++)n[0][e]=e;for(let r=1;r<=e.length;r++)for(let i=1;i<=t.length;i++)e[r-1]===t[i-1]?n[r][i]=n[r-1][i-1]:n[r][i]=1+Math.min(n[r-1][i],n[r][i-1],n[r-1][i-1]);return n[e.length][t.length]}function w(e,t){let n=e.toLowerCase(),r=t.toLowerCase(),i=C(n,r),a=Math.max(n.length,r.length);return a===0?1:1-i/a}var T=class{constructor(e,t,n={}){this.citations=e,this.text=t,this.options={scopeStrategy:n.scopeStrategy??`paragraph`,autoDetectParagraphs:n.autoDetectParagraphs??!0,paragraphBoundaryPattern:n.paragraphBoundaryPattern??/\n\n+/g,fuzzyPartyMatching:n.fuzzyPartyMatching??!0,partyMatchThreshold:n.partyMatchThreshold??.8,allowNestedResolution:n.allowNestedResolution??!1,reportUnresolved:n.reportUnresolved??!0},this.context={citationIndex:0,allCitations:e,lastFullCitation:void 0,fullCitationHistory:new Map,paragraphMap:new Map},this.options.autoDetectParagraphs&&(this.context.paragraphMap=x(t,e,this.options.paragraphBoundaryPattern))}resolve(){let e=[];for(let t=0;t<this.citations.length;t++){this.context.citationIndex=t;let n=this.citations[t],r;switch(n.type){case`id`:r=this.resolveId(n);break;case`supra`:r=this.resolveSupra(n);break;case`shortFormCase`:r=this.resolveShortFormCase(n);break;default:this.isFullCitation(n)&&(this.context.lastFullCitation=t,this.trackFullCitation(n,t));break}e.push({...n,resolution:r})}return e}resolveId(e){let t=this.context.citationIndex,n;for(let e=t-1;e>=0;e--)if(this.citations[e].type===`case`){n=e;break}return n===void 0?this.createFailureResult(`No preceding full case citation found`):this.isWithinScope(n,t)?{resolvedTo:n,confidence:1}:this.createFailureResult(`Antecedent citation outside scope boundary`)}resolveSupra(e){let t=this.context.citationIndex,n=this.normalizePartyName(e.partyName),r;for(let[e,i]of this.context.fullCitationHistory){if(!this.isWithinScope(i,t))continue;let a=w(n,e);(!r||a>r.similarity)&&(r={index:i,similarity:a})}if(!r)return this.createFailureResult(`No full citation found in scope`);if(r.similarity<this.options.partyMatchThreshold)return this.createFailureResult(`Party name similarity ${r.similarity.toFixed(2)} below threshold ${this.options.partyMatchThreshold}`);let i=[];return r.similarity<1&&i.push(`Fuzzy match: similarity ${r.similarity.toFixed(2)}`),{resolvedTo:r.index,confidence:r.similarity,warnings:i.length>0?i:void 0}}resolveShortFormCase(e){let t=this.context.citationIndex;for(let n=t-1;n>=0;n--){let r=this.citations[n];if(r.type!==`case`)continue;let i=r;if(i.volume===e.volume&&this.normalizeReporter(i.reporter)===this.normalizeReporter(e.reporter))return this.isWithinScope(n,t)?{resolvedTo:n,confidence:.95}:this.createFailureResult(`Matching citation outside scope boundary`)}return this.createFailureResult(`No matching full case citation found`)}isFullCitation(e){return e.type===`case`||e.type===`statute`||e.type===`journal`||e.type===`neutral`||e.type===`publicLaw`||e.type===`federalRegister`}trackFullCitation(e,t){if(e.type===`case`){let n=e,r=this.extractPartyName(n);if(r){let e=this.normalizePartyName(r);this.context.fullCitationHistory.set(e,t)}}}extractPartyName(e){let t=e.span.originalStart,n=Math.max(0,t-100),r=this.text.substring(n,t),i=r.match(/([A-Z][a-zA-Z]*(?:\s+[A-Z][a-zA-Z]*)*)\s+v\.?\s+[A-Z][a-zA-Z]*(?:\s+[A-Z][a-zA-Z]*)*,\s*$/);return i?i[1].trim():r.match(/([A-Z][a-zA-Z]*(?:\s+[A-Z][a-zA-Z]*)*),\s*$/)?.[1].trim()}normalizePartyName(e){return e.toLowerCase().replace(/\s+/g,` `).trim()}normalizeReporter(e){return e.toLowerCase().replace(/\s+/g,``).replace(/\./g,``)}isWithinScope(e,t){return S(e,t,this.context.paragraphMap,this.options.scopeStrategy)}createFailureResult(e){if(this.options.reportUnresolved)return{resolvedTo:void 0,failureReason:e,confidence:0}}};function E(e,t,n){return new T(e,t,n).resolve()}function D(e,t){let n=performance.now(),{cleaned:r,transformationMap:a,warnings:x}=i(e,t?.cleaners),S=d(r,t?.patterns||[...l,...u,...o,...s,...c]),C=[],w=new Set;for(let e of S){let t=`${e.span.cleanStart}-${e.span.cleanEnd}`;w.has(t)||(w.add(t),C.push(e))}let T=[];for(let e of C){let t;switch(e.type){case`case`:t=e.patternId===`id`||e.patternId===`ibid`?v(e,a):e.patternId===`supra`?y(e,a):e.patternId===`shortFormCase`?b(e,a):f(e,a);break;case`statute`:t=p(e,a);break;case`journal`:t=m(e,a);break;case`neutral`:t=h(e,a);break;case`publicLaw`:t=g(e,a);break;case`federalRegister`:t=_(e,a);break;default:continue}x.length>0&&(t.warnings=[...t.warnings||[],...x]),t.processTimeMs=performance.now()-n,T.push(t)}return t?.resolve?E(T,e,t.resolutionOptions):T}async function O(e,t){return D(e,t)}exports.DocumentResolver=T,exports.cleanText=i,exports.extractCase=f,exports.extractCitations=D,exports.extractCitationsAsync=O,exports.extractFederalRegister=_,exports.extractJournal=m,exports.extractNeutral=h,exports.extractPublicLaw=g,exports.extractStatute=p,exports.resolveCitations=E,exports.tokenize=d;
+//# sourceMappingURL=index.cjs.map

package/dist/index.cjs.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.cjs","names":[],"sources":["../src/clean/cleaners.ts","../src/clean/cleanText.ts","../src/patterns/casePatterns.ts","../src/patterns/statutePatterns.ts","../src/patterns/journalPatterns.ts","../src/patterns/neutralPatterns.ts","../src/patterns/shortForm.ts","../src/tokenize/tokenizer.ts","../src/extract/extractCase.ts","../src/extract/extractStatute.ts","../src/extract/extractJournal.ts","../src/extract/extractNeutral.ts","../src/extract/extractPublicLaw.ts","../src/extract/extractFederalRegister.ts","../src/extract/extractShortForms.ts","../src/resolve/scopeBoundary.ts","../src/resolve/levenshtein.ts","../src/resolve/DocumentResolver.ts","../src/resolve/index.ts","../src/extract/extractCitations.ts"],"sourcesContent":["/**\n * Built-in text cleaner functions for preprocessing legal documents.\n *\n * Each cleaner is a simple transformation: (text: string) => string\n * Cleaners can be composed via the cleanText() pipeline.\n */\n\n/**\n * Remove all HTML tags from text.\n *\n * @example\n * stripHtmlTags(\"Smith v. <b>Doe</b>, 500 F.2d 123\")\n * // => \"Smith v. Doe, 500 F.2d 123\"\n */\nexport function stripHtmlTags(text: string): string {\n\treturn text.replace(/<[^>]+>/g, \"\")\n}\n\n/**\n * Normalize whitespace: convert tabs/newlines to spaces, collapse multiple spaces.\n *\n * @example\n * normalizeWhitespace(\"Smith  v.  Doe,   500 F.2d 123\")\n * // => \"Smith v. Doe, 500 F.2d 123\"\n */\nexport function normalizeWhitespace(text: string): string {\n\treturn text.replace(/[\\t\\n\\r]+/g, \" \").replace(/ {2,}/g, \" \")\n}\n\n/**\n * Apply Unicode NFKC normalization (ligatures → separate chars).\n *\n * @example\n * normalizeUnicode(\"Smith v. Doe, 500 F.2d 123\") // with ligature \"fi\"\n * // => \"Smith v. Doe, 500 F.2d 123\" // normalized\n */\nexport function normalizeUnicode(text: string): string {\n\treturn text.normalize(\"NFKC\")\n}\n\n/**\n * Replace curly quotes and apostrophes with straight quotes.\n *\n * @example\n * fixSmartQuotes(\"\"Smith\" v. 'Doe', 500 F.2d 123\")\n * // => \"\\\"Smith\\\" v. 'Doe', 500 F.2d 123\"\n */\nexport function fixSmartQuotes(text: string): string {\n\treturn text\n\t\t.replace(/[\\u201C\\u201D]/g, '\"') // curly double quotes\n\t\t.replace(/[\\u2018\\u2019]/g, \"'\") // curly single quotes/apostrophes\n}\n\n/**\n * Remove underscore OCR artifacts (common in scanned documents).\n *\n * @example\n * removeOcrArtifacts(\"Smith v. Doe, 500 F._2d 123\")\n * // => \"Smith v. Doe, 500 F.2d 123\"\n */\nexport function removeOcrArtifacts(text: string): string {\n\treturn text.replace(/_/g, \"\")\n}\n","import type { TransformationMap } from \"../types/span\"\nimport {\n\tfixSmartQuotes,\n\tnormalizeUnicode,\n\tnormalizeWhitespace,\n\tstripHtmlTags,\n} from \"./cleaners\"\n\n/**\n * Result of text cleaning operation.\n */\nexport interface CleanTextResult {\n\t/** Cleaned text after all transformations */\n\tcleaned: string\n\n\t/** Position mappings between cleaned and original text */\n\ttransformationMap: TransformationMap\n\n\t/** Warnings generated during cleaning (currently unused) */\n\twarnings: Warning[]\n}\n\n/**\n * Warning generated during text cleaning.\n */\nexport interface Warning {\n\tlevel: \"error\" | \"warning\" | \"info\"\n\tmessage: string\n\tposition: { start: number; end: number }\n}\n\n/**\n * Clean text using a pipeline of transformation functions.\n *\n * Applies cleaners sequentially while maintaining accurate position mappings\n * between the original and cleaned text. This enables citation extraction from\n * cleaned text while reporting positions in the original text.\n *\n * @param original - Original input text\n * @param cleaners - Array of cleaner functions to apply (default: stripHtmlTags, normalizeWhitespace, normalizeUnicode, fixSmartQuotes)\n * @returns Cleaned text with position mappings and warnings\n *\n * @example\n * const result = cleanText(\"Smith v. <b>Doe</b>, 500 F.2d 123\")\n * // result.cleaned: \"Smith v. Doe, 500 F.2d 123\"\n * // result.transformationMap tracks position shifts from HTML removal\n */\nexport function cleanText(\n\toriginal: string,\n\tcleaners: Array<(text: string) => string> = [\n\t\tstripHtmlTags,\n\t\tnormalizeWhitespace,\n\t\tnormalizeUnicode,\n\t\tfixSmartQuotes,\n\t],\n): CleanTextResult {\n\t// Initialize 1:1 position mapping\n\tlet currentText = original\n\tlet cleanToOriginal = new Map<number, number>()\n\tlet originalToClean = new Map<number, number>()\n\n\t// Identity mapping: cleanToOriginal[i] = i, originalToClean[i] = i\n\tfor (let i = 0; i <= original.length; i++) {\n\t\tcleanToOriginal.set(i, i)\n\t\toriginalToClean.set(i, i)\n\t}\n\n\t// Apply each cleaner sequentially, rebuilding position maps\n\tfor (const cleaner of cleaners) {\n\t\tconst beforeText = currentText\n\t\tconst afterText = cleaner(currentText)\n\n\t\tif (beforeText !== afterText) {\n\t\t\t// Text changed - rebuild position maps\n\t\t\tconst { newCleanToOriginal, newOriginalToClean } = rebuildPositionMaps(\n\t\t\t\tbeforeText,\n\t\t\t\tafterText,\n\t\t\t\tcleanToOriginal,\n\t\t\t\toriginalToClean,\n\t\t\t)\n\n\t\t\tcleanToOriginal = newCleanToOriginal\n\t\t\toriginalToClean = newOriginalToClean\n\t\t\tcurrentText = afterText\n\t\t}\n\t}\n\n\tconst transformationMap: TransformationMap = {\n\t\tcleanToOriginal,\n\t\toriginalToClean,\n\t}\n\n\treturn {\n\t\tcleaned: currentText,\n\t\ttransformationMap,\n\t\twarnings: [],\n\t}\n}\n\n/**\n * Rebuild position maps after a text transformation.\n *\n * Uses a simplified algorithm that scans through both strings, matching\n * characters where possible and tracking the offset accumulation.