xml-tokenizer 0.0.35 → 0.0.37

package/dist/cjs/tokenizer/XmlStream.js

@@ -1 +1 @@
-"use strict";var h=require("./ascii-constants.js"),c=require("./utils.js"),n=require("./XmlError.js"),S=Object.defineProperty,C=Object.getOwnPropertySymbols,E=Object.prototype.hasOwnProperty,y=Object.prototype.propertyIsEnumerable,f=(d,t,e)=>t in d?S(d,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):d[t]=e,x=(d,t)=>{for(var e in t||(t={}))E.call(t,e)&&f(d,e,t[e]);if(C)for(var e of C(t))y.call(t,e)&&f(d,e,t[e]);return d},m=(d,t,e)=>f(d,typeof t!="symbol"?t+"":t,e);class g{constructor(t,e={}){m(this,"_text"),m(this,"_pos"),m(this,"_end"),m(this,"config");const{pos:i=0,strictDocument:s=!0,allowDtd:o=!0,rawTextElements:l=null,implicitSelfClosingElements:a=null,contextSliceSize:r=!1,debug:u=!1}=e;this._text=t,this._pos=i,this._end=this._text.length,this.config={strictDocument:s,allowDtd:o,rawTextElements:l,implicitSelfClosingElements:a,contextSliceSize:r,debug:u},this.config.debug&&this.debugLog("XmlStream initialized",{textLength:t.length,startPos:this._pos,platform:typeof process!="undefined"?process.arch:"unknown",nodeVersion:typeof process!="undefined"?process.version:"unknown",textSample:t.slice(0,100),charCodes:t.slice(0,20).split("").map((p,_)=>({char:p,code:p.charCodeAt(0),pos:_}))})}clone(){return new g(this._text,x({pos:this._pos},this.config))}debugLog(t,e){this.config.debug&&console.info(`[XML-TOKENIZER-DEBUG] ${t}`,e?JSON.stringify(e,null,2):"")}getPos(){return this._pos}atEnd(){return this._pos>=this._end}currCodeUnit(){if(this._pos>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos)}currCodeUnitUnchecked(){return this._text.charCodeAt(this._pos)}nextCodeUnit(){if(this._pos+1>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new n.XmlError({type:"InvalidChar",expected:t,actual:e},this.genTextPos());this._pos+=1}tryConsumeCodeUnit(t){return this.currCodeUnit()===t?(this._pos+=1,!0):!1}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked(),this);){if(!c.isXmlChar(this.currCodeUnitUnchecked()))throw new n.XmlError({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}goToEnd(){this._pos=this._end}startsWith(t){return this._text.startsWith(t,this._pos)}startsWithIgnoreCase(t){if(this._pos+t.length>this._end)return!1;for(let e=0;e<t.length;e++){const i=this._text.charCodeAt(this._pos+e),s=t.charCodeAt(e);if((i>=97&&i<=122?i-32:i)!==s)return!1}return!0}skipString(t){if(!this.startsWith(t))throw new n.XmlError({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}sliceBack(t){return this._text.slice(t,this._pos)}rangeFrom(t){return{start:t,end:this._pos}}skipSpaces(){for(;this.startsWithSpace();)this._pos+=1}startsWithSpace(){return this._pos<this._end&&c.isXmlSpaceByte(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new n.XmlError({type:"InvalidChar",expected:"a whitespace",actual:this.currCodeUnitUnchecked()},this.genTextPos());this.skipSpaces()}tryConsumeReference(){const t=this._pos,e=this.clone(),i=e.consumeReference();return i!=null?(this._pos+=e.getPos()-t,i):null}consumeReference(){if(!this.tryConsumeCodeUnit(h.AMPERSAND))return null;let t;if(this.tryConsumeCodeUnit(h.HASH)){let e,i;this.tryConsumeCodeUnit(h.LOWERCASE_X)?(e=this.consumeCodeUnitsWhile(o=>o>=h.ZERO&&o<=h.NINE||o>=h.UPPERCASE_A&&o<=h.UPPERCASE_F||o>=h.LOWERCASE_A&&o<=h.LOWERCASE_F),i=16):(e=this.consumeCodeUnitsWhile(o=>c.isAsciiDigit(o)),i=10);const s=parseInt(e,i);isNaN(s)||!c.isXmlChar(s)?t=null:t={type:"Char",value:String.fromCodePoint(s)}}else{const e=this.consumeName();switch(e){case"quot":t={type:"Char",value:'"'};break;case"amp":t={type:"Char",value:"&"};break;case"apos":t={type:"Char",value:"'"};break;case"lt":t={type:"Char",value:"<"};break;case"gt":t={type:"Char",value:">"};break;default:t={type:"Entity",value:e}}}return this.tryConsumeCodeUnit(h.SEMICOLON)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new n.XmlError({type:"InvalidName"},this.genTextPosFrom(t));return e}skipName(){const t=this._pos;for(;this._pos<this._end;){const e=this.currCodeUnitUnchecked();if(this._pos===t){if(!c.isXmlNameStart(e))throw new n.XmlError({type:"InvalidName"},this.genTextPosFrom(t))}else if(!c.isXmlName(e))break;this._pos+=1}}consumeQName(){var t,e,i;const s=this._pos;let o=null;if(this.debugLog("consumeQName started",{position:this._pos,totalLength:this._end,context:this.getTextAround(this._pos,30),textPos:this.genTextPosFrom(this._pos)}),this._pos<this._end){const r=this.currCodeUnitUnchecked(),u=c.isXmlNameStart(r);if(this.debugLog("consumeQName first character check",{char:String.fromCharCode(r),charCode:r,charHex:"0x"+r.toString(16),isXmlNameStart:u,isXmlName:c.isXmlName(r),position:this._pos,platform:typeof process!="undefined"?process.arch:"unknown"}),!u)throw this.debugLog("consumeQName failed: invalid first character",{char:String.fromCharCode(r),charCode:r,charHex:"0x"+r.toString(16),context:this.getTextAround(this._pos,50),surroundingBytes:Array.from(this._text.slice(Math.max(0,this._pos-5),this._pos+5)).map((p,_)=>({char:p,code:p.charCodeAt(0),hex:"0x"+p.charCodeAt(0).toString(16),relativePos:_-5}))}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s))}for(;this._pos<this._end;){const r=this.currCodeUnitUnchecked();if(r===h.COLON)if(o==null)o=this._pos,this._pos+=1,this.debugLog("consumeQName found colon separator",{position:this._pos-1});else throw this.debugLog("consumeQName failed: multiple colons",{position:this._pos}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s));else if(c.isXmlName(r))this._pos+=1;else{this.debugLog("consumeQName loop break - invalid character",{char:String.fromCharCode(r),charCode:r,charHex:"0x"+r.toString(16),position:this._pos,reason:"not XML name character",isXmlName:c.isXmlName(r),consumedSoFar:this._text.slice(s,this._pos)});break}}let l,a;if(o!=null?(l=this._text.slice(s,o),a=this.sliceBack(o+1)):(l="",a=this.sliceBack(s)),l.length>0&&!c.isXmlNameStart((t=l[0])==null?void 0:t.codePointAt(0)))throw this.debugLog("consumeQName failed: invalid prefix start",{prefix:l,local:a}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s));if(a.length>0){if(!c.isXmlNameStart((e=a[0])==null?void 0:e.codePointAt(0)))throw this.debugLog("consumeQName failed: invalid local name start",{local:a,prefix:l,char:a[0],charCode:(i=a[0])==null?void 0:i.codePointAt(0),platform:typeof process!="undefined"?process.arch:"unknown"}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s))}else throw this.debugLog("consumeQName failed: empty local name",{local:a,prefix:l}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s));return this.debugLog("consumeQName succeeded",{prefix:l,local:a,startPos:s,endPos:this._pos,consumedText:this._text.slice(s,this._pos)}),[l,a]}consumeQuote(){const t=this.currCodeUnit();if(t===h.SINGLE_QUOTE||t===h.DOUBLE_QUOTE)return this._pos+=1,t;throw new n.XmlError({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}getTextAround(t,e){const i=Math.floor(e/2),s=Math.max(0,t-i),o=Math.min(this._end,t+i+e%2);return this._text.slice(s,o)}genTextPos(){return this.genTextPosFrom(this._pos)}genTextPosFrom(t){const e=Math.min(t,this._end);let i=1,s=1;for(let o=0;o<e;o++)this._text.charCodeAt(o)===h.LINE_FEED?(i++,s=1):s++;return x({row:i,col:s},this.config.contextSliceSize?{contextSlice:this.getTextAround(t,this.config.contextSliceSize)}:{})}}exports.XmlStream=g;
