xml-tokenizer 0.0.32 → 0.0.34

package/dist/esm/tokenizer/tokenize.js

@@ -1,417 +1,2 @@
-import { getQName } from '../get-q-name.js';
-import { LESS_THAN, GREATER_THAN, SLASH, EQUALS, SINGLE_QUOTE, DOUBLE_QUOTE, HYPHEN, QUESTION_MARK, OPEN_SQUARE_BRACKET, PERCENT, EXCLAMATION_MARK, UPPERCASE_S, UPPERCASE_P, CLOSE_SQUARE_BRACKET } from './ascii-constants.js';
-import { isXmlSpaceByte } from './utils.js';
-import { XmlError } from './XmlError.js';
-import { XmlStream } from './XmlStream.js';
-
-const BOM = "\uFEFF";
-const XML_DECLARATION_START = "<?xml ";
-const XML_DECLARATION_END = "?>";
-const ENTITY_DECLARATION_START = "<!ENTITY";
-const ELEMENT_DECLARATION_START = "<!ELEMENT";
-const ATTLIST_DECLARATION_START = "<!ATTLIST";
-const NOTATION_DECLARATION_START = "<!NOTATION";
-const DOCTYPE_START = "<!DOCTYPE";
-const PI_START = "<?";
-const PI_END = "?>";
-const COMMENT_START = "<!--";
-const COMMENT_END = "-->";
-const CDATA_START = "<![CDATA[";
-const CDATA_END = "]]>";
-const NDATA = "NDATA";
-const PUBLIC = "PUBLIC";
-const SYSTEM = "SYSTEM";
-const VERSION = "version";
-const ENCODING = "encoding";
-const STANDALONE = "standalone";
-function tokenize(xmlString, tokenCallback, options = {}) {
-  tokenizeXmlStream(new XmlStream(xmlString, options), tokenCallback);
-}
-function tokenizeXmlStream(s, tokenCallback) {
-  if (s.startsWith(BOM)) {
-    s.advance(1);
-  }
-  if (s.startsWith(XML_DECLARATION_START)) {
-    parseDeclaration(s);
-  }
-  parseMisc(s, tokenCallback);
-  s.skipSpaces();
-  if (s.startsWith(DOCTYPE_START) || s.startsWith(DOCTYPE_START.toLowerCase())) {
-    if (!s.config.allowDtd) {
-      throw new XmlError({ type: "DtdDetected" });
-    }
-    parseDoctype(s, tokenCallback);
-    parseMisc(s, tokenCallback);
-  }
-  s.skipSpaces();
-  if (s.config.strictDocument) {
-    if (!s.atEnd() && s.currCodeUnit() === LESS_THAN) {
-      parseElement(s, tokenCallback);
-    }
-    parseMisc(s, tokenCallback);
-    if (!s.atEnd()) {
-      throw new XmlError({ type: "UnknownToken", message: "Not at end" }, s.genTextPos());
-    }
-  } else {
-    while (!s.atEnd()) {
-      if (s.currCodeUnit() === LESS_THAN) {
-        parseElement(s, tokenCallback);
-      } else {
-        parseText(s, tokenCallback);
-      }
-      parseMisc(s, tokenCallback);
-    }
-  }
-}
-function parseMisc(s, tokenCallback) {
-  while (!s.atEnd()) {
-    s.skipSpaces();
-    if (s.startsWith(COMMENT_START)) {
-      parseComment(s, tokenCallback);
-    } else if (s.startsWith(PI_START)) {
-      parsePi(s, tokenCallback);
-    } else {
-      break;
-    }
-  }
-}
-function parseDeclaration(s) {
-  function consumeSpaces(_s) {
-    if (_s.startsWithSpace()) {
-      _s.skipSpaces();
-    } else if (!_s.startsWith(XML_DECLARATION_END) && !_s.atEnd()) {
-      throw new XmlError(
-        { type: "InvalidChar", expected: "a whitespace", actual: _s.currCodeUnitUnchecked() },
-        _s.genTextPos()
-      );
-    }
-  }
-  s.advance(5);
-  consumeSpaces(s);
-  if (!s.startsWith(VERSION)) {
-    throw new XmlError({ type: "InvalidString", expected: "version" }, s.genTextPos());
-  }
-  parseAttribute(s);
-  consumeSpaces(s);
-  if (s.startsWith(ENCODING)) {
-    parseAttribute(s);
-    consumeSpaces(s);
-  }
-  if (s.startsWith(STANDALONE)) {
-    parseAttribute(s);
-  }
-  s.skipSpaces();
-  s.skipString(XML_DECLARATION_END);
-}
-function parseComment(s, tokenCallback) {
-  const start = s.getPos();
-  s.advance(4);
-  const text = s.consumeCodeUnitsWhile((c, _s) => !(c === HYPHEN && _s.startsWith(COMMENT_END)));
-  s.skipString(COMMENT_END);
-  if (text.includes("--") || text.endsWith("-")) {
-    throw new XmlError({ type: "InvalidComment" }, s.genTextPosFrom(start));
-  }
-  tokenCallback({ type: "Comment", text, range: s.rangeFrom(start) }, s);
-}
-function parsePi(s, tokenCallback) {
-  if (s.startsWith(XML_DECLARATION_START)) {
-    throw new XmlError({ type: "UnexpectedDeclaration" }, s.genTextPos());
-  }
-  const start = s.getPos();
-  s.advance(2);
-  const target = s.consumeName();
-  s.skipSpaces();
-  const content = s.consumeCodeUnitsWhile(
-    (c, _s) => !(c === QUESTION_MARK && _s.startsWith(PI_END))
-  );
-  s.skipString(PI_END);
-  tokenCallback(
-    {
-      type: "ProcessingInstruction",
-      target,
-      content: content.length === 0 ? void 0 : content,
-      range: s.rangeFrom(start)