\n *\n * @param beforeText - Text before transformation\n * @param afterText - Text after transformation\n * @param oldCleanToOriginal - Previous clean-to-original mapping\n * @param oldOriginalToClean - Previous original-to-clean mapping\n * @returns New position maps\n */\nfunction rebuildPositionMaps(\n\tbeforeText: string,\n\tafterText: string,\n\toldCleanToOriginal: Map<number, number>,\n\toldOriginalToClean: Map<number, number>,\n): {\n\tnewCleanToOriginal: Map<number, number>\n\tnewOriginalToClean: Map<number, number>\n} {\n\tconst newCleanToOriginal = new Map<number, number>()\n\tconst newOriginalToClean = new Map<number, number>()\n\n\tlet beforeIdx = 0\n\tlet afterIdx = 0\n\n\t// Scan through both strings, matching characters where possible\n\twhile (beforeIdx <= beforeText.length || afterIdx <= afterText.length) {\n\t\t// Both at end\n\t\tif (beforeIdx >= beforeText.length && afterIdx >= afterText.length) {\n\t\t\tconst originalPos = oldCleanToOriginal.get(beforeIdx) ?? beforeIdx\n\t\t\tnewCleanToOriginal.set(afterIdx, originalPos)\n\t\t\tnewOriginalToClean.set(originalPos, afterIdx)\n\t\t\tbreak\n\t\t}\n\n\t\t// Before text exhausted (expansion case)\n\t\tif (beforeIdx >= beforeText.length) {\n\t\t\tconst originalPos = oldCleanToOriginal.get(beforeIdx) ?? beforeIdx\n\t\t\tnewCleanToOriginal.set(afterIdx, originalPos)\n\t\t\tafterIdx++\n\t\t\tcontinue\n\t\t}\n\n\t\t// After text exhausted (removal case)\n\t\tif (afterIdx >= afterText.length) {\n\t\t\tconst originalPos = oldCleanToOriginal.get(beforeIdx) ?? beforeIdx\n\t\t\tnewOriginalToClean.set(originalPos, afterIdx)\n\t\t\tbeforeIdx++\n\t\t\tcontinue\n\t\t}\n\n\t\t// Characters match - carry forward the mapping\n\t\tif (beforeText[beforeIdx] === afterText[afterIdx]) {\n\t\t\tconst originalPos = oldCleanToOriginal.get(beforeIdx) ?? beforeIdx\n\t\t\tnewCleanToOriginal.set(afterIdx, originalPos)\n\t\t\tnewOriginalToClean.set(originalPos, afterIdx)\n\t\t\tbeforeIdx++\n\t\t\tafterIdx++\n\t\t} else {\n\t\t\t// Characters differ - need to determine if this is insertion/deletion/replacement\n\t\t\t// Look ahead to find next match\n\t\t\tlet foundMatch = false\n\t\t\tconst maxLookAhead = 20 // Limit lookahead to avoid performance issues\n\n\t\t\t// Check if something was deleted from before text\n\t\t\tfor (let lookAhead = 1; lookAhead <= maxLookAhead; lookAhead++) {\n\t\t\t\tif (beforeIdx + lookAhead >= beforeText.length) break\n\n\t\t\t\tif (beforeText[beforeIdx + lookAhead] === afterText[afterIdx]) {\n\t\t\t\t\t// Found a match - characters were deleted from before text\n\t\t\t\t\tfor (let i = 0; i < lookAhead; i++) {\n\t\t\t\t\t\tconst originalPos =\n\t\t\t\t\t\t\toldCleanToOriginal.get(beforeIdx + i) ?? beforeIdx + i\n\t\t\t\t\t\tnewOriginalToClean.set(originalPos, afterIdx)\n\t\t\t\t\t}\n\t\t\t\t\tbeforeIdx += lookAhead\n\t\t\t\t\tfoundMatch = true\n\t\t\t\t\tbreak\n\t\t\t\t}\n\t\t\t}\n\n\t\t\tif (foundMatch) continue\n\n\t\t\t// Check if something was inserted into after text\n\t\t\tfor (let lookAhead = 1; lookAhead <= maxLookAhead; lookAhead++) {\n\t\t\t\tif (afterIdx + lookAhead >= afterText.length) break\n\n\t\t\t\tif (beforeText[beforeIdx] === afterText[afterIdx + lookAhead]) {\n\t\t\t\t\t// Found a match - characters were inserted into after text\n\t\t\t\t\tconst originalPos = oldCleanToOriginal.get(beforeIdx) ?? beforeIdx\n\t\t\t\t\tfor (let i = 0; i < lookAhead; i++) {\n\t\t\t\t\t\tnewCleanToOriginal.set(afterIdx + i, originalPos)\n\t\t\t\t\t}\n\t\t\t\t\tafterIdx += lookAhead\n\t\t\t\t\tfoundMatch = true\n\t\t\t\t\tbreak\n\t\t\t\t}\n\t\t\t}\n\n\t\t\tif (foundMatch) continue\n\n\t\t\t// No match found within lookahead - treat as replacement\n\t\t\tconst originalPos = oldCleanToOriginal.get(beforeIdx) ?? beforeIdx\n\t\t\tnewCleanToOriginal.set(afterIdx, originalPos)\n\t\t\tnewOriginalToClean.set(originalPos, afterIdx)\n\t\t\tbeforeIdx++\n\t\t\tafterIdx++\n\t\t}\n\t}\n\n\treturn { newCleanToOriginal, newOriginalToClean }\n}\n","/**\n * Case Citation Regex Patterns\n *\n * These patterns are designed for tokenization (broad matching) not extraction.\n * They identify potential case citations in text for the tokenizer (Plan 3).\n * Metadata parsing and validation against reporters-db happens in Phase 2 Plan 5 (extraction layer).\n *\n * Pattern Design Principles (from RESEARCH.md):\n * - Use \\b word boundaries to avoid matching \"F.\" in \"F.B.I.\"\n * - Avoid nested quantifiers: (a+)+ causes ReDoS\n * - Keep patterns simple: tokenization only needs to find candidates\n * - Use global flag /g for matchAll()\n */\n\nexport interface Pattern {\n  id: string\n  regex: RegExp\n  description: string\n  type: 'case' | 'statute' | 'journal' | 'neutral' | 'publicLaw' | 'federalRegister'\n}\n\nexport const casePatterns: Pattern[] = [\n  {\n    id: 'federal-reporter',\n    regex: /\\b(\\d+)\\s+(F\\.|F\\.2d|F\\.3d|F\\.\\s?Supp\\.|F\\.\\s?Supp\\.\\s?2d|F\\.\\s?Supp\\.\\s?3d)\\s+(\\d+)\\b/g,\n    description: 'Federal Reporter (F., F.2d, F.3d, F.Supp., etc.)',\n    type: 'case',\n  },\n  {\n    id: 'supreme-court',\n    regex: /\\b(\\d+)\\s+(U\\.S\\.|S\\.\\s?Ct\\.|L\\.\\s?Ed\\.(?:\\s?2d)?)\\s+(\\d+)\\b/g,\n    description: 'U.S. Supreme Court reporters',\n    type: 'case',\n  },\n  {\n    id: 'state-reporter',\n    regex: /\\b(\\d+)\\s+([A-Z][A-Za-z\\.]+(?:\\s?2d|\\s?3d)?)\\s+(\\d+)\\b/g,\n    description: 'State reporters (broad pattern, validated against reporters-db in Phase 3)',\n    type: 'case',\n  },\n]\n","/**\n * Statute Citation Regex Patterns\n *\n * Patterns for U.S. Code and state code citations.\n * These are intentionally broad for tokenization - validation against\n * actual code databases happens in Phase 2 Plan 5 (extraction layer).\n *\n * Pattern Design:\n * - Simple structure to avoid ReDoS\n * - Matches both \"§\" and \"Section\" formats\n * - State codes use broad pattern (validated later)\n */\n\nimport type { Pattern } from './casePatterns'\n\nexport const statutePatterns: Pattern[] = [\n  {\n    id: 'usc',\n    regex: /\\b(\\d+)\\s+U\\.S\\.C\\.?\\s+§+\\s*(\\d+)\\b/g,\n    description: 'U.S. Code citations (e.g., \"42 U.S.C. § 1983\")',\n    type: 'statute',\n  },\n  {\n    id: 'state-code',\n    regex: /\\b([A-Z][a-z]+\\.?\\s+[A-Za-z\\.]+\\s+Code)\\s+§\\s*(\\d+)\\b/g,\n    description: 'State code citations (broad pattern, e.g., \"Cal. Penal Code § 187\")',\n    type: 'statute',\n  },\n]\n","/**\n * Journal Citation Regex Patterns\n *\n * Patterns for law review and journal citations.\n * These are intentionally broad for tokenization - validation against\n * journals-db happens in Phase 3 (extraction layer).\n *\n * Pattern Design:\n * - Matches volume-journal-page format\n * - Broad journal name matching (validated later)\n * - Simple structure to avoid ReDoS\n */\n\nimport type { Pattern } from './casePatterns'\n\nexport const journalPatterns: Pattern[] = [\n  {\n    id: 'law-review',\n    regex: /\\b(\\d+)\\s+([A-Z][A-Za-z\\.\\s]+)\\s+(\\d+)\\b/g,\n    description: 'Law review citations (e.g., \"120 Harv. L. Rev. 500\"), validated against journals-db in Phase 3',\n    type: 'journal',\n  },\n]\n","/**\n * Neutral and Online Citation Regex Patterns\n *\n * Patterns for WestLaw, LexisNexis, public laws, and Federal Register citations.\n * These have predictable formats and don't require external validation.\n *\n * Pattern Design:\n * - Matches year-database-number format for online citations\n * - Matches Pub. L. No. format for public laws\n * - Matches volume-Fed. Reg.-page for Federal Register\n * - Simple structure to avoid ReDoS\n */\n\nimport type { Pattern } from './casePatterns'\n\nexport const neutralPatterns: Pattern[] = [\n  {\n    id: 'westlaw',\n    regex: /\\b(\\d{4})\\s+WL\\s+(\\d+)\\b/g,\n    description: 'WestLaw citations (e.g., \"2021 WL 123456\")',\n    type: 'neutral',\n  },\n  {\n    id: 'lexis',\n    regex: /\\b(\\d{4})\\s+U\\.S\\.\\s+LEXIS\\s+(\\d+)\\b/g,\n    description: 'LexisNexis citations (e.g., \"2021 U.S. LEXIS 5000\")',\n    type: 'neutral',\n  },\n  {\n    id: 'public-law',\n    regex: /\\bPub\\.\\s?L\\.\\s?No\\.\\s?(\\d+-\\d+)\\b/g,\n    description: 'Public Law citations (e.g., \"Pub. L. No. 117-58\")',\n    type: 'publicLaw',\n  },\n  {\n    id: 'federal-register',\n    regex: /\\b(\\d+)\\s+Fed\\.\\s?Reg\\.\\s+(\\d+)\\b/g,\n    description: 'Federal Register citations (e.g., \"86 Fed. Reg. 12345\")',\n    type: 'federalRegister',\n  },\n]\n","/**\n * Short-form Citation Regex Patterns\n *\n * Patterns for Id., Ibid., supra, and short-form case citations.\n * These refer to earlier citations in the document.\n *\n * Pattern Design:\n * - Simple structure to avoid ReDoS (no nested quantifiers)\n * - Broad matching for tokenization; validation happens in extraction layer\n * - Word boundaries to prevent false positives (e.g., \"Idaho\" vs \"Id.\")\n */\n\nimport type { Pattern } from './casePatterns'\n\n/** Id. with optional pincite: \"Id.\" or \"Id. at 253\" */\nexport const ID_PATTERN: RegExp = /\\b[Ii]d\\.(?:\\s+at\\s+(\\d+))?/g\n\n/** Ibid. with optional pincite (less common variant) */\nexport const IBID_PATTERN: RegExp = /\\b[Ii]bid\\.(?:\\s+at\\s+(\\d+))?/g\n\n/**\n * Supra with party name and optional pincite.\n * Pattern: word(s), supra [, at page]\n * Captures: (1) party name, (2) pincite\n * Note: Matches party names including \"v.\" (e.g., \"Smith v. Jones, supra\")\n */\nexport const SUPRA_PATTERN: RegExp = /\\b([A-Z][a-zA-Z]+(?:(?:\\s+v\\.?\\s+|\\s+)[A-Z][a-zA-Z]+)*),?\\s+supra(?:,?\\s+at\\s+(\\d+))?/g\n\n/**\n * Short-form case: volume reporter at page\n * Pattern: number space abbreviation space \"at\" space number\n * Simplified detection; full parsing in extraction layer\n */\nexport const SHORT_FORM_CASE_PATTERN: RegExp = /\\b(\\d+)\\s+([A-Z][A-Za-z.\\s]+?(?:\\d[a-z])?)\\s+at\\s+(\\d+)\\b/g\n\n/** All short-form patterns for tokenization */\nexport const SHORT_FORM_PATTERNS: readonly RegExp[] = [\n  ID_PATTERN,\n  IBID_PATTERN,\n  SUPRA_PATTERN,\n  SHORT_FORM_CASE_PATTERN,\n] as const\n\n/** Pattern objects for consistency with other pattern modules */\nexport const shortFormPatterns: Pattern[] = [\n  {\n    id: 'id',\n    regex: ID_PATTERN,\n    description: 'Id. citations (e.g., \"Id.\" or \"Id. at 253\")',\n    type: 'case', // Will be typed as 'id' in extraction layer\n  },\n  {\n    id: 'ibid',\n    regex: IBID_PATTERN,\n    description: 'Ibid. citations (e.g., \"Ibid.\" or \"Ibid. at 125\")',\n    type: 'case', // Will be typed as 'id' in extraction layer\n  },\n  {\n    id: 'supra',\n    regex: SUPRA_PATTERN,\n    description: 'Supra citations (e.g., \"Smith, supra\" or \"Smith, supra, at 460\")',\n    type: 'case', // Will be typed as 'supra' in extraction layer\n  },\n  {\n    id: 'shortFormCase',\n    regex: SHORT_FORM_CASE_PATTERN,\n    description: 'Short-form case citations (e.g., \"500 F.2d at 125\")',\n    type: 'case', // Will be typed as 'shortFormCase' in extraction layer\n  },\n]\n","/**\n * Tokenization Layer for Citation Extraction\n *\n * Applies regex patterns to cleaned text to produce citation candidate tokens.\n * This is the second stage of the parsing pipeline:\n *   1. Clean text (remove HTML, normalize Unicode)\n *   2. Tokenize (apply patterns to find candidates) ← THIS MODULE\n *   3. Extract (parse metadata, validate against reporters-db)\n *\n * Tokenization is intentionally broad - it finds potential citations without\n * validating them. The extraction layer (Plan 5) validates tokens against\n * reporters-db and parses metadata.\n *\n * @module tokenize\n */\n\nimport type { Span } from '@/types/span'\nimport type { Pattern } from '@/patterns'\nimport {\n  casePatterns,\n  statutePatterns,\n  journalPatterns,\n  neutralPatterns,\n} from '@/patterns'\nimport { shortFormPatterns } from '@/patterns/shortForm'\n\n/**\n * A token representing a potential citation found in cleaned text.\n *\n * Tokens are produced by applying regex patterns to cleaned text.