+"use strict";var o=require("./ascii-constants.js"),c=require("./utils.js"),n=require("./XmlError.js"),C=Object.defineProperty,g=Object.getOwnPropertySymbols,x=Object.prototype.hasOwnProperty,E=Object.prototype.propertyIsEnumerable,m=(a,t,e)=>t in a?C(a,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):a[t]=e,f=(a,t)=>{for(var e in t||(t={}))x.call(t,e)&&m(a,e,t[e]);if(g)for(var e of g(t))E.call(t,e)&&m(a,e,t[e]);return a},d=(a,t,e)=>m(a,typeof t!="symbol"?t+"":t,e);class _{constructor(t,e={}){d(this,"_text"),d(this,"_pos"),d(this,"_end"),d(this,"config");const{pos:i=0,strictDocument:r=!0,allowDtd:s=!0,rawTextElements:p=null,implicitSelfClosingElements:h=null,contextSliceSize:l=!1,debugLogger:u}=e;this._text=t,this._pos=i,this._end=this._text.length,this.config={strictDocument:r,allowDtd:s,rawTextElements:p,implicitSelfClosingElements:h,contextSliceSize:l,debugLogger:u},this.debug("XmlStream initialized",{textLength:this._text.length,initialPos:i,platform:process.platform,nodeVersion:process.version,textSample:this._text.slice(0,Math.min(100,this._text.length))+(this._text.length>100?"...":"")})}debug(t,...e){this.config.debugLogger&&this.config.debugLogger(t,...e)}clone(){return new _(this._text,f({pos:this._pos},this.config))}getPos(){return this._pos}atEnd(){return this._pos>=this._end}currCodeUnit(){if(this._pos>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos)}currCodeUnitUnchecked(){return this._text.charCodeAt(this._pos)}nextCodeUnit(){if(this._pos+1>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new n.XmlError({type:"InvalidChar",expected:t,actual:e},this.genTextPos());this._pos+=1}tryConsumeCodeUnit(t){return this.currCodeUnit()===t?(this._pos+=1,!0):!1}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked(),this);){if(!c.isXmlChar(this.currCodeUnitUnchecked()))throw new n.XmlError({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}goToEnd(){this._pos=this._end}startsWith(t){return this._text.startsWith(t,this._pos)}startsWithIgnoreCase(t){if(this._pos+t.length>this._end)return!1;for(let e=0;e<t.length;e++){const i=this._text.charCodeAt(this._pos+e),r=t.charCodeAt(e);if((i>=97&&i<=122?i-32:i)!==r)return!1}return!0}skipString(t){if(!this.startsWith(t))throw new n.XmlError({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}sliceBack(t){return this._text.slice(t,this._pos)}rangeFrom(t){return{start:t,end:this._pos}}skipSpaces(){for(;this.startsWithSpace();)this._pos+=1}startsWithSpace(){return this._pos<this._end&&c.isXmlSpaceByte(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new n.XmlError({type:"InvalidChar",expected:"a whitespace",actual:this.currCodeUnitUnchecked()},this.genTextPos());this.skipSpaces()}tryConsumeReference(){const t=this._pos,e=this.clone(),i=e.consumeReference();return i!=null?(this._pos+=e.getPos()-t,i):null}consumeReference(){if(!this.tryConsumeCodeUnit(o.AMPERSAND))return null;let t;if(this.tryConsumeCodeUnit(o.HASH)){let e,i;this.tryConsumeCodeUnit(o.LOWERCASE_X)?(e=this.consumeCodeUnitsWhile(s=>s>=o.ZERO&&s<=o.NINE||s>=o.UPPERCASE_A&&s<=o.UPPERCASE_F||s>=o.LOWERCASE_A&&s<=o.LOWERCASE_F),i=16):(e=this.consumeCodeUnitsWhile(s=>c.isAsciiDigit(s)),i=10);const r=parseInt(e,i);isNaN(r)||!c.isXmlChar(r)?t=null:t={type:"Char",value:String.fromCodePoint(r)}}else{const e=this.consumeName();switch(e){case"quot":t={type:"Char",value:'"'};break;case"amp":t={type:"Char",value:"&"};break;case"apos":t={type:"Char",value:"'"};break;case"lt":t={type:"Char",value:"<"};break;case"gt":t={type:"Char",value:">"};break;default:t={type:"Entity",value:e}}}return this.tryConsumeCodeUnit(o.SEMICOLON)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new n.XmlError({type:"InvalidName"},this.genTextPosFrom(t));return e}skipName(){const t=this._pos;for(;this._pos<this._end;){const e=this.currCodeUnitUnchecked();if(this._pos===t){if(!c.isXmlNameStart(e))throw new n.XmlError({type:"InvalidName"},this.genTextPosFrom(t))}else if(!c.isXmlName(e))break;this._pos+=1}}consumeQName(){var t,e,i,r;const s=this._pos;let p=null;for(this.debug("Starting consumeQName",{position:this._pos,nextChars:this._text.slice(this._pos,this._pos+10)});this._pos<this._end;){const u=this.currCodeUnitUnchecked();if(u===o.COLON)if(p==null)p=this._pos,this._pos+=1,this.debug("Found colon splitter",{position:this._pos});else throw this.debug("Multiple colons found - error",{position:this._pos}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s));else if(c.isXmlName(u))this._pos+=1;else break}let h,l;if(p!=null?(h=this._text.slice(s,p),l=this.sliceBack(p+1)):(h="",l=this.sliceBack(s)),this.debug("Finished consumeQName",{prefix:h,local:l,startPos:s,endPos:this._pos}),h.length>0&&!c.isXmlNameStart((t=h[0])==null?void 0:t.codePointAt(0)))throw this.debug("Invalid prefix start character",{prefix:h,firstChar:h[0],codePoint:(e=h[0])==null?void 0:e.codePointAt(0)}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s));if(l.length>0){if(!c.isXmlNameStart((i=l[0])==null?void 0:i.codePointAt(0)))throw this.debug("Invalid local name start character",{local:l,firstChar:l[0],codePoint:(r=l[0])==null?void 0:r.codePointAt(0)}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s))}else throw this.debug("Empty local name - error"),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(s));return[h,l]}consumeQuote(){const t=this.currCodeUnit();if(t===o.SINGLE_QUOTE||t===o.DOUBLE_QUOTE)return this._pos+=1,t;throw new n.XmlError({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}getTextAround(t,e){const i=Math.floor(e/2),r=Math.max(0,t-i),s=Math.min(this._end,t+i+e%2);return this._text.slice(r,s)}genTextPos(){return this.genTextPosFrom(this._pos)}genTextPosFrom(t){const e=Math.min(t,this._end);let i=1,r=1;for(let s=0;s<e;s++)this._text.charCodeAt(s)===o.LINE_FEED?(i++,r=1):r++;return f({row:i,col:r},this.config.contextSliceSize?{contextSlice:this.getTextAround(t,this.config.contextSliceSize)}:{})}getTextSnippet(t,e){return this._text.slice(t,t+e)}}exports.XmlStream=_;

