@rethinkhealth/hl7v2-parser 0.2.4 → 0.2.5

package/dist/index.js

@@ -1,3 +1,48 @@
+// src/parser.ts
+import { DEFAULT_DELIMITERS } from "@rethinkhealth/hl7v2-utils";
+
+// src/preprocessor.ts
+var stripBOM = (ctx) => {
+  if (ctx.input.charCodeAt(0) === 65279) {
+    ctx.input = ctx.input.slice(1);
+  }
+  return ctx;
+};
+var normalizeNewlines = (ctx) => {
+  ctx.input = ctx.input.replace(/\r?\n/g, "\r");
+  return ctx;
+};
+var detectDelimiters = (ctx) => {
+  if (ctx.input.startsWith("MSH")) {
+    const fieldDelim = ctx.input.charAt(3) || "|";
+    const enc = ctx.input.slice(4, 8);
+    const component = enc.charAt(0) || "^";
+    const repetition = enc.charAt(1) || "~";
+    const _escape = enc.charAt(2) || "\\";
+    const subcomponent = enc.charAt(3) || "&";
+    ctx.delimiters = {
+      field: fieldDelim,
+      component,
+      repetition,
+      escape: _escape,
+      subcomponent,
+      segment: "\r"
+    };
+  }
+  return ctx;
+};
+var defaultPreprocessors = [
+  stripBOM,
+  normalizeNewlines,
+  detectDelimiters
+];
+function runPreprocessors(ctx, steps) {
+  for (const step of steps) {
+    ctx = step(ctx);
+  }
+  return ctx;
+}
+
 // src/processor.ts
 function createParserCore() {
   const root = { type: "root", children: [] };
@@ -155,9 +200,6 @@ function parseHL7v2FromIterator(tokens) {
 }
 
 // src/tokenizer.ts
-import {
-  DEFAULT_DELIMITERS
-} from "@rethinkhealth/hl7v2-utils";
 var HL7v2Tokenizer = class {
   input = "";
   i = 0;
@@ -168,9 +210,9 @@ var HL7v2Tokenizer = class {
   didMshBootstrap = false;
   pendingBootstrap = null;
   // queue to emit TEXT('MSH'), FIELD_DELIM, TEXT('^~\\&')
-  reset(input, opts) {
-    this.input = input;
-    this.delims = opts.delimiters || DEFAULT_DELIMITERS;
+  reset(ctx) {
+    this.input = ctx.input;
+    this.delims = ctx.delimiters;
     this.i = 0;
     this.line = 1;
     this.col = 1;
@@ -178,11 +220,13 @@ var HL7v2Tokenizer = class {
     this.pendingBootstrap = null;
     if (this.input.startsWith("MSH")) {
       const msh = this._slice(0, 3);
+      const msh1 = this._slice(3, 4);
       const msh2 = this._slice(4, 8);
       this.pendingBootstrap = [
         { kind: "TEXT", value: msh, advance: msh.length },
-        { kind: "FIELD_DELIM", advance: 1 },
-        // consume the single field delimiter char at index 3
+        { kind: "FIELD_DELIM", advance: 0 },
+        { kind: "TEXT", value: msh1, advance: msh1.length },
+        { kind: "FIELD_DELIM", advance: 0 },
         { kind: "TEXT", value: msh2, advance: msh2.length }
       ];
     }
@@ -304,62 +348,19 @@ function* iterateTokenizerSync(t) {
 
 // src/parser.ts
 function parseHL7v2(input, opts) {
+  let ctx = {
+    input,
+    delimiters: opts.delimiters || DEFAULT_DELIMITERS
+  };
+  ctx = runPreprocessors(ctx, opts.preprocess || defaultPreprocessors);
   const tokenizer = new HL7v2Tokenizer();
-  tokenizer.reset(input, opts);
+  tokenizer.reset(ctx);
   return parseHL7v2FromIterator(iterateTokenizerSync(tokenizer));
 }
 var hl7v2Parser = function(options = {}) {
-  const self = this;
-  self.parser = (value) => parseHL7v2(value, options);
+  this.parser = (document) => parseHL7v2(document, options);
 };
 var parser_default = hl7v2Parser;
-
-// src/preprocessor.ts
-var stripBOM = (ctx) => {
-  if (ctx.input.charCodeAt(0) === 65279) {
-    ctx.input = ctx.input.slice(1);
-  }
-  return ctx;
-};
-var normalizeNewlines = (ctx) => {
-  ctx.input = ctx.input.replace(/\r?\n/g, "\r");
-  return ctx;
-};
-var detectDelimiters = (ctx) => {
-  if (ctx.input.startsWith("MSH")) {
-    const fieldDelim = ctx.input.charAt(3) || "|";
-    const enc = ctx.input.slice(4, 8);
-    const component = enc.charAt(0) || "^";
-    const repetition = enc.charAt(1) || "~";
-    const _escape = enc.charAt(2) || "\\";
-    const subcomponent = enc.charAt(3) || "&";
-    ctx.options.delimiters = {
-      field: fieldDelim,
-      component,
-      repetition,
-      escape: _escape,
-      subcomponent,
-      segment: "\r"
-    };
-  }
-  return ctx;
-};
-var defaultPreprocessors = [
-  stripBOM,
-  normalizeNewlines,
-  detectDelimiters
-];
-function runPreprocessors(raw, opts = {}, steps = defaultPreprocessors) {
-  let ctx = {
-    input: raw,
-    options: opts,
-    metadata: {}
-  };
-  for (const step of steps) {
-    ctx = step(ctx);
-  }
-  return ctx;
-}
 export {
   HL7v2Tokenizer,
   defaultPreprocessors,