\n * They include matched text, position in cleaned text, and pattern metadata\n * for use in the extraction layer.\n */\nexport interface Token {\n  /** Matched text from input */\n  text: string\n\n  /** Position in cleaned text (cleanStart/cleanEnd only, no original positions yet) */\n  span: Pick<Span, 'cleanStart' | 'cleanEnd'>\n\n  /** Pattern type that matched this token */\n  type: Pattern['type']\n\n  /** Pattern ID that matched this token */\n  patternId: string\n}\n\n/**\n * Tokenizes cleaned text by applying regex patterns to find citation candidates.\n *\n * For each pattern in the patterns array:\n *   1. Apply pattern.regex.matchAll(cleanedText)\n *   2. Create Token for each match with position, text, and pattern metadata\n *   3. Collect all tokens from all patterns\n *   4. Sort by cleanStart position (ascending)\n *\n * Timeout protection: If a pattern throws (e.g., ReDoS), skip it and continue\n * with remaining patterns. Logs warning to console.\n *\n * Note: This function is synchronous because regex matching is inherently\n * synchronous. This enables both sync (extractCitations) and async\n * (extractCitationsAsync) APIs in Plan 6.\n *\n * @param cleanedText - Text that has been cleaned by cleanText() from Plan 1\n * @param patterns - Regex patterns to apply (defaults to all patterns from Plan 2)\n * @returns Array of tokens sorted by position (cleanStart ascending)\n *\n * @example\n * ```typescript\n * import { tokenize } from '@/tokenize'\n * import { cleanText } from '@/clean'\n *\n * const original = \"See Smith v. Doe, 500 F.2d 123 (9th Cir. 2020)\"\n * const { cleanedText } = cleanText(original)\n * const tokens = tokenize(cleanedText)\n * // tokens[0] = {\n * //   text: \"500 F.2d 123\",\n * //   span: { cleanStart: 18, cleanEnd: 30 },\n * //   type: \"case\",\n * //   patternId: \"federal-reporter\"\n * // }\n * ```\n */\nexport function tokenize(\n  cleanedText: string,\n  patterns: Pattern[] = [\n    ...casePatterns,\n    ...statutePatterns,\n    ...journalPatterns,\n    ...neutralPatterns,\n    ...shortFormPatterns,\n  ]\n): Token[] {\n  const tokens: Token[] = []\n\n  for (const pattern of patterns) {\n    try {\n      // Apply pattern to cleaned text\n      const matches = cleanedText.matchAll(pattern.regex)\n\n      for (const match of matches) {\n        // Create token from match\n        tokens.push({\n          text: match[0],\n          span: {\n            cleanStart: match.index!,\n            cleanEnd: match.index! + match[0].length,\n          },\n          type: pattern.type,\n          patternId: pattern.id,\n        })\n      }\n    } catch (error) {\n      // Timeout protection: If pattern throws (ReDoS, etc.), skip it\n      console.warn(\n        `Pattern ${pattern.id} threw error, skipping:`,\n        error instanceof Error ? error.message : String(error)\n      )\n      continue\n    }\n  }\n\n  // Sort tokens by position (cleanStart ascending)\n  tokens.sort((a, b) => a.span.cleanStart - b.span.cleanStart)\n\n  return tokens\n}\n","/**\n * Case Citation Extraction\n *\n * Parses tokenized case citations to extract volume, reporter, page, and\n * optional metadata (pincite, court, year). This is the third stage of\n * the parsing pipeline:\n *   1. Clean text (remove HTML, normalize Unicode)\n *   2. Tokenize (apply patterns to find candidates)\n *   3. Extract (parse metadata, validate) ← THIS MODULE\n *\n * Extraction parses structured data from token text. Validation against\n * reporters-db happens in Phase 3 (resolution layer).\n *\n * @module extract/extractCase\n */\n\nimport type { Token } from '@/tokenize'\nimport type { FullCaseCitation } from '@/types/citation'\nimport type { TransformationMap } from '@/types/span'\n\n/**\n * Extracts case citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Volume: Leading digits (e.g., \"500\" from \"500 F.2d 123\")\n * - Reporter: Alphabetic abbreviation (e.g., \"F.2d\")\n * - Page: Trailing digits after reporter (e.g., \"123\")\n * - Pincite: Optional page reference after comma (e.g., \", 125\")\n * - Court: Optional court abbreviation in parentheses (e.g., \"(9th Cir.)\")\n * - Year: Optional year in parentheses (e.g., \"(2020)\")\n *\n * Confidence scoring:\n * - Base: 0.5\n * - Common reporter pattern (F., U.S., etc.): +0.3\n * - Valid year (not future): +0.2\n * - Capped at 1.0\n *\n * Position translation:\n * - Uses TransformationMap to convert clean positions → original positions\n * - cleanStart/cleanEnd from token span\n * - originalStart/originalEnd via transformationMap.cleanToOriginal\n *\n * Note: This function does NOT validate against reporters-db. That happens\n * in Phase 3 (resolution layer). Phase 2 extraction only parses structure.\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns FullCaseCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"500 F.2d 123, 125\",\n *   span: { cleanStart: 10, cleanEnd: 27 },\n *   type: \"case\",\n *   patternId: \"federal-reporter\"\n * }\n * const citation = extractCase(token, transformationMap)\n * // citation = {\n * //   type: \"case\",\n * //   text: \"500 F.2d 123, 125\",\n * //   volume: 500,\n * //   reporter: \"F.2d\",\n * //   page: 123,\n * //   pincite: 125,\n * //   span: { cleanStart: 10, cleanEnd: 27, originalStart: 10, originalEnd: 27 },\n * //   confidence: 0.8,\n * //   ...\n * // }\n * ```\n */\nexport function extractCase(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): FullCaseCitation {\n\tconst { text, span } = token\n\n\t// Parse volume-reporter-page using regex\n\t// Pattern: volume (digits) + reporter (letters/periods/spaces/numbers) + page (digits)\n\t// Use greedy matching for reporter to capture full abbreviation including spaces\n\tconst volumeReporterPageRegex = /^(\\d+)\\s+([A-Za-z0-9.\\s]+)\\s+(\\d+)/\n\tconst match = volumeReporterPageRegex.exec(text)\n\n\tif (!match) {\n\t\t// Fallback if pattern doesn't match (shouldn't happen if tokenizer is correct)\n\t\tthrow new Error(`Failed to parse case citation: ${text}`)\n\t}\n\n\tconst volume = Number.parseInt(match[1], 10)\n\tconst reporter = match[2].trim()\n\tconst page = Number.parseInt(match[3], 10)\n\n\t// Extract optional pincite (page reference after comma)\n\t// Pattern: \", digits\" (e.g., \", 125\")\n\tconst pinciteRegex = /,\\s*(\\d+)/\n\tconst pinciteMatch = pinciteRegex.exec(text)\n\tconst pincite = pinciteMatch ? Number.parseInt(pinciteMatch[1], 10) : undefined\n\n\t// Extract optional year in parentheses (extract first for better matching)\n\t// Pattern: 4-digit year anywhere in parentheses\n\tconst yearRegex = /\\((?:[^)]*\\s)?(\\d{4})\\)/\n\tconst yearMatch = yearRegex.exec(text)\n\tconst year = yearMatch ? Number.parseInt(yearMatch[1], 10) : undefined\n\n\t// Extract optional court abbreviation in parentheses\n\t// Pattern: \"(text)\" where text contains letters (captures full parenthetical)\n\tconst courtRegex = /\\(([^)]*[A-Za-z][^)]*)\\)/\n\tconst courtMatch = courtRegex.exec(text)\n\tconst court = courtMatch ? courtMatch[1].trim() : undefined\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Calculate confidence score\n\tlet confidence = 0.5 // Base confidence\n\n\t// Common reporter patterns (F., U.S., S. Ct., etc.)\n\tconst commonReporters = [\n\t\t'F.',\n\t\t'F.2d',\n\t\t'F.3d',\n\t\t'F.4th',\n\t\t'U.S.',\n\t\t'S. Ct.',\n\t\t'L. Ed.',\n\t\t'P.',\n\t\t'P.2d',\n\t\t'P.3d',\n\t\t'A.',\n\t\t'A.2d',\n\t\t'A.3d',\n\t\t'N.E.',\n\t\t'N.E.2d',\n\t\t'N.E.3d',\n\t\t'N.W.',\n\t\t'N.W.2d',\n\t\t'S.E.',\n\t\t'S.E.2d',\n\t\t'S.W.',\n\t\t'S.W.2d',\n\t\t'S.W.3d',\n\t\t'So.',\n\t\t'So. 2d',\n\t\t'So. 3d',\n\t]\n\n\tif (commonReporters.some((r) => reporter.includes(r))) {\n\t\tconfidence += 0.3\n\t}\n\n\t// Valid year check (not in future)\n\tif (year !== undefined) {\n\t\tconst currentYear = new Date().getFullYear()\n\t\tif (year <= currentYear) {\n\t\t\tconfidence += 0.2\n\t\t}\n\t}\n\n\t// Cap at 1.0\n\tconfidence = Math.min(confidence, 1.0)\n\n\treturn {\n\t\ttype: 'case',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0, // Placeholder - timing handled by orchestration layer\n\t\tpatternsChecked: 1, // Single token processed\n\t\tvolume,\n\t\treporter,\n\t\tpage,\n\t\tpincite,\n\t\tcourt,\n\t\tyear,\n\t}\n}\n","/**\n * Statute Citation Extraction\n *\n * Parses tokenized statute citations to extract title, code, section, and\n * optional subsections. Examples: \"42 U.S.C. § 1983\", \"Cal. Civ. Code § 1234(a)(1)\"\n *\n * @module extract/extractStatute\n */\n\nimport type { Token } from '@/tokenize'\nimport type { StatuteCitation } from '@/types/citation'\nimport type { TransformationMap } from '@/types/span'\n\n/**\n * Extracts statute citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Title: Optional leading digits (e.g., \"42\" from \"42 U.S.C. § 1983\")\n * - Code: Statutory code abbreviation (e.g., \"U.S.C.\", \"Cal. Civ. Code\")\n * - Section: Section number after § symbol (e.g., \"1983\")\n * - Subsections: Optional parenthetical subdivisions (e.g., \"(a)(1)\")\n *\n * Confidence scoring:\n * - Base: 0.5\n * - Known code pattern (U.S.C., C.F.R., state codes): +0.3\n * - Capped at 1.0\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns StatuteCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"42 U.S.C. § 1983\",\n *   span: { cleanStart: 10, cleanEnd: 26 },\n *   type: \"statute\",\n *   patternId: \"usc\"\n * }\n * const citation = extractStatute(token, transformationMap)\n * // citation = {\n * //   type: \"statute\",\n * //   title: 42,\n * //   code: \"U.S.C.\",\n * //   section: \"1983\",\n * //   ...\n * // }\n * ```\n */\nexport function extractStatute(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): StatuteCitation {\n\tconst { text, span } = token\n\n\t// Parse title-code-section using regex\n\t// Pattern: optional title (digits) + code (letters/periods/spaces) + § + section\n\tconst statuteRegex = /^(?:(\\d+)\\s+)?([A-Za-z.\\s]+?)\\s*§\\s*(\\d+[A-Za-z0-9\\-]*)/\n\tconst match = statuteRegex.exec(text)\n\n\tif (!match) {\n\t\tthrow new Error(`Failed to parse statute citation: ${text}`)\n\t}\n\n\tconst title = match[1] ? Number.parseInt(match[1], 10) : undefined\n\tconst code = match[2].trim()\n\tconst section = match[3]\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Calculate confidence score\n\tlet confidence = 0.5 // Base confidence\n\n\t// Known statutory code patterns\n\tconst knownCodes = [\n\t\t'U.S.C.',\n\t\t'C.F.R.',\n\t\t'Cal. Civ. Code',\n\t\t'Cal. Penal Code',\n\t\t'N.Y. Civ. Prac. L. & R.',\n\t\t'Tex. Civ. Prac. & Rem. Code',\n\t]\n\n\tif (knownCodes.some((c) => code.includes(c))) {\n\t\tconfidence += 0.3\n\t}\n\n\tconfidence = Math.min(confidence, 1.0)\n\n\treturn {\n\t\ttype: 'statute',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0,\n\t\tpatternsChecked: 1,\n\t\ttitle,\n\t\tcode,\n\t\tsection,\n\t}\n}\n","/**\n * Journal Citation Extraction\n *\n * Parses tokenized journal citations to extract volume, journal name, page,\n * and optional metadata. Examples: \"123 Harv. L. Rev. 456\", \"75 Yale L.J. 789, 791\"\n *\n * @module extract/extractJournal\n */\n\nimport type { Token } from '@/tokenize'\nimport type { JournalCitation } from '@/types/citation'\nimport type { TransformationMap } from '@/types/span'\n\n/**\n * Extracts journal citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Volume: Leading digits (e.g., \"123\" from \"123 Harv. L. Rev. 456\")\n * - Journal: Journal abbreviation (e.g., \"Harv. L. Rev.