package/dist/cjs/tokenizer/tokenize.js

@@ -1,2 +1 @@
-"use strict";var k=require("../get-q-name.js"),s=require("./ascii-constants.js"),I=require("./utils.js"),a=require("./XmlError.js"),D=require("./XmlStream.js");const H="\uFEFF",U="<?xml ",y="?>",Q="<!ENTITY",F="<!ELEMENT",O="<!ATTLIST",b="<!NOTATION",m="<!DOCTYPE",A="<?",w="?>",f="<!--",P="-->",G="<![CDATA[",g="]]>",B="NDATA",x="PUBLIC",M="SYSTEM",q="version",z="encoding",K="standalone";function Y(t,n,e={}){N(new D.XmlStream(t,e),n)}function N(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(H)&&t.advance(1),t.startsWith(U)&&J(t),p(t,n),t.config.strictDocument){if(t.skipSpaces(),t.startsWithIgnoreCase(m)&&(S(t,n),p(t,n)),t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===s.LESS_THAN&&T(t,n),p(t,n),!t.atEnd())throw new a.XmlError({type:"UnknownToken",message:"Not at end"},t.genTextPos())}else for(;!t.atEnd();)t.currCodeUnit()===s.LESS_THAN?T(t,n):L(t,n),p(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 a.XmlError?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 p(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith(f))E(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(m))S(t,n);else if(t.startsWith(A))h(t,n);else break}function J(t){function n(e){if(e.startsWithSpace())e.skipSpaces();else if(!e.startsWith(y)&&!e.atEnd())throw new a.XmlError({type:"InvalidChar",expected:"a whitespace",actual:e.currCodeUnitUnchecked()},e.genTextPos())}if(t.advance(5),n(t),!t.startsWith(q))throw new a.XmlError({type:"InvalidString",expected:"version"},t.genTextPos());d(t),n(t),t.startsWith(z)&&(d(t),n(t)),t.startsWith(K)&&d(t),t.skipSpaces(),t.skipString(y)}function E(t,n){const e=t.getPos();t.advance(4);const o=t.consumeCodeUnitsWhile((i,c)=>!(i===s.HYPHEN&&c.startsWith(P)));if(t.skipString(P),o.includes("--")||o.endsWith("-"))throw new a.XmlError({type:"InvalidComment"},t.genTextPosFrom(e));n({type:"Comment",text:o,range:t.rangeFrom(e)},t)}function h(t,n){if(t.startsWith(U))throw new a.XmlError({type:"UnexpectedDeclaration"},t.genTextPos());const e=t.getPos();t.advance(2);const o=t.consumeName();t.skipSpaces();const i=t.consumeCodeUnitsWhile((c,r)=>!(c===s.QUESTION_MARK&&r.startsWith(w)));t.skipString(w),n({type:"ProcessingInstruction",target:o,content:i.length===0?void 0:i,range:t.rangeFrom(e)},t)}function S(t,n){if(!t.config.allowDtd)throw new a.XmlError({type:"DtdDetected"});const e=t.getPos();if(V(t),t.skipSpaces(),t.currCodeUnit()===s.GREATER_THAN){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(Q))$(t,n);else if(t.startsWith(f))E(t,n);else if(t.startsWith(A))h(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===s.GREATER_THAN){t.advance(1);break}else throw new a.XmlError({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith(F)||t.startsWith(O)||t.startsWith(b))try{Z(t)}catch(o){throw new a.XmlError({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(e))}else throw new a.XmlError({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function V(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),v(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==s.OPEN_SQUARE_BRACKET&&n!==s.GREATER_THAN)throw new a.XmlError({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function v(t){if(t.startsWith(M)||t.startsWith(x)){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===x){t.consumeSpaces();const i=t.consumeQuote();t.consumeCodeUnitsWhile(c=>c!==i),t.consumeCodeUnit(i)}return!0}return!1}function $(t,n){t.advance(8),t.consumeSpaces();const e=!t.tryConsumeCodeUnit(s.PERCENT);e||t.consumeSpaces();const o=t.consumeName();t.consumeSpaces();const i=j(t,e);i!==null&&n({type:"EntityDeclaration",name:o,definition:i},t),t.skipSpaces(),t.consumeCodeUnit(s.GREATER_THAN)}function j(t,n){const e=t.currCodeUnit();if(e===s.DOUBLE_QUOTE||e===s.SINGLE_QUOTE){const o=t.consumeQuote(),i=t.getPos();t.skipCodeUnitsWhile(r=>r!==o);const c=t.sliceBack(i);return t.consumeCodeUnit(o),c}else if(e===s.UPPERCASE_S||e===s.UPPERCASE_P){if(v(t))return n&&(t.skipSpaces(),t.startsWith(B)&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new a.XmlError({type:"InvalidExternalID"},t.genTextPos())}else throw new a.XmlError({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:e},t.genTextPos())}function Z(t){t.skipCodeUnitsWhile(n=>n!==s.GREATER_THAN),t.consumeCodeUnit(s.GREATER_THAN)}function T(t,n){const e=t.getPos();if(t.config.debug){const r=t.currCodeUnit();t.debugLog("parseElement started",JSON.stringify({position:e,character:String.fromCharCode(r),charCode:r,charHex:"0x"+r.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(r){throw t.debugLog("parseElement consumeQName FAILED",JSON.stringify({error:r instanceof Error?r.message:String(r),errorType:r instanceof a.XmlError?r.variant.type:"unknown",position:t.getPos(),startPosition:e,context:t.getTextAround(t.getPos(),50),platform:typeof process!="undefined"?process.arch:"unknown",stackTrace:r instanceof Error?r.stack:void 0},null,2)),r}let c=!1;for(;!t.atEnd();){const r=t.startsWithSpace();t.skipSpaces();const l=t.getPos(),C=t.currCodeUnit();if(C===s.SLASH){t.advance(1),t.consumeCodeUnit(s.GREATER_THAN);const u=t.rangeFrom(l);n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else if(C===s.GREATER_THAN){t.advance(1);const u=t.rangeFrom(l);if(t.config.implicitSelfClosingElements!=null&&t.config.implicitSelfClosingElements.includes(k.getQName(i,o))){n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else{n({type:"ElementEnd",end:{type:"Open"},range:u},t),c=!0;break}}else{if(!r)throw new a.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());const[u,R,X]=d(t),_=t.getPos();n({type:"Attribute",range:{start:l,end:_},prefix:u,local:R,value:X},t)}}if(c){if(t.config.rawTextElements!=null){const r=k.getQName(i,o);if(t.config.rawTextElements.includes(r)){W(t,n,`</${r}>`);return}}W(t,n)}}function d(t){const[n,e]=t.consumeQName();let o;const i=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(s.EQUALS)){t.skipSpaces();const c=t.currCodeUnit();if(c===s.SINGLE_QUOTE||c===s.DOUBLE_QUOTE){const r=t.consumeQuote();o=t.consumeCodeUnitsWhile(l=>l!==r&&l!==s.LESS_THAN),t.consumeCodeUnit(r)}else if(!t.config.strictDocument)o=t.consumeCodeUnitsWhile(r=>!I.isXmlSpaceByte(r)&&r!==s.GREATER_THAN&&r!==s.SLASH&&r!