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/processor.ts","../src/tokenizer.ts","../src/utils.ts","../src/parser.ts","../src/preprocessor.ts"],"sourcesContent":["// src/parser.ts\n\nimport type {\n  Component,\n  Field,\n  FieldRepetition,\n  Root,\n  Segment,\n  Subcomponent,\n} from '@rethinkhealth/hl7v2-ast';\nimport type { Token } from './types';\n\n// Shared core: process a single token into mutable parse state\nfunction createParserCore() {\n  const root: Root = { type: 'root', children: [] };\n\n  let seg: Segment | null = null;\n  let field: Field | null = null;\n  let rep: FieldRepetition | null = null;\n  let comp: Component | null = null;\n  let currentSub: Subcomponent | null = null;\n  let segmentHasContent = false;\n\n  const openSegment = () => {\n    seg = { type: 'segment', children: [] };\n    root.children.push(seg);\n    field = null;\n    rep = null;\n    comp = null;\n    currentSub = null;\n    segmentHasContent = false;\n  };\n\n  const ensureSegment = () => {\n    if (!seg) {\n      openSegment();\n    }\n  };\n\n  const openField = () => {\n    ensureSegment();\n    field = { type: 'field', children: [] };\n    // biome-ignore lint/style/noNonNullAssertion: seg is ensured above\n    seg!.children.push(field);\n    rep = { type: 'field-repetition', children: [] };\n    field.children.push(rep);\n    comp = { type: 'component', children: [] };\n    rep.children.push(comp);\n    currentSub = null;\n    segmentHasContent = true;\n  };\n\n  const openRepetition = () => {\n    if (!field) {\n      openField();\n    }\n    rep = { type: 'field-repetition', children: [] };\n    // biome-ignore lint/style/noNonNullAssertion: field is ensured above\n    field!.children.push(rep);\n    comp = { type: 'component', children: [] };\n    rep.children.push(comp);\n    currentSub = null;\n    segmentHasContent = true;\n  };\n\n  const openComponent = () => {\n    if (!field) {\n      openField();\n    }\n    if (!rep) {\n      rep = { type: 'field-repetition', children: [] };\n      // biome-ignore lint/style/noNonNullAssertion: field is ensured above\n      field!.children.push(rep);\n    }\n    comp = { type: 'component', children: [] };\n    // biome-ignore lint/style/noNonNullAssertion: rep is ensured above\n    rep!.children.push(comp);\n    currentSub = null;\n    segmentHasContent = true;\n  };\n\n  const ensureForText = () => {\n    if (!field) {\n      openField();\n    }\n    if (!rep) {\n      openRepetition();\n    }\n    if (!comp) {\n      openComponent();\n    }\n    if (!currentSub) {\n      currentSub = { type: 'subcomponent', value: '' };\n      // biome-ignore lint/style/noNonNullAssertion: comp is ensured above\n      comp!.children.push(currentSub);\n      segmentHasContent = true;\n    }\n  };\n\n  const processToken = (tok: Token) => {\n    switch (tok.type) {\n      case 'SEGMENT_END': {\n        dropTrailingEmptyFieldIfPresent();\n        // Close current segment (if any) and reset; do not auto-open next segment\n        seg = null;\n        field = null;\n        rep = null;\n        comp = null;\n        currentSub = null;\n        segmentHasContent = false;\n        return;\n      }\n      case 'FIELD_DELIM': {\n        // Leading field delimiter implies an empty first field\n        if (!field) {\n          // Open an empty first field and record an empty subcomponent slot\n          openField();\n          // biome-ignore lint/style/noNonNullAssertion: comp is initialized in openField\n          comp!.children.push({ type: 'subcomponent', value: '' });\n          segmentHasContent = true;\n        }\n        openField();\n        return;\n      }\n      case 'REPETITION_DELIM': {\n        if (!field) {\n          openField();\n        }\n        openRepetition();\n        return;\n      }\n      case 'COMPONENT_DELIM': {\n        openComponent();\n        return;\n      }\n      case 'SUBCOMP_DELIM': {\n        if (!comp) {\n          openComponent();\n        }\n        // Start a new empty subcomponent slot\n        currentSub = { type: 'subcomponent', value: '' };\n        // biome-ignore lint/style/noNonNullAssertion: comp is ensured above\n        comp!.children.push(currentSub);\n        segmentHasContent = true;\n        return;\n      }\n      case 'TEXT': {\n        const val = tok.value ?? '';\n        ensureForText();\n        // biome-ignore lint/style/noNonNullAssertion: ensured above\n        currentSub!.value += val;\n        segmentHasContent = true;\n        return;\n      }\n      default: {\n        throw new Error(`Unexpected token type: ${tok.type}`);\n      }\n    }\n  };\n\n  // Do not pre-open a segment; lazily open upon first non-SEGMENT_END token.\n\n  const finalize = () => {\n    // Handle input that ends without an explicit segment delimiter as above.\n    dropTrailingEmptyFieldIfPresent();\n    if (seg && !segmentHasContent) {\n      // Drop trailing empty segment if no content was added\n      root.children.pop();\n    }\n    return root;\n  };\n\n  function dropTrailingEmptyFieldIfPresent() {\n    if (!seg || seg.children.length === 0) {\n      return;\n    }\n    // Drop only the final trailing empty field (created by the last delimiter),\n    // preserving any intentional empty fields immediately before it.\n    const lastField = seg.children.at(-1) as Field;\n    const hasAnySubcomponents = lastField.children.some((r) =>\n      r.children.some((c) => c.children.length > 0)\n    );\n    if (!hasAnySubcomponents) {\n      seg.children.pop();\n    }\n  }\n\n  return { processToken, finalize, root };\n}\n\n// Sync convenience wrapper over a sync Iterable token source\nexport function parseHL7v2FromIterator(tokens: Iterable<Token>): Root {\n  const core = createParserCore();\n  for (const tok of tokens) {\n    core.processToken(tok);\n  }\n  return core.finalize();\n}\n","// src/tokenizer.ts\nimport {\n  DEFAULT_DELIMITERS,\n  type HL7v2Delimiters,\n} from '@rethinkhealth/hl7v2-utils';\nimport type {\n  ParseOptions,\n  Position,\n  Token,\n  Tokenizer,\n  TokenType,\n} from './types';\n\nexport class HL7v2Tokenizer implements Tokenizer, Iterable<Token> {\n  private input = '';\n  private i = 0;\n  private line = 1;\n  private col = 1;\n  private delims!: HL7v2Delimiters;\n\n  // Only-run-once MSH bootstrap at the start of the file\n  private didMshBootstrap = false;\n  private pendingBootstrap: null | Array<\n    | { kind: 'TEXT'; value: string; advance: number }\n    | { kind: 'FIELD_DELIM'; advance: number }\n  > = null; // queue to emit TEXT('MSH'), FIELD_DELIM, TEXT('^~\\\\&')\n\n  reset(input: string, opts: ParseOptions) {\n    this.input = input;\n    this.delims = opts.delimiters || DEFAULT_DELIMITERS;\n    this.i = 0;\n    this.line = 1;\n    this.col = 1;\n    this.didMshBootstrap = false;\n    this.pendingBootstrap = null;\n\n    // Prepare a one-time bootstrap if file starts with MSH\n    if (this.input.startsWith('MSH')) {\n      // Precompute MSH and MSH.2; MSH.1 is the field delimiter char at index 3\n      const msh = this._slice(0, 3);\n      const msh2 = this._slice(4, 8); // may be shorter than 4 if truncated\n      this.pendingBootstrap = [\n        { kind: 'TEXT', value: msh, advance: msh.length },\n        { kind: 'FIELD_DELIM', advance: 1 }, // consume the single field delimiter char at index 3\n        { kind: 'TEXT', value: msh2, advance: msh2.length },\n      ];\n      // NOTE: we have not advanced indices yet; we will as we emit tokens\n    }\n  }\n\n  // biome-ignore lint/complexity/noExcessiveCognitiveComplexity: The cognitive complexity of this method is justified because it must handle the HL7v2 MSH segment bootstrap logic and multiple tokenization cases in a single method for performance and maintainability.