\")\n * - Page: Starting page number (e.g., \"456\")\n * - Pincite: Optional specific page reference after comma (e.g., \", 458\")\n *\n * Confidence scoring:\n * - Base: 0.6 (journal validation happens in Phase 3)\n *\n * Note: Author and title extraction from preceding text is not implemented\n * in Phase 2. That requires context analysis in Phase 3.\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns JournalCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"123 Harv. L. Rev. 456\",\n *   span: { cleanStart: 10, cleanEnd: 31 },\n *   type: \"journal\",\n *   patternId: \"journal-standard\"\n * }\n * const citation = extractJournal(token, transformationMap)\n * // citation = {\n * //   type: \"journal\",\n * //   volume: 123,\n * //   journal: \"Harv. L. Rev.\",\n * //   abbreviation: \"Harv. L. Rev.\",\n * //   page: 456,\n * //   ...\n * // }\n * ```\n */\nexport function extractJournal(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): JournalCitation {\n\tconst { text, span } = token\n\n\t// Parse volume-journal-page using regex\n\t// Pattern: volume (digits) + journal (letters/periods/spaces) + page (digits)\n\tconst journalRegex = /^(\\d+)\\s+([A-Za-z.\\s]+?)\\s+(\\d+)/\n\tconst match = journalRegex.exec(text)\n\n\tif (!match) {\n\t\tthrow new Error(`Failed to parse journal citation: ${text}`)\n\t}\n\n\tconst volume = Number.parseInt(match[1], 10)\n\tconst journal = match[2].trim()\n\tconst page = Number.parseInt(match[3], 10)\n\n\t// Extract optional pincite (page reference after comma)\n\tconst pinciteRegex = /,\\s*(\\d+)/\n\tconst pinciteMatch = pinciteRegex.exec(text)\n\tconst pincite = pinciteMatch ? Number.parseInt(pinciteMatch[1], 10) : undefined\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Confidence: 0.6 base (journal validation against database happens in Phase 3)\n\tconst confidence = 0.6\n\n\treturn {\n\t\ttype: 'journal',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0,\n\t\tpatternsChecked: 1,\n\t\tvolume,\n\t\tjournal,\n\t\tabbreviation: journal, // For Phase 2, abbreviation = journal name\n\t\tpage,\n\t\tpincite,\n\t}\n}\n","/**\n * Neutral Citation Extraction\n *\n * Parses tokenized neutral (vendor-neutral) citations to extract year, court,\n * and document number. Examples: \"2020 WL 123456\", \"2020 U.S. LEXIS 456\"\n *\n * @module extract/extractNeutral\n */\n\nimport type { Token } from '@/tokenize'\nimport type { NeutralCitation } from '@/types/citation'\nimport type { TransformationMap } from '@/types/span'\n\n/**\n * Extracts neutral citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Year: 4-digit year (e.g., \"2020\")\n * - Court: Vendor identifier (e.g., \"WL\", \"U.S. LEXIS\")\n * - Document number: Unique document identifier (e.g., \"123456\")\n *\n * Confidence scoring:\n * - 1.0 (neutral format is unambiguous and standardized)\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns NeutralCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"2020 WL 123456\",\n *   span: { cleanStart: 10, cleanEnd: 24 },\n *   type: \"neutral\",\n *   patternId: \"westlaw-neutral\"\n * }\n * const citation = extractNeutral(token, transformationMap)\n * // citation = {\n * //   type: \"neutral\",\n * //   year: 2020,\n * //   court: \"WL\",\n * //   documentNumber: \"123456\",\n * //   confidence: 1.0,\n * //   ...\n * // }\n * ```\n */\nexport function extractNeutral(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): NeutralCitation {\n\tconst { text, span } = token\n\n\t// Parse year-court-documentNumber using regex\n\t// Pattern: 4-digit year + court identifier (WL, LEXIS, etc.) + document number\n\tconst neutralRegex = /^(\\d{4})\\s+(WL|LEXIS|U\\.S\\.\\s+LEXIS)\\s+(\\d+)/\n\tconst match = neutralRegex.exec(text)\n\n\tif (!match) {\n\t\tthrow new Error(`Failed to parse neutral citation: ${text}`)\n\t}\n\n\tconst year = Number.parseInt(match[1], 10)\n\tconst court = match[2]\n\tconst documentNumber = match[3]\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Confidence: 1.0 (neutral format is unambiguous)\n\tconst confidence = 1.0\n\n\treturn {\n\t\ttype: 'neutral',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0,\n\t\tpatternsChecked: 1,\n\t\tyear,\n\t\tcourt,\n\t\tdocumentNumber,\n\t}\n}\n","/**\n * Public Law Citation Extraction\n *\n * Parses tokenized public law citations to extract congress number and law number.\n * Examples: \"Pub. L. No. 116-283\", \"Pub. L. 117-58\"\n *\n * @module extract/extractPublicLaw\n */\n\nimport type { Token } from '@/tokenize'\nimport type { PublicLawCitation } from '@/types/citation'\nimport type { TransformationMap } from '@/types/span'\n\n/**\n * Extracts public law citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Congress: Congress number (e.g., \"116\" from \"Pub. L. No. 116-283\")\n * - Law number: Law number within that Congress (e.g., \"283\")\n *\n * Confidence scoring:\n * - 0.9 (public law format is fairly standard)\n *\n * Note: Bill title extraction from nearby text is not implemented in Phase 2.\n * That requires context analysis in Phase 3.\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns PublicLawCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"Pub. L. No. 116-283\",\n *   span: { cleanStart: 10, cleanEnd: 29 },\n *   type: \"publicLaw\",\n *   patternId: \"public-law\"\n * }\n * const citation = extractPublicLaw(token, transformationMap)\n * // citation = {\n * //   type: \"publicLaw\",\n * //   congress: 116,\n * //   lawNumber: 283,\n * //   confidence: 0.9,\n * //   ...\n * // }\n * ```\n */\nexport function extractPublicLaw(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): PublicLawCitation {\n\tconst { text, span } = token\n\n\t// Parse congress-lawNumber using regex\n\t// Pattern: \"Pub. L.\" (with optional \"No.\") + congress number + \"-\" + law number\n\tconst publicLawRegex = /Pub\\.\\s?L\\.(?:\\s?No\\.)?\\s?(\\d+)-(\\d+)/\n\tconst match = publicLawRegex.exec(text)\n\n\tif (!match) {\n\t\tthrow new Error(`Failed to parse public law citation: ${text}`)\n\t}\n\n\tconst congress = Number.parseInt(match[1], 10)\n\tconst lawNumber = Number.parseInt(match[2], 10)\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Confidence: 0.9 (public law format is fairly standard)\n\tconst confidence = 0.9\n\n\treturn {\n\t\ttype: 'publicLaw',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0,\n\t\tpatternsChecked: 1,\n\t\tcongress,\n\t\tlawNumber,\n\t}\n}\n","/**\n * Federal Register Citation Extraction\n *\n * Parses tokenized Federal Register citations to extract volume, page, and\n * optional year. Examples: \"85 Fed. Reg. 12345\", \"86 Fed. Reg. 56789 (Jan. 15, 2021)\"\n *\n * @module extract/extractFederalRegister\n */\n\nimport type { Token } from '@/tokenize'\nimport type { FederalRegisterCitation } from '@/types/citation'\nimport type { TransformationMap } from '@/types/span'\n\n/**\n * Extracts Federal Register citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Volume: Federal Register volume number (e.g., \"85\")\n * - Page: Page number (e.g., \"12345\")\n * - Year: Optional publication year in parentheses (e.g., \"(2021)\")\n *\n * Confidence scoring:\n * - 0.9 (Federal Register format is standardized)\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns FederalRegisterCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"85 Fed. Reg. 12345\",\n *   span: { cleanStart: 10, cleanEnd: 28 },\n *   type: \"federalRegister\",\n *   patternId: \"federal-register\"\n * }\n * const citation = extractFederalRegister(token, transformationMap)\n * // citation = {\n * //   type: \"federalRegister\",\n * //   volume: 85,\n * //   page: 12345,\n * //   confidence: 0.9,\n * //   ...\n * // }\n * ```\n */\nexport function extractFederalRegister(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): FederalRegisterCitation {\n\tconst { text, span } = token\n\n\t// Parse volume-page using regex\n\t// Pattern: volume (digits) + \"Fed. Reg.\" + page (digits)\n\tconst federalRegisterRegex = /^(\\d+)\\s+Fed\\.\\s?Reg\\.\\s+(\\d+)/\n\tconst match = federalRegisterRegex.exec(text)\n\n\tif (!match) {\n\t\tthrow new Error(`Failed to parse Federal Register citation: ${text}`)\n\t}\n\n\tconst volume = Number.parseInt(match[1], 10)\n\tconst page = Number.parseInt(match[2], 10)\n\n\t// Extract optional year in parentheses\n\t// Pattern: \"(year)\" or \"(month day, year)\"\n\tconst yearRegex = /\\((?:.*?\\s)?(\\d{4})\\)/\n\tconst yearMatch = yearRegex.exec(text)\n\tconst year = yearMatch ? Number.parseInt(yearMatch[1], 10) : undefined\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Confidence: 0.9 (Federal Register format is standardized)\n\tconst confidence = 0.9\n\n\treturn {\n\t\ttype: 'federalRegister',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0,\n\t\tpatternsChecked: 1,\n\t\tvolume,\n\t\tpage,\n\t\tyear,\n\t}\n}\n","/**\n * Short-form Citation Extraction\n *\n * Parses tokenized short-form citations (Id., supra, short-form case) to extract\n * metadata. Short-form citations refer to earlier citations in the document.\n *\n * @module extract/extractShortForms\n */\n\nimport type { Token } from '@/tokenize'\nimport type { IdCitation, SupraCitation, ShortFormCaseCitation } from '@/types/citation'\nimport type { TransformationMap } from '@/types/span'\n\n/**\n * Extracts Id. citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Pincite: Optional page reference (e.g., \"253\" from \"Id. at 253\")\n *\n * Confidence scoring:\n * - 1.0 (Id. format is unambiguous and standardized)\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns IdCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"Id. at 253\",\n *   span: { cleanStart: 10, cleanEnd: 20 },\n *   type: \"case\",\n *   patternId: \"id\"\n * }\n * const citation = extractId(token, transformationMap)\n * // citation = {\n * //   type: \"id\",\n * //   pincite: 253,\n * //   confidence: 1.0,\n * //   ...\n * // }\n * ```\n */\nexport function extractId(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): IdCitation {\n\tconst { text, span } = token\n\n\t// Parse Id. with optional pincite\n\t// Pattern: Id. or Ibid. with optional \"at [page]\"\n\tconst idRegex = /[Ii](?:d|bid)\\.(?:\\s+at\\s+(\\d+))?/\n\tconst match = idRegex.exec(text)\n\n\tif (!match) {\n\t\tthrow new Error(`Failed to parse Id. citation: ${text}`)\n\t}\n\n\t// Extract pincite if present\n\tconst pincite = match[1] ? Number.parseInt(match[1], 10) : undefined\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Confidence: 1.0 (Id. format is unambiguous)\n\tconst confidence = 1.0\n\n\treturn {\n\t\ttype: 'id',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0,\n\t\tpatternsChecked: 1,\n\t\tpincite,\n\t}\n}\n\n/**\n * Extracts supra citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Party name: Name preceding \"supra\" (e.g., \"Smith\" from \"Smith, supra\")\n * - Pincite: Optional page reference (e.g., \"460\" from \"Smith, supra, at 460\")\n *\n * Confidence scoring:\n * - 0.9 (supra format is fairly standard but party name extraction can vary)\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns SupraCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"Smith, supra, at 460\",\n *   span: { cleanStart: 10, cleanEnd: 30 },\n *   type: \"case\",\n *   patternId: \"supra\"\n * }\n * const citation = extractSupra(token, transformationMap)\n * // citation = {\n * //   type: \"supra\",\n * //   partyName: \"Smith\",\n * //   pincite: 460,\n * //   confidence: 0.9,\n * //   ...