xml-tokenizer 0.0.3 → 0.0.5

package/README.md

@@ -82,6 +82,13 @@ The following token types are supported:
 - **Text**: Text content between elements, including whitespace.
 - **Cdata**: `<![CDATA[text]]>`
 
+## 👀 Differences from [XML 1.0 Specification](https://www.w3.org/TR/xml/)
+
+- **Attribute Value Handling:**
+  - **XML 1.0:** Attributes must be explicitly assigned a value in the format `Name="Value"`. An attribute without a value is not valid XML.
+  - **Parser Behavior:** Attributes without an explicit value are interpreted as `true` (e.g., `<element attribute/>` is parsed as `attribute="true"`). 
+  - **Reason**: This behavior aligns with HTML-style parsing, which was necessary to handle HTML attributes without explicit values.
+
 ## 🚀 Benchmark
 
 The performance of `xml-tokenizer` was benchmarked against other popular XML parsers. These tests focus on XML to object conversion and node counting. Interestingly, the version of `xml-tokenizer` imported directly from npm performed significantly better. The reason for this discrepancy is unclear, but the results seem accurate based on external testing. 
@@ -128,7 +135,7 @@ The `roxmltree` package with the Rust implementation can be found in the `_depre
 | roxmltree:wasmMix   | 28.17                           | 34.83         | 36.71         | 35.49          | ±0.91%                         | 
 | roxmltree:wasm      | 109.30                          | 8.30          | 13.16         | 9.15           | ±3.31%                         | 
 
-### Why Port `tokenizer.rs` to TypeScript?
+### Why ported `tokenizer.rs` to TypeScript?
 
 We ported [`tokenizer.rs`](https://github.com/RazrFalcon/roxmltree/blob/master/src/tokenizer.rs) to TypeScript because frequent communication between Rust and TypeScript negated Rust's performance benefits. The stream architecture required constant interaction between Rust and TypeScript via the `tokenCallback`, reducing overall efficiency.
 

package/dist/cjs/tokenizer/XmlStream.js

@@ -1 +1 @@
-"use strict";var s=require("./utils.js"),o=require("./XmlError.js"),u=Object.defineProperty,m=(h,t,e)=>t in h?u(h,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):h[t]=e,l=(h,t,e)=>m(h,typeof t!="symbol"?t+"":t,e);class c{constructor(t,e=0){l(this,"_text"),l(this,"_pos"),l(this,"_end"),this._text=t,this._pos=e,this._end=this._text.length}clone(){return new c(this._text,this._pos)}getPos(){return this._pos}atEnd(){return this._pos>=this._end}currCodeUnit(){if(this._pos>=this._end)throw new o.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 o.XmlError({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}advance(t){this._pos+=t}startsWith(t){return this._text.startsWith(t,this._pos)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new o.XmlError({type:"InvalidChar",expected:t,actual:e},this.genTextPos());this._pos+=1}tryConsumeCodeUnit(t){return this.currCodeUnit()===t?(this._pos+=1,!0):!1}skipString(t){if(!this.startsWith(t))throw new o.XmlError({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked());)this._pos+=1}consumeCharsWhile(t){const e=this._pos;return this.skipCharsWhile(t),this.sliceBack(e)}skipCharsWhile(t){for(;this._pos<this._end;){const e=this._text[this._pos];if(e==null||!s.isXmlChar(e.codePointAt(0)))throw new o.XmlError({type:"NonXmlChar",char:e!=null?e:"\uFFFD"},this.genTextPos());if(t(this,e))this._pos+=e.length;else break}}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&&s.isXmlSpaceByte(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new o.XmlError({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new o.XmlError({type:"InvalidChar",expected:"a whitespace",actual:this.currCodeUnitUnchecked()},this.genTextPos());this.skipSpaces()}tryConsumeReference(){const t=this._pos,e=this.clone(),r=e.consumeReference();return r!