xml-tokenizer 0.0.26 → 0.0.27

package/dist/cjs/tokenizer/tokenize.js

@@ -1,408 +1 @@
-'use strict';
-
-var getQName = require('../get-q-name.js');
-var asciiConstants = require('./ascii-constants.js');
-var utils = require('./utils.js');
-var XmlError = require('./XmlError.js');
-var XmlStream = require('./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.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.XmlError({ type: "DtdDetected" });
-    }
-    parseDoctype(s, tokenCallback);
-    parseMisc(s, tokenCallback);
-  }
-  s.skipSpaces();
-  if (s.config.strictDocument) {
-    if (!s.atEnd() && s.currCodeUnit() === asciiConstants.LESS_THAN) {
-      parseElement(s, tokenCallback);
-    }
-    parseMisc(s, tokenCallback);
-    if (!s.atEnd()) {
-      throw new XmlError.XmlError({ type: "UnknownToken", message: "Not at end" }, s.genTextPos());
-    }
-  } else {
-    while (!s.atEnd()) {
-      if (s.currCodeUnit() === asciiConstants.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.XmlError(
-        { type: "InvalidChar", expected: "a whitespace", actual: _s.currCodeUnitUnchecked() },
-        _s.genTextPos()
-      );
-    }
-  }
-  s.advance(5);
-  consumeSpaces(s);
-  if (!s.startsWith(VERSION)) {
-    throw new XmlError.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 === asciiConstants.HYPHEN && _s.startsWith(COMMENT_END)));
-  s.skipString(COMMENT_END);
-  if (text.includes("--") || text.endsWith("-")) {
-    throw new XmlError.XmlError({ type: "InvalidComment" }, s.genTextPosFrom(start));
-  }
-  tokenCallback({ type: "Comment", text, range: s.rangeFrom(start) });
-}
-function parsePi(s, tokenCallback) {
-  if (s.startsWith(XML_DECLARATION_START)) {
-    throw new XmlError.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 === asciiConstants.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)
-  });
-}
-function parseDoctype(s, tokenCallback) {
-  const start = s.getPos();
-  parseDoctypeStart(s);
-  s.skipSpaces();
-  if (s.currCodeUnit() === asciiConstants.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() === asciiConstants.GREATER_THAN) {
-        s.advance(1);
-        break;
-      } else {
-        throw new XmlError.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.XmlError(
-          { type: "UnknownToken", message: "Failed to consume declaration" },
-          s.genTextPosFrom(start)
-        );
-      }
-    } else {
-      throw new XmlError.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 !== asciiConstants.OPEN_SQUARE_BRACKET && currCodeUnit !== asciiConstants.GREATER_THAN) {
-    throw new XmlError.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(asciiConstants.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.skipSpaces();
-  s.consumeCodeUnit(asciiConstants.GREATER_THAN);
-}
-function parseEntityDef(s, isGe) {
-  const currCodeUnit = s.currCodeUnit();
-  if (currCodeUnit === asciiConstants.DOUBLE_QUOTE || currCodeUnit === asciiConstants.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 === asciiConstants.UPPERCASE_S || currCodeUnit === asciiConstants.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.XmlError({ type: "InvalidExternalID" }, s.genTextPos());
-  } else {
-    throw new XmlError.XmlError(
-      { type: "InvalidChar", expected: "a quote, SYSTEM or PUBLIC", actual: currCodeUnit },
-      s.genTextPos()
-    );
-  }
-}
-function consumeDecl(s) {
-  s.skipCodeUnitsWhile((c) => c !== asciiConstants.GREATER_THAN);
-  s.consumeCodeUnit(asciiConstants.GREATER_THAN);
-}
-function parseElement(s, tokenCallback) {
-  const start = s.getPos();