\n  next(): Token | null {\n    const s = this.input;\n    const n = s.length;\n    if (this.i >= n && !this.pendingBootstrap?.length) {\n      return null;\n    }\n\n    const start = { offset: this.i, line: this.line, column: this.col };\n\n    // ---- One-time MSH bootstrap at file start ----\n    if (!this.didMshBootstrap && this.pendingBootstrap) {\n      const step = this.pendingBootstrap.shift();\n      if (step) {\n        if (step.kind === 'TEXT') {\n          const take = Math.min(step.advance, n - this.i);\n          const out = step.value.slice(0, take);\n          this._fastAdvance(out);\n          if (this.pendingBootstrap.length === 0) {\n            this.pendingBootstrap = null;\n            this.didMshBootstrap = true;\n          }\n          return this._tok('TEXT', out, start);\n        }\n        // FIELD_DELIM: advance exactly one char (the delimiter) and emit a FIELD_DELIM token\n        this._advance(Math.min(step.advance, n - this.i));\n        if (this.pendingBootstrap.length === 0) {\n          this.pendingBootstrap = null;\n          this.didMshBootstrap = true;\n        }\n        return this._tok('FIELD_DELIM', undefined, start);\n      }\n    }\n\n    // ---- Normal tokenization (delimiters + text) ----\n    if (this.i >= n) {\n      return null;\n    }\n    const ch = s.charAt(this.i);\n\n    if (ch === this.delims.segment) {\n      this._advance(1, true);\n      return this._tok('SEGMENT_END', undefined, start);\n    }\n    if (ch === this.delims.field) {\n      this._advance(1);\n      return this._tok('FIELD_DELIM', undefined, start);\n    }\n    if (ch === this.delims.repetition) {\n      this._advance(1);\n      return this._tok('REPETITION_DELIM', undefined, start);\n    }\n    if (ch === this.delims.component) {\n      this._advance(1);\n      return this._tok('COMPONENT_DELIM', undefined, start);\n    }\n    if (ch === this.delims.subcomponent) {\n      this._advance(1);\n      return this._tok('SUBCOMP_DELIM', undefined, start);\n    }\n\n    // TEXT until next delimiter or end\n    let j = this.i;\n    const seg = this.delims.segment,\n      fld = this.delims.field,\n      rep = this.delims.repetition,\n      cmp = this.delims.component,\n      sub = this.delims.subcomponent;\n\n    while (j < s.length) {\n      const c = s.charAt(j);\n      if (c === seg || c === fld || c === rep || c === cmp || c === sub) {\n        break;\n      }\n      j++;\n    }\n\n    const val = s.slice(this.i, j);\n    this._fastAdvance(val);\n    return this._tok('TEXT', val, start);\n  }\n\n  // ---- helpers ----\n  private _slice(start: number, end: number): string {\n    return this.input.slice(start, Math.min(end, this.input.length));\n  }\n\n  private _advance(n: number, isNewline = false) {\n    this.i += n;\n    if (isNewline) {\n      this.line += 1;\n      this.col = 1;\n    } else {\n      this.col += n;\n    }\n  }\n\n  private _fastAdvance(chunk: string) {\n    const parts = chunk.split('\\r');\n    if (parts.length > 1) {\n      this.line += parts.length - 1;\n      this.col = (parts.at(-1)?.length ?? 0) + 1;\n    } else {\n      this.col += chunk.length;\n    }\n    this.i += chunk.length;\n  }\n\n  private _tok(\n    type: TokenType,\n    value: string | undefined,\n    start: Position['start']\n  ): Token {\n    const end = { offset: this.i, line: this.line, column: this.col };\n    return { type, value, position: { start, end } };\n  }\n\n  // Iterable protocol (sync)\n  [Symbol.iterator](): Iterator<Token> {\n    return {\n      next: () => {\n        const t = this.next();\n        if (t == null) {\n          return { done: true, value: undefined as unknown as Token };\n        }\n        return { done: false, value: t };\n      },\n    };\n  }\n\n  // Async iteration support removed to keep the API synchronous.\n}\n","import type { Token, Tokenizer } from './types';\n\nexport function* iterateTokenizerSync(t: Tokenizer): Iterable<Token> {\n  for (let tok = t.next(); tok; tok = t.next()) {\n    yield tok;\n  }\n}\n\n// Removed async stream support; the parser now operates synchronously only.\n","import type { Root } from '@rethinkhealth/hl7v2-ast';\nimport type { Plugin } from 'unified';\nimport { parseHL7v2FromIterator } from './processor';\nimport { HL7v2Tokenizer } from './tokenizer';\nimport type { ParseOptions } from './types';\nimport { iterateTokenizerSync } from './utils';\n\n// Back-compat convenience API: parse from string using the built-in tokenizer (sync)\nexport function parseHL7v2(input: string, opts: ParseOptions): Root {\n  const tokenizer = new HL7v2Tokenizer();\n  tokenizer.reset(input, opts);\n  return parseHL7v2FromIterator(iterateTokenizerSync(tokenizer));\n}\n\nconst hl7v2Parser: Plugin<[ParseOptions?], string, Root> = function (\n  options = {}\n) {\n  const self = this as { parser?: (value: string) => Root };\n  self.parser = (value: string) => parseHL7v2(value, options);\n};\n\nexport default hl7v2Parser;\n","import type { ParseOptions, ParserContext } from './types';\n\n/**\n * A preprocessing step: takes a ParserContext, mutates or replaces it.\n */\nexport type PreprocessorStep = (ctx: ParserContext) => ParserContext;\n\n/**\n * Step: strip UTF-8 BOM.\n */\nexport const stripBOM: PreprocessorStep = (ctx) => {\n  if (ctx.input.charCodeAt(0) === 0xfe_ff) {\n    ctx.input = ctx.input.slice(1);\n  }\n  return ctx;\n};\n\n/**\n * Step: normalize newlines to CR (\\r).\n */\nexport const normalizeNewlines: PreprocessorStep = (ctx) => {\n  ctx.input = ctx.input.replace(/\\r?\\n/g, '\\r');\n  return ctx;\n};\n\n/**\n * Step: auto-detect delimiters from MSH.\n */\nexport const detectDelimiters: PreprocessorStep = (ctx) => {\n  if (ctx.input.startsWith('MSH')) {\n    const fieldDelim = ctx.input.charAt(3) || '|';\n    const enc = ctx.input.slice(4, 8);\n    const component = enc.charAt(0) || '^';\n    const repetition = enc.charAt(1) || '~';\n    const _escape = enc.charAt(2) || '\\\\';\n    const subcomponent = enc.charAt(3) || '&';\n\n    ctx.options.delimiters = {\n      field: fieldDelim,\n      component,\n      repetition,\n      escape: _escape,\n      subcomponent,\n      segment: '\\r',\n    };\n  }\n  return ctx;\n};\n\n/**\n * Default preprocessing pipeline.\n */\nexport const defaultPreprocessors: PreprocessorStep[] = [\n  stripBOM,\n  normalizeNewlines,\n  detectDelimiters,\n];\n\n/**\n * Run the pipeline to initialize ParserContext.\n */\nexport function runPreprocessors(\n  raw: string,\n  opts: ParseOptions = {},\n  steps: PreprocessorStep[] = defaultPreprocessors\n): ParserContext {\n  let ctx: ParserContext = {\n    input: raw,\n    options: opts,\n    metadata: {},\n  };\n\n  for (const step of steps) {\n    ctx = step(ctx);\n  }\n\n  return ctx;\n}\n"],"mappings":";AAaA,SAAS,mBAAmB;AAC1B,QAAM,OAAa,EAAE,MAAM,QAAQ,UAAU,CAAC,EAAE;AAEhD,MAAI,MAAsB;AAC1B,MAAI,QAAsB;AAC1B,MAAI,MAA8B;AAClC,MAAI,OAAyB;AAC7B,MAAI,aAAkC;AACtC,MAAI,oBAAoB;AAExB,QAAM,cAAc,MAAM;AACxB,UAAM,EAAE,MAAM,WAAW,UAAU,CAAC,EAAE;AACtC,SAAK,SAAS,KAAK,GAAG;AACtB,YAAQ;AACR,UAAM;AACN,WAAO;AACP,iBAAa;AACb,wBAAoB;AAAA,EACtB;AAEA,QAAM,gBAAgB,MAAM;AAC1B,QAAI,CAAC,KAAK;AACR,kBAAY;AAAA,IACd;AAAA,EACF;AAEA,QAAM,YAAY,MAAM;AACtB,kBAAc;AACd,YAAQ,EAAE,MAAM,SAAS,UAAU,CAAC,EAAE;AAEtC,QAAK,SAAS,KAAK,KAAK;AACxB,UAAM,EAAE,MAAM,oBAAoB,UAAU,CAAC,EAAE;AAC/C,UAAM,SAAS,KAAK,GAAG;AACvB,WAAO,EAAE,MAAM,aAAa,UAAU,CAAC,EAAE;AACzC,QAAI,SAAS,KAAK,IAAI;AACtB,iBAAa;AACb,wBAAoB;AAAA,EACtB;AAEA,QAAM,iBAAiB,MAAM;AAC3B,QAAI,CAAC,OAAO;AACV,gBAAU;AAAA,IACZ;AACA,UAAM,EAAE,MAAM,oBAAoB,UAAU,CAAC,EAAE;AAE/C,UAAO,SAAS,KAAK,GAAG;AACxB,WAAO,EAAE,MAAM,aAAa,UAAU,CAAC,EAAE;AACzC,QAAI,SAAS,KAAK,IAAI;AACtB,iBAAa;AACb,wBAAoB;AAAA,EACtB;AAEA,QAAM,gBAAgB,MAAM;AAC1B,QAAI,CAAC,OAAO;AACV,gBAAU;AAAA,IACZ;AACA,QAAI,CAAC,KAAK;AACR,YAAM,EAAE,MAAM,oBAAoB,UAAU,CAAC,EAAE;AAE/C,YAAO,SAAS,KAAK,GAAG;AAAA,IAC1B;AACA,WAAO,EAAE,MAAM,aAAa,UAAU,CAAC,EAAE;AAEzC,QAAK,SAAS,KAAK,IAAI;AACvB,iBAAa;AACb,wBAAoB;AAAA,EACtB;AAEA,QAAM,gBAAgB,MAAM;AAC1B,QAAI,CAAC,OAAO;AACV,gBAAU;AAAA,IACZ;AACA,QAAI,CAAC,KAAK;AACR,qBAAe;AAAA,IACjB;AACA,QAAI,CAAC,MAAM;AACT,oBAAc;AAAA,IAChB;AACA,QAAI,CAAC,YAAY;AACf,mBAAa,EAAE,MAAM,gBAAgB,OAAO,GAAG;AAE/C,WAAM,SAAS,KAAK,UAAU;AAC9B,0BAAoB;AAAA,IACtB;AAAA,EACF;AAEA,QAAM,eAAe,CAAC,QAAe;AACnC,YAAQ,IAAI,MAAM;AAAA,MAChB,KAAK,eAAe;AAClB,wCAAgC;AAEhC,cAAM;AACN,gBAAQ;AACR,cAAM;AACN,eAAO;AACP,qBAAa;AACb,4BAAoB;AACpB;AAAA,MACF;AAAA,MACA,KAAK,eAAe;AAElB,YAAI,CAAC,OAAO;AAEV,oBAAU;AAEV,eAAM,SAAS,KAAK,EAAE,MAAM,gBAAgB,OAAO,GAAG,CAAC;AACvD,8BAAoB;AAAA,QACtB;AACA,kBAAU;AACV;AAAA,MACF;AAAA,MACA,KAAK,oBAAoB;AACvB,YAAI,CAAC,OAAO;AACV,oBAAU;AAAA,QACZ;AACA,uBAAe;AACf;AAAA,MACF;AAAA,MACA,KAAK,mBAAmB;AACtB,sBAAc;AACd;AAAA,MACF;AAAA,MACA,KAAK,iBAAiB;AACpB,YAAI,CAAC,MAAM;AACT,wBAAc;AAAA,QAChB;AAEA,qBAAa,EAAE,MAAM,gBAAgB,OAAO,GAAG;AAE/C,aAAM,SAAS,KAAK,UAAU;AAC9B,4BAAoB;AACpB;AAAA,MACF;AAAA,MACA,KAAK,QAAQ;AACX,cAAM,MAAM,IAAI,SAAS;AACzB,sBAAc;AAEd,mBAAY,SAAS;AACrB,4BAAoB;AACpB;AAAA,MACF;AAAA,MACA,SAAS;AACP,cAAM,IAAI,MAAM,0BAA0B,IAAI,IAAI,EAAE;AAAA,MACtD;AAAA,IACF;AAAA,EACF;AAIA,QAAM,WAAW,MAAM;AAErB,oCAAgC;AAChC,QAAI,OAAO,CAAC,mBAAmB;AAE7B,WAAK,SAAS,IAAI;AAAA,IACpB;AACA,WAAO;AAAA,EACT;AAEA,WAAS,kCAAkC;AACzC,QAAI,CAAC,OAAO,IAAI,SAAS,WAAW,GAAG;AACrC;AAAA,IACF;AAGA,UAAM,YAAY,IAAI,SAAS,GAAG,EAAE;AACpC,UAAM,sBAAsB,UAAU,SAAS;AAAA,MAAK,CAAC,MACnD,EAAE,SAAS,KAAK,CAAC,MAAM,EAAE,SAAS,SAAS,CAAC;AAAA,IAC9C;AACA,QAAI,CAAC,qBAAqB;AACxB,UAAI,SAAS,IAAI;AAAA,IACnB;AAAA,EACF;AAEA,SAAO,EAAE,cAAc,UAAU,KAAK;AACxC;AAGO,SAAS,uBAAuB,QAA+B;AACpE,QAAM,OAAO,iBAAiB;AAC9B,aAAW,OAAO,QAAQ;AACxB,SAAK,aAAa,GAAG;AAAA,EACvB;AACA,SAAO,KAAK,SAAS;AACvB;;;ACpMA;AAAA,EACE;AAAA,OAEK;AASA,IAAM,iBAAN,MAA2D;AAAA,EACxD,QAAQ;AAAA,EACR,IAAI;AAAA,EACJ,OAAO;AAAA,EACP,MAAM;AAAA,EACN;AAAA;AAAA,EAGA,kBAAkB;AAAA,EAClB,mBAGJ;AAAA;AAAA,EAEJ,MAAM,OAAe,MAAoB;AACvC,SAAK,QAAQ;AACb,SAAK,