=null?(this._pos+=e.getPos()-t,r):null}consumeReference(){if(!this.tryConsumeCodeUnit(s.AMPERSAND))return null;let t;if(this.tryConsumeCodeUnit(s.HASH)){let e,r;this.tryConsumeCodeUnit(s.LOWERCASE_X)?(e=this.consumeCodeUnitsWhile(i=>i>=s.ZERO&&i<=s.NINE||i>=s.UPPERCASE_A&&i<=s.UPPERCASE_F||i>=s.LOWERCASE_A&&i<=s.LOWERCASE_F),r=16):(e=this.consumeCodeUnitsWhile(i=>s.isAsciiDigit(i)),r=10);const n=parseInt(e,r);isNaN(n)||!s.isXmlChar(n)?t=null:t={type:"Char",value:String.fromCodePoint(n)}}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(s.SEMICOLON)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new o.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(!s.isXmlNameStart(e))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(t))}else if(!s.isXmlName(e))break;this._pos+=1}}consumeQName(){var t,e;const r=this._pos;let n=null;for(;this._pos<this._end;){const p=this.currCodeUnitUnchecked();if(p===s.COLON)if(n==null)n=this._pos,this._pos+=1;else throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));else if(s.isXmlName(p))this._pos+=1;else break}let i,a;if(n!=null?(i=this._text.slice(r,n),a=this.sliceBack(n+1)):(i="",a=this.sliceBack(r)),i.length>0&&!s.isXmlNameStart((t=i[0])==null?void 0:t.codePointAt(0)))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));if(a.length>0){if(!s.isXmlNameStart((e=a[0])==null?void 0:e.codePointAt(0)))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r))}else throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));return[i,a]}consumeEq(){this.skipSpaces(),this.consumeCodeUnit(s.EQUALS),this.skipSpaces()}consumeQuote(){const t=this.currCodeUnit();if(t===s.SINGLE_QUOTE||t===s.DOUBLE_QUOTE)return this._pos+=1,t;throw new o.XmlError({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}genTextPos(){return this.genTextPosFrom(this._pos)}genTextPosFrom(t){const e=Math.min(t,this._end);let r=1,n=1;for(let i=0;i<e;i++)this._text.charCodeAt(i)===s.LINE_FEED?(r++,n=1):n++;return{row:r,col:n}}}exports.XmlStream=c;
+"use strict";var s=require("./utils.js"),o=require("./XmlError.js"),u=Object.defineProperty,m=(h,t,e)=>t in h?u(h,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):h[t]=e,l=(h,t,e)=>m(h,typeof t!="symbol"?t+"":t,e);class c{constructor(t,e=0){l(this,"_text"),l(this,"_pos"),l(this,"_end"),this._text=t,this._pos=e,this._end=this._text.length}clone(){return new c(this._text,this._pos)}getPos(){return this._pos}atEnd(){return this._pos>=this._end}currCodeUnit(){if(this._pos>=this._end)throw new o.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 o.XmlError({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}advance(t){this._pos+=t}goTo(t){this._pos=t}startsWith(t){return this._text.startsWith(t,this._pos)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new o.XmlError({type:"InvalidChar",expected:t,actual:e},this.genTextPos());this._pos+=1}tryConsumeCodeUnit(t){return this.currCodeUnit()===t?(this._pos+=1,!0):!1}skipString(t){if(!this.startsWith(t))throw new o.XmlError({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked());)this._pos+=1}consumeCharsWhile(t){const e=this._pos;return this.skipCharsWhile(t),this.sliceBack(e)}skipCharsWhile(t){for(;this._pos<this._end;){const e=this._text[this._pos];if(e==null||!s.isXmlChar(e.codePointAt(0)))throw new o.XmlError({type:"NonXmlChar",char:e!