-  s.advance(1);
-  const [prefix, local] = s.consumeQName();
-  tokenCallback({ type: "ElementStart", prefix, local, start });
-  let open = false;
-  while (!s.atEnd()) {
-    const hasSpace = s.startsWithSpace();
-    s.skipSpaces();
-    const _start = s.getPos();
-    const currCodeUnit = s.currCodeUnit();
-    if (currCodeUnit === asciiConstants.SLASH) {
-      s.advance(1);
-      s.consumeCodeUnit(asciiConstants.GREATER_THAN);
-      const range = s.rangeFrom(_start);
-      tokenCallback({ type: "ElementEnd", end: { type: "Empty" }, range });
-      break;
-    } else if (currCodeUnit === asciiConstants.GREATER_THAN) {
-      s.advance(1);
-      const range = s.rangeFrom(_start);
-      tokenCallback({ type: "ElementEnd", end: { type: "Open" }, range });
-      open = true;
-      break;
-    } else {
-      if (!hasSpace) {
-        throw new XmlError.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
-      });
-    }
-  }
-  if (open) {
-    if (s.config.rawTextElements != null) {
-      const qName = getQName.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(asciiConstants.EQUALS)) {
-    s.skipSpaces();
-    const currCodeUnit = s.currCodeUnit();
-    if (currCodeUnit === asciiConstants.SINGLE_QUOTE || currCodeUnit === asciiConstants.DOUBLE_QUOTE) {
-      const quote = s.consumeQuote();
-      value = s.consumeCodeUnitsWhile((c) => c !== quote && c !== asciiConstants.LESS_THAN);
-      s.consumeCodeUnit(quote);
-    } else if (!s.config.strictDocument) {
-      value = s.consumeCodeUnitsWhile(
-        (c) => !utils.isXmlSpaceByte(c) && c !== asciiConstants.GREATER_THAN && c !== asciiConstants.SLASH && c !== asciiConstants.LESS_THAN
-      );
-    } else {
-      throw new XmlError.XmlError(
-        { type: "InvalidChar", expected: "a quote", actual: currCodeUnit },
-        s.genTextPos()
-      );
-    }
-  } else if (!s.config.strictDocument) {
-    s.goTo(start);
-    value = "true";
-  } else {
-    throw new XmlError.XmlError(
-      { type: "InvalidChar", expected: asciiConstants.EQUALS, actual: s.currCodeUnit() },
-      s.genTextPos()
-    );
-  }
-  return [prefix, local, value];
-}
-function parseContent(s, tokenCallback, parseContentUntil = asciiConstants.LESS_THAN) {
-  while (!s.atEnd()) {
-    if (typeof parseContentUntil === "number" ? s.currCodeUnit() === parseContentUntil : s.startsWith(parseContentUntil)) {
-      const nextCodeUnit = s.nextCodeUnit();
-      if (nextCodeUnit === asciiConstants.EXCLAMATION_MARK) {
-        if (s.startsWith(COMMENT_START)) {
-          parseComment(s, tokenCallback);
-        } else if (s.startsWith(CDATA_START)) {
-          parseCdata(s, tokenCallback);
-        } else {
-          throw new XmlError.XmlError(
-            { type: "UnknownToken", message: "Failed to parse content" },
-            s.genTextPos()
-          );
-        }
-      } else if (nextCodeUnit === asciiConstants.QUESTION_MARK) {
-        parsePi(s, tokenCallback);
-      } else if (nextCodeUnit === asciiConstants.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 === asciiConstants.CLOSE_SQUARE_BRACKET && _s.startsWith(CDATA_END))
-  );
-  s.skipString(CDATA_END);
-  const range = s.rangeFrom(start);
-  tokenCallback({ type: "Cdata", text, range });
-}
-function parseCloseElement(s, tokenCallback) {
-  const start = s.getPos();
-  s.advance(2);
-  const [prefix, local] = s.consumeQName();
-  s.skipSpaces();
-  s.consumeCodeUnit(asciiConstants.GREATER_THAN);
-  const range = s.rangeFrom(start);
-  tokenCallback({ type: "ElementEnd", end: { type: "Close", prefix, local }, range });