==s.LESS_THAN);else throw new a.XmlError({type:"InvalidChar",expected:"a quote",actual:c},t.genTextPos())}else if(!t.config.strictDocument)t.goTo(i),o="true";else throw new a.XmlError({type:"InvalidChar",expected:s.EQUALS,actual:t.currCodeUnit()},t.genTextPos());return[n,e,o]}function W(t,n,e=s.LESS_THAN){for(;!t.atEnd();)if(typeof e=="number"?t.currCodeUnit()===e:t.startsWith(e)){const o=t.nextCodeUnit();if(o===s.EXCLAMATION_MARK)if(t.startsWith(f))E(t,n);else if(t.startsWith(G))tt(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(m))S(t,n);else throw new a.XmlError({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(o===s.QUESTION_MARK)h(t,n);else if(o===s.SLASH){nt(t,n);break}else T(t,n)}else L(t,n,e)}function tt(t,n){const e=t.getPos();t.advance(9);const o=t.consumeCodeUnitsWhile((c,r)=>!(c===s.CLOSE_SQUARE_BRACKET&&r.startsWith(g)));t.skipString(g);const i=t.rangeFrom(e);n({type:"Cdata",text:o,range:i},t)}function nt(t,n){const e=t.getPos();t.advance(2);const[o,i]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(s.GREATER_THAN);const c=t.rangeFrom(e);n({type:"ElementEnd",end:{type:"Close",prefix:o,local:i},range:c},t)}function L(t,n,e=s.LESS_THAN){const o=t.getPos(),i=typeof e=="number"?t.consumeCodeUnitsWhile(c=>c!==e):t.consumeCodeUnitsWhile((c,r)=>!r.startsWith(e));if(t.config.strictDocument&&i.includes(g))throw new a.XmlError({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:i,range:t.rangeFrom(o)},t)}exports.tokenize=Y,exports.tokenizeXmlStream=N;
+"use strict";var g=require("../get-q-name.js"),o=require("./ascii-constants.js"),I=require("./utils.js"),c=require("./XmlError.js"),L=require("./XmlStream.js");const F="\uFEFF",P="<?xml ",w="?>",H="<!ENTITY",Q="<!ELEMENT",D="<!ATTLIST",O="<!NOTATION",E="<!DOCTYPE",x="<?",W="?>",f="<!--",y="-->",G="<![CDATA[",h="]]>",M="NDATA",N="PUBLIC",B="SYSTEM",q="version",z="encoding",Y="standalone";function K(t,n,e={}){const s=new L.XmlStream(t,e);s.debug("Starting tokenization",{stringLength:t.length}),v(s,n)}function v(t,n){if(t.debug("Starting XML stream tokenization"),t.startsWith(F)&&(t.debug("Skipping BOM"),t.advance(1)),t.startsWith(P)&&(t.debug("Found XML declaration"),$(t)),m(t,n),t.config.strictDocument){if(t.debug("Parsing in strict document mode"),t.skipSpaces(),t.startsWithIgnoreCase(E)&&(t.debug("Found DOCTYPE declaration"),C(t,n),m(t,n)),t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===o.LESS_THAN&&(t.debug("Starting root element parsing"),U(t,n)),m(t,n),!t.atEnd())throw t.debug("Unexpected content after root element",{position:t.getPos(),nextChars:t.getTextSnippet(t.getPos(),10)}),new c.XmlError({type:"UnknownToken",message:"Not at end"},t.genTextPos())}else for(t.debug("Parsing in non-strict mode");!t.atEnd();)t.currCodeUnit()===o.LESS_THAN?U(t,n):_(t,n),m(t,n);t.debug("Finished XML stream tokenization")}function m(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith(f))S(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(E))C(t,n);else if(t.startsWith(x))T(t,n);else break}function $(t){function n(e){if(e.startsWithSpace())e.skipSpaces();else if(!e.startsWith(w)&&!e.atEnd())throw new c.XmlError({type:"InvalidChar",expected:"a whitespace",actual:e.currCodeUnitUnchecked()},e.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(z)&&(p(t),n(t)),t.startsWith(Y)&&p(t),t.skipSpaces(),t.skipString(w)}function S(t,n){const e=t.getPos();t.advance(4);const s=t.consumeCodeUnitsWhile((i,a)=>!(i===o.HYPHEN&&a.startsWith(y)));if(t.skipString(y),s.includes("--")||s.endsWith("-"))throw new c.XmlError({type:"InvalidComment"},t.genTextPosFrom(e));n({type:"Comment",text:s,range:t.rangeFrom(e)},t)}function T(t,n){if(t.startsWith(P))throw new c.XmlError({type:"UnexpectedDeclaration"},t.genTextPos());const e=t.getPos();t.advance(2);const s=t.consumeName();t.skipSpaces();const i=t.consumeCodeUnitsWhile((a,r)=>!(a===o.QUESTION_MARK&&r.startsWith(W)));t.skipString(W),n({type:"ProcessingInstruction",target:s,content:i.length===0?void 0:i,range:t.rangeFrom(e)},t)}function C(t,n){if(!t.config.allowDtd)throw new c.XmlError({type:"DtdDetected"});const e=t.getPos();if(j(t),t.skipSpaces(),t.currCodeUnit()===o.GREATER_THAN){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(H))J(t,n);else if(t.startsWith(f))S(t,n);else if(t.startsWith(x))T(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===o.GREATER_THAN){t.advance(1);break}else throw new c.XmlError({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith(Q)||t.startsWith(D)||t.startsWith(O))try{Z(t)}catch(s){throw new c.XmlError({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(e))}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(),b(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==o.OPEN_SQUARE_BRACKET&&n!==o.GREATER_THAN)throw new c.XmlError({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function b(t){if(t.startsWith(B)||t.startsWith(N)){const n=t.getPos();t.advance(6);const e=t.sliceBack(n);t.consumeSpaces();const s=t.consumeQuote();if(t.consumeCodeUnitsWhile(i=>i!==s),t.consumeCodeUnit(s),e===N){t.consumeSpaces();const i=t.consumeQuote();t.consumeCodeUnitsWhile(a=>a!==i),t.consumeCodeUnit(i)}return!0}return!1}function J(t,n){t.advance(8),t.consumeSpaces();const e=!t.tryConsumeCodeUnit(o.PERCENT);e||t.consumeSpaces();const s=t.consumeName();t.consumeSpaces();const i=V(t,e);i!==null&&n({type:"EntityDeclaration",name:s,definition:i},t),t.skipSpaces(),t.consumeCodeUnit(o.GREATER_THAN)}function V(t,n){const e=t.currCodeUnit();if(e===o.DOUBLE_QUOTE||e===o.SINGLE_QUOTE){const s=t.consumeQuote(),i=t.getPos();t.skipCodeUnitsWhile(r=>r!==s);const a=t.sliceBack(i);return t.consumeCodeUnit(s),a}else if(e===o.UPPERCASE_S||e===o.UPPERCASE_P){if(b(t))return n&&(t.skipSpaces(),t.startsWith(M)&&(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:e},t.genTextPos())}function Z(t){t.skipCodeUnitsWhile(n=>n!==o.GREATER_THAN),t.consumeCodeUnit(o.GREATER_THAN)}function U(t,n){const e=t.getPos();t.advance(1),t.debug("Starting element parsing",{position:e,nextChars:t.getTextSnippet(e,10)});const[s,i]=t.consumeQName();t.debug("Element name parsed",{prefix:s,local:i}),n({type:"ElementStart",prefix:s,local:i,start:e},t);let a=!1;for(;!t.atEnd();){const r=t.startsWithSpace();t.skipSpaces();const l=t.getPos(),u=t.currCodeUnit();if(t.debug("Processing element content",{position:l,currCodeUnit:u,hasSpace:r}),u===o.SLASH){t.debug("Found self-closing tag"),t.advance(1),t.consumeCodeUnit(o.GREATER_THAN);const d=t.rangeFrom(l);n({type:"ElementEnd",end:{type:"Empty"},range:d},t);break}else if(u===o.GREATER_THAN){t.advance(1);const d=t.rangeFrom(l);if(t.config.implicitSelfClosingElements!