=null?e:"\uFFFD"},this.genTextPos());if(t(this,e))this._pos+=e.length;else break}}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&&s.isXmlSpaceByte(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new o.XmlError({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new o.XmlError({type:"InvalidChar",expected:"a whitespace",actual:this.currCodeUnitUnchecked()},this.genTextPos());this.skipSpaces()}tryConsumeReference(){const t=this._pos,e=this.clone(),r=e.consumeReference();return r!=null?(this._pos+=e.getPos()-t,r):null}consumeReference(){if(!this.tryConsumeCodeUnit(s.AMPERSAND))return null;let t;if(this.tryConsumeCodeUnit(s.HASH)){let e,r;this.tryConsumeCodeUnit(s.LOWERCASE_X)?(e=this.consumeCodeUnitsWhile(i=>i>=s.ZERO&&i<=s.NINE||i>=s.UPPERCASE_A&&i<=s.UPPERCASE_F||i>=s.LOWERCASE_A&&i<=s.LOWERCASE_F),r=16):(e=this.consumeCodeUnitsWhile(i=>s.isAsciiDigit(i)),r=10);const n=parseInt(e,r);isNaN(n)||!s.isXmlChar(n)?t=null:t={type:"Char",value:String.fromCodePoint(n)}}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(s.SEMICOLON)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new o.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(!s.isXmlNameStart(e))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(t))}else if(!s.isXmlName(e))break;this._pos+=1}}consumeQName(){var t,e;const r=this._pos;let n=null;for(;this._pos<this._end;){const p=this.currCodeUnitUnchecked();if(p===s.COLON)if(n==null)n=this._pos,this._pos+=1;else throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));else if(s.isXmlName(p))this._pos+=1;else break}let i,a;if(n!=null?(i=this._text.slice(r,n),a=this.sliceBack(n+1)):(i="",a=this.sliceBack(r)),i.length>0&&!s.isXmlNameStart((t=i[0])==null?void 0:t.codePointAt(0)))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));if(a.length>0){if(!s.isXmlNameStart((e=a[0])==null?void 0:e.codePointAt(0)))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r))}else throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));return[i,a]}consumeQuote(){const t=this.currCodeUnit();if(t===s.SINGLE_QUOTE||t===s.DOUBLE_QUOTE)return this._pos+=1,t;throw new o.XmlError({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}genTextPos(){return this.genTextPosFrom(this._pos)}genTextPosFrom(t){const e=Math.min(t,this._end);let r=1,n=1;for(let i=0;i<e;i++)this._text.charCodeAt(i)===s.LINE_FEED?(r++,n=1):n++;return{row:r,col:n}}}exports.XmlStream=c;

package/dist/cjs/tokenizer/tokenizer.js

@@ -1 +1 @@
-"use strict";var r=require("./utils.js"),i=require("./XmlError.js");function U(t,n,s){if(t.startsWith("\uFEFF")&&t.advance(1),t.startsWith("<?xml ")&&P(t),m(t,s),t.skipSpaces(),t.startsWith("<!DOCTYPE")){if(!n)throw new i.XmlError({type:"DtdDetected"});w(t,s),m(t,s)}if(t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===r.LESS_THAN&&g(t,s),m(t,s),!t.atEnd())throw new i.XmlError({type:"UnknownToken",message:"Not at end"},t.genTextPos())}function m(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith("<!--"))h(t,n);else if(t.startsWith("<?"))f(t,n);else break}function P(t){if(t.advance(5),E(t),!t.startsWith("version"))throw new i.XmlError({type:"InvalidString",expected:"version"},t.genTextPos());u(t),E(t),t.startsWith("encoding")&&(u(t),E(t)),t.startsWith("standalone")&&u(t),t.skipSpaces(),t.skipString("?>")}function E(t){if(t.startsWithSpace())t.skipSpaces();else if(!t.startsWith("?>")&&!t.atEnd())throw new i.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos())}function h(t,n){const s=t.getPos();t.advance(4);const o=t.consumeCharsWhile((e,a)=>!