-    },
-    s
-  );
-}
-function parseDoctype(s, tokenCallback) {
-  const start = s.getPos();
-  parseDoctypeStart(s);
-  s.skipSpaces();
-  if (s.currCodeUnit() === GREATER_THAN) {
-    s.advance(1);
-    return;
-  }
-  s.advance(1);
-  while (!s.atEnd()) {
-    s.skipSpaces();
-    if (s.startsWith(ENTITY_DECLARATION_START)) {
-      parseEntityDecl(s, tokenCallback);
-    } else if (s.startsWith(COMMENT_START)) {
-      parseComment(s, tokenCallback);
-    } else if (s.startsWith(PI_START)) {
-      parsePi(s, tokenCallback);
-    } else if (s.startsWith("]")) {
-      s.advance(1);
-      s.skipSpaces();
-      if (s.currCodeUnit() === GREATER_THAN) {
-        s.advance(1);
-        break;
-      } else {
-        throw new XmlError(
-          { type: "InvalidChar", expected: "'>'", actual: s.currCodeUnitUnchecked() },
-          s.genTextPos()
-        );
-      }
-    } else if (s.startsWith(ELEMENT_DECLARATION_START) || s.startsWith(ATTLIST_DECLARATION_START) || s.startsWith(NOTATION_DECLARATION_START)) {
-      try {
-        consumeDecl(s);
-      } catch (e) {
-        throw new XmlError(
-          { type: "UnknownToken", message: "Failed to consume declaration" },
-          s.genTextPosFrom(start)
-        );
-      }
-    } else {
-      throw new XmlError(
-        { type: "UnknownToken", message: "Failed to parse doctype" },
-        s.genTextPos()
-      );
-    }
-  }
-}
-function parseDoctypeStart(s) {
-  s.advance(9);
-  s.consumeSpaces();
-  s.skipName();
-  s.skipSpaces();
-  parseExternalId(s);
-  s.skipSpaces();
-  const currCodeUnit = s.currCodeUnit();
-  if (currCodeUnit !== OPEN_SQUARE_BRACKET && currCodeUnit !== GREATER_THAN) {
-    throw new XmlError(
-      { type: "InvalidChar", expected: "'[' or '>'", actual: currCodeUnit },
-      s.genTextPos()
-    );
-  }
-}
-function parseExternalId(s) {
-  if (s.startsWith(SYSTEM) || s.startsWith(PUBLIC)) {
-    const start = s.getPos();
-    s.advance(6);
-    const id = s.sliceBack(start);
-    s.consumeSpaces();
-    const quote = s.consumeQuote();
-    s.consumeCodeUnitsWhile((c) => c !== quote);
-    s.consumeCodeUnit(quote);
-    if (id === PUBLIC) {
-      s.consumeSpaces();
-      const _quote = s.consumeQuote();
-      s.consumeCodeUnitsWhile((c) => c !== _quote);
-      s.consumeCodeUnit(_quote);
-    }
-    return true;
-  }
-  return false;
-}
-function parseEntityDecl(s, tokenCallback) {
-  s.advance(8);
-  s.consumeSpaces();
-  const isGe = !s.tryConsumeCodeUnit(PERCENT);
-  if (!isGe) {
-    s.consumeSpaces();
-  }
-  const name = s.consumeName();
-  s.consumeSpaces();
-  const definition = parseEntityDef(s, isGe);
-  if (definition !== null) {
-    tokenCallback({ type: "EntityDeclaration", name, definition }, s);
-  }
-  s.skipSpaces();
-  s.consumeCodeUnit(GREATER_THAN);
-}
-function parseEntityDef(s, isGe) {
-  const currCodeUnit = s.currCodeUnit();
-  if (currCodeUnit === DOUBLE_QUOTE || currCodeUnit === SINGLE_QUOTE) {
-    const quote = s.consumeQuote();
-    const start = s.getPos();
-    s.skipCodeUnitsWhile((c) => c !== quote);
-    const value = s.sliceBack(start);
-    s.consumeCodeUnit(quote);
-    return value;
-  } else if (currCodeUnit === UPPERCASE_S || currCodeUnit === UPPERCASE_P) {
-    if (parseExternalId(s)) {
-      if (isGe) {
-        s.skipSpaces();
-        if (s.startsWith(NDATA)) {
-          s.advance(5);
-          s.consumeSpaces();
-          s.skipName();
-        }
-      }
-      return null;
-    }
-    throw new XmlError({ type: "InvalidExternalID" }, s.genTextPos());
-  } else {
-    throw new XmlError(
-      { type: "InvalidChar", expected: "a quote, SYSTEM or PUBLIC", actual: currCodeUnit },
-      s.genTextPos()
-    );
-  }
-}
-function consumeDecl(s) {
-  s.skipCodeUnitsWhile((c) => c !== GREATER_THAN);
-  s.consumeCodeUnit(GREATER_THAN);
-}
-function parseElement(s, tokenCallback) {
-  const start = s.getPos();
-  s.advance(1);
-  const [prefix, local] = s.consumeQName();
-  tokenCallback({ type: "ElementStart", prefix, local, start }, s);
-  let open = false;
-  while (!s.atEnd()) {
-    const hasSpace = s.startsWithSpace();
-    s.skipSpaces();
-    const _start = s.getPos();
-    const currCodeUnit = s.currCodeUnit();
-    if (currCodeUnit === SLASH) {
-      s.advance(1);
-      s.consumeCodeUnit(GREATER_THAN);
-      const range = s.rangeFrom(_start);