-}
-function parseText(s, tokenCallback, parseTextUntil = asciiConstants.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.XmlError({ type: "InvalidCharacterData" }, s.genTextPos());
-  }
-  tokenCallback({ type: "Text", text, range: s.rangeFrom(start) });
-}
-
-exports.tokenize = tokenize;
-exports.tokenizeXmlStream = tokenizeXmlStream;
+"use strict";var X=require("../get-q-name.js"),e=require("./ascii-constants.js"),I=require("./utils.js"),c=require("./XmlError.js"),L=require("./XmlStream.js");const H="\uFEFF",g="<?xml ",C="?>",Q="<!ENTITY",D="<!ELEMENT",O="<!ATTLIST",F="<!NOTATION",U="<!DOCTYPE",A="<?",k="?>",m="<!--",w="-->",G="<![CDATA[",E="]]>",B="NDATA",W="PUBLIC",b="SYSTEM",q="version",M="encoding",z="standalone";function K(t,n,s={}){P(new L.XmlStream(t,s),n)}function P(t,n){if(t.startsWith(H)&&t.advance(1),t.startsWith(g)&&Y(t),d(t,n),t.skipSpaces(),t.startsWith(U)||t.startsWith(U.toLowerCase())){if(!t.config.allowDtd)throw new c.XmlError({type:"DtdDetected"});$(t,n),d(t,n)}if(t.skipSpaces(),t.config.strictDocument){if(!t.atEnd()&&t.currCodeUnit()===e.LESS_THAN&&S(t,n),d(t,n),!t.atEnd())throw new c.XmlError({type:"UnknownToken",message:"Not at end"},t.genTextPos())}else for(;!t.atEnd();)t.currCodeUnit()===e.LESS_THAN?S(t,n):x(t,n),d(t,n)}function d(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith(m))f(t,n);else if(t.startsWith(A))h(t,n);else break}function Y(t){function n(s){if(s.startsWithSpace())s.skipSpaces();else if(!s.startsWith(C)&&!s.atEnd())throw new c.XmlError({type:"InvalidChar",expected:"a whitespace",actual:s.currCodeUnitUnchecked()},s.genTextPos())}if(t.advance(5),n(t),!t.startsWith(q))throw new c.XmlError({type:"InvalidString",expected:"version"},t.genTextPos());p(t),n(t),t.startsWith(M)&&(p(t),n(t)),t.startsWith(z)&&p(t),t.skipSpaces(),t.skipString(C)}function f(t,n){const s=t.getPos();t.advance(4);const o=t.consumeCodeUnitsWhile((r,i)=>!(r===e.HYPHEN&&i.startsWith(w)));if(t.skipString(w),o.includes("--")||o.endsWith("-"))throw new c.XmlError({type:"InvalidComment"},t.genTextPosFrom(s));n({type:"Comment",text:o,range:t.rangeFrom(s)},t)}function h(t,n){if(t.startsWith(g))throw new c.XmlError({type:"UnexpectedDeclaration"},t.genTextPos());const s=t.getPos();t.advance(2);const o=t.consumeName();t.skipSpaces();const r=t.consumeCodeUnitsWhile((i,a)=>!(i===e.QUESTION_MARK&&a.startsWith(k)));t.skipString(k),n({type:"ProcessingInstruction",target:o,content:r.length===0?void 0:r,range:t.rangeFrom(s)},t)}function $(t,n){const s=t.getPos();if(j(t),t.skipSpaces(),t.currCodeUnit()===e.GREATER_THAN){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(Q))J(t,n);else if(t.startsWith(m))f(t,n);else if(t.startsWith(A))h(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===e.GREATER_THAN){t.advance(1);break}else throw new c.XmlError({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith(D)||t.startsWith(O)||t.startsWith(F))try{Z(t)}catch(o){throw new c.XmlError({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(s))}else throw new c.XmlError({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function j(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),y(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==e.OPEN_SQUARE_BRACKET&&n!==e.GREATER_THAN)throw new c.XmlError({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function y(t){if(t.startsWith(b)||t.startsWith(W)){const n=t.getPos();t.advance(6);const s=t.sliceBack(n);t.consumeSpaces();const o=t.consumeQuote();if(t.consumeCodeUnitsWhile(r=>r!==o),t.consumeCodeUnit(o),s===W){t.consumeSpaces();const r=t.consumeQuote();t.consumeCodeUnitsWhile(i=>i!