SAAS,KAAK,cAAc;AACjC,SAAK,IAAI;AACT,SAAK,OAAO;AACZ,SAAK,MAAM;AACX,SAAK,kBAAkB;AACvB,SAAK,mBAAmB;AAGxB,QAAI,KAAK,MAAM,WAAW,KAAK,GAAG;AAEhC,YAAM,MAAM,KAAK,OAAO,GAAG,CAAC;AAC5B,YAAM,OAAO,KAAK,OAAO,GAAG,CAAC;AAC7B,WAAK,mBAAmB;AAAA,QACtB,EAAE,MAAM,QAAQ,OAAO,KAAK,SAAS,IAAI,OAAO;AAAA,QAChD,EAAE,MAAM,eAAe,SAAS,EAAE;AAAA;AAAA,QAClC,EAAE,MAAM,QAAQ,OAAO,MAAM,SAAS,KAAK,OAAO;AAAA,MACpD;AAAA,IAEF;AAAA,EACF;AAAA;AAAA,EAGA,OAAqB;AACnB,UAAM,IAAI,KAAK;AACf,UAAM,IAAI,EAAE;AACZ,QAAI,KAAK,KAAK,KAAK,CAAC,KAAK,kBAAkB,QAAQ;AACjD,aAAO;AAAA,IACT;AAEA,UAAM,QAAQ,EAAE,QAAQ,KAAK,GAAG,MAAM,KAAK,MAAM,QAAQ,KAAK,IAAI;AAGlE,QAAI,CAAC,KAAK,mBAAmB,KAAK,kBAAkB;AAClD,YAAM,OAAO,KAAK,iBAAiB,MAAM;AACzC,UAAI,MAAM;AACR,YAAI,KAAK,SAAS,QAAQ;AACxB,gBAAM,OAAO,KAAK,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC;AAC9C,gBAAM,MAAM,KAAK,MAAM,MAAM,GAAG,IAAI;AACpC,eAAK,aAAa,GAAG;AACrB,cAAI,KAAK,iBAAiB,WAAW,GAAG;AACtC,iBAAK,mBAAmB;AACxB,iBAAK,kBAAkB;AAAA,UACzB;AACA,iBAAO,KAAK,KAAK,QAAQ,KAAK,KAAK;AAAA,QACrC;AAEA,aAAK,SAAS,KAAK,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,CAAC;AAChD,YAAI,KAAK,iBAAiB,WAAW,GAAG;AACtC,eAAK,mBAAmB;AACxB,eAAK,kBAAkB;AAAA,QACzB;AACA,eAAO,KAAK,KAAK,eAAe,QAAW,KAAK;AAAA,MAClD;AAAA,IACF;AAGA,QAAI,KAAK,KAAK,GAAG;AACf,aAAO;AAAA,IACT;AACA,UAAM,KAAK,EAAE,OAAO,KAAK,CAAC;AAE1B,QAAI,OAAO,KAAK,OAAO,SAAS;AAC9B,WAAK,SAAS,GAAG,IAAI;AACrB,aAAO,KAAK,KAAK,eAAe,QAAW,KAAK;AAAA,IAClD;AACA,QAAI,OAAO,KAAK,OAAO,OAAO;AAC5B,WAAK,SAAS,CAAC;AACf,aAAO,KAAK,KAAK,eAAe,QAAW,KAAK;AAAA,IAClD;AACA,QAAI,OAAO,KAAK,OAAO,YAAY;AACjC,WAAK,SAAS,CAAC;AACf,aAAO,KAAK,KAAK,oBAAoB,QAAW,KAAK;AAAA,IACvD;AACA,QAAI,OAAO,KAAK,OAAO,WAAW;AAChC,WAAK,SAAS,CAAC;AACf,aAAO,KAAK,KAAK,mBAAmB,QAAW,KAAK;AAAA,IACtD;AACA,QAAI,OAAO,KAAK,OAAO,cAAc;AACnC,WAAK,SAAS,CAAC;AACf,aAAO,KAAK,KAAK,iBAAiB,QAAW,KAAK;AAAA,IACpD;AAGA,QAAI,IAAI,KAAK;AACb,UAAM,MAAM,KAAK,OAAO,SACtB,MAAM,KAAK,OAAO,OAClB,MAAM,KAAK,OAAO,YAClB,MAAM,KAAK,OAAO,WAClB,MAAM,KAAK,OAAO;AAEpB,WAAO,IAAI,EAAE,QAAQ;AACnB,YAAM,IAAI,EAAE,OAAO,CAAC;AACpB,UAAI,MAAM,OAAO,MAAM,OAAO,MAAM,OAAO,MAAM,OAAO,MAAM,KAAK;AACjE;AAAA,MACF;AACA;AAAA,IACF;AAEA,UAAM,MAAM,EAAE,MAAM,KAAK,GAAG,CAAC;AAC7B,SAAK,aAAa,GAAG;AACrB,WAAO,KAAK,KAAK,QAAQ,KAAK,KAAK;AAAA,EACrC;AAAA;AAAA,EAGQ,OAAO,OAAe,KAAqB;AACjD,WAAO,KAAK,MAAM,MAAM,OAAO,KAAK,IAAI,KAAK,KAAK,MAAM,MAAM,CAAC;AAAA,EACjE;AAAA,EAEQ,SAAS,GAAW,YAAY,OAAO;AAC7C,SAAK,KAAK;AACV,QAAI,WAAW;AACb,WAAK,QAAQ;AACb,WAAK,MAAM;AAAA,IACb,OAAO;AACL,WAAK,OAAO;AAAA,IACd;AAAA,EACF;AAAA,EAEQ,aAAa,OAAe;AAClC,UAAM,QAAQ,MAAM,MAAM,IAAI;AAC9B,QAAI,MAAM,SAAS,GAAG;AACpB,WAAK,QAAQ,MAAM,SAAS;AAC5B,WAAK,OAAO,MAAM,GAAG,EAAE,GAAG,UAAU,KAAK;AAAA,IAC3C,OAAO;AACL,WAAK,OAAO,MAAM;AAAA,IACpB;AACA,SAAK,KAAK,MAAM;AAAA,EAClB;AAAA,EAEQ,KACN,MACA,OACA,OACO;AACP,UAAM,MAAM,EAAE,QAAQ,KAAK,GAAG,MAAM,KAAK,MAAM,QAAQ,KAAK,IAAI;AAChE,WAAO,EAAE,MAAM,OAAO,UAAU,EAAE,OAAO,IAAI,EAAE;AAAA,EACjD;AAAA;AAAA,EAGA,CAAC,OAAO,QAAQ,IAAqB;AACnC,WAAO;AAAA,MACL,MAAM,MAAM;AACV,cAAM,IAAI,KAAK,KAAK;AACpB,YAAI,KAAK,MAAM;AACb,iBAAO,EAAE,MAAM,MAAM,OAAO,OAA8B;AAAA,QAC5D;AACA,eAAO,EAAE,MAAM,OAAO,OAAO,EAAE;AAAA,MACjC;AAAA,IACF;AAAA,EACF;AAAA;AAGF;;;ACnLO,UAAU,qBAAqB,GAA+B;AACnE,WAAS,MAAM,EAAE,KAAK,GAAG,KAAK,MAAM,EAAE,KAAK,GAAG;AAC5C,UAAM;AAAA,EACR;AACF;;;ACEO,SAAS,WAAW,OAAe,MAA0B;AAClE,QAAM,YAAY,IAAI,eAAe;AACrC,YAAU,MAAM,OAAO,IAAI;AAC3B,SAAO,uBAAuB,qBAAqB,SAAS,CAAC;AAC/D;AAEA,IAAM,cAAqD,SACzD,UAAU,CAAC,GACX;AACA,QAAM,OAAO;AACb,OAAK,SAAS,CAAC,UAAkB,WAAW,OAAO,OAAO;AAC5D;AAEA,IAAO,iBAAQ;;;ACXR,IAAM,WAA6B,CAAC,QAAQ;AACjD,MAAI,IAAI,MAAM,WAAW,CAAC,MAAM,OAAS;AACvC,QAAI,QAAQ,IAAI,MAAM,MAAM,CAAC;AAAA,EAC/B;AACA,SAAO;AACT;AAKO,IAAM,oBAAsC,CAAC,QAAQ;AAC1D,MAAI,QAAQ,IAAI,MAAM,QAAQ,UAAU,IAAI;AAC5C,SAAO;AACT;AAKO,IAAM,mBAAqC,CAAC,QAAQ;AACzD,MAAI,IAAI,MAAM,WAAW,KAAK,GAAG;AAC/B,UAAM,aAAa,IAAI,MAAM,OAAO,CAAC,KAAK;AAC1C,UAAM,MAAM,IAAI,MAAM,MAAM,GAAG,CAAC;AAChC,UAAM,YAAY,IAAI,OAAO,CAAC,KAAK;AACnC,UAAM,aAAa,IAAI,OAAO,CAAC,KAAK;AACpC,UAAM,UAAU,IAAI,OAAO,CAAC,KAAK;AACjC,UAAM,eAAe,IAAI,OAAO,CAAC,KAAK;AAEtC,QAAI,QAAQ,aAAa;AAAA,MACvB,OAAO;AAAA,MACP;AAAA,MACA;AAAA,MACA,QAAQ;AAAA,MACR;AAAA,MACA,SAAS;AAAA,IACX;AAAA,EACF;AACA,SAAO;AACT;AAKO,IAAM,uBAA2C;AAAA,EACtD;AAAA,EACA;AAAA,EACA;AACF;AAKO,SAAS,iBACd,KACA,OAAqB,CAAC,GACtB,QAA4B,sBACb;AACf,MAAI,MAAqB;AAAA,IACvB,OAAO;AAAA,IACP,SAAS;AAAA,IACT,UAAU,CAAC;AAAA,EACb;AAEA,aAAW,QAAQ,OAAO;AACxB,UAAM,KAAK,GAAG;AAAA,EAChB;AAEA,SAAO;AACT;","names":[]}
+{"version":3,"sources":["../src/parser.ts","../src/preprocessor.ts","../src/processor.ts","../src/tokenizer.ts","../src/utils.ts"],"sourcesContent":["import type { Root } from '@rethinkhealth/hl7v2-ast';\nimport { DEFAULT_DELIMITERS } from '@rethinkhealth/hl7v2-utils';\nimport type { Plugin } from 'unified';\nimport { defaultPreprocessors, runPreprocessors } from './preprocessor';\nimport { parseHL7v2FromIterator } from './processor';\nimport { HL7v2Tokenizer } from './tokenizer';\nimport type { ParseOptions, ParserContext } from './types';\nimport { iterateTokenizerSync } from './utils';\n\nexport function parseHL7v2(input: string, opts: ParseOptions): Root {\n  let ctx: ParserContext = {\n    input,\n    delimiters: opts.delimiters || DEFAULT_DELIMITERS,\n  };\n  // Run preprocessing\n  ctx = runPreprocessors(ctx, opts.preprocess || defaultPreprocessors);\n  // Run tokenizer\n  const tokenizer = new HL7v2Tokenizer();\n  // Reset tokenizer.\n  tokenizer.reset(ctx);\n  // Parse\n  return parseHL7v2FromIterator(iterateTokenizerSync(tokenizer));\n}\n\nconst hl7v2Parser: Plugin<[ParseOptions?], string, Root> = function (\n  options = {}\n) {\n  this.parser = (document: string) => parseHL7v2(document, options);\n};\n\nexport default hl7v2Parser;\n","import type { ParserContext, PreprocessorStep } from './types';\n\n// PreprocessorStep is declared in `types.ts` to avoid circular imports\n\n/**\n * Step: strip UTF-8 BOM.\n */\nexport const stripBOM: PreprocessorStep = (ctx) => {\n  if (ctx.input.charCodeAt(0) === 0xfe_ff) {\n    ctx.input = ctx.input.slice(1);\n  }\n  return ctx;\n};\n\n/**\n * Step: normalize newlines to CR (\\r).\n */\nexport const normalizeNewlines: PreprocessorStep = (ctx) => {\n  ctx.input = ctx.input.replace(/\\r?\\n/g, '\\r');\n  return ctx;\n};\n\n/**\n * Step: auto-detect delimiters from MSH.\n */\nexport const detectDelimiters: PreprocessorStep = (ctx) => {\n  if (ctx.input.startsWith('MSH')) {\n    const fieldDelim = ctx.input.charAt(3) || '|';\n    const enc = ctx.input.slice(4, 8);\n    const component = enc.charAt(0) || '^';\n    const repetition = enc.charAt(1) || '~';\n    const _escape = enc.charAt(2) || '\\\\';\n    const subcomponent = enc.charAt(3) || '&';\n\n    ctx.delimiters = {\n      field: fieldDelim,\n      component,\n      repetition,\n      escape: _escape,\n      subcomponent,\n      segment: '\\r',\n    };\n  }\n  return ctx;\n};\n\n/**\n * Default preprocessing pipeline.\n */\nexport const defaultPreprocessors: PreprocessorStep[] = [\n  stripBOM,\n  normalizeNewlines,\n  detectDelimiters,\n];\n\n/**\n * Run the pipeline to initialize ParserContext.\n */\nexport function runPreprocessors(\n  ctx: ParserContext,\n  steps: PreprocessorStep[]\n): ParserContext {\n  for (const step of steps) {\n    // biome-ignore lint/style/noParameterAssign: this is necessary to keep no-copy\n    ctx = step(ctx);\n  }\n\n  return ctx;\n}\n","// src/parser.ts\n\nimport type {\n  Component,\n  Field,\n  FieldRepetition,\n  Root,\n  Segment,\n  Subcomponent,\n} from '@rethinkhealth/hl7v2-ast';\nimport type { Token } from './types';\n\n// Shared core: process a single token into mutable parse state\nfunction createParserCore() {\n  const root: Root = { type: 'root', children: [] };\n\n  let seg: Segment | null = null;\n  let field: Field | null = null;\n  let rep: FieldRepetition | null = null;\n  let comp: Component | null = null;\n  let currentSub: Subcomponent | null = null;\n  let segmentHasContent = false;\n\n  const openSegment = () => {\n    seg = { type: 'segment', children: [] };\n    root.children.push(seg);\n    field = null;\n    rep = null;\n    comp = null;\n    currentSub = null;\n    segmentHasContent = false;\n  };\n\n  const ensureSegment = () => {\n    if (!seg) {\n      openSegment();\n    }\n  };\n\n  const openField = () => {\n    ensureSegment();\n    field = { type: 'field', children: [] };\n    // biome-ignore lint/style/noNonNullAssertion: seg is ensured above\n    seg!.children.push(field);\n    rep = { type: 'field-repetition', children: [] };\n    field.children.push(rep);\n    comp = { type: 'component', children: [] };\n    rep.children.push(comp);\n    currentSub = null;\n    segmentHasContent = true;\n  };\n\n  const openRepetition = () => {\n    if (!field) {\n      openField();\n    }\n    rep = { type: 'field-repetition', children: [] };\n    // biome-ignore lint/style/noNonNullAssertion: field is ensured above\n    field!.children.push(rep);\n    comp = { type: 