-      tokenCallback({ type: "ElementEnd", end: { type: "Empty" }, range }, s);
-      break;
-    } else if (currCodeUnit === GREATER_THAN) {
-      s.advance(1);
-      const range = s.rangeFrom(_start);
-      if (s.config.implicitSelfClosingElements != null && s.config.implicitSelfClosingElements.includes(getQName(local, prefix))) {
-        tokenCallback({ type: "ElementEnd", end: { type: "Empty" }, range }, s);
-        break;
-      } else {
-        tokenCallback({ type: "ElementEnd", end: { type: "Open" }, range }, s);
-        open = true;
-        break;
-      }
-    } else {
-      if (!hasSpace) {
-        throw new XmlError(
-          { type: "InvalidChar", expected: "a whitespace", actual: s.currCodeUnitUnchecked() },
-          s.genTextPos()
-        );
-      }
-      const [_prefix, _local, value] = parseAttribute(s);
-      const end = s.getPos();
-      tokenCallback(
-        {
-          type: "Attribute",
-          range: { start: _start, end },
-          prefix: _prefix,
-          local: _local,
-          value
-        },
-        s
-      );
-    }
-  }
-  if (open) {
-    if (s.config.rawTextElements != null) {
-      const qName = getQName(local, prefix);
-      if (s.config.rawTextElements.includes(qName)) {
-        parseContent(s, tokenCallback, `</${qName}>`);
-        return;
-      }
-    }
-    parseContent(s, tokenCallback);
-  }
-}
-function parseAttribute(s) {
-  const [prefix, local] = s.consumeQName();
-  let value;
-  const start = s.getPos();
-  s.skipSpaces();
-  if (s.tryConsumeCodeUnit(EQUALS)) {
-    s.skipSpaces();
-    const currCodeUnit = s.currCodeUnit();
-    if (currCodeUnit === SINGLE_QUOTE || currCodeUnit === DOUBLE_QUOTE) {
-      const quote = s.consumeQuote();
-      value = s.consumeCodeUnitsWhile((c) => c !== quote && c !== LESS_THAN);
-      s.consumeCodeUnit(quote);
-    } else if (!s.config.strictDocument) {
-      value = s.consumeCodeUnitsWhile(
-        (c) => !isXmlSpaceByte(c) && c !== GREATER_THAN && c !== SLASH && c !== LESS_THAN
-      );
-    } else {
-      throw new XmlError(
-        { type: "InvalidChar", expected: "a quote", actual: currCodeUnit },
-        s.genTextPos()
-      );
-    }
-  } else if (!s.config.strictDocument) {
-    s.goTo(start);
-    value = "true";
-  } else {
-    throw new XmlError(
-      { type: "InvalidChar", expected: EQUALS, actual: s.currCodeUnit() },
-      s.genTextPos()
-    );
-  }
-  return [prefix, local, value];
-}
-function parseContent(s, tokenCallback, parseContentUntil = LESS_THAN) {
-  while (!s.atEnd()) {
-    if (typeof parseContentUntil === "number" ? s.currCodeUnit() === parseContentUntil : s.startsWith(parseContentUntil)) {
-      const nextCodeUnit = s.nextCodeUnit();
-      if (nextCodeUnit === EXCLAMATION_MARK) {
-        if (s.startsWith(COMMENT_START)) {
-          parseComment(s, tokenCallback);
-        } else if (s.startsWith(CDATA_START)) {
-          parseCdata(s, tokenCallback);
-        } else {
-          throw new XmlError(
-            { type: "UnknownToken", message: "Failed to parse content" },
-            s.genTextPos()
-          );
-        }
-      } else if (nextCodeUnit === QUESTION_MARK) {
-        parsePi(s, tokenCallback);
-      } else if (nextCodeUnit === SLASH) {
-        parseCloseElement(s, tokenCallback);
-        break;
-      } else {
-        parseElement(s, tokenCallback);
-      }
-    } else {
-      parseText(s, tokenCallback, parseContentUntil);
-    }
-  }
-}
-function parseCdata(s, tokenCallback) {
-  const start = s.getPos();
-  s.advance(9);
-  const text = s.consumeCodeUnitsWhile(
-    (c, _s) => !(c === CLOSE_SQUARE_BRACKET && _s.startsWith(CDATA_END))
-  );
-  s.skipString(CDATA_END);
-  const range = s.rangeFrom(start);
-  tokenCallback({ type: "Cdata", text, range }, s);
-}
-function parseCloseElement(s, tokenCallback) {
-  const start = s.getPos();
-  s.advance(2);
-  const [prefix, local] = s.consumeQName();
-  s.skipSpaces();
-  s.consumeCodeUnit(GREATER_THAN);
-  const range = s.rangeFrom(start);
-  tokenCallback({ type: "ElementEnd", end: { type: "Close", prefix, local }, range }, s);
-}
-function parseText(s, tokenCallback, parseTextUntil = LESS_THAN) {
-  const start = s.getPos();
-  const text = typeof parseTextUntil === "number" ? s.consumeCodeUnitsWhile((c) => c !== parseTextUntil) : s.consumeCodeUnitsWhile((_, _s) => !_s.startsWith(parseTextUntil));