==r),t.consumeCodeUnit(r)}return!0}return!1}function J(t,n){t.advance(8),t.consumeSpaces();const s=!t.tryConsumeCodeUnit(e.PERCENT);s||t.consumeSpaces();const o=t.consumeName();t.consumeSpaces();const r=V(t,s);r!==null&&n({type:"EntityDeclaration",name:o,definition:r},t),t.skipSpaces(),t.consumeCodeUnit(e.GREATER_THAN)}function V(t,n){const s=t.currCodeUnit();if(s===e.DOUBLE_QUOTE||s===e.SINGLE_QUOTE){const o=t.consumeQuote(),r=t.getPos();t.skipCodeUnitsWhile(a=>a!==o);const i=t.sliceBack(r);return t.consumeCodeUnit(o),i}else if(s===e.UPPERCASE_S||s===e.UPPERCASE_P){if(y(t))return n&&(t.skipSpaces(),t.startsWith(B)&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new c.XmlError({type:"InvalidExternalID"},t.genTextPos())}else throw new c.XmlError({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:s},t.genTextPos())}function Z(t){t.skipCodeUnitsWhile(n=>n!==e.GREATER_THAN),t.consumeCodeUnit(e.GREATER_THAN)}function S(t,n){const s=t.getPos();t.advance(1);const[o,r]=t.consumeQName();n({type:"ElementStart",prefix:o,local:r,start:s},t);let i=!1;for(;!t.atEnd();){const a=t.startsWithSpace();t.skipSpaces();const l=t.getPos(),T=t.currCodeUnit();if(T===e.SLASH){t.advance(1),t.consumeCodeUnit(e.GREATER_THAN);const u=t.rangeFrom(l);n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else if(T===e.GREATER_THAN){t.advance(1);const u=t.rangeFrom(l);n({type:"ElementEnd",end:{type:"Open"},range:u},t),i=!0;break}else{if(!a)throw new c.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());const[u,v,R]=p(t),_=t.getPos();n({type:"Attribute",range:{start:l,end:_},prefix:u,local:v,value:R},t)}}if(i){if(t.config.rawTextElements!=null){const a=X.getQName(r,o);if(t.config.rawTextElements.includes(a)){N(t,n,`</${a}>`);return}}N(t,n)}}function p(t){const[n,s]=t.consumeQName();let o;const r=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(e.EQUALS)){t.skipSpaces();const i=t.currCodeUnit();if(i===e.SINGLE_QUOTE||i===e.DOUBLE_QUOTE){const a=t.consumeQuote();o=t.consumeCodeUnitsWhile(l=>l!==a&&l!==e.LESS_THAN),t.consumeCodeUnit(a)}else if(!t.config.strictDocument)o=t.consumeCodeUnitsWhile(a=>!I.isXmlSpaceByte(a)&&a!==e.GREATER_THAN&&a!==e.SLASH&&a!==e.LESS_THAN);else throw new c.XmlError({type:"InvalidChar",expected:"a quote",actual:i},t.genTextPos())}else if(!t.config.strictDocument)t.goTo(r),o="true";else throw new c.XmlError({type:"InvalidChar",expected:e.EQUALS,actual:t.currCodeUnit()},t.genTextPos());return[n,s,o]}function N(t,n,s=e.LESS_THAN){for(;!t.atEnd();)if(typeof s=="number"?t.currCodeUnit()===s:t.startsWith(s)){const o=t.nextCodeUnit();if(o===e.EXCLAMATION_MARK)if(t.startsWith(m))f(t,n);else if(t.startsWith(G))tt(t,n);else throw new c.XmlError({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(o===e.QUESTION_MARK)h(t,n);else if(o===e.SLASH){nt(t,n);break}else S(t,n)}else x(t,n,s)}function tt(t,n){const s=t.getPos();t.advance(9);const o=t.consumeCodeUnitsWhile((i,a)=>!(i===e.CLOSE_SQUARE_BRACKET&&a.startsWith(E)));t.skipString(E);const r=t.rangeFrom(s);n({type:"Cdata",text:o,range:r},t)}function nt(t,n){const s=t.getPos();t.advance(2);const[o,r]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(e.GREATER_THAN);const i=t.rangeFrom(s);n({type:"ElementEnd",end:{type:"Close",prefix:o,local:r},range:i},t)}function x(t,n,s=e.LESS_THAN){const o=t.getPos(),r=typeof s=="number"?t.consumeCodeUnitsWhile(i=>i!==s):t.consumeCodeUnitsWhile((i,a)=>!a.startsWith(s));if(r.includes(E))throw new c.XmlError({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:r,range:t.rangeFrom(o)},t)}exports.tokenize=K,exports.tokenizeXmlStream=P;