'component', children: [] };\n    rep.children.push(comp);\n    currentSub = null;\n    segmentHasContent = true;\n  };\n\n  const openComponent = () => {\n    if (!field) {\n      openField();\n    }\n    if (!rep) {\n      rep = { type: 'field-repetition', children: [] };\n      // biome-ignore lint/style/noNonNullAssertion: field is ensured above\n      field!.children.push(rep);\n    }\n    comp = { type: 'component', children: [] };\n    // biome-ignore lint/style/noNonNullAssertion: rep is ensured above\n    rep!.children.push(comp);\n    currentSub = null;\n    segmentHasContent = true;\n  };\n\n  const ensureForText = () => {\n    if (!field) {\n      openField();\n    }\n    if (!rep) {\n      openRepetition();\n    }\n    if (!comp) {\n      openComponent();\n    }\n    if (!currentSub) {\n      currentSub = { type: 'subcomponent', value: '' };\n      // biome-ignore lint/style/noNonNullAssertion: comp is ensured above\n      comp!.children.push(currentSub);\n      segmentHasContent = true;\n    }\n  };\n\n  const processToken = (tok: Token) => {\n    switch (tok.type) {\n      case 'SEGMENT_END': {\n        dropTrailingEmptyFieldIfPresent();\n        // Close current segment (if any) and reset; do not auto-open next segment\n        seg = null;\n        field = null;\n        rep = null;\n        comp = null;\n        currentSub = null;\n        segmentHasContent = false;\n        return;\n      }\n      case 'FIELD_DELIM': {\n        // Leading field delimiter implies an empty first field\n        if (!field) {\n          // Open an empty first field and record an empty subcomponent slot\n          openField();\n          // biome-ignore lint/style/noNonNullAssertion: comp is initialized in openField\n          comp!.children.push({ type: 'subcomponent', value: '' });\n          segmentHasContent = true;\n        }\n        openField();\n        return;\n      }\n      case 'REPETITION_DELIM': {\n        if (!field) {\n          openField();\n        }\n        openRepetition();\n        return;\n      }\n      case 'COMPONENT_DELIM': {\n        openComponent();\n        return;\n      }\n      case 'SUBCOMP_DELIM': {\n        if (!comp) {\n          openComponent();\n        }\n        // Start a new empty subcomponent slot\n        currentSub = { type: 'subcomponent', value: '' };\n        // biome-ignore lint/style/noNonNullAssertion: comp is ensured above\n        comp!.children.push(currentSub);\n        segmentHasContent = true;\n        return;\n      }\n      case 'TEXT': {\n        const val = tok.value ?? '';\n        ensureForText();\n        // biome-ignore lint/style/noNonNullAssertion: ensured above\n        currentSub!.value += val;\n        segmentHasContent = true;\n        return;\n      }\n      default: {\n        throw new Error(`Unexpected token type: ${tok.type}`);\n      }\n    }\n  };\n\n  // Do not pre-open a segment; lazily open upon first non-SEGMENT_END token.\n\n  const finalize = () => {\n    // Handle input that ends without an explicit segment delimiter as above.\n    dropTrailingEmptyFieldIfPresent();\n    if (seg && !segmentHasContent) {\n      // Drop trailing empty segment if no content was added\n      root.children.pop();\n    }\n    return root;\n  };\n\n  function dropTrailingEmptyFieldIfPresent() {\n    if (!seg || seg.children.length === 0) {\n      return;\n    }\n    // Drop only the final trailing empty field (created by the last delimiter),\n    // preserving any intentional empty fields immediately before it.\n    const lastField = seg.children.at(-1) as Field;\n    const hasAnySubcomponents = lastField.children.some((r) =>\n      r.children.some((c) => c.children.length > 0)\n    );\n    if (!hasAnySubcomponents) {\n      seg.children.pop();\n    }\n  }\n\n  return { processToken, finalize, root };\n}\n\n// Sync convenience wrapper over a sync Iterable token source\nexport function parseHL7v2FromIterator(tokens: Iterable<Token>): Root {\n  const core = createParserCore();\n  for (const tok of tokens) {\n    core.processToken(tok);\n  }\n  return core.finalize();\n}\n","// src/tokenizer.ts\nimport type { HL7v2Delimiters } from '@rethinkhealth/hl7v2-utils';\nimport type {\n  ParserContext,\n  Position,\n  Token,\n  Tokenizer,\n  TokenType,\n} from './types';\n\nexport class HL7v2Tokenizer implements Tokenizer, Iterable<Token> {\n  private input = '';\n  private i = 0;\n  private line = 1;\n  private col = 1;\n  private delims!: HL7v2Delimiters;\n\n  // Only-run-once MSH bootstrap at the start of the file\n  private didMshBootstrap = false;\n  private pendingBootstrap: null | Array<\n    | { kind: 'TEXT'; value: string; advance: number }\n    | { kind: 'FIELD_DELIM'; advance: number }\n  > = null; // queue to emit TEXT('MSH'), FIELD_DELIM, TEXT('^~\\\\&')\n\n  reset(ctx: ParserContext) {\n    this.input = ctx.input;\n    this.delims = ctx.delimiters;\n    this.i = 0;\n    this.line = 1;\n    this.col = 1;\n    this.didMshBootstrap = false;\n    this.pendingBootstrap = null;\n\n    // Prepare a one-time bootstrap if file starts with MSH\n    if (this.input.startsWith('MSH')) {\n      // Precompute MSH and MSH.2; MSH.1 is the field delimiter char at index 3\n      const msh = this._slice(0, 3);\n      const msh1 = this._slice(3, 4); // the field delimiter char at index 3\n      const msh2 = this._slice(4, 8); // may be shorter than 4 if truncated\n      this.pendingBootstrap = [\n        { kind: 'TEXT', value: msh, advance: msh.length },\n        { kind: 'FIELD_DELIM', advance: 0 },\n        { kind: 'TEXT', value: msh1, advance: msh1.length },\n        { kind: 'FIELD_DELIM', advance: 0 },\n        { kind: 'TEXT', value: msh2, advance: msh2.length },\n      ];\n      // NOTE: we have not advanced indices yet; we will as we emit tokens\n    }\n  }\n\n  // biome-ignore lint/complexity/noExcessiveCognitiveComplexity: The cognitive complexity of this method is justified because it must handle the HL7v2 MSH segment bootstrap logic and multiple tokenization cases in a single method for performance and maintainability.\n  next(): Token | null {\n    const s = this.input;\n    const n = s.length;\n    if (this.i >= n && !this.pendingBootstrap?.length) {\n      return null;\n    }\n\n    const start = { offset: this.i, line: this.line, column: this.col };\n\n    // ---- One-time MSH bootstrap at file start ----\n    if (!this.didMshBootstrap && this.pendingBootstrap) {\n      const step = this.pendingBootstrap.shift();\n      if (step) {\n        if (step.kind === 'TEXT') {\n          const take = Math.min(step.advance, n - this.i);\n          const out = step.value.slice(0, take);\n          this._fastAdvance(out);\n          if (this.pendingBootstrap.length === 0) {\n            this.pendingBootstrap = null;\n            this.didMshBootstrap = true;\n          }\n          return this._tok('TEXT', out, start);\n        }\n        // FIELD_DELIM: advance exactly one char (the delimiter) and emit a FIELD_DELIM token\n        this._advance(Math.min(step.advance, n - this.i));\n        if (this.pendingBootstrap.length === 0) {\n          this.pendingBootstrap = null;\n          this.didMshBootstrap = true;\n        }\n        return this._tok('FIELD_DELIM', undefined, start);\n      }\n    }\n\n    // ---- Normal tokenization (delimiters + text) ----\n    if (this.i >= n) {\n      return null;\n    }\n    const ch = s.charAt(this.i);\n\n    if (ch === this.delims.segment) {\n      this._advance(1, true);\n      return this._tok('SEGMENT_END', undefined, start);\n    }\n    if (ch === this.delims.field) {\n      this._advance(1);\n      return this._tok('FIELD_DELIM', undefined, start);\n    }\n    if (ch === this.delims.repetition) {\n      this._advance(1);\n      return this._tok('REPETITION_DELIM', undefined, start);\n    }\n    if (ch === this.delims.component) {\n      this._advance(1);\n      return this._tok('COMPONENT_DELIM', undefined, start);\n    }\n    if (ch === this.delims.subcomponent) {\n      this._advance(1);\n      return this._tok('SUBCOMP_DELIM', undefined, start);\n    }\n\n    // TEXT until next delimiter or end\n    let j = this.i;\n    const seg = this.delims.segment,\n      fld = this.delims.field,\n      rep = this.delims.repetition,\n      cmp = this.delims.component,\n      sub = this.delims.subcomponent;\n\n    while (j < s.length) {\n      const c = s.charAt(j);\n      if (c === seg || c === fld || c === rep || c === cmp || c === sub) {\n        break;\n      }\n      j++;\n    }\n\n    const val = s.slice(this.i, j);\n    this._fastAdvance(val);\n    return this._tok('TEXT', val, start);\n  }\n\n  // ---- helpers ----\n  private _slice(start: number, end: number): string {\n    return this.input.slice(start, Math.min(end, this.input.length));\n  }\n\n  private _advance(n: number, isNewline = false) {\n    this.i += n;\n    if (isNewline) {\n      this.line += 1;\n      this.col = 1;\n    } else {\n      this.col += n;\n    }\n  }\n\n  private _fastAdvance(chunk: string) {\n    const parts = chunk.split('\\r');\n    if (parts.length > 1) {\n      this.line += parts.length - 1;\n      this.col = (parts.at(-1)?.length ?? 