-  if (text.includes(CDATA_END)) {
-    throw new XmlError({ type: "InvalidCharacterData" }, s.genTextPos());
-  }
-  tokenCallback({ type: "Text", text, range: s.rangeFrom(start) }, s);
-}
-
-export { tokenize, tokenizeXmlStream };
-//# sourceMappingURL=tokenize.js.map
+import{getQName as U}from"../get-q-name.js";import{LESS_THAN as l,GREATER_THAN as a,SLASH as f,EQUALS as w,SINGLE_QUOTE as x,DOUBLE_QUOTE as P,HYPHEN as B,QUESTION_MARK as W,OPEN_SQUARE_BRACKET as M,PERCENT as H,EXCLAMATION_MARK as z,UPPERCASE_S as K,UPPERCASE_P as Y,CLOSE_SQUARE_BRACKET as J}from"./ascii-constants.js";import{isXmlSpaceByte as q}from"./utils.js";import{XmlError as c}from"./XmlError.js";import{XmlStream as G}from"./XmlStream.js";const V="\uFEFF",v="<?xml ",N="?>",$="<!ENTITY",j="<!ELEMENT",Z="<!ATTLIST",tt="<!NOTATION",g="<!DOCTYPE",A="<?",I="?>",h="<!--",D="-->",nt="<![CDATA[",C="]]>",et="NDATA",L="PUBLIC",ot="SYSTEM",st="version",it="encoding",rt="standalone";function ct(t,n,e={}){F(new G(t,e),n)}function F(t,n){var e;t.debugLog("tokenizeXmlStream started",JSON.stringify({strictDocument:t.config.strictDocument,textLength:t.getPos()===0?"full":"partial",platform:typeof process!="undefined"?process.arch:"unknown",nodeVersion:typeof process!="undefined"?process.version:"unknown"},null,2));try{if(t.startsWith(V)&&t.advance(1),t.startsWith(v)&&at(t),d(t,n),t.config.strictDocument){if(t.skipSpaces(),t.startsWithIgnoreCase(g)&&(k(t,n),d(t,n)),t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===l&&y(t,n),d(t,n),!t.atEnd())throw new c({type:"UnknownToken",message:"Not at end"},t.genTextPos())}else for(;!t.atEnd();)t.currCodeUnit()===l?y(t,n):Q(t,n),d(t,n);t.debugLog("tokenizeXmlStream completed successfully")}catch(o){throw t.debugLog("tokenizeXmlStream FAILED",{error:o instanceof Error?o.message:String(o),errorType:o instanceof c?o.variant.type:"unknown",position:t.getPos(),context:t.getTextAround(t.getPos(),50),textPos:t.genTextPosFrom(t.getPos()),platform:typeof process!="undefined"?process.arch:"unknown",stackTrace:o instanceof Error?(e=o.stack)==null?void 0:e.split(`
+`).slice(0,10):void 0}),o}}function d(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith(h))E(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(g))k(t,n);else if(t.startsWith(A))S(t,n);else break}function at(t){function n(e){if(e.startsWithSpace())e.skipSpaces();else if(!e.startsWith(N)&&!e.atEnd())throw new c({type:"InvalidChar",expected:"a whitespace",actual:e.currCodeUnitUnchecked()},e.genTextPos())}if(t.advance(5),n(t),!t.startsWith(st))throw new c({type:"InvalidString",expected:"version"},t.genTextPos());m(t),n(t),t.startsWith(it)&&(m(t),n(t)),t.startsWith(rt)&&m(t),t.skipSpaces(),t.skipString(N)}function E(t,n){const e=t.getPos();t.advance(4);const o=t.consumeCodeUnitsWhile((i,r)=>!(i===B&&r.startsWith(D)));if(t.skipString(D),o.includes("--")||o.endsWith("-"))throw new c({type:"InvalidComment"},t.genTextPosFrom(e));n({type:"Comment",text:o,range:t.rangeFrom(e)},t)}function S(t,n){if(t.startsWith(v))throw new c({type:"UnexpectedDeclaration"},t.genTextPos());const e=t.getPos();t.advance(2);const o=t.consumeName();t.skipSpaces();const i=t.consumeCodeUnitsWhile((r,s)=>!(r===W&&s.startsWith(I)));t.skipString(I),n({type:"ProcessingInstruction",target:o,content:i.length===0?void 0:i,range:t.rangeFrom(e)},t)}function k(t,n){if(!t.config.allowDtd)throw new c({type:"DtdDetected"});const e=t.getPos();if(pt(t),t.skipSpaces(),t.currCodeUnit()===a){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith($))ut(t,n);else if(t.startsWith(h))E(t,n);else if(t.startsWith(A))S(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===a){t.advance(1);break}else throw new c({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith(j)||t.startsWith(Z)||t.startsWith(tt))try{dt(t)}catch(o){throw new c({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(e))}else throw new c({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function pt(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),b(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==M&&n!==a)throw new c({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function b(t){if(t.startsWith(ot)||t.startsWith(L)){const n=t.getPos();t.advance(6);const e=t.sliceBack(n);t.consumeSpaces();const o=t.consumeQuote();if(t.consumeCodeUnitsWhile(i=>i!