(a==="-"&&e.startsWith("-->")));if(t.skipString("-->"),o.includes("--")||o.endsWith("-"))throw new i.XmlError({type:"InvalidComment"},t.genTextPosFrom(s));n({type:"Comment",text:o,range:t.rangeFrom(s)})}function f(t,n){if(t.startsWith("<?xml "))throw new i.XmlError({type:"UnexpectedDeclaration"},t.genTextPos());const s=t.getPos();t.advance(2);const o=t.consumeName();t.skipSpaces();const e=t.consumeCharsWhile((a,c)=>!(c==="?"&&a.startsWith("?>")));t.skipString("?>"),n({type:"ProcessingInstruction",target:o,content:e.length===0?void 0:e,range:t.rangeFrom(s)})}function w(t,n){const s=t.getPos();if(A(t),t.skipSpaces(),t.currCodeUnit()===r.GREATER_THAN){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith("<!ENTITY"))W(t,n);else if(t.startsWith("<!--"))h(t,n);else if(t.startsWith("<?"))f(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===r.GREATER_THAN){t.advance(1);break}else throw new i.XmlError({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith("<!ELEMENT")||t.startsWith("<!ATTLIST")||t.startsWith("<!NOTATION"))try{y(t)}catch(o){throw new i.XmlError({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(s))}else throw new i.XmlError({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function A(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),T(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==r.OPEN_BRACKET&&n!==r.GREATER_THAN)throw new i.XmlError({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function T(t){if(t.startsWith("SYSTEM")||t.startsWith("PUBLIC")){const n=t.getPos();t.advance(6);const s=t.sliceBack(n);t.consumeSpaces();const o=t.consumeQuote();if(t.consumeCodeUnitsWhile(e=>e!==o),t.consumeCodeUnit(o),s==="PUBLIC"){t.consumeSpaces();const e=t.consumeQuote();t.consumeCodeUnitsWhile(a=>a!==e),t.consumeCodeUnit(e)}return!0}return!1}function W(t,n){t.advance(8),t.consumeSpaces();const s=!t.tryConsumeCodeUnit(r.PERCENT);s||t.consumeSpaces();const o=t.consumeName();t.consumeSpaces();const e=v(t,s);e!==null&&n({type:"EntityDeclaration",name:o,definition:e}),t.skipSpaces(),t.consumeCodeUnit(r.GREATER_THAN)}function v(t,n){const s=t.currCodeUnit();if(s===r.DOUBLE_QUOTE||s===r.SINGLE_QUOTE){const o=t.consumeQuote(),e=t.getPos();t.skipCodeUnitsWhile(c=>c!==o);const a=t.sliceBack(e);return t.consumeCodeUnit(o),a}else if(s===r.UPPERCASE_S||s===r.UPPERCASE_P){if(T(t))return n&&(t.skipSpaces(),t.startsWith("NDATA")&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new i.XmlError({type:"InvalidExternalID"},t.genTextPos())}else throw new i.XmlError({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:s},t.genTextPos())}function y(t){t.skipCodeUnitsWhile(n=>n!==r.GREATER_THAN),t.consumeCodeUnit(r.GREATER_THAN)}function g(t,n){const s=t.getPos();t.advance(1);const[o,e]=t.consumeQName();n({type:"ElementStart",prefix:o,local:e,start:s});let a=!1;for(;!t.atEnd();){const c=t.startsWithSpace();t.skipSpaces();const d=t.getPos(),l=t.currCodeUnit();if(l===r.SLASH){t.advance(1),t.consumeCodeUnit(r.GREATER_THAN);const p=t.rangeFrom(d);n({type:"ElementEnd",end:{type:"Empty"},range:p});break}else if(l===r.GREATER_THAN){t.advance(1);const p=t.rangeFrom(d);n({type:"ElementEnd",end:{type:"Open"},range:p}),a=!0;break}else{if(!c)throw new i.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());const[p,C,S]=u(t),k=t.getPos();n({type:"Attribute",range:{start:d,end:k},prefix:p,local:C,value:S})}}a&&x(t,n)}function u(t){const[n,s]=t.consumeQName();t.consumeEq();const o=t.consumeQuote(),e=String.fromCharCode(o),a=t.getPos();t.skipCharsWhile((d,l)=>l!==e&&l!