=null&&t.config.implicitSelfClosingElements.includes(g.getQName(i,s))){t.debug("Found implicit self-closing element",{element:g.getQName(i,s)}),n({type:"ElementEnd",end:{type:"Empty"},range:d},t);break}else{t.debug("Found opening tag end"),n({type:"ElementEnd",end:{type:"Open"},range:d},t),a=!0;break}}else{if(!r)throw t.debug("Missing whitespace before attribute",{position:t.getPos(),currCodeUnit:String.fromCharCode(u)}),new c.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());t.debug("Parsing attribute");const[d,k,A]=p(t),X=t.getPos();t.debug("Attribute parsed",{prefix:d,local:k,value:A}),n({type:"Attribute",range:{start:l,end:X},prefix:d,local:k,value:A},t)}}if(a){if(t.config.rawTextElements!=null){const r=g.getQName(i,s);if(t.config.rawTextElements.includes(r)){t.debug("Processing raw text element content",{element:r}),R(t,n,`</${r}>`);return}}t.debug("Processing element content"),R(t,n)}t.debug("Finished element parsing",{prefix:s,local:i})}function p(t){const[n,e]=t.consumeQName();let s;const i=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(o.EQUALS)){t.skipSpaces();const a=t.currCodeUnit();if(a===o.SINGLE_QUOTE||a===o.DOUBLE_QUOTE){const r=t.consumeQuote();s=t.consumeCodeUnitsWhile(l=>l!==r&&l!==o.LESS_THAN),t.consumeCodeUnit(r)}else if(!t.config.strictDocument)s=t.consumeCodeUnitsWhile(r=>!I.isXmlSpaceByte(r)&&r!==o.GREATER_THAN&&r!==o.SLASH&&r!==o.LESS_THAN);else throw new c.XmlError({type:"InvalidChar",expected:"a quote",actual:a},t.genTextPos())}else if(!t.config.strictDocument)t.goTo(i),s="true";else throw new c.XmlError({type:"InvalidChar",expected:o.EQUALS,actual:t.currCodeUnit()},t.genTextPos());return[n,e,s]}function R(t,n,e=o.LESS_THAN){for(;!t.atEnd();)if(typeof e=="number"?t.currCodeUnit()===e:t.startsWith(e)){const s=t.nextCodeUnit();if(s===o.EXCLAMATION_MARK)if(t.startsWith(f))S(t,n);else if(t.startsWith(G))tt(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(E))C(t,n);else throw new c.XmlError({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===o.QUESTION_MARK)T(t,n);else if(s===o.SLASH){nt(t,n);break}else U(t,n)}else _(t,n,e)}function tt(t,n){const e=t.getPos();t.advance(9);const s=t.consumeCodeUnitsWhile((a,r)=>!(a===o.CLOSE_SQUARE_BRACKET&&r.startsWith(h)));t.skipString(h);const i=t.rangeFrom(e);n({type:"Cdata",text:s,range:i},t)}function nt(t,n){const e=t.getPos();t.advance(2);const[s,i]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(o.GREATER_THAN);const a=t.rangeFrom(e);n({type:"ElementEnd",end:{type:"Close",prefix:s,local:i},range:a},t)}function _(t,n,e=o.LESS_THAN){const s=t.getPos(),i=typeof e=="number"?t.consumeCodeUnitsWhile(a=>a!==e):t.consumeCodeUnitsWhile((a,r)=>!r.startsWith(e));if(t.config.strictDocument&&i.includes(h))throw new c.XmlError({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:i,range:t.rangeFrom(s)},t)}exports.tokenize=K,exports.tokenizeXmlStream=v;

package/dist/esm/tokenizer/XmlStream.js

@@ -1 +1 @@
-import{AMPERSAND as y,HASH as w,LOWERCASE_X as U,ZERO as N,NINE as v,UPPERCASE_A as P,UPPERCASE_F as E,LOWERCASE_A as b,LOWERCASE_F as k,SEMICOLON as T,COLON as A,SINGLE_QUOTE as L,DOUBLE_QUOTE as O,LINE_FEED as I}from"./ascii-constants.js";import{isXmlChar as x,isXmlSpaceByte as F,isAsciiDigit as W,isXmlNameStart as d,isXmlName as u}from"./utils.js";import{XmlError as r}from"./XmlError.js";var Q=Object.defineProperty,C=Object.getOwnPropertySymbols,X=Object.prototype.hasOwnProperty,R=Object.prototype.propertyIsEnumerable,f=(c,t,e)=>t in c?Q(c,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):c[t]=e,S=(c,t)=>{for(var e in t||(t={}))X.call(t,e)&&f(c,e,t[e]);if(C)for(var e of C(t))R.call(t,e)&&f(c,e,t[e]);return c},m=(c,t,e)=>f(c,typeof t!="symbol"?t+"":t,e);class g{constructor(t,e={}){m(this,"_text"),m(this,"_pos"),m(this,"_end"),m(this,"config");const{pos:i=0,strictDocument:s=!0,allowDtd:o=!0,rawTextElements:a=null,implicitSelfClosingElements:h=null,contextSliceSize:n=!1,debug:p=!1}=e;this._text=t,this._pos=i,this._end=this._text.length,this.config={strictDocument:s,allowDtd:o,rawTextElements:a,implicitSelfClosingElements:h,contextSliceSize:n,debug:p},this.config.debug&&this.debugLog("XmlStream initialized",{textLength:t.length,startPos:this._pos,platform:typeof process!="undefined"?process.arch:"unknown",nodeVersion:typeof process!="undefined"?process.version:"unknown",textSample:t.slice(0,100),charCodes:t.slice(0,20).split("").map((l,_)=>({char:l,code:l.charCodeAt(0),pos:_}))})}clone(){return new g(this._text,S({pos:this._pos},this.config))}debugLog(t,e){this.config.debug&&console.info(`[XML-TOKENIZER-DEBUG] ${t}`,e?JSON.stringify(e,null,2):"")}getPos(){return this._pos}atEnd(){return this._pos>=this._end}currCodeUnit(){if(this._pos>=this._end)throw new r({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos)}currCodeUnitUnchecked(){return this._text.charCodeAt(this._pos)}nextCodeUnit(){if(this._pos+1>=this._end)throw new r({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new r({type:"InvalidChar",expected:t,actual:e},this.genTextPos());this._pos+=1}tryConsumeCodeUnit(t){return this.currCodeUnit()===t?(this._pos+=1,!0):!1}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked(),this);){if(!x(this.currCodeUnitUnchecked()))throw new r({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}goToEnd(){this._pos=this._end}startsWith(t){return this._text.startsWith(t,this._pos)}startsWithIgnoreCase(t){if(this._pos+t.length>this._end)return!1;for(let e=0;e<t.length;e++){const i=this._text.charCodeAt(this._pos+e),s=t.charCodeAt(e);if((i>=97&&i<=122?i-32:i)!==s)return!1}return!0}skipString(t){if(!this.startsWith(t))throw new r({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}sliceBack(t){return this._text.slice(t,this._pos)}rangeFrom(t){return{start:t,end:this._pos}}skipSpaces(){for(;this.startsWithSpace();)this._pos+=1}startsWithSpace(){return this._pos<this._end&&F(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new r({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new r({type:"InvalidChar",expected:"a whitespace",actual:this.currCodeUnitUnchecked()},this.genTextPos());this.skipSpaces()}tryConsumeReference(){const t=this._pos,e=this.clone(),i=e.consumeReference();return i!=null?(this._pos+=e.getPos()-t,i):null}consumeReference(){if(!this.tryConsumeCodeUnit(y))return null;let t;if(this.tryConsumeCodeUnit(w)){let e,i;this.tryConsumeCodeUnit(U)?