==o),t.consumeCodeUnit(o),e===L){t.consumeSpaces();const i=t.consumeQuote();t.consumeCodeUnitsWhile(r=>r!==i),t.consumeCodeUnit(i)}return!0}return!1}function ut(t,n){t.advance(8),t.consumeSpaces();const e=!t.tryConsumeCodeUnit(H);e||t.consumeSpaces();const o=t.consumeName();t.consumeSpaces();const i=lt(t,e);i!==null&&n({type:"EntityDeclaration",name:o,definition:i},t),t.skipSpaces(),t.consumeCodeUnit(a)}function lt(t,n){const e=t.currCodeUnit();if(e===P||e===x){const o=t.consumeQuote(),i=t.getPos();t.skipCodeUnitsWhile(s=>s!==o);const r=t.sliceBack(i);return t.consumeCodeUnit(o),r}else if(e===K||e===Y){if(b(t))return n&&(t.skipSpaces(),t.startsWith(et)&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new c({type:"InvalidExternalID"},t.genTextPos())}else throw new c({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:e},t.genTextPos())}function dt(t){t.skipCodeUnitsWhile(n=>n!==a),t.consumeCodeUnit(a)}function y(t,n){const e=t.getPos();if(t.config.debug){const s=t.currCodeUnit();t.debugLog("parseElement started",JSON.stringify({position:e,character:String.fromCharCode(s),charCode:s,charHex:"0x"+s.toString(16),context:t.getTextAround(e,30),textPos:t.genTextPosFrom(e),platform:typeof process!="undefined"?process.arch:"unknown"},null,2))}t.advance(1);let o,i;try{[o,i]=t.consumeQName(),t.debugLog("parseElement consumeQName succeeded",JSON.stringify({prefix:o,local:i,position:t.getPos()},null,2)),n({type:"ElementStart",prefix:o,local:i,start:e},t)}catch(s){throw t.debugLog("parseElement consumeQName FAILED",JSON.stringify({error:s instanceof Error?s.message:String(s),errorType:s instanceof c?s.variant.type:"unknown",position:t.getPos(),startPosition:e,context:t.getTextAround(t.getPos(),50),platform:typeof process!="undefined"?process.arch:"unknown",stackTrace:s instanceof Error?s.stack:void 0},null,2)),s}let r=!1;for(;!t.atEnd();){const s=t.startsWithSpace();t.skipSpaces();const p=t.getPos(),T=t.currCodeUnit();if(T===f){t.advance(1),t.consumeCodeUnit(a);const u=t.rangeFrom(p);n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else if(T===a){t.advance(1);const u=t.rangeFrom(p);if(t.config.implicitSelfClosingElements!=null&&t.config.implicitSelfClosingElements.includes(U(i,o))){n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else{n({type:"ElementEnd",end:{type:"Open"},range:u},t),r=!0;break}}else{if(!s)throw new c({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());const[u,_,R]=m(t),X=t.getPos();n({type:"Attribute",range:{start:p,end:X},prefix:u,local:_,value:R},t)}}if(r){if(t.config.rawTextElements!=null){const s=U(i,o);if(t.config.rawTextElements.includes(s)){O(t,n,`</${s}>`);return}}O(t,n)}}function m(t){const[n,e]=t.consumeQName();let o;const i=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(w)){t.skipSpaces();const r=t.currCodeUnit();if(r===x||r===P){const s=t.consumeQuote();o=t.consumeCodeUnitsWhile(p=>p!==s&&p!==l),t.consumeCodeUnit(s)}else if(!t.config.strictDocument)o=t.consumeCodeUnitsWhile(s=>!q(s)&&s!==a&&s!==f&&s!==l);else throw new c({type:"InvalidChar",expected:"a quote",actual:r},t.genTextPos())}else if(!t.config.strictDocument)t.goTo(i),o="true";else throw new c({type:"InvalidChar",expected:w,actual:t.currCodeUnit()},t.genTextPos());return[n,e,o]}function O(t,n,e=l){for(;!t.atEnd();)if(typeof e=="number"?t.currCodeUnit()===e:t.startsWith(e)){const o=t.nextCodeUnit();if(o===z)if(t.startsWith(h))E(t,n);else if(t.startsWith(nt))mt(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(g))k(t,n);else throw new c({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(o===W)S(t,n);else if(o===f){ft(t,n);break}else y(t,n)}else Q(t,n,e)}function mt(t,n){const e=t.getPos();t.advance(9);const o=t.consumeCodeUnitsWhile((r,s)=>!(r===J&&s.startsWith(C)));t.skipString(C);const i=t.rangeFrom(e);n({type:"Cdata",text:o,range:i},t)}function ft(t,n){const e=t.getPos();t.advance(2);const[o,i]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(a);const r=t.rangeFrom(e);n({type:"ElementEnd",end:{type:"Close",prefix:o,local:i},range:r},t)}function Q(t,n,e=l){const o=t.getPos(),i=typeof e=="number"?t.consumeCodeUnitsWhile(r=>r!==e):t.consumeCodeUnitsWhile((r,s)=>!s.startsWith(e));if(t.config.strictDocument&&i.includes(C))throw new c({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:i,range:t.rangeFrom(o)},t)}export{ct as tokenize,F as tokenizeXmlStream};