=="<");const c=t.sliceBack(a);return t.consumeCodeUnit(o),[n,s,c]}function x(t,n){for(;!t.atEnd();)if(t.currCodeUnit()===r.LESS_THAN){const s=t.nextCodeUnit();if(s===r.EXCLAMATION_MARK)if(t.startsWith("<!--"))h(t,n);else if(t.startsWith("<![CDATA["))N(t,n);else throw new i.XmlError({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===r.QUESTION_MARK)f(t,n);else if(s===r.SLASH){R(t,n);break}else g(t,n)}else I(t,n)}function N(t,n){const s=t.getPos();t.advance(9);const o=t.consumeCharsWhile((a,c)=>!(c==="]"&&a.startsWith("]]>")));t.skipString("]]>");const e=t.rangeFrom(s);n({type:"Cdata",text:o,range:e})}function R(t,n){const s=t.getPos();t.advance(2);const[o,e]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(r.GREATER_THAN);const a=t.rangeFrom(s);n({type:"ElementEnd",end:{type:"Close",prefix:o,local:e},range:a})}function I(t,n){const s=t.getPos(),o=t.consumeCharsWhile((e,a)=>a!=="<");if(o.includes(">")&&o.includes("]]>"))throw new i.XmlError({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:o,range:t.rangeFrom(s)})}exports.parseXmlStream=U;
+"use strict";var r=require("./utils.js"),i=require("./XmlError.js");function U(t,n,s){if(t.startsWith("\uFEFF")&&t.advance(1),t.startsWith("<?xml ")&&P(t),m(t,s),t.skipSpaces(),t.startsWith("<!DOCTYPE")){if(!n)throw new i.XmlError({type:"DtdDetected"});w(t,s),m(t,s)}if(t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===r.LESS_THAN&&g(t,s),m(t,s),!t.atEnd())throw new i.XmlError({type:"UnknownToken",message:"Not at end"},t.genTextPos())}function m(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith("<!--"))h(t,n);else if(t.startsWith("<?"))f(t,n);else break}function P(t){if(t.advance(5),E(t),!t.startsWith("version"))throw new i.XmlError({type:"InvalidString",expected:"version"},t.genTextPos());d(t),E(t),t.startsWith("encoding")&&(d(t),E(t)),t.startsWith("standalone")&&d(t),t.skipSpaces(),t.skipString("?>")}function E(t){if(t.startsWithSpace())t.skipSpaces();else if(!t.startsWith("?>")&&!t.atEnd())throw new i.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos())}function h(t,n){const s=t.getPos();t.advance(4);const o=t.consumeCharsWhile((e,a)=>!(a==="-"&&e.startsWith("-->")));if(t.skipString("-->"),o.includes("--")||o.endsWith("-"))throw new i.XmlError({type:"InvalidComment"},t.genTextPosFrom(s));n({type:"Comment",text:o,range:t.rangeFrom(s)})}function f(t,n){if(t.startsWith("<?xml "))throw new i.XmlError({type:"UnexpectedDeclaration"},t.genTextPos());const s=t.getPos();t.advance(2);const o=t.consumeName();t.skipSpaces();const e=t.consumeCharsWhile((a,c)=>!(c==="?"&&a.startsWith("?>")));t.skipString("?>"),n({type:"ProcessingInstruction",target:o,content:e.length===0?void 0:e,range:t.rangeFrom(s)})}function w(t,n){const s=t.getPos();if(A(t),t.skipSpaces(),t.currCodeUnit()===r.GREATER_THAN){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith("<!ENTITY"))W(t,n);else if(t.startsWith("<!--"))h(t,n);else if(t.startsWith("<?"))f(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===r.GREATER_THAN){t.advance(1);break}else throw new i.XmlError({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith("<!ELEMENT")||t.startsWith("<!ATTLIST")||t.startsWith("<!NOTATION"))try{v(t)}catch(o){throw new i.XmlError({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(s))}else throw new i.XmlError({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function A(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),T(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==r.OPEN_BRACKET&&n!