\n * // }\n * ```\n */\nexport function extractSupra(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): SupraCitation {\n\tconst { text, span } = token\n\n\t// Parse party name and optional pincite\n\t// Pattern: word(s), supra [, at page]\n\t// Note: Matches party names including \"v.\" (e.g., \"Smith v. Jones\")\n\tconst supraRegex = /\\b([A-Z][a-zA-Z]+(?:(?:\\s+v\\.?\\s+|\\s+)[A-Z][a-zA-Z]+)*),?\\s+supra(?:,?\\s+at\\s+(\\d+))?/\n\tconst match = supraRegex.exec(text)\n\n\tif (!match) {\n\t\tthrow new Error(`Failed to parse supra citation: ${text}`)\n\t}\n\n\tconst partyName = match[1]\n\tconst pincite = match[2] ? Number.parseInt(match[2], 10) : undefined\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Confidence: 0.9 (supra format is fairly standard)\n\tconst confidence = 0.9\n\n\treturn {\n\t\ttype: 'supra',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0,\n\t\tpatternsChecked: 1,\n\t\tpartyName,\n\t\tpincite,\n\t}\n}\n\n/**\n * Extracts short-form case citation metadata from a tokenized citation.\n *\n * Parses token text to extract:\n * - Volume: Volume number\n * - Reporter: Reporter abbreviation\n * - Pincite: Page reference (from \"at [page]\" pattern)\n *\n * Confidence scoring:\n * - 0.7 (short-form case citations are more ambiguous than full citations)\n *\n * @param token - Token from tokenizer containing matched text and clean positions\n * @param transformationMap - Position mapping from clean → original text\n * @returns ShortFormCaseCitation with parsed metadata and translated positions\n *\n * @example\n * ```typescript\n * const token = {\n *   text: \"500 F.2d at 125\",\n *   span: { cleanStart: 10, cleanEnd: 25 },\n *   type: \"case\",\n *   patternId: \"short-form-case\"\n * }\n * const citation = extractShortFormCase(token, transformationMap)\n * // citation = {\n * //   type: \"shortFormCase\",\n * //   volume: 500,\n * //   reporter: \"F.2d\",\n * //   pincite: 125,\n * //   confidence: 0.7,\n * //   ...\n * // }\n * ```\n */\nexport function extractShortFormCase(\n\ttoken: Token,\n\ttransformationMap: TransformationMap,\n): ShortFormCaseCitation {\n\tconst { text, span } = token\n\n\t// Parse volume-reporter-at-page\n\t// Pattern: number space abbreviation space \"at\" space number\n\tconst shortFormRegex = /(\\d+)\\s+([A-Z][A-Za-z.\\s]+?(?:\\d[a-z])?)\\s+at\\s+(\\d+)/\n\tconst match = shortFormRegex.exec(text)\n\n\tif (!match) {\n\t\tthrow new Error(`Failed to parse short-form case citation: ${text}`)\n\t}\n\n\tconst volume = Number.parseInt(match[1], 10)\n\tconst reporter = match[2].trim() // Remove trailing spaces\n\tconst pincite = Number.parseInt(match[3], 10)\n\n\t// Translate positions from clean → original\n\tconst originalStart =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanStart) ?? span.cleanStart\n\tconst originalEnd =\n\t\ttransformationMap.cleanToOriginal.get(span.cleanEnd) ?? span.cleanEnd\n\n\t// Confidence: 0.7 (short-form citations are more ambiguous)\n\tconst confidence = 0.7\n\n\treturn {\n\t\ttype: 'shortFormCase',\n\t\ttext,\n\t\tspan: {\n\t\t\tcleanStart: span.cleanStart,\n\t\t\tcleanEnd: span.cleanEnd,\n\t\t\toriginalStart,\n\t\t\toriginalEnd,\n\t\t},\n\t\tconfidence,\n\t\tmatchedText: text,\n\t\tprocessTimeMs: 0,\n\t\tpatternsChecked: 1,\n\t\tvolume,\n\t\treporter,\n\t\tpincite,\n\t}\n}\n","/**\n * Scope Boundary Detection\n *\n * Detects paragraph/section boundaries in text and validates whether\n * an antecedent citation is within the resolution scope.\n */\n\nimport type { Citation } from '../types/citation'\nimport type { ScopeStrategy } from './types'\n\n/**\n * Detects paragraph boundaries from text and assigns each citation to a paragraph.\n *\n * @param text - Original document text\n * @param citations - Extracted citations with position spans\n * @param boundaryPattern - Regex pattern to detect boundaries (default: /\\n\\n+/)\n * @returns Map of citation index to paragraph number (0-based)\n */\nexport function detectParagraphBoundaries(\n  text: string,\n  citations: Citation[],\n  boundaryPattern: RegExp = /\\n\\n+/g\n): Map<number, number> {\n  const paragraphMap = new Map<number, number>()\n\n  // Find all paragraph boundaries (positions in text)\n  const boundaries: number[] = [0] // Start of document is first boundary\n  let match: RegExpExecArray | null\n\n  while ((match = boundaryPattern.exec(text)) !== null) {\n    // Boundary is at end of match (start of next paragraph)\n    boundaries.push(match.index + match[0].length)\n  }\n\n  boundaries.push(text.length) // End of document\n\n  // Assign each citation to a paragraph\n  for (let i = 0; i < citations.length; i++) {\n    const citation = citations[i]\n    const citationStart = citation.span.originalStart\n\n    // Find which paragraph this citation belongs to\n    let paragraphNum = 0\n    for (let j = 0; j < boundaries.length - 1; j++) {\n      if (citationStart >= boundaries[j] && citationStart < boundaries[j + 1]) {\n        paragraphNum = j\n        break\n      }\n    }\n\n    paragraphMap.set(i, paragraphNum)\n  }\n\n  return paragraphMap\n}\n\n/**\n * Checks if an antecedent citation is within resolution scope.\n *\n * @param antecedentIndex - Index of the antecedent citation\n * @param currentIndex - Index of current citation being resolved\n * @param paragraphMap - Map of citation index to paragraph number\n * @param strategy - Scope boundary strategy\n * @returns true if antecedent is within scope, false otherwise\n */\nexport function isWithinBoundary(\n  antecedentIndex: number,\n  currentIndex: number,\n  paragraphMap: Map<number, number>,\n  strategy: ScopeStrategy\n): boolean {\n  if (strategy === 'none') {\n    // No boundary restriction - can resolve across entire document\n    return true\n  }\n\n  // Get paragraph numbers for both citations\n  const antecedentParagraph = paragraphMap.get(antecedentIndex)\n  const currentParagraph = paragraphMap.get(currentIndex)\n\n  // If either is undefined, default to allowing resolution\n  if (antecedentParagraph === undefined || currentParagraph === undefined) {\n    return true\n  }\n\n  // For paragraph/section/footnote strategies, citations must be in same boundary\n  // (In this MVP, section and footnote behave same as paragraph - future enhancement)\n  return antecedentParagraph === currentParagraph\n}\n","/**\n * Levenshtein Distance\n *\n * Calculates edit distance between strings for fuzzy party name matching\n * in supra citation resolution.\n *\n * Uses dynamic programming for O(m*n) time complexity.\n */\n\n/**\n * Calculates Levenshtein distance (edit distance) between two strings.\n *\n * The edit distance is the minimum number of single-character edits\n * (insertions, deletions, substitutions) needed to change one string into the other.\n *\n * @param a - First string\n * @param b - Second string\n * @returns Number of edits required (0 = identical)\n */\nexport function levenshteinDistance(a: string, b: string): number {\n  // Handle empty strings\n  if (a.length === 0) return b.length\n  if (b.length === 0) return a.length\n\n  // Create 2D array for dynamic programming\n  // dp[i][j] = edit distance between a[0...i-1] and b[0...j-1]\n  const dp: number[][] = Array.from({ length: a.length + 1 }, () =>\n    Array(b.length + 1).fill(0)\n  )\n\n  // Initialize base cases\n  for (let i = 0; i <= a.length; i++) {\n    dp[i][0] = i // Distance from a[0...i-1] to empty string\n  }\n  for (let j = 0; j <= b.length; j++) {\n    dp[0][j] = j // Distance from empty string to b[0...j-1]\n  }\n\n  // Fill the DP table\n  for (let i = 1; i <= a.length; i++) {\n    for (let j = 1; j <= b.length; j++) {\n      if (a[i - 1] === b[j - 1]) {\n        // Characters match - no edit needed\n        dp[i][j] = dp[i - 1][j - 1]\n      } else {\n        // Characters differ - take minimum of:\n        // 1. Insert: dp[i][j-1] + 1\n        // 2. Delete: dp[i-1][j] + 1\n        // 3. Substitute: dp[i-1][j-1] + 1\n        dp[i][j] = 1 + Math.min(\n          dp[i - 1][j],     // Delete from a\n          dp[i][j - 1],     // Insert into a\n          dp[i - 1][j - 1]  // Substitute\n        )\n      }\n    }\n  }\n\n  return dp[a.length][b.length]\n}\n\n/**\n * Calculates normalized Levenshtein similarity (0-1 scale).\n *\n * Returns similarity score where:\n * - 1.0 = identical strings\n * - 0.0 = completely different\n *\n * Comparison is case-insensitive.\n *\n * @param a - First string\n * @param b - Second string\n * @returns Similarity score from 0 to 1\n */\nexport function normalizedLevenshteinDistance(a: string, b: string): number {\n  // Normalize to lowercase for case-insensitive comparison\n  const lowerA = a.toLowerCase()\n  const lowerB = b.toLowerCase()\n\n  // Calculate raw edit distance\n  const distance = levenshteinDistance(lowerA, lowerB)\n\n  // Normalize by max length\n  const maxLength = Math.max(lowerA.length, lowerB.length)\n  if (maxLength === 0) return 1.0 // Both empty strings\n\n  // Convert distance to similarity: 1 - (distance / maxLength)\n  return 1 - distance / maxLength\n}\n","/**\n * Document-Scoped Citation Resolver\n *\n * Resolves short-form citations (Id./supra/short-form case) to their full antecedent citations\n * by maintaining resolution context and enforcing scope boundaries.\n *\n * Resolution rules:\n * - Id. resolves to immediately preceding full citation (within scope)\n * - Supra resolves to full citation with matching party name (within scope)\n * - Short-form case resolves to full case with matching volume/reporter (within scope)\n */\n\nimport type {\n  Citation,\n  IdCitation,\n  SupraCitation,\n  ShortFormCaseCitation,\n  FullCaseCitation,\n} from '../types/citation'\nimport type {\n  ResolutionOptions,\n  ResolutionResult,\n  ResolvedCitation,\n  ResolutionContext,\n} from './types'\nimport { detectParagraphBoundaries, isWithinBoundary } from './scopeBoundary'\nimport { normalizedLevenshteinDistance } from './levenshtein'\n\n/**\n * Document-scoped resolver that processes citations sequentially\n * and resolves short-form citations to their antecedents.\n */\nexport class DocumentResolver {\n  private readonly citations: Citation[]\n  private readonly text: string\n  private readonly options: Required<ResolutionOptions>\n  private readonly context: ResolutionContext\n\n  /**\n   * Creates a new DocumentResolver.\n   *\n   * @param citations - All citations in document (in order of appearance)\n   * @param text - Original document text\n   * @param options - Resolution options\n   */\n  constructor(\n    citations: Citation[],\n    text: string,\n    options: ResolutionOptions = {}\n  ) {\n    this.citations = citations\n    this.text = text\n\n    // Apply defaults to options\n    this.options = {\n      scopeStrategy: options.scopeStrategy ?? 'paragraph',\n      autoDetectParagraphs: options.autoDetectParagraphs ?? true,\n      paragraphBoundaryPattern: options.paragraphBoundaryPattern ?? /\\n\\n+/g,\n      fuzzyPartyMatching: options.fuzzyPartyMatching ?? true,\n      partyMatchThreshold: options.partyMatchThreshold ?? 