package/dist/esm/tokenizer/utils.js

@@ -1,41 +1 @@
-import { ZERO, NINE, UPPERCASE_A, UPPERCASE_Z, LOWERCASE_A, LOWERCASE_Z, COLON, UNDERSCORE, SPACE, HORIZONTAL_TAB, LINE_FEED, CARRIAGE_RETURN, HYPHEN, PERIOD } from './ascii-constants.js';
-
-function isAsciiDigit(byte) {
-  return byte >= ZERO && byte <= NINE;
-}
-function isXmlNameStart(codePoint) {
-  if (codePoint == null) {
-    return false;
-  }
-  if (codePoint <= 128) {
-    return codePoint >= UPPERCASE_A && codePoint <= UPPERCASE_Z || codePoint >= LOWERCASE_A && codePoint <= LOWERCASE_Z || codePoint === COLON || codePoint === UNDERSCORE;
-  }
-  return codePoint >= 192 && codePoint <= 214 || codePoint >= 216 && codePoint <= 246 || codePoint >= 248 && codePoint <= 767 || codePoint >= 880 && codePoint <= 893 || codePoint >= 895 && codePoint <= 8191 || codePoint >= 8204 && codePoint <= 8205 || codePoint >= 8304 && codePoint <= 8591 || codePoint >= 11264 && codePoint <= 12271 || codePoint >= 12289 && codePoint <= 55295 || codePoint >= 63744 && codePoint <= 64975 || codePoint >= 65008 && codePoint <= 65533 || codePoint >= 65536 && codePoint <= 983039;
-}
-function isXmlName(codePoint) {
-  if (codePoint == null) {
-    return false;
-  }
-  if (codePoint <= 128) {
-    return isXmlNameByte(codePoint);
-  }
-  return codePoint === 183 || codePoint >= 192 && codePoint <= 214 || codePoint >= 216 && codePoint <= 246 || codePoint >= 248 && codePoint <= 767 || codePoint >= 768 && codePoint <= 879 || codePoint >= 880 && codePoint <= 893 || codePoint >= 895 && codePoint <= 8191 || codePoint >= 8204 && codePoint <= 8205 || codePoint >= 8255 && codePoint <= 8256 || codePoint >= 8304 && codePoint <= 8591 || codePoint >= 11264 && codePoint <= 12271 || codePoint >= 12289 && codePoint <= 55295 || codePoint >= 63744 && codePoint <= 64975 || codePoint >= 65008 && codePoint <= 65533 || codePoint >= 65536 && codePoint <= 983039;
-}
-function isXmlChar(codePoint) {
-  if (codePoint == null) {
-    return false;
-  }
-  if (codePoint < 32) {
-    return isXmlSpaceByte(codePoint);
-  }
-  return codePoint !== 65535 && codePoint !== 65534;
-}
-function isXmlSpaceByte(byte) {
-  return byte === SPACE || byte === HORIZONTAL_TAB || byte === LINE_FEED || byte === CARRIAGE_RETURN;
-}
-function isXmlNameByte(byte) {
-  return byte >= ZERO && byte <= NINE || byte >= UPPERCASE_A && byte <= UPPERCASE_Z || byte >= LOWERCASE_A && byte <= LOWERCASE_Z || byte === COLON || byte === UNDERSCORE || byte === HYPHEN || byte === PERIOD;
-}
-
-export { isAsciiDigit, isXmlChar, isXmlName, isXmlNameByte, isXmlNameStart, isXmlSpaceByte };
-//# sourceMappingURL=utils.js.map
+import{ZERO as E,NINE as t,UPPERCASE_A as i,UPPERCASE_Z as r,LOWERCASE_A as e,LOWERCASE_Z as u,COLON as l,UNDERSCORE as R,SPACE as m,HORIZONTAL_TAB as s,LINE_FEED as a,CARRIAGE_RETURN as o,HYPHEN as C,PERIOD as O}from"./ascii-constants.js";function c(n){return n>=E&&n<=t}function S(n){return n==null?!1:n<=128?n>=i&&n<=r||n>=e&&n<=u||n===l||n===R:n>=192&&n<=214||n>=216&&n<=246||n>=248&&n<=767||n>=880&&n<=893||n>=895&&n<=8191||n>=8204&&n<=8205||n>=8304&&n<=8591||n>=11264&&n<=12271||n>=12289&&n<=55295||n>=63744&&n<=64975||n>=65008&&n<=65533||n>=65536&&n<=983039}function P(n){return n==null?!1:n<=128?N(n):n===183||n>=192&&n<=214||n>=216&&n<=246||n>=248&&n<=767||n>=768&&n<=879||n>=880&&n<=893||n>=895&&n<=8191||n>=8204&&n<=8205||n>=8255&&n<=8256||n>=8304&&n<=8591||n>=11264&&n<=12271||n>=12289&&n<=55295||n>=63744&&n<=64975||n>=65008&&n<=65533||n>=65536&&n<=983039}function _(n){return n==null?!1:n<32?A(n):n!==65535&&n!==65534}function A(n){return n===m||n===s||n===a||n===o}function N(n){return n>=E&&n<=t||n>=i&&n<=r||n>=e&&n<=u||n===l||n===R||n===C||n===O}export{c as isAsciiDigit,_ as isXmlChar,P as isXmlName,N as isXmlNameByte,S as isXmlNameStart,A as isXmlSpaceByte};