0) + 1;\n    } else {\n      this.col += chunk.length;\n    }\n    this.i += chunk.length;\n  }\n\n  private _tok(\n    type: TokenType,\n    value: string | undefined,\n    start: Position['start']\n  ): Token {\n    const end = { offset: this.i, line: this.line, column: this.col };\n    return { type, value, position: { start, end } };\n  }\n\n  // Iterable protocol (sync)\n  [Symbol.iterator](): Iterator<Token> {\n    return {\n      next: () => {\n        const t = this.next();\n        if (t == null) {\n          return { done: true, value: undefined as unknown as Token };\n        }\n        return { done: false, value: t };\n      },\n    };\n  }\n\n  // Async iteration support removed to keep the API synchronous.\n}\n","import type { Token, Tokenizer } from './types';\n\nexport function* iterateTokenizerSync(t: Tokenizer): Iterable<Token> {\n  for (let tok = t.next(); tok; tok = t.next()) {\n    yield tok;\n  }\n}\n\n// Removed async stream support; the parser now operates synchronously only.\n"],"mappings":";AACA,SAAS,0BAA0B;;;ACM5B,IAAM,WAA6B,CAAC,QAAQ;AACjD,MAAI,IAAI,MAAM,WAAW,CAAC,MAAM,OAAS;AACvC,QAAI,QAAQ,IAAI,MAAM,MAAM,CAAC;AAAA,EAC/B;AACA,SAAO;AACT;AAKO,IAAM,oBAAsC,CAAC,QAAQ;AAC1D,MAAI,QAAQ,IAAI,MAAM,QAAQ,UAAU,IAAI;AAC5C,SAAO;AACT;AAKO,IAAM,mBAAqC,CAAC,QAAQ;AACzD,MAAI,IAAI,MAAM,WAAW,KAAK,GAAG;AAC/B,UAAM,aAAa,IAAI,MAAM,OAAO,CAAC,KAAK;AAC1C,UAAM,MAAM,IAAI,MAAM,MAAM,GAAG,CAAC;AAChC,UAAM,YAAY,IAAI,OAAO,CAAC,KAAK;AACnC,UAAM,aAAa,IAAI,OAAO,CAAC,KAAK;AACpC,UAAM,UAAU,IAAI,OAAO,CAAC,KAAK;AACjC,UAAM,eAAe,IAAI,OAAO,CAAC,KAAK;AAEtC,QAAI,aAAa;AAAA,MACf,OAAO;AAAA,MACP;AAAA,MACA;AAAA,MACA,QAAQ;AAAA,MACR;AAAA,MACA,SAAS;AAAA,IACX;AAAA,EACF;AACA,SAAO;AACT;AAKO,IAAM,uBAA2C;AAAA,EACtD;AAAA,EACA;AAAA,EACA;AACF;AAKO,SAAS,iBACd,KACA,OACe;AACf,aAAW,QAAQ,OAAO;AAExB,UAAM,KAAK,GAAG;AAAA,EAChB;AAEA,SAAO;AACT;;;ACvDA,SAAS,mBAAmB;AAC1B,QAAM,OAAa,EAAE,MAAM,QAAQ,UAAU,CAAC,EAAE;AAEhD,MAAI,MAAsB;AAC1B,MAAI,QAAsB;AAC1B,MAAI,MAA8B;AAClC,MAAI,OAAyB;AAC7B,MAAI,aAAkC;AACtC,MAAI,oBAAoB;AAExB,QAAM,cAAc,MAAM;AACxB,UAAM,EAAE,MAAM,WAAW,UAAU,CAAC,EAAE;AACtC,SAAK,SAAS,KAAK,GAAG;AACtB,YAAQ;AACR,UAAM;AACN,WAAO;AACP,iBAAa;AACb,wBAAoB;AAAA,EACtB;AAEA,QAAM,gBAAgB,MAAM;AAC1B,QAAI,CAAC,KAAK;AACR,kBAAY;AAAA,IACd;AAAA,EACF;AAEA,QAAM,YAAY,MAAM;AACtB,kBAAc;AACd,YAAQ,EAAE,MAAM,SAAS,UAAU,CAAC,EAAE;AAEtC,QAAK,SAAS,KAAK,KAAK;AACxB,UAAM,EAAE,MAAM,oBAAoB,UAAU,CAAC,EAAE;AAC/C,UAAM,SAAS,KAAK,GAAG;AACvB,WAAO,EAAE,MAAM,aAAa,UAAU,CAAC,EAAE;AACzC,QAAI,SAAS,KAAK,IAAI;AACtB,iBAAa;AACb,wBAAoB;AAAA,EACtB;AAEA,QAAM,iBAAiB,MAAM;AAC3B,QAAI,CAAC,OAAO;AACV,gBAAU;AAAA,IACZ;AACA,UAAM,EAAE,MAAM,oBAAoB,UAAU,CAAC,EAAE;AAE/C,UAAO,SAAS,KAAK,GAAG;AACxB,WAAO,EAAE,MAAM,aAAa,UAAU,CAAC,EAAE;AACzC,QAAI,SAAS,KAAK,IAAI;AACtB,iBAAa;AACb,wBAAoB;AAAA,EACtB;AAEA,QAAM,gBAAgB,MAAM;AAC1B,QAAI,CAAC,OAAO;AACV,gBAAU;AAAA,IACZ;AACA,QAAI,CAAC,KAAK;AACR,YAAM,EAAE,MAAM,oBAAoB,UAAU,CAAC,EAAE;AAE/C,YAAO,SAAS,KAAK,GAAG;AAAA,IAC1B;AACA,WAAO,EAAE,MAAM,aAAa,UAAU,CAAC,EAAE;AAEzC,QAAK,SAAS,KAAK,IAAI;AACvB,iBAAa;AACb,wBAAoB;AAAA,EACtB;AAEA,QAAM,gBAAgB,MAAM;AAC1B,QAAI,CAAC,OAAO;AACV,gBAAU;AAAA,IACZ;AACA,QAAI,CAAC,KAAK;AACR,qBAAe;AAAA,IACjB;AACA,QAAI,CAAC,MAAM;AACT,oBAAc;AAAA,IAChB;AACA,QAAI,CAAC,YAAY;AACf,mBAAa,EAAE,MAAM,gBAAgB,OAAO,GAAG;AAE/C,WAAM,SAAS,KAAK,UAAU;AAC9B,0BAAoB;AAAA,IACtB;AAAA,EACF;AAEA,QAAM,eAAe,CAAC,QAAe;AACnC,YAAQ,IAAI,MAAM;AAAA,MAChB,KAAK,eAAe;AAClB,wCAAgC;AAEhC,cAAM;AACN,gBAAQ;AACR,cAAM;AACN,eAAO;AACP,qBAAa;AACb,4BAAoB;AACpB;AAAA,MACF;AAAA,MACA,KAAK,eAAe;AAElB,YAAI,CAAC,OAAO;AAEV,oBAAU;AAEV,eAAM,SAAS,KAAK,EAAE,MAAM,gBAAgB,OAAO,GAAG,CAAC;AACvD,8BAAoB;AAAA,QACtB;AACA,kBAAU;AACV;AAAA,MACF;AAAA,MACA,KAAK,oBAAoB;AACvB,YAAI,CAAC,OAAO;AACV,oBAAU;AAAA,QACZ;AACA,uBAAe;AACf;AAAA,MACF;AAAA,MACA,KAAK,mBAAmB;AACtB,sBAAc;AACd;AAAA,MACF;AAAA,MACA,KAAK,iBAAiB;AACpB,YAAI,CAAC,MAAM;AACT,wBAAc;AAAA,QAChB;AAEA,qBAAa,EAAE,MAAM,gBAAgB,OAAO,GAAG;AAE/C,aAAM,SAAS,KAAK,UAAU;AAC9B,4BAAoB;AACpB;AAAA,MACF;AAAA,MACA,KAAK,QAAQ;AACX,cAAM,MAAM,IAAI,SAAS;AACzB,sBAAc;AAEd,mBAAY,SAAS;AACrB,4BAAoB;AACpB;AAAA,MACF;AAAA,MACA,SAAS;AACP,cAAM,IAAI,MAAM,0BAA0B,IAAI,IAAI,EAAE;AAAA,MACtD;AAAA,IACF;AAAA,EACF;AAIA,QAAM,WAAW,MAAM;AAErB,oCAAgC;AAChC,QAAI,OAAO,CAAC,mBAAmB;AAE7B,WAAK,SAAS,IAAI;AAAA,IACpB;AACA,WAAO;AAAA,EACT;AAEA,WAAS,kCAAkC;AACzC,QAAI,CAAC,OAAO,IAAI,SAAS,WAAW,GAAG;AACrC;AAAA,IACF;AAGA,UAAM,YAAY,IAAI,SAAS,GAAG,EAAE;AACpC,UAAM,sBAAsB,UAAU,SAAS;AAAA,MAAK,CAAC,MACnD,EAAE,SAAS,KAAK,CAAC,MAAM,EAAE,SAAS,SAAS,CAAC;AAAA,IAC9C;AACA,QAAI,CAAC,qBAAqB;AACxB,UAAI,SAAS,IAAI;AAAA,IACnB;AAAA,EACF;AAEA,SAAO,EAAE,cAAc,UAAU,KAAK;AACxC;AAGO,SAAS,uBAAuB,QAA+B;AACpE,QAAM,OAAO,iBAAiB;AAC9B,aAAW,OAAO,QAAQ;AACxB,SAAK,aAAa,GAAG;AAAA,EACvB;AACA,SAAO,KAAK,SAAS;AACvB;;;AC3LO,IAAM,iBAAN,MAA2D;AAAA,EACxD,QAAQ;AAAA,EACR,IAAI;AAAA,EACJ,OAAO;AAAA,EACP,MAAM;AAAA,EACN;AAAA;AAAA,EAGA,kBAAkB;AAAA,EAClB,mBAGJ;AAAA;AAAA,EAEJ,MAAM,KAAoB;AACxB,SAAK,QAAQ,IAAI;AACjB,SAAK,SAAS,IAAI;AAClB,SAAK,IAAI;AACT,SAAK,OAAO;AACZ,SAAK,MAAM;AACX,SAAK,kBAAkB;AACvB,SAAK,mBAAmB;AAGxB,QAAI,KAAK,MAAM,WAAW,KAAK,GAAG;AAEhC,YAAM,MAAM,KAAK,OAAO,GAAG,CAAC;AAC5B,YAAM,OAAO,KAAK,OAAO,GAAG,CAAC;AAC7B,YAAM,OAAO,KAAK,OAAO,GAAG,CAAC;AAC7B,WAAK,mBAAmB;AAAA,QACtB,EAAE,MAAM,QAAQ,OAAO,KAAK,SAAS,IAAI,OAAO;AAAA,QAChD,EAAE,MAAM,eAAe,SAAS,EAAE;AAAA,QAClC,EAAE,MAAM,QAAQ,OAAO,MAAM,SAAS,KAAK,OAAO;AAAA,QAClD,EAAE,MAAM,eAAe,SAAS,EAAE;AAAA,QAClC,EAAE,MAAM,QAAQ,OAAO,MAAM,SAAS,KAAK,OAAO;AAAA,MACpD;AAAA,IAEF;AAAA,EACF;AAAA;AAAA,EAGA,OAAqB;AACnB,UAAM,IAAI,KAAK;AACf,UAAM,IAAI,EAAE;AACZ,QAAI,KAAK,KAAK,KAAK,CAAC,KAAK,kBAAkB,QAAQ;AACjD,aAAO;AAAA,IACT;AAEA,UAAM,QAAQ,EAAE,QAAQ,KAAK,GAAG,MAAM,KAAK,MAAM,QAAQ,KAAK,IAAI;AAGlE,QAAI,CAAC,KAAK,mBAAmB,KAAK,kBAAkB;AAClD,YAAM,OAAO,KAAK,iBAAiB,MAAM;AACzC,UAAI,MAAM;AACR,YAAI,KAAK,SAAS,QAAQ;AACxB,gBAAM,OAAO,KAAK,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC;AAC9C,gBAAM,MAAM,KAAK,MAAM,MAAM,GAAG,IAAI;AACpC,eAAK,aAAa,GAAG;AACrB,cAAI,KAAK,iBAAiB,WAAW,GAAG;AACtC,iBAAK,mBAAmB;AACxB,iBAAK,kBAAkB;AAAA,UACzB;AACA,iBAAO,KAAK,KAAK,QAAQ,KAAK,KAAK;AAAA,QACrC;AAEA,aAAK,SAAS,KAAK,IAAI,KAAK,SAAS,IAAI,KAAK,CAAC,CAAC;AAChD,YAAI,KAAK,iBAAiB,WAAW,GAAG;AACtC,eAAK,mBAAmB;AACxB,eAAK,kBAAkB;AAAA,QACzB;AACA,eAAO,KAAK,KAAK,eAAe,QAAW,KAAK;AAAA,MAClD;AAAA,IACF;AAGA,QAAI,KAAK,KAAK,GAAG;AACf,aAAO;AAAA,IACT;AACA,UAAM,KAAK,EAAE,OAAO,KAAK,CAAC;AAE1B,QAAI,OAAO,KAAK,OAAO,SAAS;AAC9B,WAAK,SAAS,GAAG,IAAI;AACrB,aAAO,KAAK,KAAK,eAAe,QAAW,KAAK;AAAA,IAClD;AACA,QAAI,OAAO,KAAK,OAAO,OAAO;AAC5B,WAAK,SAAS,CAAC;AACf,aAAO,KAAK,KAAK,eAAe,QAAW,KAAK;AAAA,IAClD;AACA,QAAI,OAAO,KAAK,OAAO,YAAY;AACjC,WAAK,SAAS,CAAC;AACf,aAAO,KAAK,KAAK,oBAAoB,QAAW,KAAK;AAAA,IACvD;AACA,QAAI,OAAO,KAAK,OAAO,WAAW;AAChC,WAAK,SAAS,CAAC;AACf,aAAO,KAAK,KAAK,mBAAmB,QAAW,KAAK;AAAA,IACtD;AACA,QAAI,OAAO,KAAK,OAAO,cAAc;AACnC,WAAK,SAAS,CAAC;AACf,aAAO,KAAK,KAAK,iBAAiB,QAAW,KAAK;AAAA,IACpD;AAGA,QAAI,IAAI,KAAK;AACb,UAAM,MAAM,KAAK,OAAO,SACtB,MAAM,KAAK,OAAO,OAClB,MAAM,KAAK,OAAO,YAClB,MAAM,KAAK,OAAO,WAClB,MAAM,KAAK,OAAO;AAEpB,WAAO,IAAI,EAAE,QAAQ;AACnB,YAAM,IAAI,EAAE,OAAO,CAAC;AACpB,UAAI,MAAM,OAAO,MAAM,OAAO,MAAM,OAAO,MAAM,OAAO,MAAM,KAAK;AACjE;AAAA,MACF;AACA;AAAA,IACF;AAEA,UAAM,MAAM,EAAE,MAAM,KAAK,GAAG,CAAC;AAC7B,SAAK,aAAa,GAAG;AACrB,WAAO,KAAK,KAAK,QAAQ,KAAK,KAAK;AAAA,EACrC;AAAA;AAAA,EAGQ,OAAO,OAAe,KAAqB;AACjD,WAAO,KAAK,MAAM,MAAM,OAAO,KAAK,IAAI,KAAK,KAAK,MAAM,MAAM,CAAC;AAAA,EACjE;AAAA,EAEQ,SAAS,GAAW,YAAY,OAAO;AAC7C,SAAK,KAAK;AACV,QAAI,WAAW;AACb,WAAK,QAAQ;AACb,WAAK,MAAM;AAAA,IACb,OAAO;AACL,WAAK,OAAO;AAAA,IACd;AAAA,EACF;AAAA,EAEQ,aAAa,OAAe;AAClC,UAAM,QAAQ,MAAM,MAAM,IAAI;AAC9B,QAAI,MAAM,SAAS,GAAG;AACpB,WAAK,QAAQ,MAAM,SAAS;AAC5B,WAAK,OAAO,MAAM,GAAG,EAAE,GAAG,UAAU,KAAK;AAAA,IAC3C,OAAO;AACL,WAAK,OAAO,MAAM;AAAA,IACpB;AACA,SAAK,KAAK,MAAM;AAAA,EAClB;AAAA,EAEQ,KACN,MACA,OACA,OACO;AACP,UAAM,MAAM,EAAE,QAAQ,KAAK,GAAG,MAAM,KAAK,MAAM,QAAQ,KAAK,IAAI;AAChE,WAAO,EAAE,MAAM,OAAO,UAAU,EAAE,OAAO,IAAI,EAAE;AAAA,EACjD;AAAA;AAAA,EAGA,CAAC,OAAO,QAAQ,IAAqB;AACnC,WAAO;AAAA,MACL,MAAM,MAAM;AACV,cAAM,IAAI,KAAK,KAAK;AACpB,YAAI,KAAK,MAAM;AACb,iBAAO,EAAE,MAAM,MAAM,OAAO,OAA8B;AAAA,QAC5D;AACA,eAAO,EAAE,MAAM,OAAO,OAAO,EAAE;AAAA,MACjC;AAAA,IACF;AAAA,EACF;AAAA;AAGF;;;ACnLO,UAAU,qBAAqB,GAA+B;AACnE,WAAS,MAAM,EAAE,KAAK,GAAG,KAAK,MAAM,EAAE,KAAK,GAAG;AAC5C,UAAM;AAAA,EACR;AACF;;;AJGO,SAAS,WAAW,OAAe,MAA0B;AAClE,MAAI,MAAqB;AAAA,IACvB;AAAA,IACA,YAAY,KAAK,cAAc;AAAA,EACjC;AAEA,QAAM,iBAAiB,KAAK,KAAK,cAAc,oBAAoB;AAEnE,QAAM,YAAY,IAAI,eAAe;AAErC,YAAU,MAAM,GAAG;AAEnB,SAAO,uBAAuB,qBAAqB,SAAS,CAAC;AAC/D;AAEA,IAAM,cAAqD,SACzD,UAAU,CAAC,GACX;AACA,OAAK,SAAS,CAAC,aAAqB,WAAW,UAAU,OAAO;AAClE;AAEA,IAAO,iBAAQ;","names":[]}