(e=this.consumeCodeUnitsWhile(o=>o>=N&&o<=v||o>=P&&o<=E||o>=b&&o<=k),i=16):(e=this.consumeCodeUnitsWhile(o=>W(o)),i=10);const s=parseInt(e,i);isNaN(s)||!x(s)?t=null:t={type:"Char",value:String.fromCodePoint(s)}}else{const e=this.consumeName();switch(e){case"quot":t={type:"Char",value:'"'};break;case"amp":t={type:"Char",value:"&"};break;case"apos":t={type:"Char",value:"'"};break;case"lt":t={type:"Char",value:"<"};break;case"gt":t={type:"Char",value:">"};break;default:t={type:"Entity",value:e}}}return this.tryConsumeCodeUnit(T)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new r({type:"InvalidName"},this.genTextPosFrom(t));return e}skipName(){const t=this._pos;for(;this._pos<this._end;){const e=this.currCodeUnitUnchecked();if(this._pos===t){if(!d(e))throw new r({type:"InvalidName"},this.genTextPosFrom(t))}else if(!u(e))break;this._pos+=1}}consumeQName(){var t,e,i;const s=this._pos;let o=null;if(this.debugLog("consumeQName started",{position:this._pos,totalLength:this._end,context:this.getTextAround(this._pos,30),textPos:this.genTextPosFrom(this._pos)}),this._pos<this._end){const n=this.currCodeUnitUnchecked(),p=d(n);if(this.debugLog("consumeQName first character check",{char:String.fromCharCode(n),charCode:n,charHex:"0x"+n.toString(16),isXmlNameStart:p,isXmlName:u(n),position:this._pos,platform:typeof process!="undefined"?process.arch:"unknown"}),!p)throw this.debugLog("consumeQName failed: invalid first character",{char:String.fromCharCode(n),charCode:n,charHex:"0x"+n.toString(16),context:this.getTextAround(this._pos,50),surroundingBytes:Array.from(this._text.slice(Math.max(0,this._pos-5),this._pos+5)).map((l,_)=>({char:l,code:l.charCodeAt(0),hex:"0x"+l.charCodeAt(0).toString(16),relativePos:_-5}))}),new r({type:"InvalidName"},this.genTextPosFrom(s))}for(;this._pos<this._end;){const n=this.currCodeUnitUnchecked();if(n===A)if(o==null)o=this._pos,this._pos+=1,this.debugLog("consumeQName found colon separator",{position:this._pos-1});else throw this.debugLog("consumeQName failed: multiple colons",{position:this._pos}),new r({type:"InvalidName"},this.genTextPosFrom(s));else if(u(n))this._pos+=1;else{this.debugLog("consumeQName loop break - invalid character",{char:String.fromCharCode(n),charCode:n,charHex:"0x"+n.toString(16),position:this._pos,reason:"not XML name character",isXmlName:u(n),consumedSoFar:this._text.slice(s,this._pos)});break}}let a,h;if(o!=null?(a=this._text.slice(s,o),h=this.sliceBack(o+1)):(a="",h=this.sliceBack(s)),a.length>0&&!d((t=a[0])==null?void 0:t.codePointAt(0)))throw this.debugLog("consumeQName failed: invalid prefix start",{prefix:a,local:h}),new r({type:"InvalidName"},this.genTextPosFrom(s));if(h.length>0){if(!d((e=h[0])==null?void 0:e.codePointAt(0)))throw this.debugLog("consumeQName failed: invalid local name start",{local:h,prefix:a,char:h[0],charCode:(i=h[0])==null?void 0:i.codePointAt(0),platform:typeof process!="undefined"?process.arch:"unknown"}),new r({type:"InvalidName"},this.genTextPosFrom(s))}else throw this.debugLog("consumeQName failed: empty local name",{local:h,prefix:a}),new r({type:"InvalidName"},this.genTextPosFrom(s));return this.debugLog("consumeQName succeeded",{prefix:a,local:h,startPos:s,endPos:this._pos,consumedText:this._text.slice(s,this._pos)}),[a,h]}consumeQuote(){const t=this.currCodeUnit();if(t===L||t===O)return this._pos+=1,t;throw new r({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}getTextAround(t,e){const i=Math.floor(e/2),s=Math.max(0,t-i),o=Math.min(this._end,t+i+e%2);return this._text.slice(s,o)}genTextPos(){return this.genTextPosFrom(this._pos)}genTextPosFrom(t){const e=Math.min(t,this._end);let i=1,s=1;for(let o=0;o<e;o++)this._text.charCodeAt(o)===I?(i++,s=1):s++;return S({row:i,col:s},this.config.contextSliceSize?{contextSlice:this.getTextAround(t,this.config.contextSliceSize)}:{})}}export{g as XmlStream};
+import{AMPERSAND as x,HASH as S,LOWERCASE_X as U,ZERO as w,NINE as y,UPPERCASE_A as P,UPPERCASE_F as v,LOWERCASE_A as E,LOWERCASE_F as b,SEMICOLON as k,COLON as N,SINGLE_QUOTE as T,DOUBLE_QUOTE as A,LINE_FEED as O}from"./ascii-constants.js";import{isXmlChar as m,isXmlSpaceByte as I,isAsciiDigit as W,isXmlNameStart as u,isXmlName as g}from"./utils.js";import{XmlError as n}from"./XmlError.js";var F=Object.defineProperty,f=Object.getOwnPropertySymbols,L=Object.prototype.hasOwnProperty,R=Object.prototype.propertyIsEnumerable,d=(c,t,e)=>t in c?F(c,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):c[t]=e,C=(c,t)=>{for(var e in t||(t={}))L.call(t,e)&&d(c,e,t[e]);if(f)for(var e of f(t))R.call(t,e)&&d(c,e,t[e]);return c},p=(c,t,e)=>d(c,typeof t!="symbol"?t+"":t,e);class _{constructor(t,e={}){p(this,"_text"),p(this,"_pos"),p(this,"_end"),p(this,"config");const{pos:i=0,strictDocument:o=!0,allowDtd:s=!0,rawTextElements:a=null,implicitSelfClosingElements:r=null,contextSliceSize:h=!1,debugLogger:l}=e;this._text=t,this._pos=i,this._end=this._text.length,this.config={strictDocument:o,allowDtd:s,rawTextElements:a,implicitSelfClosingElements:r,contextSliceSize:h,debugLogger:l},this.debug("XmlStream initialized",{textLength:this._text.length,initialPos:i,platform:process.platform,nodeVersion:process.version,textSample:this._text.slice(0,Math.min(100,this._text.length))+(this._text.length>100?"...":"")})}debug(t,...e){this.config.debugLogger&&this.config.debugLogger(t,...e)}clone(){return new _(this._text,C({pos:this._pos},this.config))}getPos(){return this._pos}atEnd(){return this._pos>=this._end}currCodeUnit(){if(this._pos>=this._end)throw new n({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos)}currCodeUnitUnchecked(){return this._text.charCodeAt(this._pos)}nextCodeUnit(){if(this._pos+1>=this._end)throw new n({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new n({type:"InvalidChar",expected:t,actual:e},this.genTextPos());this._pos+=1}tryConsumeCodeUnit(t){return this.currCodeUnit()===t?(this._pos+=1,!0):!1}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked(),this);){if(!m(this.currCodeUnitUnchecked()))throw new n({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}goToEnd(){this._pos=this._end}startsWith(t){return this._text.startsWith(t,this._pos)}startsWithIgnoreCase(t){if(this._pos+t.length>this._end)return!1;for(let e=0;e<t.length;e++){const i=this._text.charCodeAt(this._pos+e),o=t.charCodeAt(e);if((i>=97&&i<=122?i-32:i)!==o)return!1}return!0}skipString(t){if(!this.startsWith(t))throw new n({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}sliceBack(t){return this._text.slice(t,this._pos)}rangeFrom(t){return{start:t,end:this._pos}}skipSpaces(){for(;this.startsWithSpace();)this._pos+=1}startsWithSpace(){return this._pos<this._end&&I(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new n({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new n({type:"InvalidChar",expected:"a whitespace",actual:this.currCodeUnitUnchecked()},this.genTextPos());this.skipSpaces()}tryConsumeReference(){const t=this._pos,e=this.clone(),i=e.consumeReference();return i!