package/dist/esm/tokens-to-xml.js

@@ -1,12 +1 @@
-import { tokenToXml } from './token-to-xml.js';
-
-function tokensToXml(tokens) {
-  let xmlString = "";
-  tokens.forEach((token) => {
-    xmlString += tokenToXml(token);
-  });
-  return xmlString;
-}
-
-export { tokensToXml };
-//# sourceMappingURL=tokens-to-xml.js.map
+import{tokenToXml as e}from"./token-to-xml.js";function n(t){let o="";return t.forEach(r=>{o+=e(r)}),o}export{n as tokensToXml};

package/dist/esm/xml-to-object.js

@@ -1,67 +1 @@
-import { tokenize } from './tokenizer/tokenize.js';
-
-function xmlToObject(xmlString, options = {}) {
-  const root = {
-    local: "root",
-    attributes: [],
-    content: []
-  };
-  const stack = [root];
-  tokenize(
-    xmlString,
-    (token) => {
-      processTokenForObject(token, stack);
-    },
-    options
-  );
-  return root.content[0];
-}
-function processTokenForObject(token, stack) {
-  switch (token.type) {
-    case "ElementStart": {
-      const newNode = {
-        local: token.local,
-        prefix: token.prefix.length > 0 ? token.prefix : void 0,
-        attributes: [],
-        content: []
-      };
-      const currentNode = stack[stack.length - 1];
-      if (currentNode != null) {
-        currentNode.content.push(newNode);
-      }
-      stack.push(newNode);
-      break;
-    }
-    case "ElementEnd": {
-      if (token.end.type === "Close" || token.end.type === "Empty") {
-        stack.pop();
-      }
-      break;
-    }
-    case "Attribute": {
-      const currentNode = stack[stack.length - 1];
-      if (currentNode != null) {
-        currentNode.attributes.push({
-          local: token.local,
-          prefix: token.prefix.length > 0 ? token.prefix : void 0,
-          value: token.value
-        });
-      }
-      break;
-    }
-    case "Text":
-    case "Cdata": {
-      const currentNode = stack[stack.length - 1];
-      if (currentNode != null) {
-        const trimmedText = token.text.trim();
-        if (trimmedText.length > 0) {
-          currentNode.content.push(token.text);
-        }
-      }
-      break;
-    }
-  }
-}
-
-export { processTokenForObject, xmlToObject };
-//# sourceMappingURL=xml-to-object.js.map
+import{tokenize as c}from"./tokenizer/tokenize.js";function a(t,o={}){const e={local:"root",attributes:[],content:[]},l=[e];return c(t,r=>{n(r,l)},o),e.content[0]}function n(t,o){switch(t.type){case"ElementStart":{const e={local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,attributes:[],content:[]},l=o[o.length-1];l!=null&&l.content.push(e),o.push(e);break}case"ElementEnd":{(t.end.type==="Close"||t.end.type==="Empty")&&o.pop();break}case"Attribute":{const e=o[o.length-1];e!=null&&e.attributes.push({local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,value:t.value});break}case"Text":case"Cdata":{const e=o[o.length-1];e!=null&&t.text.trim().length>0&&e.content.push(t.text);break}}}export{n as processTokenForObject,a as xmlToObject};