==r.GREATER_THAN)throw new i.XmlError({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function T(t){if(t.startsWith("SYSTEM")||t.startsWith("PUBLIC")){const n=t.getPos();t.advance(6);const s=t.sliceBack(n);t.consumeSpaces();const o=t.consumeQuote();if(t.consumeCodeUnitsWhile(e=>e!==o),t.consumeCodeUnit(o),s==="PUBLIC"){t.consumeSpaces();const e=t.consumeQuote();t.consumeCodeUnitsWhile(a=>a!==e),t.consumeCodeUnit(e)}return!0}return!1}function W(t,n){t.advance(8),t.consumeSpaces();const s=!t.tryConsumeCodeUnit(r.PERCENT);s||t.consumeSpaces();const o=t.consumeName();t.consumeSpaces();const e=y(t,s);e!==null&&n({type:"EntityDeclaration",name:o,definition:e}),t.skipSpaces(),t.consumeCodeUnit(r.GREATER_THAN)}function y(t,n){const s=t.currCodeUnit();if(s===r.DOUBLE_QUOTE||s===r.SINGLE_QUOTE){const o=t.consumeQuote(),e=t.getPos();t.skipCodeUnitsWhile(c=>c!==o);const a=t.sliceBack(e);return t.consumeCodeUnit(o),a}else if(s===r.UPPERCASE_S||s===r.UPPERCASE_P){if(T(t))return n&&(t.skipSpaces(),t.startsWith("NDATA")&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new i.XmlError({type:"InvalidExternalID"},t.genTextPos())}else throw new i.XmlError({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:s},t.genTextPos())}function v(t){t.skipCodeUnitsWhile(n=>n!==r.GREATER_THAN),t.consumeCodeUnit(r.GREATER_THAN)}function g(t,n){const s=t.getPos();t.advance(1);const[o,e]=t.consumeQName();n({type:"ElementStart",prefix:o,local:e,start:s});let a=!1;for(;!t.atEnd();){const c=t.startsWithSpace();t.skipSpaces();const l=t.getPos(),u=t.currCodeUnit();if(u===r.SLASH){t.advance(1),t.consumeCodeUnit(r.GREATER_THAN);const p=t.rangeFrom(l);n({type:"ElementEnd",end:{type:"Empty"},range:p});break}else if(u===r.GREATER_THAN){t.advance(1);const p=t.rangeFrom(l);n({type:"ElementEnd",end:{type:"Open"},range:p}),a=!0;break}else{if(!c)throw new i.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());const[p,C,S]=d(t),k=t.getPos();n({type:"Attribute",range:{start:l,end:k},prefix:p,local:C,value:S})}}a&&x(t,n)}function d(t){const[n,s]=t.consumeQName();let o;const e=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(r.EQUALS)){t.skipSpaces();const a=t.consumeQuote(),c=String.fromCharCode(a),l=t.getPos();t.skipCharsWhile((u,p)=>p!==c&&p!=="<"),o=t.sliceBack(l),t.consumeCodeUnit(a)}else t.goTo(e),o="true";return[n,s,o]}function x(t,n){for(;!t.atEnd();)if(t.currCodeUnit()===r.LESS_THAN){const s=t.nextCodeUnit();if(s===r.EXCLAMATION_MARK)if(t.startsWith("<!--"))h(t,n);else if(t.startsWith("<![CDATA["))N(t,n);else throw new i.XmlError({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===r.QUESTION_MARK)f(t,n);else if(s===r.SLASH){R(t,n);break}else g(t,n)}else I(t,n)}function N(t,n){const s=t.getPos();t.advance(9);const o=t.consumeCharsWhile((a,c)=>!(c==="]"&&a.startsWith("]]>")));t.skipString("]]>");const e=t.rangeFrom(s);n({type:"Cdata",text:o,range:e})}function R(t,n){const s=t.getPos();t.advance(2);const[o,e]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(r.GREATER_THAN);const a=t.rangeFrom(s);n({type:"ElementEnd",end:{type:"Close",prefix:o,local:e},range:a})}function I(t,n){const s=t.getPos(),o=t.consumeCharsWhile((e,a)=>a!=="<");if(o.includes(">")&&o.includes("]]>"))throw new i.XmlError({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:o,range:t.rangeFrom(s)})}exports.parseXmlStream=U;

package/dist/esm/tokenizer/XmlStream.js

@@ -1 +1 @@