package/dist/parser.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"parser.d.ts","sourceRoot":"","sources":["../src/parser.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,0BAA0B,CAAC;AACrD,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,SAAS,CAAC;AAGtC,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,SAAS,CAAC;AAI5C,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,YAAY,GAAG,IAAI,CAIlE;AAED,QAAA,MAAM,WAAW,EAAE,MAAM,CAAC,CAAC,YAAY,CAAC,CAAC,EAAE,MAAM,EAAE,IAAI,CAKtD,CAAC;AAEF,eAAe,WAAW,CAAC"}
+{"version":3,"file":"parser.d.ts","sourceRoot":"","sources":["../src/parser.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,IAAI,EAAE,MAAM,0BAA0B,CAAC;AAErD,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,SAAS,CAAC;AAItC,OAAO,KAAK,EAAE,YAAY,EAAiB,MAAM,SAAS,CAAC;AAG3D,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,YAAY,GAAG,IAAI,CAalE;AAED,QAAA,MAAM,WAAW,EAAE,MAAM,CAAC,CAAC,YAAY,CAAC,CAAC,EAAE,MAAM,EAAE,IAAI,CAItD,CAAC;AAEF,eAAe,WAAW,CAAC"}

package/dist/preprocessor.d.ts

@@ -1,8 +1,4 @@
-import type { ParseOptions, ParserContext } from './types';
-/**
- * A preprocessing step: takes a ParserContext, mutates or replaces it.
- */
-export type PreprocessorStep = (ctx: ParserContext) => ParserContext;
+import type { ParserContext, PreprocessorStep } from './types';
 /**
  * Step: strip UTF-8 BOM.
  */
@@ -22,5 +18,5 @@ export declare const defaultPreprocessors: PreprocessorStep[];
 /**
  * Run the pipeline to initialize ParserContext.
  */
-export declare function runPreprocessors(raw: string, opts?: ParseOptions, steps?: PreprocessorStep[]): ParserContext;
+export declare function runPreprocessors(ctx: ParserContext, steps: PreprocessorStep[]): ParserContext;
 //# sourceMappingURL=preprocessor.d.ts.map