package/dist/esm/xml-to-simplified-object.js

@@ -1,84 +1 @@
-import { getQName } from './get-q-name.js';
-import { tokenize } from './tokenizer/tokenize.js';
-
-function xmlToSimplifiedObject(xmlString, options = {}) {
-  const result = {};
-  const stack = [result];
-  tokenize(
-    xmlString,
-    (token) => {
-      processTokenForSimplifiedObject(token, stack);
-    },
-    options
-  );
-  return result;
-}
-function processTokenForSimplifiedObject(token, stack) {
-  switch (token.type) {
-    case "ElementStart": {
-      const tagName = getQName(token.local, token.prefix);
-      const tagNameWithPrefix = `_${tagName}`;
-      const newNode = {};
-      const currentNode = stack[stack.length - 1];
-      if (currentNode != null) {
-        if (currentNode.content != null) {
-          newNode.tag = tagName;
-          currentNode.content.push(newNode);
-        } else if (currentNode.text != null) {
-          newNode.tag = tagName;
-          currentNode.content = [currentNode.text, newNode];
-          delete currentNode.text;
-        } else if (currentNode[tagNameWithPrefix] == null) {
-          currentNode[tagNameWithPrefix] = [newNode];
-        } else if (Array.isArray(currentNode[tagNameWithPrefix])) {
-          currentNode[tagNameWithPrefix].push(newNode);
-        } else {
-          currentNode[tagNameWithPrefix] = [currentNode[tagNameWithPrefix], newNode];
-        }
-      }
-      stack.push(newNode);
-      break;
-    }
-    case "ElementEnd": {
-      if (token.end.type === "Close" || token.end.type === "Empty") {
-        stack.pop();
-      }
-      break;
-    }
-    case "Attribute": {
-      const currentNode = stack[stack.length - 1];
-      if (currentNode != null) {
-        const attrName = getQName(token.local, token.prefix);
-        if (currentNode.attributes != null) {
-          currentNode.attributes[attrName] = token.value;
-        } else {
-          currentNode.attributes = {
-            [attrName]: token.value
-          };
-        }
-      }
-      break;
-    }
-    case "Text":
-    case "Cdata": {
-      const currentNode = stack[stack.length - 1];
-      if (currentNode != null) {
-        const trimmedText = token.text.trim();
-        if (trimmedText.length > 0) {
-          if (currentNode.content != null) {
-            currentNode.content.push(trimmedText);
-          } else if (currentNode.text != null) {
-            currentNode.content = [currentNode.text, trimmedText];
-            delete currentNode.text;
-          } else {
-            currentNode.text = trimmedText;
-          }
-        }
-      }
-      break;
-    }
-  }
-}
-
-export { processTokenForSimplifiedObject, xmlToSimplifiedObject };
-//# sourceMappingURL=xml-to-simplified-object.js.map
+import{getQName as c}from"./get-q-name.js";import{tokenize as i}from"./tokenizer/tokenize.js";function s(o,l={}){const t={},e=[t];return i(o,r=>{a(r,e)},l),t}function a(o,l){switch(o.type){case"ElementStart":{const t=c(o.local,o.prefix),e=`_${t}`,r={},n=l[l.length-1];n!=null&&(n.content!=null?(r.tag=t,n.content.push(r)):n.text!=null?(r.tag=t,n.content=[n.text,r],delete n.text):n[e]==null?n[e]=[r]:Array.isArray(n[e])?n[e].push(r):n[e]=[n[e],r]),l.push(r);break}case"ElementEnd":{(o.end.type==="Close"||o.end.type==="Empty")&&l.pop();break}case"Attribute":{const t=l[l.length-1];if(t!=null){const e=c(o.local,o.prefix);t.attributes!=null?t.attributes[e]=o.value:t.attributes={[e]:o.value}}break}case"Text":case"Cdata":{const t=l[l.length-1];if(t!=null){const e=o.text.trim();e.length>0&&(t.content!=null?t.content.push(e):t.text!=null?(t.content=[t.text,e],delete t.text):t.text=e)}break}}}export{a as processTokenForSimplifiedObject,s as xmlToSimplifiedObject};

package/dist/types/config.d.ts

@@ -14,4 +14,3 @@ export declare const htmlConfig: TXmlStreamOptions;
  * SVG configuration with SVG-specific settings.
  */
 export declare const svgConfig: TXmlStreamOptions;
-//# sourceMappingURL=config.d.ts.map

package/dist/types/get-q-name.d.ts

@@ -4,4 +4,3 @@
  * https://www.w3.org/2001/tag/doc/qnameids.html
  */
 export declare function getQName(local: string, prefix?: string | null): string;
-//# sourceMappingURL=get-q-name.d.ts.map

package/dist/types/index.d.ts

@@ -1,9 +1,9 @@
 export * from './config';
 export * from './get-q-name';
+export * from './processor';
 export * from './selector';
 export * from './token-to-xml';
 export * from './tokenizer';
 export * from './tokens-to-xml';
 export * from './xml-to-object';
 export * from './xml-to-simplified-object';
-//# sourceMappingURL=index.d.ts.map

package/dist/types/processor/index.d.ts

@@ -0,0 +1,4 @@
+export * from './process';
+export * from './processors';
+export * from './types';
+export * from './validate-processor-order';