package/dist/cjs/tokenizer/utils.js

@@ -1,47 +1 @@
-'use strict';
-
-var asciiConstants = require('./ascii-constants.js');
-
-function isAsciiDigit(byte) {
-  return byte >= asciiConstants.ZERO && byte <= asciiConstants.NINE;
-}
-function isXmlNameStart(codePoint) {
-  if (codePoint == null) {
-    return false;
-  }
-  if (codePoint <= 128) {
-    return codePoint >= asciiConstants.UPPERCASE_A && codePoint <= asciiConstants.UPPERCASE_Z || codePoint >= asciiConstants.LOWERCASE_A && codePoint <= asciiConstants.LOWERCASE_Z || codePoint === asciiConstants.COLON || codePoint === asciiConstants.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 === asciiConstants.SPACE || byte === asciiConstants.HORIZONTAL_TAB || byte === asciiConstants.LINE_FEED || byte === asciiConstants.CARRIAGE_RETURN;
-}
-function isXmlNameByte(byte) {
-  return byte >= asciiConstants.ZERO && byte <= asciiConstants.NINE || byte >= asciiConstants.UPPERCASE_A && byte <= asciiConstants.UPPERCASE_Z || byte >= asciiConstants.LOWERCASE_A && byte <= asciiConstants.LOWERCASE_Z || byte === asciiConstants.COLON || byte === asciiConstants.UNDERSCORE || byte === asciiConstants.HYPHEN || byte === asciiConstants.PERIOD;
-}
-
-exports.isAsciiDigit = isAsciiDigit;
-exports.isXmlChar = isXmlChar;
-exports.isXmlName = isXmlName;
-exports.isXmlNameByte = isXmlNameByte;
-exports.isXmlNameStart = isXmlNameStart;
-exports.isXmlSpaceByte = isXmlSpaceByte;
+"use strict";var i=require("./ascii-constants.js");function n(E){return E>=i.ZERO&&E<=i.NINE}function s(E){return E==null?!1:E<=128?E>=i.UPPERCASE_A&&E<=i.UPPERCASE_Z||E>=i.LOWERCASE_A&&E<=i.LOWERCASE_Z||E===i.COLON||E===i.UNDERSCORE:E>=192&&E<=214||E>=216&&E<=246||E>=248&&E<=767||E>=880&&E<=893||E>=895&&E<=8191||E>=8204&&E<=8205||E>=8304&&E<=8591||E>=11264&&E<=12271||E>=12289&&E<=55295||E>=63744&&E<=64975||E>=65008&&E<=65533||E>=65536&&E<=983039}function r(E){return E==null?!1:E<=128?e(E):E===183||E>=192&&E<=214||E>=216&&E<=246||E>=248&&E<=767||E>=768&&E<=879||E>=880&&E<=893||E>=895&&E<=8191||E>=8204&&E<=8205||E>=8255&&E<=8256||E>=8304&&E<=8591||E>=11264&&E<=12271||E>=12289&&E<=55295||E>=63744&&E<=64975||E>=65008&&E<=65533||E>=65536&&E<=983039}function R(E){return E==null?!1:E<32?t(E):E!==65535&&E!==65534}function t(E){return E===i.SPACE||E===i.HORIZONTAL_TAB||E===i.LINE_FEED||E===i.CARRIAGE_RETURN}function e(E){return E>=i.ZERO&&E<=i.NINE||E>=i.UPPERCASE_A&&E<=i.UPPERCASE_Z||E>=i.LOWERCASE_A&&E<=i.LOWERCASE_Z||E===i.COLON||E===i.UNDERSCORE||E===i.HYPHEN||E===i.PERIOD}exports.isAsciiDigit=n,exports.isXmlChar=R,exports.isXmlName=r,exports.isXmlNameByte=e,exports.isXmlNameStart=s,exports.isXmlSpaceByte=t;

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

@@ -1,13 +1 @@
-'use strict';
-
-var tokenToXml = require('./token-to-xml.js');
-
-function tokensToXml(tokens) {
-  let xmlString = "";
-  tokens.forEach((token) => {
-    xmlString += tokenToXml.tokenToXml(token);
-  });
-  return xmlString;
-}
-
-exports.tokensToXml = tokensToXml;
+"use strict";var n=require("./token-to-xml.js");function r(e){let o="";return e.forEach(t=>{o+=n.tokenToXml(t)}),o}exports.tokensToXml=r;