package/dist/preprocessor.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"preprocessor.d.ts","sourceRoot":"","sources":["../src/preprocessor.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,SAAS,CAAC;AAE3D;;GAEG;AACH,MAAM,MAAM,gBAAgB,GAAG,CAAC,GAAG,EAAE,aAAa,KAAK,aAAa,CAAC;AAErE;;GAEG;AACH,eAAO,MAAM,QAAQ,EAAE,gBAKtB,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,iBAAiB,EAAE,gBAG/B,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,gBAAgB,EAAE,gBAmB9B,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,oBAAoB,EAAE,gBAAgB,EAIlD,CAAC;AAEF;;GAEG;AACH,wBAAgB,gBAAgB,CAC9B,GAAG,EAAE,MAAM,EACX,IAAI,GAAE,YAAiB,EACvB,KAAK,GAAE,gBAAgB,EAAyB,GAC/C,aAAa,CAYf"}
+{"version":3,"file":"preprocessor.d.ts","sourceRoot":"","sources":["../src/preprocessor.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,aAAa,EAAE,gBAAgB,EAAE,MAAM,SAAS,CAAC;AAI/D;;GAEG;AACH,eAAO,MAAM,QAAQ,EAAE,gBAKtB,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,iBAAiB,EAAE,gBAG/B,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,gBAAgB,EAAE,gBAmB9B,CAAC;AAEF;;GAEG;AACH,eAAO,MAAM,oBAAoB,EAAE,gBAAgB,EAIlD,CAAC;AAEF;;GAEG;AACH,wBAAgB,gBAAgB,CAC9B,GAAG,EAAE,aAAa,EAClB,KAAK,EAAE,gBAAgB,EAAE,GACxB,aAAa,CAOf"}

package/dist/tokenizer.d.ts

@@ -1,4 +1,4 @@
-import type { ParseOptions, Token, Tokenizer } from './types';
+import type { ParserContext, Token, Tokenizer } from './types';
 export declare class HL7v2Tokenizer implements Tokenizer, Iterable<Token> {
     private input;
     private i;
@@ -7,7 +7,7 @@ export declare class HL7v2Tokenizer implements Tokenizer, Iterable<Token> {
     private delims;
     private didMshBootstrap;
     private pendingBootstrap;
-    reset(input: string, opts: ParseOptions): void;
+    reset(ctx: ParserContext): void;
     next(): Token | null;
     private _slice;
     private _advance;

package/dist/tokenizer.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"tokenizer.d.ts","sourceRoot":"","sources":["../src/tokenizer.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EACV,YAAY,EAEZ,KAAK,EACL,SAAS,EAEV,MAAM,SAAS,CAAC;AAEjB,qBAAa,cAAe,YAAW,SAAS,EAAE,QAAQ,CAAC,KAAK,CAAC;IAC/D,OAAO,CAAC,KAAK,CAAM;IACnB,OAAO,CAAC,CAAC,CAAK;IACd,OAAO,CAAC,IAAI,CAAK;IACjB,OAAO,CAAC,GAAG,CAAK;IAChB,OAAO,CAAC,MAAM,CAAmB;IAGjC,OAAO,CAAC,eAAe,CAAS;IAChC,OAAO,CAAC,gBAAgB,CAGf;IAET,KAAK,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,YAAY;IAwBvC,IAAI,IAAI,KAAK,GAAG,IAAI;IAkFpB,OAAO,CAAC,MAAM;IAId,OAAO,CAAC,QAAQ;IAUhB,OAAO,CAAC,YAAY;IAWpB,OAAO,CAAC,IAAI;IAUZ,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,QAAQ,CAAC,KAAK,CAAC;CAarC"}
+{"version":3,"file":"tokenizer.d.ts","sourceRoot":"","sources":["../src/tokenizer.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EACV,aAAa,EAEb,KAAK,EACL,SAAS,EAEV,MAAM,SAAS,CAAC;AAEjB,qBAAa,cAAe,YAAW,SAAS,EAAE,QAAQ,CAAC,KAAK,CAAC;IAC/D,OAAO,CAAC,KAAK,CAAM;IACnB,OAAO,CAAC,CAAC,CAAK;IACd,OAAO,CAAC,IAAI,CAAK;IACjB,OAAO,CAAC,GAAG,CAAK;IAChB,OAAO,CAAC,MAAM,CAAmB;IAGjC,OAAO,CAAC,eAAe,CAAS;IAChC,OAAO,CAAC,gBAAgB,CAGf;IAET,KAAK,CAAC,GAAG,EAAE,aAAa;IA2BxB,IAAI,IAAI,KAAK,GAAG,IAAI;IAkFpB,OAAO,CAAC,MAAM;IAId,OAAO,CAAC,QAAQ;IAUhB,OAAO,CAAC,YAAY;IAWpB,OAAO,CAAC,IAAI;IAUZ,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,QAAQ,CAAC,KAAK,CAAC;CAarC"}

package/dist/types.d.ts

@@ -1,10 +1,12 @@
 import type { HL7v2Delimiters } from '@rethinkhealth/hl7v2-utils';
+export type PreprocessorStep = (ctx: ParserContext) => ParserContext;
 export type ParseOptions = {
     delimiters?: HL7v2Delimiters;
+    preprocess?: PreprocessorStep[];
 };
 export interface ParserContext {
     input: string;
-    options: ParseOptions;
+    delimiters: HL7v2Delimiters;
     metadata?: Record<string, unknown>;
 }
 export type Position = {
@@ -26,7 +28,7 @@ export type Token = {
 };
 export type TokenType = 'SEGMENT_START' | 'SEGMENT_END' | 'FIELD_DELIM' | 'REPETITION_DELIM' | 'COMPONENT_DELIM' | 'SUBCOMP_DELIM' | 'TEXT';
 export interface Tokenizer {
-    reset(input: string, opts: ParseOptions): void;
+    reset(ctx: ParserContext): void;
     next(): Token | null;
 }
 //# sourceMappingURL=types.d.ts.map

package/dist/types.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAC;AAElE,MAAM,MAAM,YAAY,GAAG;IACzB,UAAU,CAAC,EAAE,eAAe,CAAC;CAC9B,CAAC;AAEF,MAAM,WAAW,aAAa;IAC5B,KAAK,EAAE,MAAM,CAAC;IACd,OAAO,EAAE,YAAY,CAAC;IAEtB,QAAQ,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CACpC;AAGD,MAAM,MAAM,QAAQ,GAAG;IACrB,KAAK,EAAE;QAAE,MAAM,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC;IACxD,GAAG,EAAE;QAAE,MAAM,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC;CACvD,CAAC;AAEF,MAAM,MAAM,KAAK,GAAG;IAClB,IAAI,EAAE,SAAS,CAAC;IAChB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,EAAE,QAAQ,CAAC;CACpB,CAAC;AAGF,MAAM,MAAM,SAAS,GACjB,eAAe,GACf,aAAa,GACb,aAAa,GACb,kBAAkB,GAClB,iBAAiB,GACjB,eAAe,GACf,MAAM,CAAC;AAEX,MAAM,WAAW,SAAS;IACxB,KAAK,CAAC,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,YAAY,GAAG,IAAI,CAAC;IAC/C,IAAI,IAAI,KAAK,GAAG,IAAI,CAAC;CACtB"}
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../src/types.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,4BAA4B,CAAC;AAIlE,MAAM,MAAM,gBAAgB,GAAG,CAAC,GAAG,EAAE,aAAa,KAAK,aAAa,CAAC;AAErE,MAAM,MAAM,YAAY,GAAG;IACzB,UAAU,CAAC,EAAE,eAAe,CAAC;IAE7B,UAAU,CAAC,EAAE,gBAAgB,EAAE,CAAC;CACjC,CAAC;AAEF,MAAM,WAAW,aAAa;IAC5B,KAAK,EAAE,MAAM,CAAC;IACd,UAAU,EAAE,eAAe,CAAC;IAC5B,QAAQ,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CACpC;AAGD,MAAM,MAAM,QAAQ,GAAG;IACrB,KAAK,EAAE;QAAE,MAAM,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC;IACxD,GAAG,EAAE;QAAE,MAAM,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAC;QAAC,MAAM,EAAE,MAAM,CAAA;KAAE,CAAC;CACvD,CAAC;AAEF,MAAM,MAAM,KAAK,GAAG;IAClB,IAAI,EAAE,SAAS,CAAC;IAChB,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,QAAQ,EAAE,QAAQ,CAAC;CACpB,CAAC;AAGF,MAAM,MAAM,SAAS,GACjB,eAAe,GACf,aAAa,GACb,aAAa,GACb,kBAAkB,GAClB,iBAAiB,GACjB,eAAe,GACf,MAAM,CAAC;AAEX,MAAM,WAAW,SAAS;IACxB,KAAK,CAAC,GAAG,EAAE,aAAa,GAAG,IAAI,CAAC;IAChC,IAAI,IAAI,KAAK,GAAG,IAAI,CAAC;CACtB"}

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@rethinkhealth/hl7v2-parser",
   "description": "hl7v2 plugin to parse hl7v2 messages",
-  "version": "0.2.4",
+  "version": "0.2.5",
   "license": "MIT",
   "author": {
     "name": "Melek Somai",
@@ -17,7 +17,7 @@
   },
   "dependencies": {
     "unified": "11.0.1",
-    "@rethinkhealth/hl7v2-utils": "0.2.4"
+    "@rethinkhealth/hl7v2-utils": "0.2.5"
   },
   "devDependencies": {
     "@types/node": "22.15.31",
@@ -27,7 +27,7 @@
     "tsup": "8.5.0",
     "typescript": "^5.8.3",
     "vitest": "^3.2.4",
-    "@rethinkhealth/hl7v2-ast": "0.2.4",
+    "@rethinkhealth/hl7v2-ast": "0.2.5",
     "@rethinkhealth/testing": "0.0.0",
     "@rethinkhealth/tsconfig": "0.0.0"
   },