0.8,\n      allowNestedResolution: options.allowNestedResolution ?? false,\n      reportUnresolved: options.reportUnresolved ?? true,\n    }\n\n    // Initialize resolution context\n    this.context = {\n      citationIndex: 0,\n      allCitations: citations,\n      lastFullCitation: undefined,\n      fullCitationHistory: new Map(),\n      paragraphMap: new Map(),\n    }\n\n    // Detect paragraph boundaries if enabled\n    if (this.options.autoDetectParagraphs) {\n      this.context.paragraphMap = detectParagraphBoundaries(\n        text,\n        citations,\n        this.options.paragraphBoundaryPattern\n      )\n    }\n  }\n\n  /**\n   * Resolves all citations in the document.\n   *\n   * @returns Array of citations with resolution metadata\n   */\n  resolve(): ResolvedCitation[] {\n    const resolved: ResolvedCitation[] = []\n\n    for (let i = 0; i < this.citations.length; i++) {\n      this.context.citationIndex = i\n      const citation = this.citations[i]\n\n      // Resolve based on citation type\n      let resolution: ResolutionResult | undefined\n\n      switch (citation.type) {\n        case 'id':\n          resolution = this.resolveId(citation as IdCitation)\n          break\n        case 'supra':\n          resolution = this.resolveSupra(citation as SupraCitation)\n          break\n        case 'shortFormCase':\n          resolution = this.resolveShortFormCase(citation as ShortFormCaseCitation)\n          break\n        default:\n          // Full citation - update context for future resolutions\n          if (this.isFullCitation(citation)) {\n            this.context.lastFullCitation = i\n            this.trackFullCitation(citation, i)\n          }\n          break\n      }\n\n      // Add citation with optional resolution metadata\n      resolved.push({\n        ...citation,\n        resolution,\n      })\n    }\n\n    return resolved\n  }\n\n  /**\n   * Resolves Id. citation to immediately preceding full case citation.\n   */\n  private resolveId(citation: IdCitation): ResolutionResult | undefined {\n    const currentIndex = this.context.citationIndex\n\n    // Find most recent full case citation (Id. only resolves to case citations, not statutes/journals)\n    let antecedentIndex: number | undefined\n    for (let i = currentIndex - 1; i >= 0; i--) {\n      const candidate = this.citations[i]\n      if (candidate.type === 'case') {\n        antecedentIndex = i\n        break\n      }\n    }\n\n    // Check if we have a previous case citation\n    if (antecedentIndex === undefined) {\n      return this.createFailureResult('No preceding full case citation found')\n    }\n\n    // Check scope boundary\n    if (!this.isWithinScope(antecedentIndex, currentIndex)) {\n      return this.createFailureResult('Antecedent citation outside scope boundary')\n    }\n\n    return {\n      resolvedTo: antecedentIndex,\n      confidence: 1.0, // Id. resolution is unambiguous when successful\n    }\n  }\n\n  /**\n   * Resolves supra citation by matching party name.\n   */\n  private resolveSupra(citation: SupraCitation): ResolutionResult | undefined {\n    const currentIndex = this.context.citationIndex\n    const targetPartyName = this.normalizePartyName(citation.partyName)\n\n    // Search full citation history for matching party name\n    let bestMatch: { index: number; similarity: number } | undefined\n\n    for (const [partyName, citationIndex] of this.context.fullCitationHistory) {\n      // Check scope boundary\n      if (!this.isWithinScope(citationIndex, currentIndex)) {\n        continue\n      }\n\n      // Calculate similarity\n      const similarity = normalizedLevenshteinDistance(targetPartyName, partyName)\n\n      // Update best match if this is better\n      if (!bestMatch || similarity > bestMatch.similarity) {\n        bestMatch = { index: citationIndex, similarity }\n      }\n    }\n\n    // Check if we found a match above threshold\n    if (!bestMatch) {\n      return this.createFailureResult('No full citation found in scope')\n    }\n\n    if (bestMatch.similarity < this.options.partyMatchThreshold) {\n      return this.createFailureResult(\n        `Party name similarity ${bestMatch.similarity.toFixed(2)} below threshold ${this.options.partyMatchThreshold}`\n      )\n    }\n\n    // Return successful resolution with confidence based on similarity\n    const warnings: string[] = []\n    if (bestMatch.similarity < 1.0) {\n      warnings.push(`Fuzzy match: similarity ${bestMatch.similarity.toFixed(2)}`)\n    }\n\n    return {\n      resolvedTo: bestMatch.index,\n      confidence: bestMatch.similarity,\n      warnings: warnings.length > 0 ? warnings : undefined,\n    }\n  }\n\n  /**\n   * Resolves short-form case citation by matching volume/reporter.\n   */\n  private resolveShortFormCase(citation: ShortFormCaseCitation): ResolutionResult | undefined {\n    const currentIndex = this.context.citationIndex\n\n    // Search backwards for matching full case citation\n    for (let i = currentIndex - 1; i >= 0; i--) {\n      const candidate = this.citations[i]\n\n      // Only match against full case citations\n      if (candidate.type !== 'case') {\n        continue\n      }\n\n      const fullCase = candidate as FullCaseCitation\n\n      // Check if volume and reporter match\n      if (\n        fullCase.volume === citation.volume &&\n        this.normalizeReporter(fullCase.reporter) === this.normalizeReporter(citation.reporter)\n      ) {\n        // Check scope boundary\n        if (!this.isWithinScope(i, currentIndex)) {\n          return this.createFailureResult('Matching citation outside scope boundary')\n        }\n\n        // Found a match\n        return {\n          resolvedTo: i,\n          confidence: 0.95, // High confidence but not perfect (multiple cases could have same volume/reporter)\n        }\n      }\n    }\n\n    return this.createFailureResult('No matching full case citation found')\n  }\n\n  /**\n   * Checks if a citation is a full citation (not short-form).\n   */\n  private isFullCitation(citation: Citation): boolean {\n    return (\n      citation.type === 'case' ||\n      citation.type === 'statute' ||\n      citation.type === 'journal' ||\n      citation.type === 'neutral' ||\n      citation.type === 'publicLaw' ||\n      citation.type === 'federalRegister'\n    )\n  }\n\n  /**\n   * Tracks a full citation in the resolution history.\n   * Extracts party name for supra resolution.\n   */\n  private trackFullCitation(citation: Citation, index: number): void {\n    // Only case citations have party names for supra resolution\n    if (citation.type === 'case') {\n      const fullCase = citation as FullCaseCitation\n      const partyName = this.extractPartyName(fullCase)\n      if (partyName) {\n        const normalized = this.normalizePartyName(partyName)\n        this.context.fullCitationHistory.set(normalized, index)\n      }\n    }\n  }\n\n  /**\n   * Extracts party name from full case citation text.\n   * Handles \"Party v. Party\" format by looking at text before citation span.\n   */\n  private extractPartyName(citation: FullCaseCitation): string | undefined {\n    // Look at text before citation span to find party names\n    // Case citations typically appear as: \"Smith v. Jones, 100 F.2d 10\"\n    // But tokenizer only captures \"100 F.2d 10\" - we need to look backwards in text\n\n    const citationStart = citation.span.originalStart\n    // Look backwards up to 100 characters for party name\n    const lookbackStart = Math.max(0, citationStart - 100)\n    const beforeText = this.text.substring(lookbackStart, citationStart)\n\n    // Match pattern: \"FirstParty v. SecondParty, \" before the citation\n    // Capture the first party name (handles single-letter party names like \"A\" or \"B\")\n    const vMatch = beforeText.match(/([A-Z][a-zA-Z]*(?:\\s+[A-Z][a-zA-Z]*)*)\\s+v\\.?\\s+[A-Z][a-zA-Z]*(?:\\s+[A-Z][a-zA-Z]*)*,\\s*$/)\n    if (vMatch) {\n      return vMatch[1].trim()\n    }\n\n    // Fallback: try to find any capitalized word(s) before comma\n    const beforeComma = beforeText.match(/([A-Z][a-zA-Z]*(?:\\s+[A-Z][a-zA-Z]*)*),\\s*$/)\n    return beforeComma?.[1].trim()\n  }\n\n  /**\n   * Normalizes party name for matching.\n   */\n  private normalizePartyName(name: string): string {\n    return name\n      .toLowerCase()\n      .replace(/\\s+/g, ' ') // Normalize whitespace\n      .trim()\n  }\n\n  /**\n   * Normalizes reporter abbreviation for matching.\n   */\n  private normalizeReporter(reporter: string): string {\n    return reporter\n      .toLowerCase()\n      .replace(/\\s+/g, '') // Remove spaces (F.2d vs F. 2d)\n      .replace(/\\./g, '')   // Remove periods\n  }\n\n  /**\n   * Checks if antecedent citation is within scope boundary.\n   */\n  private isWithinScope(antecedentIndex: number, currentIndex: number): boolean {\n    return isWithinBoundary(\n      antecedentIndex,\n      currentIndex,\n      this.context.paragraphMap,\n      this.options.scopeStrategy\n    )\n  }\n\n  /**\n   * Creates a failure result for unresolved citations.\n   */\n  private createFailureResult(reason: string): ResolutionResult | undefined {\n    if (this.options.reportUnresolved) {\n      return {\n        resolvedTo: undefined,\n        failureReason: reason,\n        confidence: 0.0,\n      }\n    }\n    return undefined\n  }\n}\n","/**\n * Citation Resolution\n *\n * Resolves short-form citations (Id./supra/short-form case) to their full antecedents.\n *\n * @example\n * ```ts\n * import { resolveCitations } from 'eyecite-ts/resolve'\n * import { extractCitations } from 'eyecite-ts'\n *\n * const text = 'See Smith v. Jones, 500 F.2d 100 (1974). Id. at 105.'\n * const citations = extractCitations(text)\n * const resolved = resolveCitations(citations, text)\n *\n * // resolved[1] is Id. citation with resolution.resolvedTo = 0\n * console.log(resolved[1].resolution?.resolvedTo) // 0 (points to Smith v. Jones)\n * ```\n */\n\nimport type { Citation } from '../types/citation'\nimport type { ResolutionOptions, ResolvedCitation } from './types'\nimport { DocumentResolver } from './DocumentResolver'\n\n/**\n * Resolves short-form citations to their full antecedents.\n *\n * Convenience wrapper around DocumentResolver that handles common use cases.\n *\n * @param citations - Extracted citations in order of appearance\n * @param text - Original document text\n * @param options - Resolution options\n * @returns Citations with resolution metadata\n */\nexport function resolveCitations(\n  citations: Citation[],\n  text: string,\n  options?: ResolutionOptions\n): ResolvedCitation[] {\n  const resolver = new DocumentResolver(citations, text, options)\n  return resolver.resolve()\n}\n\n// Re-export core types and classes\nexport { DocumentResolver } from './DocumentResolver'\nexport type {\n  ResolutionOptions,\n  ResolutionResult,\n  ResolvedCitation,\n  ScopeStrategy,\n} from './types'\n","/**\n * Main Citation Extraction Pipeline\n *\n * Orchestrates the complete citation extraction flow:\n *   1. Clean text (remove HTML, normalize Unicode)\n *   2. Tokenize (apply patterns to find candidates)\n *   3. Extract (parse metadata from tokens)\n *\n * This is the primary public API for citation extraction.