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

@@ -1,69 +1 @@
-'use strict';
-
-var tokenize = require('./tokenizer/tokenize.js');
-
-function xmlToObject(xmlString, options = {}) {
-  const root = {
-    local: "root",
-    attributes: [],
-    content: []
-  };
-  const stack = [root];
-  tokenize.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;
-    }
-  }
-}
-
-exports.processTokenForObject = processTokenForObject;
-exports.xmlToObject = xmlToObject;
+"use strict";var r=require("./tokenizer/tokenize.js");function s(t,o={}){const e={local:"root",attributes:[],content:[]},n=[e];return r.tokenize(t,l=>{c(l,n)},o),e.content[0]}function c(t,o){switch(t.type){case"ElementStart":{const e={local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,attributes:[],content:[]},n=o[o.length-1];n!=null&&n.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}}}exports.processTokenForObject=c,exports.xmlToObject=s;

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

@@ -1,86 +1 @@
-'use strict';
-
-var getQName = require('./get-q-name.js');
-var tokenize = require('./tokenizer/tokenize.js');
-
-function xmlToSimplifiedObject(xmlString, options = {}) {
-  const result = {};
-  const stack = [result];
-  tokenize.tokenize(
-    xmlString,
-    (token) => {
-      processTokenForSimplifiedObject(token, stack);
-    },
-    options
-  );
-  return result;
-}
-function processTokenForSimplifiedObject(token, stack) {
-  switch (token.type) {
-    case "ElementStart": {
-      const tagName = getQName.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.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;
-    }
-  }
-}
-
-exports.processTokenForSimplifiedObject = processTokenForSimplifiedObject;
-exports.xmlToSimplifiedObject = xmlToSimplifiedObject;
+"use strict";var c=require("./get-q-name.js"),o=require("./tokenizer/tokenize.js");function s(i,l={}){const e={},t=[e];return o.tokenize(i,r=>{a(r,t)},l),e}function a(i,l){switch(i.type){case"ElementStart":{const e=c.getQName(i.local,i.prefix),t=`_${e}`,r={},n=l[l.length-1];n!=null&&(n.content!=null?(r.tag=e,n.content.push(r)):n.text!=null?(r.tag=e,n.content=[n.text,r],delete n.text):n[t]==null?n[t]=[r]:Array.isArray(n[t])?n[t].push(r):n[t]=[n[t],r]),l.push(r);break}case"ElementEnd":{(i.end.type==="Close"||i.end.type==="Empty")&&l.pop();break}case"Attribute":{const e=l[l.length-1];if(e!=null){const t=c.getQName(i.local,i.prefix);e.attributes!=null?e.attributes[t]=i.value:e.attributes={[t]:i.value}}break}case"Text":case"Cdata":{const e=l[l.length-1];if(e!=null){const t=i.text.trim();t.length>0&&(e.content!=null?e.content.push(t):e.text!=null?(e.content=[e.text,t],delete e.text):e.text=t)}break}}}exports.processTokenForSimplifiedObject=a,exports.xmlToSimplifiedObject=s;

package/dist/esm/get-q-name.js

@@ -1,5 +1 @@
-function getQName(local, prefix = null) {
-  return prefix != null && prefix.length > 0 ? `${prefix}:${local}` : local;
-}
-
-export { getQName };
+function n(l,e=null){return e!=null&&e.length>0?`${e}:${l}`:l}export{n as getQName};