=null?(this._pos+=e.getPos()-t,i):null}consumeReference(){if(!this.tryConsumeCodeUnit(x))return null;let t;if(this.tryConsumeCodeUnit(S)){let e,i;this.tryConsumeCodeUnit(U)?(e=this.consumeCodeUnitsWhile(s=>s>=w&&s<=y||s>=P&&s<=v||s>=E&&s<=b),i=16):(e=this.consumeCodeUnitsWhile(s=>W(s)),i=10);const o=parseInt(e,i);isNaN(o)||!m(o)?t=null:t={type:"Char",value:String.fromCodePoint(o)}}else{const e=this.consumeName();switch(e){case"quot":t={type:"Char",value:'"'};break;case"amp":t={type:"Char",value:"&"};break;case"apos":t={type:"Char",value:"'"};break;case"lt":t={type:"Char",value:"<"};break;case"gt":t={type:"Char",value:">"};break;default:t={type:"Entity",value:e}}}return this.tryConsumeCodeUnit(k)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new n({type:"InvalidName"},this.genTextPosFrom(t));return e}skipName(){const t=this._pos;for(;this._pos<this._end;){const e=this.currCodeUnitUnchecked();if(this._pos===t){if(!u(e))throw new n({type:"InvalidName"},this.genTextPosFrom(t))}else if(!g(e))break;this._pos+=1}}consumeQName(){var t,e,i,o;const s=this._pos;let a=null;for(this.debug("Starting consumeQName",{position:this._pos,nextChars:this._text.slice(this._pos,this._pos+10)});this._pos<this._end;){const l=this.currCodeUnitUnchecked();if(l===N)if(a==null)a=this._pos,this._pos+=1,this.debug("Found colon splitter",{position:this._pos});else throw this.debug("Multiple colons found - error",{position:this._pos}),new n({type:"InvalidName"},this.genTextPosFrom(s));else if(g(l))this._pos+=1;else break}let r,h;if(a!=null?(r=this._text.slice(s,a),h=this.sliceBack(a+1)):(r="",h=this.sliceBack(s)),this.debug("Finished consumeQName",{prefix:r,local:h,startPos:s,endPos:this._pos}),r.length>0&&!u((t=r[0])==null?void 0:t.codePointAt(0)))throw this.debug("Invalid prefix start character",{prefix:r,firstChar:r[0],codePoint:(e=r[0])==null?void 0:e.codePointAt(0)}),new n({type:"InvalidName"},this.genTextPosFrom(s));if(h.length>0){if(!u((i=h[0])==null?void 0:i.codePointAt(0)))throw this.debug("Invalid local name start character",{local:h,firstChar:h[0],codePoint:(o=h[0])==null?void 0:o.codePointAt(0)}),new n({type:"InvalidName"},this.genTextPosFrom(s))}else throw this.debug("Empty local name - error"),new n({type:"InvalidName"},this.genTextPosFrom(s));return[r,h]}consumeQuote(){const t=this.currCodeUnit();if(t===T||t===A)return this._pos+=1,t;throw new n({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}getTextAround(t,e){const i=Math.floor(e/2),o=Math.max(0,t-i),s=Math.min(this._end,t+i+e%2);return this._text.slice(o,s)}genTextPos(){return this.genTextPosFrom(this._pos)}genTextPosFrom(t){const e=Math.min(t,this._end);let i=1,o=1;for(let s=0;s<e;s++)this._text.charCodeAt(s)===O?(i++,o=1):o++;return C({row:i,col:o},this.config.contextSliceSize?{contextSlice:this.getTextAround(t,this.config.contextSliceSize)}:{})}getTextSnippet(t,e){return this._text.slice(t,t+e)}}export{_ as XmlStream};

package/dist/esm/tokenizer/tokenize.js

@@ -1,2 +1 @@
-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};
+import{getQName as f}from"../get-q-name.js";import{LESS_THAN as l,GREATER_THAN as r,SLASH as h,EQUALS as W,SINGLE_QUOTE as y,DOUBLE_QUOTE as b,HYPHEN as B,QUESTION_MARK as v,OPEN_SQUARE_BRACKET as X,PERCENT as Y,EXCLAMATION_MARK as z,UPPERCASE_S as H,UPPERCASE_P as K,CLOSE_SQUARE_BRACKET as q}from"./ascii-constants.js";import{isXmlSpaceByte as G}from"./utils.js";import{XmlError as c}from"./XmlError.js";import{XmlStream as $}from"./XmlStream.js";const j="\uFEFF",I="<?xml ",A="?>",J="<!ENTITY",V="<!ELEMENT",Z="<!ATTLIST",tt="<!NOTATION",C="<!DOCTYPE",F="<?",N="?>",S="<!--",D="-->",nt="<![CDATA[",E="]]>",et="NDATA",L="PUBLIC",st="SYSTEM",ot="version",it="encoding",at="standalone";function ct(t,n,e={}){const s=new $(t,e);s.debug("Starting tokenization",{stringLength:t.length}),O(s,n)}function O(t,n){if(t.debug("Starting XML stream tokenization"),t.startsWith(j)&&(t.debug("Skipping BOM"),t.advance(1)),t.startsWith(I)&&(t.debug("Found XML declaration"),rt(t)),g(t,n),t.config.strictDocument){if(t.debug("Parsing in strict document mode"),t.skipSpaces(),t.startsWithIgnoreCase(C)&&(t.debug("Found DOCTYPE declaration"),P(t,n),g(t,n)),t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===l&&(t.debug("Starting root element parsing"),T(t,n)),g(t,n),!t.atEnd())throw t.debug("Unexpected content after root element",{position:t.getPos(),nextChars:t.getTextSnippet(t.getPos(),10)}),new c({type:"UnknownToken",message:"Not at end"},t.genTextPos())}else for(t.debug("Parsing in non-strict mode");!t.atEnd();)t.currCodeUnit()===l?T(t,n):R(t,n),g(t,n);t.debug("Finished XML stream tokenization")}function g(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith(S))U(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(C))P(t,n);else if(t.startsWith(F))k(t,n);else break}function rt(t){function n(e){if(e.startsWithSpace())e.skipSpaces();else if(!e.startsWith(A)&&!e.atEnd())throw new c({type:"InvalidChar",expected:"a whitespace",actual:e.currCodeUnitUnchecked()},e.genTextPos())}if(t.advance(5),n(t),!t.startsWith(ot))throw new c({type:"InvalidString",expected:"version"},t.genTextPos());m(t),n(t),t.startsWith(it)&&(m(t),n(t)),t.startsWith(at)&&m(t),t.skipSpaces(),t.skipString(A)}function U(t,n){const e=t.getPos();t.advance(4);const s=t.consumeCodeUnitsWhile((o,a)=>!(o===B&&a.startsWith(D)));if(t.skipString(D),s.includes("--")||s.endsWith("-"))throw new c({type:"InvalidComment"},t.genTextPosFrom(e));n({type:"Comment",text:s,range:t.rangeFrom(e)},t)}function k(t,n){if(t.startsWith(I))throw new c({type:"UnexpectedDeclaration"},t.genTextPos());const e=t.getPos();t.advance(2);const s=t.consumeName();t.skipSpaces();const o=t.consumeCodeUnitsWhile((a,i)=>!(a===v&&i.startsWith(N)));t.skipString(N),n({type:"ProcessingInstruction",target:s,content:o.length===0?void 0:o,range:t.rangeFrom(e)},t)}function P(t,n){if(!t.config.allowDtd)throw new c({type:"DtdDetected"});const e=t.getPos();if(dt(t),t.skipSpaces(),t.currCodeUnit()===r){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(J))ut(t,n);else if(t.startsWith(S))U(t,n);else if(t.startsWith(F))k(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===r){t.advance(1);break}else throw new c({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith(V)||t.startsWith(Z)||t.startsWith(tt))try{pt(t)}catch(s){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 dt(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),Q(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==X&&n!