\n *\n * @module extract/extractCitations\n */\n\nimport { cleanText } from '@/clean'\nimport { tokenize } from '@/tokenize'\nimport {\n\textractCase,\n\textractStatute,\n\textractJournal,\n\textractNeutral,\n\textractPublicLaw,\n\textractFederalRegister,\n} from '@/extract'\nimport { extractId, extractSupra, extractShortFormCase } from './extractShortForms'\nimport {\n\tcasePatterns,\n\tstatutePatterns,\n\tjournalPatterns,\n\tneutralPatterns,\n\tshortFormPatterns,\n} from '@/patterns'\nimport { resolveCitations } from '../resolve'\nimport type { Citation } from '@/types/citation'\nimport type { Pattern } from '@/patterns'\nimport type { ResolutionOptions, ResolvedCitation } from '../resolve/types'\n\n/**\n * Options for customizing citation extraction behavior.\n */\nexport interface ExtractOptions {\n\t/**\n\t * Custom text cleaners (overrides defaults).\n\t *\n\t * If provided, these cleaners replace the default pipeline:\n\t * [stripHtmlTags, normalizeWhitespace, normalizeUnicode, fixSmartQuotes]\n\t *\n\t * @example\n\t * ```typescript\n\t * // Use only HTML stripping, skip Unicode normalization\n\t * const citations = extractCitations(text, {\n\t *   cleaners: [stripHtmlTags]\n\t * })\n\t * ```\n\t */\n\tcleaners?: Array<(text: string) => string>\n\n\t/**\n\t * Custom regex patterns (overrides defaults).\n\t *\n\t * If provided, these patterns replace the default pattern set:\n\t * [casePatterns, statutePatterns, journalPatterns, neutralPatterns, shortFormPatterns]\n\t *\n\t * @example\n\t * ```typescript\n\t * // Extract only case citations\n\t * const citations = extractCitations(text, {\n\t *   patterns: casePatterns\n\t * })\n\t * ```\n\t */\n\tpatterns?: Pattern[]\n\n\t/**\n\t * Resolve short-form citations to their full antecedents (default: false).\n\t *\n\t * If true, returns ResolvedCitation[] with resolution metadata for short-form citations\n\t * (Id., supra, short-form case). Full citations are unchanged.\n\t *\n\t * @example\n\t * ```typescript\n\t * const text = \"Smith v. Jones, 500 F.2d 100 (1974). Id. at 105.\"\n\t * const citations = extractCitations(text, { resolve: true })\n\t * // citations[1].resolution.resolvedTo === 0 (points to Smith v. Jones)\n\t * ```\n\t */\n\tresolve?: boolean\n\n\t/**\n\t * Options for citation resolution (only used if resolve: true).\n\t *\n\t * @example\n\t * ```typescript\n\t * const citations = extractCitations(text, {\n\t *   resolve: true,\n\t *   resolutionOptions: {\n\t *     scopeStrategy: 'paragraph',\n\t *     fuzzyPartyMatching: true\n\t *   }\n\t * })\n\t * ```\n\t */\n\tresolutionOptions?: ResolutionOptions\n}\n\n/**\n * Extracts legal citations from text using the full parsing pipeline.\n *\n * Pipeline flow:\n * 1. **Clean:** Remove HTML tags, normalize Unicode, fix smart quotes\n * 2. **Tokenize:** Apply regex patterns to find citation candidates\n * 3. **Extract:** Parse metadata (volume, reporter, page, etc.)\n * 4. **Translate:** Map positions from cleaned text back to original text\n *\n * This function is synchronous because all stages (cleaning, tokenization,\n * extraction) are synchronous. For async operations (e.g., future reporters-db\n * lookups), use extractCitationsAsync().\n *\n * Position tracking:\n * - TransformationMap is built during cleaning\n * - Tokens contain positions in cleaned text (cleanStart/cleanEnd)\n * - Extraction translates cleaned positions → original positions\n * - Final citations have originalStart/originalEnd pointing to input text\n *\n * Warnings from cleaning layer are attached to all extracted citations.\n *\n * @param text - Raw text to extract citations from (may contain HTML, Unicode)\n * @param options - Optional customization (cleaners, patterns)\n * @returns Array of citations with parsed metadata and accurate positions\n *\n * @example\n * ```typescript\n * const text = \"See Smith v. Doe, 500 F.2d 123 (9th Cir. 2020)\"\n * const citations = extractCitations(text)\n * // citations[0] = {\n * //   type: \"case\",\n * //   volume: 500,\n * //   reporter: \"F.2d\",\n * //   page: 123,\n * //   court: \"9th Cir.\",\n * //   year: 2020,\n * //   span: { originalStart: 18, originalEnd: 30, ... }\n * // }\n * ```\n *\n * @example\n * ```typescript\n * // Extract from HTML\n * const html = \"<p>In <b>Smith</b>, 500 F.2d 123, the court held...</p>\"\n * const citations = extractCitations(html)\n * // HTML is stripped, positions point to original HTML\n * ```\n *\n * @example\n * ```typescript\n * // Extract multiple citation types\n * const text = \"See 42 U.S.C. § 1983; Smith, 500 F.2d 123; 123 Harv. L. Rev. 456\"\n * const citations = extractCitations(text)\n * // citations[0].type === \"statute\"\n * // citations[1].type === \"case\"\n * // citations[2].type === \"journal\"\n * ```\n */\nexport function extractCitations(\n\ttext: string,\n\toptions?: ExtractOptions,\n): Citation[] | ResolvedCitation[] {\n\tconst startTime = performance.now()\n\n\t// Step 1: Clean text\n\tconst { cleaned, transformationMap, warnings } = cleanText(\n\t\ttext,\n\t\toptions?.cleaners,\n\t)\n\n\t// Step 2: Tokenize (synchronous)\n\t// Note: Pattern order matters for deduplication - more specific patterns first\n\tconst allPatterns = options?.patterns || [\n\t\t...neutralPatterns,      // Most specific (year-based format)\n\t\t...shortFormPatterns,    // Short-form (requires \" at \" keyword)\n\t\t...casePatterns,         // Case citations (reporter-specific)\n\t\t...statutePatterns,      // Statutes (code-specific)\n\t\t...journalPatterns,      // Least specific (broad pattern)\n\t]\n\tconst tokens = tokenize(cleaned, allPatterns)\n\n\t// Step 3: Deduplicate overlapping tokens\n\t// Multiple patterns may match the same text (e.g., \"500 F.2d 123\" matches both federal-reporter and state-reporter)\n\t// Keep only the most specific match for each position\n\tconst deduplicatedTokens: typeof tokens = []\n\tconst seenPositions = new Set<string>()\n\n\tfor (const token of tokens) {\n\t\tconst posKey = `${token.span.cleanStart}-${token.span.cleanEnd}`\n\t\tif (!seenPositions.has(posKey)) {\n\t\t\tseenPositions.add(posKey)\n\t\t\tdeduplicatedTokens.push(token)\n\t\t}\n\t}\n\n\t// Step 4: Extract citations from deduplicated tokens\n\tconst citations: Citation[] = []\n\tfor (const token of deduplicatedTokens) {\n\t\tlet citation: Citation\n\n\t\tswitch (token.type) {\n\t\t\tcase 'case':\n\t\t\t\t// Check pattern ID to distinguish short-form from full citations\n\t\t\t\tif (token.patternId === 'id' || token.patternId === 'ibid') {\n\t\t\t\t\tcitation = extractId(token, transformationMap)\n\t\t\t\t} else if (token.patternId === 'supra') {\n\t\t\t\t\tcitation = extractSupra(token, transformationMap)\n\t\t\t\t} else if (token.patternId === 'shortFormCase') {\n\t\t\t\t\tcitation = extractShortFormCase(token, transformationMap)\n\t\t\t\t} else {\n\t\t\t\t\tcitation = extractCase(token, transformationMap)\n\t\t\t\t}\n\t\t\t\tbreak\n\t\t\tcase 'statute':\n\t\t\t\tcitation = extractStatute(token, transformationMap)\n\t\t\t\tbreak\n\t\t\tcase 'journal':\n\t\t\t\tcitation = extractJournal(token, transformationMap)\n\t\t\t\tbreak\n\t\t\tcase 'neutral':\n\t\t\t\tcitation = extractNeutral(token, transformationMap)\n\t\t\t\tbreak\n\t\t\tcase 'publicLaw':\n\t\t\t\tcitation = extractPublicLaw(token, transformationMap)\n\t\t\t\tbreak\n\t\t\tcase 'federalRegister':\n\t\t\t\tcitation = extractFederalRegister(token, transformationMap)\n\t\t\t\tbreak\n\t\t\tdefault:\n\t\t\t\t// Unknown type - skip\n\t\t\t\tcontinue\n\t\t}\n\n\t\t// Attach cleaning warnings to citation if any\n\t\tif (warnings.length > 0) {\n\t\t\tcitation.warnings = [...(citation.warnings || []), ...warnings]\n\t\t}\n\n\t\t// Update processing time\n\t\tcitation.processTimeMs = performance.now() - startTime\n\n\t\tcitations.push(citation)\n\t}\n\n\t// Step 5: Resolve short-form citations if requested\n\tif (options?.resolve) {\n\t\treturn resolveCitations(citations, text, options.resolutionOptions)\n\t}\n\n\treturn citations\n}\n\n/**\n * Asynchronous version of extractCitations().\n *\n * Currently wraps the synchronous extractCitations() function. This API\n * exists for future extensibility when async operations are added:\n * - Async reporters-db lookups (Phase 3)\n * - Async resolution/annotation services\n * - Web Workers for parallel processing\n *\n * For now, this function immediately resolves with the same results as\n * the synchronous version.\n *\n * @param text - Raw text to extract citations from\n * @param options - Optional customization (cleaners, patterns, resolve)\n * @returns Promise resolving to array of citations (or ResolvedCitation[] if resolve: true)\n *\n * @example\n * ```typescript\n * const citations = await extractCitationsAsync(text, { resolve: true })\n * // Returns ResolvedCitation[] with resolution metadata\n * ```\n */\nexport async function extractCitationsAsync(\n\ttext: string,\n\toptions?: ExtractOptions,\n): Promise<Citation[] | ResolvedCitation[]> {\n\t// Async wrapper for future extensibility (e.g., async reporters-db lookup)\n\t// For MVP, wraps synchronous extractCitations\n\treturn extractCitations(text, options)\n}\n"],"mappings":"mEAcA,SAAgB,EAAc,EAAsB,CACnD,OAAO,EAAK,QAAQ,WAAY,GAAG,CAUpC,SAAgB,EAAoB,EAAsB,CACzD,OAAO,EAAK,QAAQ,aAAc,IAAI,CAAC,QAAQ,SAAU,IAAI,CAU9D,SAAgB,EAAiB,EAAsB,CACtD,OAAO,EAAK,UAAU,OAAO,CAU9B,SAAgB,EAAe,EAAsB,CACpD,OAAO,EACL,QAAQ,kBAAmB,IAAI,CAC/B,QAAQ,kBAAmB,IAAI,CCHlC,SAAgB,EACf,EACA,EAA4C,CAC3C,EACA,EACA,EACA,EACA,CACiB,CAElB,IAAI,EAAc,EACd,EAAkB,IAAI,IACtB,EAAkB,IAAI,IAG1B,IAAK,IAAI,EAAI,EAAG,GAAK,EAAS,OAAQ,IACrC,EAAgB,IAAI,EAAG,EAAE,CACzB,EAAgB,IAAI,EAAG,EAAE,CAI1B,IAAK,IAAM,KAAW,EAAU,CAC/B,IAAM,EAAa,EACb,EAAY,EAAQ,EAAY,CAEtC,GAAI,IAAe,EAAW,CAE7B,GAAM,CAAE,qBAAoB,sBAAuB,EAClD,EACA,EACA,EACA,EACA,CAED,EAAkB,EAClB,EAAkB,EAClB,EAAc,GAShB,MAAO,CACN,QAAS,EACT,kBAP4C,CAC5C,kBACA,kBACA,CAKA,SAAU,EAAE,CACZ,CAeF,SAAS,EACR,EACA,EACA,EACA,EAIC,CACD,IAAM,EAAqB,IAAI,IACzB,EAAqB,IAAI,IAE3B,EAAY,EACZ,EAAW,EAGf,KAAO,GAAa,EAAW,QAAU,GAAY,EAAU,QAAQ,CAEtE,GAAI,GAAa,EAAW,QAAU,GAAY,EAAU,OAAQ,CACnE,IAAM,EAAc,EAAmB,IAAI,EAAU,EAAI,EACzD,EAAmB,IAAI,EAAU,EAAY,CAC7C,EAAmB,IAAI,EAAa,EAAS,CAC7C,MAID,GAAI,GAAa,EAAW,OAAQ,CACnC,IAAM,EAAc,EAAmB,IAAI,EAAU,EAAI,EACzD,EAAmB,IAAI,EAAU,EAAY,CAC7C,IACA,SAID,GAAI,GAAY,EAAU,OAAQ,CACjC,IAAM,EAAc,EAAmB,IAAI,EAAU,EAAI,EACzD,EAAmB,IAAI,EAAa,EAAS,CAC7C,IACA,SAID,GAAI,EAAW,KAAe,EAAU,GAAW,CAClD,IAAM,EAAc,EAAmB,IAAI,EAAU,EAAI,EACzD,EAAmB,IAAI,EAAU,EAAY,CAC7C,EAAmB,IAAI,EAAa,EAAS,CAC7C,IACA,QACM,CAGN,IAAI,EAAa,GAIjB,IAAK,IAAI,EAAY,EAAG,GAAa,IAChC,IAAY,GAAa,EAAW,QADU,IAGlD,GAAI,EAAW,EAAY,KAAe,EAAU,GAAW,CAE9D,IAAK,IAAI,EAAI,EAAG,EAAI,EAAW,IAAK,CACnC,IAAM,EACL,EAAmB,IAAI,EAAY,EAAE,EAAI,EAAY,EACtD,EAAmB,IAAI,EAAa,EAAS,CAE9C,GAAa,EACb,EAAa,GACb,MAIF,GAAI,EAAY,SAGhB,IAAK,IAAI,EAAY,EAAG,GAAa,IAChC,IAAW,GAAa,EAAU,QADY,IAGlD,GAAI,EAAW,KAAe,EAAU,EAAW,GAAY,CAE9D,IAAM,EAAc,EAAmB,IAAI,EAAU,EAAI,EACzD,IAAK,IAAI,EAAI,EAAG,EAAI,EAAW,IAC9B,EAAmB,IAAI,EAAW,EAAG,EAAY,CAElD,GAAY,EACZ,EAAa,GACb,MAIF,GAAI,EAAY,SAGhB,IAAM,EAAc,EAAmB,IAAI,EAAU,EAAI,EACzD,EAAmB,IAAI,EAAU,EAAY,CAC7C,EAAmB,IAAI,EAAa,EAAS,CAC7C,IACA,KAIF,MAAO,CAAE,qBAAoB,qBAAoB,CC9LlD,MAAa,EAA0B,CACrC,CACE,GAAI,mBACJ,MAAO,0FACP,YAAa,mDACb,KAAM,OACP,CACD,CACE,GAAI,gBACJ,MAAO,gEACP,YAAa,+BACb,KAAM,OACP,CACD,CACE,GAAI,iBACJ,MAAO,0DACP,YAAa,6EACb,KAAM,OACP,CACF,CCzBY,EAA6B,CACxC,CACE,GAAI,MACJ,MAAO,uCACP,YAAa,iDACb,KAAM,UACP,CACD,CACE,GAAI,aACJ,MAAO,yDACP,YAAa,sEACb,KAAM,UACP,CACF,CCbY,EAA6B,CACxC,CACE,GAAI,aACJ,MAAO,4CACP,YAAa,iGACb,KAAM,UACP,CACF,CCPY,EAA6B,CACxC,CACE,GAAI,UACJ,MAAO,4BACP,YAAa,6CACb,KAAM,UACP,CACD,CACE,GAAI,QACJ,MAAO,wCACP,YAAa,sDACb,KAAM,UACP,CACD,CACE,GAAI,aACJ,MAAO,sCACP,YAAa,oDACb,KAAM,YACP,CACD,CACE,GAAI,mBACJ,MAAO,qCACP,YAAa,0DACb,KAAM,kBACP,CACF,CCIY,EAA+B,CAC1C,CACE,GAAI,KACJ,MAhC8B,+BAiC9B,YAAa,8CACb,KAAM,OACP,CACD,CACE,GAAI,OACJ,MAnCgC,iCAoChC,YAAa,oDACb,KAAM,OACP,CACD,CACE,GAAI,QACJ,MAjCiC,yFAkCjC,YAAa,mEACb,KAAM,OACP,CACD,CACE,GAAI,gBACJ,MAhC2C,6DAiC3C,YAAa,sDACb,KAAM,OACP,CACF,CCcD,SAAgB,EACd,EACA,EAAsB,CACpB,GAAG,EACH,GAAG,EACH,GAAG,EACH,GAAG,EACH,GAAG,EACJ,CACQ,CACT,IAAM,EAAkB,EAAE,CAE1B,IAAK,IAAM,KAAW,EACpB,GAAI,CAEF,IAAM,EAAU,EAAY,SAAS,EAAQ,MAAM,CAEnD,IAAK,IAAM,KAAS,EAElB,EAAO,KAAK,CACV,KAAM,EAAM,GACZ,KAAM,CACJ,WAAY,EAAM,MAClB,SAAU,EAAM,MAAS,EAAM,GAAG,OACnC,CACD,KAAM,EAAQ,KACd,UAAW,EAAQ,GACpB,CAAC,OAEG,EAAO,CAEd,QAAQ,KACN,WAAW,EAAQ,GAAG,yBACtB,aAAiB,MAAQ,EAAM,QAAU,OAAO,EAAM,CACvD,CACD,SAOJ,OAFA,EAAO,MAAM,EAAG,IAAM,EAAE,KAAK,WAAa,EAAE,KAAK,WAAW,CAErD,ECtDT,SAAgB,EACf,EACA,EACmB,CACnB,GAAM,CAAE,OAAM,QAAS,EAMjB,EAD0B,qCACM,KAAK,EAAK,CAEhD,GAAI,CAAC,EAEJ,MAAU,MAAM,kCAAkC,IAAO,CAG1D,IAAM,EAAS,OAAO,SAAS,EAAM,GAAI,GAAG,CACtC,EAAW,EAAM,GAAG,MAAM,CAC1B,EAAO,OAAO,SAAS,EAAM,GAAI,GAAG,CAKpC,EADe,YACa,KAAK,EAAK,CACtC,EAAU,EAAe,OAAO,SAAS,EAAa,GAAI,GAAG,CAAG,IAAA,GAKhE,EADY,0BACU,KAAK,EAAK,CAChC,EAAO,EAAY,OAAO,SAAS,EAAU,GAAI,GAAG,CAAG,IAAA,GAKvD,EADa,2BACW,KAAK,EAAK,CAClC,EAAQ,EAAa,EAAW,GAAG,MAAM,CAAG,IAAA,GAG5C,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAG1D,EAAa,GA+CjB,MA5CwB,4JA2BvB,CAEmB,KAAM,GAAM,EAAS,SAAS,EAAE,CAAC,GACpD,GAAc,IAIX,IAAS,IAAA,IAER,GADgB,IAAI,MAAM,CAAC,aAAa,GAE3C,GAAc,IAKhB,EAAa,KAAK,IAAI,EAAY,EAAI,CAE/B,CACN,KAAM,OACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,aACA,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,SACA,WACA,OACA,UACA,QACA,OACA,CCtIF,SAAgB,EACf,EACA,EACkB,CAClB,GAAM,CAAE,OAAM,QAAS,EAKjB,EADe,0DACM,KAAK,EAAK,CAErC,GAAI,CAAC,EACJ,MAAU,MAAM,qCAAqC,IAAO,CAG7D,IAAM,EAAQ,EAAM,GAAK,OAAO,SAAS,EAAM,GAAI,GAAG,CAAG,IAAA,GACnD,EAAO,EAAM,GAAG,MAAM,CACtB,EAAU,EAAM,GAGhB,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAG1D,EAAa,GAkBjB,MAfmB,CAClB,SACA,SACA,iBACA,kBACA,0BACA,8BACA,CAEc,KAAM,GAAM,EAAK,SAAS,EAAE,CAAC,GAC3C,GAAc,IAGf,EAAa,KAAK,IAAI,EAAY,EAAI,CAE/B,CACN,KAAM,UACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,aACA,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,QACA,OACA,UACA,CC1DF,SAAgB,EACf,EACA,EACkB,CAClB,GAAM,CAAE,OAAM,QAAS,EAKjB,EADe,mCACM,KAAK,EAAK,CAErC,GAAI,CAAC,EACJ,MAAU,MAAM,qCAAqC,IAAO,CAG7D,IAAM,EAAS,OAAO,SAAS,EAAM,GAAI,GAAG,CACtC,EAAU,EAAM,GAAG,MAAM,CACzB,EAAO,OAAO,SAAS,EAAM,GAAI,GAAG,CAIpC,EADe,YACa,KAAK,EAAK,CACtC,EAAU,EAAe,OAAO,SAAS,EAAa,GAAI,GAAG,CAAG,IAAA,GAGhE,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAK9D,MAAO,CACN,KAAM,UACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,WAXkB,GAYlB,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,SACA,UACA,aAAc,EACd,OACA,UACA,CCvDF,SAAgB,EACf,EACA,EACkB,CAClB,GAAM,CAAE,OAAM,QAAS,EAKjB,EADe,+CACM,KAAK,EAAK,CAErC,GAAI,CAAC,EACJ,MAAU,MAAM,qCAAqC,IAAO,CAG7D,IAAM,EAAO,OAAO,SAAS,EAAM,GAAI,GAAG,CACpC,EAAQ,EAAM,GACd,EAAiB,EAAM,GAGvB,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAK9D,MAAO,CACN,KAAM,UACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,WAXkB,EAYlB,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,OACA,QACA,iBACA,CC3CF,SAAgB,EACf,EACA,EACoB,CACpB,GAAM,CAAE,OAAM,QAAS,EAKjB,EADiB,wCACM,KAAK,EAAK,CAEvC,GAAI,CAAC,EACJ,MAAU,MAAM,wCAAwC,IAAO,CAGhE,IAAM,EAAW,OAAO,SAAS,EAAM,GAAI,GAAG,CACxC,EAAY,OAAO,SAAS,EAAM,GAAI,GAAG,CAGzC,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAK9D,MAAO,CACN,KAAM,YACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,WAXkB,GAYlB,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,WACA,YACA,CC5CF,SAAgB,EACf,EACA,EAC0B,CAC1B,GAAM,CAAE,OAAM,QAAS,EAKjB,EADuB,iCACM,KAAK,EAAK,CAE7C,GAAI,CAAC,EACJ,MAAU,MAAM,8CAA8C,IAAO,CAGtE,IAAM,EAAS,OAAO,SAAS,EAAM,GAAI,GAAG,CACtC,EAAO,OAAO,SAAS,EAAM,GAAI,GAAG,CAKpC,EADY,wBACU,KAAK,EAAK,CAChC,EAAO,EAAY,OAAO,SAAS,EAAU,GAAI,GAAG,CAAG,IAAA,GAGvD,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAK9D,MAAO,CACN,KAAM,kBACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,WAXkB,GAYlB,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,SACA,OACA,OACA,CCpDF,SAAgB,EACf,EACA,EACa,CACb,GAAM,CAAE,OAAM,QAAS,EAKjB,EADU,oCACM,KAAK,EAAK,CAEhC,GAAI,CAAC,EACJ,MAAU,MAAM,iCAAiC,IAAO,CAIzD,IAAM,EAAU,EAAM,GAAK,OAAO,SAAS,EAAM,GAAI,GAAG,CAAG,IAAA,GAGrD,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAK9D,MAAO,CACN,KAAM,KACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,WAXkB,EAYlB,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,UACA,CAmCF,SAAgB,EACf,EACA,EACgB,CAChB,GAAM,CAAE,OAAM,QAAS,EAMjB,EADa,wFACM,KAAK,EAAK,CAEnC,GAAI,CAAC,EACJ,MAAU,MAAM,mCAAmC,IAAO,CAG3D,IAAM,EAAY,EAAM,GAClB,EAAU,EAAM,GAAK,OAAO,SAAS,EAAM,GAAI,GAAG,CAAG,IAAA,GAGrD,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAK9D,MAAO,CACN,KAAM,QACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,WAXkB,GAYlB,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,YACA,UACA,CAqCF,SAAgB,EACf,EACA,EACwB,CACxB,GAAM,CAAE,OAAM,QAAS,EAKjB,EADiB,wDACM,KAAK,EAAK,CAEvC,GAAI,CAAC,EACJ,MAAU,MAAM,6CAA6C,IAAO,CAGrE,IAAM,EAAS,OAAO,SAAS,EAAM,GAAI,GAAG,CACtC,EAAW,EAAM,GAAG,MAAM,CAC1B,EAAU,OAAO,SAAS,EAAM,GAAI,GAAG,CAGvC,EACL,EAAkB,gBAAgB,IAAI,EAAK,WAAW,EAAI,EAAK,WAC1D,EACL,EAAkB,gBAAgB,IAAI,EAAK,SAAS,EAAI,EAAK,SAK9D,MAAO,CACN,KAAM,gBACN,OACA,KAAM,CACL,WAAY,EAAK,WACjB,SAAU,EAAK,SACf,gBACA,cACA,CACD,WAXkB,GAYlB,YAAa,EACb,cAAe,EACf,gBAAiB,EACjB,SACA,WACA,UACA,CCjOF,SAAgB,EACd,EACA,EACA,EAA0B,SACL,CACrB,IAAM,EAAe,IAAI,IAGnB,EAAuB,CAAC,EAAE,CAC5B,EAEJ,MAAQ,EAAQ,EAAgB,KAAK,EAAK,IAAM,MAE9C,EAAW,KAAK,EAAM,MAAQ,EAAM,GAAG,OAAO,CAGhD,EAAW,KAAK,EAAK,OAAO,CAG5B,IAAK,IAAI,EAAI,EAAG,EAAI,EAAU,OAAQ,IAAK,CAEzC,IAAM,EADW,EAAU,GACI,KAAK,cAGhC,EAAe,EACnB,IAAK,IAAI,EAAI,EAAG,EAAI,EAAW,OAAS,EAAG,IACzC,GAAI,GAAiB,EAAW,IAAM,EAAgB,EAAW,EAAI,GAAI,CACvE,EAAe,EACf,MAIJ,EAAa,IAAI,EAAG,EAAa,CAGnC,OAAO,EAYT,SAAgB,EACd,EACA,EACA,EACA,EACS,CACT,GAAI,IAAa,OAEf,MAAO,GAIT,IAAM,EAAsB,EAAa,IAAI,EAAgB,CACvD,EAAmB,EAAa,IAAI,EAAa,CASvD,OANI,IAAwB,IAAA,IAAa,IAAqB,IAAA,GACrD,GAKF,IAAwB,ECpEjC,SAAgB,EAAoB,EAAW,EAAmB,CAEhE,GAAI,EAAE,SAAW,EAAG,OAAO,EAAE,OAC7B,GAAI,EAAE,SAAW,EAAG,OAAO,EAAE,OAI7B,IAAM,EAAiB,MAAM,KAAK,CAAE,OAAQ,EAAE,OAAS,EAAG,KACxD,MAAM,EAAE,OAAS,EAAE,CAAC,KAAK,EAAE,CAC5B,CAGD,IAAK,IAAI,EAAI,EAAG,GAAK,EAAE,OAAQ,IAC7B,EAAG,GAAG,GAAK,EAEb,IAAK,IAAI,EAAI,EAAG,GAAK,EAAE,OAAQ,IAC7B,EAAG,GAAG,GAAK,EAIb,IAAK,IAAI,EAAI,EAAG,GAAK,EAAE,OAAQ,IAC7B,IAAK,IAAI,EAAI,EAAG,GAAK,EAAE,OAAQ,IACzB,EAAE,EAAI,KAAO,EAAE,EAAI,GAErB,EAAG,GAAG,GAAK,EAAG,EAAI,GAAG,EAAI,GAMzB,EAAG,GAAG,GAAK,EAAI,KAAK,IAClB,EAAG,EAAI,GAAG,GACV,EAAG,GAAG,EAAI,GACV,EAAG,EAAI,GAAG,EAAI,GACf,CAKP,OAAO,EAAG,EAAE,QAAQ,EAAE,QAgBxB,SAAgB,EAA8B,EAAW,EAAmB,CAE1E,IAAM,EAAS,EAAE,aAAa,CACxB,EAAS,EAAE,aAAa,CAGxB,EAAW,EAAoB,EAAQ,EAAO,CAG9C,EAAY,KAAK,IAAI,EAAO,OAAQ,EAAO,OAAO,CAIxD,OAHI,IAAc,EAAU,EAGrB,EAAI,EAAW,ECvDxB,IAAa,EAAb,KAA8B,CAa5B,YACE,EACA,EACA,EAA6B,EAAE,CAC/B,CACA,KAAK,UAAY,EACjB,KAAK,KAAO,EAGZ,KAAK,QAAU,CACb,cAAe,EAAQ,eAAiB,YACxC,qBAAsB,EAAQ,sBAAwB,GACtD,yBAA0B,EAAQ,0BAA4B,SAC9D,mBAAoB,EAAQ,oBAAsB,GAClD,oBAAqB,EAAQ,qBAAuB,GACpD,sBAAuB,EAAQ,uBAAyB,GACxD,iBAAkB,EAAQ,kBAAoB,GAC/C,CAGD,KAAK,QAAU,CACb,cAAe,EACf,aAAc,EACd,iBAAkB,IAAA,GAClB,oBAAqB,IAAI,IACzB,aAAc,IAAI,IACnB,CAGG,KAAK,QAAQ,uBACf,KAAK,QAAQ,aAAe,EAC1B,EACA,EACA,KAAK,QAAQ,yBACd,EASL,SAA8B,CAC5B,IAAM,EAA+B,EAAE,CAEvC,IAAK,IAAI,EAAI,EAAG,EAAI,KAAK,UAAU,OAAQ,IAAK,CAC9C,KAAK,QAAQ,cAAgB,EAC7B,IAAM,EAAW,KAAK,UAAU,GAG5B,EAEJ,OAAQ,EAAS,KAAjB,CACE,IAAK,KACH,EAAa,KAAK,UAAU,EAAuB,CACnD,MACF,IAAK,QACH,EAAa,KAAK,aAAa,EAA0B,CACzD,MACF,IAAK,gBACH,EAAa,KAAK,qBAAqB,EAAkC,CACzE,MACF,QAEM,KAAK,eAAe,EAAS,GAC/B,KAAK,QAAQ,iBAAmB,EAChC,KAAK,kBAAkB,EAAU,EAAE,EAErC,MAIJ,EAAS,KAAK,CACZ,GAAG,EACH,aACD,CAAC,CAGJ,OAAO,EAMT,UAAkB,EAAoD,CACpE,IAAM,EAAe,KAAK,QAAQ,cAG9B,EACJ,IAAK,IAAI,EAAI,EAAe,EAAG,GAAK,EAAG,IAErC,GADkB,KAAK,UAAU,GACnB,OAAS,OAAQ,CAC7B,EAAkB,EAClB,MAcJ,OATI,IAAoB,IAAA,GACf,KAAK,oBAAoB,wCAAwC,CAIrE,KAAK,cAAc,EAAiB,EAAa,CAI/C,CACL,WAAY,EACZ,WAAY,EACb,CANQ,KAAK,oBAAoB,6CAA6C,CAYjF,aAAqB,EAAuD,CAC1E,IAAM,EAAe,KAAK,QAAQ,cAC5B,EAAkB,KAAK,mBAAmB,EAAS,UAAU,CAG/D,EAEJ,IAAK,GAAM,CAAC,EAAW,KAAkB,KAAK,QAAQ,oBAAqB,CAEzE,GAAI,CAAC,KAAK,cAAc,EAAe,EAAa,CAClD,SAIF,IAAM,EAAa,EAA8B,EAAiB,EAAU,EAGxE,CAAC,GAAa,EAAa,EAAU,cACvC,EAAY,CAAE,MAAO,EAAe,aAAY,EAKpD,GAAI,CAAC,EACH,OAAO,KAAK,oBAAoB,kCAAkC,CAGpE,GAAI,EAAU,WAAa,KAAK,QAAQ,oBACtC,OAAO,KAAK,oBACV,yBAAyB,EAAU,WAAW,QAAQ,EAAE,CAAC,mBAAmB,KAAK,QAAQ,sBAC1F,CAIH,IAAM,EAAqB,EAAE,CAK7B,OAJI,EAAU,WAAa,GACzB,EAAS,KAAK,2BAA2B,EAAU,WAAW,QAAQ,EAAE,GAAG,CAGtE,CACL,WAAY,EAAU,MACtB,WAAY,EAAU,WACtB,SAAU,EAAS,OAAS,EAAI,EAAW,IAAA,GAC5C,CAMH,qBAA6B,EAA+D,CAC1F,IAAM,EAAe,KAAK,QAAQ,cAGlC,IAAK,IAAI,EAAI,EAAe,EAAG,GAAK,EAAG,IAAK,CAC1C,IAAM,EAAY,KAAK,UAAU,GAGjC,GAAI,EAAU,OAAS,OACrB,SAGF,IAAM,EAAW,EAGjB,GACE,EAAS,SAAW,EAAS,QAC7B,KAAK,kBAAkB,EAAS,SAAS,GAAK,KAAK,kBAAkB,EAAS,SAAS,CAQvF,OALK,KAAK,cAAc,EAAG,EAAa,CAKjC,CACL,WAAY,EACZ,WAAY,IACb,CAPQ,KAAK,oBAAoB,2CAA2C,CAWjF,OAAO,KAAK,oBAAoB,uCAAuC,CAMzE,eAAuB,EAA6B,CAClD,OACE,EAAS,OAAS,QAClB,EAAS,OAAS,WAClB,EAAS,OAAS,WAClB,EAAS,OAAS,WAClB,EAAS,OAAS,aAClB,EAAS,OAAS,kBAQtB,kBAA0B,EAAoB,EAAqB,CAEjE,GAAI,EAAS,OAAS,OAAQ,CAC5B,IAAM,EAAW,EACX,EAAY,KAAK,iBAAiB,EAAS,CACjD,GAAI,EAAW,CACb,IAAM,EAAa,KAAK,mBAAmB,EAAU,CACrD,KAAK,QAAQ,oBAAoB,IAAI,EAAY,EAAM,GAS7D,iBAAyB,EAAgD,CAKvE,IAAM,EAAgB,EAAS,KAAK,cAE9B,EAAgB,KAAK,IAAI,EAAG,EAAgB,IAAI,CAChD,EAAa,KAAK,KAAK,UAAU,EAAe,EAAc,CAI9D,EAAS,EAAW,MAAM,4FAA4F,CAO5H,OANI,EACK,EAAO,GAAG,MAAM,CAIL,EAAW,MAAM,8CAA8C,GAC9D,GAAG,MAAM,CAMhC,mBAA2B,EAAsB,CAC/C,OAAO,EACJ,aAAa,CACb,QAAQ,OAAQ,IAAI,CACpB,MAAM,CAMX,kBAA0B,EAA0B,CAClD,OAAO,EACJ,aAAa,CACb,QAAQ,OAAQ,GAAG,CACnB,QAAQ,MAAO,GAAG,CAMvB,cAAsB,EAAyB,EAA+B,CAC5E,OAAO,EACL,EACA,EACA,KAAK,QAAQ,aACb,KAAK,QAAQ,cACd,CAMH,oBAA4B,EAA8C,CACxE,GAAI,KAAK,QAAQ,iBACf,MAAO,CACL,WAAY,IAAA,GACZ,cAAe,EACf,WAAY,EACb,GCtTP,SAAgB,EACd,EACA,EACA,EACoB,CAEpB,OADiB,IAAI,EAAiB,EAAW,EAAM,EAAQ,CAC/C,SAAS,CC2H3B,SAAgB,EACf,EACA,EACkC,CAClC,IAAM,EAAY,YAAY,KAAK,CAG7B,CAAE,UAAS,oBAAmB,YAAa,EAChD,EACA,GAAS,SACT,CAWK,EAAS,EAAS,EAPJ,GAAS,UAAY,CACxC,GAAG,EACH,GAAG,EACH,GAAG,EACH,GAAG,EACH,GAAG,EACH,CAC4C,CAKvC,EAAoC,EAAE,CACtC,EAAgB,IAAI,IAE1B,IAAK,IAAM,KAAS,EAAQ,CAC3B,IAAM,EAAS,GAAG,EAAM,KAAK,WAAW,GAAG,EAAM,KAAK,WACjD,EAAc,IAAI,EAAO,GAC7B,EAAc,IAAI,EAAO,CACzB,EAAmB,KAAK,EAAM,EAKhC,IAAM,EAAwB,EAAE,CAChC,IAAK,IAAM,KAAS,EAAoB,CACvC,IAAI,EAEJ,OAAQ,EAAM,KAAd,CACC,IAAK,OAEJ,AAOC,EAPG,EAAM,YAAc,MAAQ,EAAM,YAAc,OACxC,EAAU,EAAO,EAAkB,CACpC,EAAM,YAAc,QACnB,EAAa,EAAO,EAAkB,CACvC,EAAM,YAAc,gBACnB,EAAqB,EAAO,EAAkB,CAE9C,EAAY,EAAO,EAAkB,CAEjD,MACD,IAAK,UACJ,EAAW,EAAe,EAAO,EAAkB,CACnD,MACD,IAAK,UACJ,EAAW,EAAe,EAAO,EAAkB,CACnD,MACD,IAAK,UACJ,EAAW,EAAe,EAAO,EAAkB,CACnD,MACD,IAAK,YACJ,EAAW,EAAiB,EAAO,EAAkB,CACrD,MACD,IAAK,kBACJ,EAAW,EAAuB,EAAO,EAAkB,CAC3D,MACD,QAEC,SAIE,EAAS,OAAS,IACrB,EAAS,SAAW,CAAC,GAAI,EAAS,UAAY,EAAE,CAAG,GAAG,EAAS,EAIhE,EAAS,cAAgB,YAAY,KAAK,CAAG,EAE7C,EAAU,KAAK,EAAS,CAQzB,OAJI,GAAS,QACL,EAAiB,EAAW,EAAM,EAAQ,kBAAkB,CAG7D,EAyBR,eAAsB,EACrB,EACA,EAC2C,CAG3C,OAAO,EAAiB,EAAM,EAAQ"}