package/dist/esm/index.js

@@ -1,14 +1 @@
-export { getQName } from './get-q-name.js';
-export { select } from './selector/select.js';
-export { TokenSelector } from './selector/TokenSelector.js';
-export { EToCacheNodeProps, ETokenMatchCriteria, TokenSelectState } from './selector/TokenSelectState.js';
-export { TokenSelectStateMachine } from './selector/TokenSelectStateMachine.js';
-export { tokenToXml } from './token-to-xml.js';
-export { AMPERSAND, CARRIAGE_RETURN, CLOSE_SQUARE_BRACKET, COLON, DOUBLE_QUOTE, EQUALS, EXCLAMATION_MARK, GREATER_THAN, HASH, HORIZONTAL_TAB, HYPHEN, LESS_THAN, LINE_FEED, LOWERCASE_A, LOWERCASE_F, LOWERCASE_X, LOWERCASE_Z, NINE, OPEN_SQUARE_BRACKET, PERCENT, PERIOD, QUESTION_MARK, SEMICOLON, SINGLE_QUOTE, SLASH, SPACE, UNDERSCORE, UPPERCASE_A, UPPERCASE_F, UPPERCASE_P, UPPERCASE_S, UPPERCASE_Z, ZERO } from './tokenizer/ascii-constants.js';
-export { tokenize, tokenizeXmlStream } from './tokenizer/tokenize.js';
-export { isAsciiDigit, isXmlChar, isXmlName, isXmlNameByte, isXmlNameStart, isXmlSpaceByte } from './tokenizer/utils.js';
-export { XmlError } from './tokenizer/XmlError.js';
-export { XmlStream } from './tokenizer/XmlStream.js';
-export { tokensToXml } from './tokens-to-xml.js';
-export { processTokenForObject, xmlToObject } from './xml-to-object.js';
-export { processTokenForSimplifiedObject, xmlToSimplifiedObject } from './xml-to-simplified-object.js';
+import{getQName as o}from"./get-q-name.js";import{select as t}from"./selector/select.js";import{TokenSelector as m}from"./selector/TokenSelector.js";import{EToCacheNodeProps as R,ETokenMatchCriteria as O,TokenSelectState as T}from"./selector/TokenSelectState.js";import{TokenSelectStateMachine as N}from"./selector/TokenSelectStateMachine.js";import{tokenToXml as i}from"./token-to-xml.js";import{AMPERSAND as l,CARRIAGE_RETURN as P,CLOSE_SQUARE_BRACKET as f,COLON as x,DOUBLE_QUOTE as a,EQUALS as c,EXCLAMATION_MARK as L,GREATER_THAN as s,HASH as U,HORIZONTAL_TAB as X,HYPHEN as n,LESS_THAN as k,LINE_FEED as I,LOWERCASE_A as H,LOWERCASE_F as M,LOWERCASE_X as Q,LOWERCASE_Z as B,NINE as D,OPEN_SQUARE_BRACKET as F,PERCENT as b,PERIOD as h,QUESTION_MARK as j,SEMICOLON as K,SINGLE_QUOTE as W,SLASH as Z,SPACE as d,UNDERSCORE as G,UPPERCASE_A as g,UPPERCASE_F as y,UPPERCASE_P as z,UPPERCASE_S as Y,UPPERCASE_Z as q,ZERO as u}from"./tokenizer/ascii-constants.js";import{tokenize as w,tokenizeXmlStream as J}from"./tokenizer/tokenize.js";import{isAsciiDigit as $,isXmlChar as ee,isXmlName as Ee,isXmlNameByte as oe,isXmlNameStart as re,isXmlSpaceByte as te}from"./tokenizer/utils.js";import{XmlError as me}from"./tokenizer/XmlError.js";import{XmlStream as Re}from"./tokenizer/XmlStream.js";import{tokensToXml as Te}from"./tokens-to-xml.js";import{processTokenForObject as Ne,xmlToObject as _e}from"./xml-to-object.js";import{processTokenForSimplifiedObject as pe,xmlToSimplifiedObject as le}from"./xml-to-simplified-object.js";export{l as AMPERSAND,P as CARRIAGE_RETURN,f as CLOSE_SQUARE_BRACKET,x as COLON,a as DOUBLE_QUOTE,c as EQUALS,R as EToCacheNodeProps,O as ETokenMatchCriteria,L as EXCLAMATION_MARK,s as GREATER_THAN,U as HASH,X as HORIZONTAL_TAB,n as HYPHEN,k as LESS_THAN,I as LINE_FEED,H as LOWERCASE_A,M as LOWERCASE_F,Q as LOWERCASE_X,B as LOWERCASE_Z,D as NINE,F as OPEN_SQUARE_BRACKET,b as PERCENT,h as PERIOD,j as QUESTION_MARK,K as SEMICOLON,W as SINGLE_QUOTE,Z as SLASH,d as SPACE,T as TokenSelectState,N as TokenSelectStateMachine,m as TokenSelector,G as UNDERSCORE,g as UPPERCASE_A,y as UPPERCASE_F,z as UPPERCASE_P,Y as UPPERCASE_S,q as UPPERCASE_Z,me as XmlError,Re as XmlStream,u as ZERO,o as getQName,$ as isAsciiDigit,ee as isXmlChar,Ee as isXmlName,oe as isXmlNameByte,re as isXmlNameStart,te as isXmlSpaceByte,Ne as processTokenForObject,pe as processTokenForSimplifiedObject,t as select,i as tokenToXml,w as tokenize,J as tokenizeXmlStream,Te as tokensToXml,_e as xmlToObject,le as xmlToSimplifiedObject};