==r)throw new c({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function Q(t){if(t.startsWith(st)||t.startsWith(L)){const n=t.getPos();t.advance(6);const e=t.sliceBack(n);t.consumeSpaces();const s=t.consumeQuote();if(t.consumeCodeUnitsWhile(o=>o!==s),t.consumeCodeUnit(s),e===L){t.consumeSpaces();const o=t.consumeQuote();t.consumeCodeUnitsWhile(a=>a!==o),t.consumeCodeUnit(o)}return!0}return!1}function ut(t,n){t.advance(8),t.consumeSpaces();const e=!t.tryConsumeCodeUnit(Y);e||t.consumeSpaces();const s=t.consumeName();t.consumeSpaces();const o=lt(t,e);o!==null&&n({type:"EntityDeclaration",name:s,definition:o},t),t.skipSpaces(),t.consumeCodeUnit(r)}function lt(t,n){const e=t.currCodeUnit();if(e===b||e===y){const s=t.consumeQuote(),o=t.getPos();t.skipCodeUnitsWhile(i=>i!==s);const a=t.sliceBack(o);return t.consumeCodeUnit(s),a}else if(e===H||e===K){if(Q(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 pt(t){t.skipCodeUnitsWhile(n=>n!==r),t.consumeCodeUnit(r)}function T(t,n){const e=t.getPos();t.advance(1),t.debug("Starting element parsing",{position:e,nextChars:t.getTextSnippet(e,10)});const[s,o]=t.consumeQName();t.debug("Element name parsed",{prefix:s,local:o}),n({type:"ElementStart",prefix:s,local:o,start:e},t);let a=!1;for(;!t.atEnd();){const i=t.startsWithSpace();t.skipSpaces();const d=t.getPos(),p=t.currCodeUnit();if(t.debug("Processing element content",{position:d,currCodeUnit:p,hasSpace:i}),p===h){t.debug("Found self-closing tag"),t.advance(1),t.consumeCodeUnit(r);const u=t.rangeFrom(d);n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else if(p===r){t.advance(1);const u=t.rangeFrom(d);if(t.config.implicitSelfClosingElements!=null&&t.config.implicitSelfClosingElements.includes(f(o,s))){t.debug("Found implicit self-closing element",{element:f(o,s)}),n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else{t.debug("Found opening tag end"),n({type:"ElementEnd",end:{type:"Open"},range:u},t),a=!0;break}}else{if(!i)throw t.debug("Missing whitespace before attribute",{position:t.getPos(),currCodeUnit:String.fromCharCode(p)}),new c({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());t.debug("Parsing attribute");const[u,w,x]=m(t),M=t.getPos();t.debug("Attribute parsed",{prefix:u,local:w,value:x}),n({type:"Attribute",range:{start:d,end:M},prefix:u,local:w,value:x},t)}}if(a){if(t.config.rawTextElements!=null){const i=f(o,s);if(t.config.rawTextElements.includes(i)){t.debug("Processing raw text element content",{element:i}),_(t,n,`</${i}>`);return}}t.debug("Processing element content"),_(t,n)}t.debug("Finished element parsing",{prefix:s,local:o})}function m(t){const[n,e]=t.consumeQName();let s;const o=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(W)){t.skipSpaces();const a=t.currCodeUnit();if(a===y||a===b){const i=t.consumeQuote();s=t.consumeCodeUnitsWhile(d=>d!==i&&d!==l),t.consumeCodeUnit(i)}else if(!t.config.strictDocument)s=t.consumeCodeUnitsWhile(i=>!G(i)&&i!==r&&i!==h&&i!==l);else throw new c({type:"InvalidChar",expected:"a quote",actual:a},t.genTextPos())}else if(!t.config.strictDocument)t.goTo(o),s="true";else throw new c({type:"InvalidChar",expected:W,actual:t.currCodeUnit()},t.genTextPos());return[n,e,s]}function _(t,n,e=l){for(;!t.atEnd();)if(typeof e=="number"?t.currCodeUnit()===e:t.startsWith(e)){const s=t.nextCodeUnit();if(s===z)if(t.startsWith(S))U(t,n);else if(t.startsWith(nt))gt(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(C))P(t,n);else throw new c({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===v)k(t,n);else if(s===h){mt(t,n);break}else T(t,n)}else R(t,n,e)}function gt(t,n){const e=t.getPos();t.advance(9);const s=t.consumeCodeUnitsWhile((a,i)=>!(a===q&&i.startsWith(E)));t.skipString(E);const o=t.rangeFrom(e);n({type:"Cdata",text:s,range:o},t)}function mt(t,n){const e=t.getPos();t.advance(2);const[s,o]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(r);const a=t.rangeFrom(e);n({type:"ElementEnd",end:{type:"Close",prefix:s,local:o},range:a},t)}function R(t,n,e=l){const s=t.getPos(),o=typeof e=="number"?t.consumeCodeUnitsWhile(a=>a!==e):t.consumeCodeUnitsWhile((a,i)=>!i.startsWith(e));if(t.config.strictDocument&&o.includes(E))throw new c({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:o,range:t.rangeFrom(s)},t)}export{ct as tokenize,O as tokenizeXmlStream};

package/dist/types/tokenizer/XmlStream.d.ts

@@ -12,19 +12,16 @@ export declare class XmlStream {
     private _end;
     readonly config: TXmlStreamConfig;
     constructor(text: string, options?: TXmlStreamOptions);
+    /**
+     * Public method to log debug information if a logger is configured
+     */
+    debug(message: unknown, ...args: unknown[]): void;
     /**
      * Creates a clone of the current stream.
      *
      * @returns A new instance of XmlStream with the same state.
      */
     clone(): XmlStream;
-    /**
-     * Logs debug information when debug logging is enabled.
-     *
-     * @param message - The debug message.
-     * @param data - Additional debug data.
-     */
-    debugLog(message: string, data?: any): void;
     /**
      * Gets the current position in the stream.
      *
@@ -218,6 +215,10 @@ export declare class XmlStream {
      * @returns The calculated text position.
      */
     genTextPosFrom(pos: number): TTextPos;
+    /**
+     * Gets a snippet of text from the current position
+     */
+    getTextSnippet(start: number, length: number): string;
 }
 export interface TXmlStreamConfig {
     /**
@@ -261,11 +262,11 @@ export interface TXmlStreamConfig {
      */
     contextSliceSize: number | false;
     /**
-     * Enable debug logging for platform-specific issue diagnosis.
-     * When enabled, detailed parsing information will be logged to console.
-     * @default false
+     * Optional debug logger function that will be called with debug information.
+     * This allows users to capture and handle debug output in their own way.
+     * @default undefined
      */
-    debug?: boolean;
+    debugLogger?: (message: unknown, ...args: unknown[]) => void;
 }
 export type TXmlStreamOptions = {
     pos?: number;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "xml-tokenizer",
-  "version": "0.0.35",
+  "version": "0.0.37",
   "private": false,
   "description": "Straightforward and typesafe XML tokenizer that streams tokens through a callback mechanism",
   "keywords": [],