-import{isXmlChar as u,isXmlSpaceByte as _,AMPERSAND as m,HASH as C,LOWERCASE_X as f,ZERO as U,NINE as x,UPPERCASE_A as w,UPPERCASE_F as y,LOWERCASE_A as k,LOWERCASE_F as g,isAsciiDigit as E,SEMICOLON as P,isXmlNameStart as a,isXmlName as d,COLON as S,EQUALS as v,SINGLE_QUOTE as N,DOUBLE_QUOTE as A,LINE_FEED as T}from"./utils.js";import{XmlError as n}from"./XmlError.js";var W=Object.defineProperty,F=(h,t,e)=>t in h?W(h,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):h[t]=e,c=(h,t,e)=>F(h,typeof t!="symbol"?t+"":t,e);class p{constructor(t,e=0){c(this,"_text"),c(this,"_pos"),c(this,"_end"),this._text=t,this._pos=e,this._end=this._text.length}clone(){return new p(this._text,this._pos)}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)}advance(t){this._pos+=t}startsWith(t){return this._text.startsWith(t,this._pos)}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}skipString(t){if(!this.startsWith(t))throw new n({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked());)this._pos+=1}consumeCharsWhile(t){const e=this._pos;return this.skipCharsWhile(t),this.sliceBack(e)}skipCharsWhile(t){for(;this._pos<this._end;){const e=this._text[this._pos];if(e==null||!u(e.codePointAt(0)))throw new n({type:"NonXmlChar",char:e!=null?e:"\uFFFD"},this.genTextPos());if(t(this,e))this._pos+=e.length;else break}}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&&_(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(m))return null;let t;if(this.tryConsumeCodeUnit(C)){let e,i;this.tryConsumeCodeUnit(f)?(e=this.consumeCodeUnitsWhile(s=>s>=U&&s<=x||s>=w&&s<=y||s>=k&&s<=g),i=16):(e=this.consumeCodeUnitsWhile(s=>E(s)),i=10);const o=parseInt(e,i);isNaN(o)||!u(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(P)?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(!a(e))throw new n({type:"InvalidName"},this.genTextPosFrom(t))}else if(!d(e))break;this._pos+=1}}consumeQName(){var t,e;const i=this._pos;let o=null;for(;this._pos<this._end;){const l=this.currCodeUnitUnchecked();if(l===S)if(o==null)o=this._pos,this._pos+=1;else throw new n({type:"InvalidName"},this.genTextPosFrom(i));else if(d(l))this._pos+=1;else break}let s,r;if(o!=null?(s=this._text.slice(i,o),r=this.sliceBack(o+1)):(s="",r=this.sliceBack(i)),s.length>0&&!a((t=s[0])==null?void 0:t.codePointAt(0)))throw new n({type:"InvalidName"},this.genTextPosFrom(i));if(r.length>0){if(!a((e=r[0])==null?void 0:e.codePointAt(0)))throw new n({type:"InvalidName"},this.genTextPosFrom(i))}else throw new n({type:"InvalidName"},this.genTextPosFrom(i));return[s,r]}consumeEq(){this.skipSpaces(),this.consumeCodeUnit(v),this.skipSpaces()}consumeQuote(){const t=this.currCodeUnit();if(t===N||t===A)return this._pos+=1,t;throw new n({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}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)===T?(i++,o=1):o++;return{row:i,col:o}}}export{p as XmlStream};
+import{isXmlChar as u,isXmlSpaceByte as _,AMPERSAND as m,HASH as C,LOWERCASE_X as f,ZERO as x,NINE as U,UPPERCASE_A as w,UPPERCASE_F as y,LOWERCASE_A as g,LOWERCASE_F as k,isAsciiDigit as E,SEMICOLON as P,isXmlNameStart as a,isXmlName as d,COLON as v,SINGLE_QUOTE as N,DOUBLE_QUOTE as S,LINE_FEED as A}from"./utils.js";import{XmlError as n}from"./XmlError.js";var T=Object.defineProperty,W=(h,t,e)=>t in h?T(h,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):h[t]=e,c=(h,t,e)=>W(h,typeof t!="symbol"?t+"":t,e);class l{constructor(t,e=0){c(this,"_text"),c(this,"_pos"),c(this,"_end"),this._text=t,this._pos=e,this._end=this._text.length}clone(){return new l(this._text,this._pos)}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)}advance(t){this._pos+=t}goTo(t){this._pos=t}startsWith(t){return this._text.startsWith(t,this._pos)}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}skipString(t){if(!this.startsWith(t))throw new n({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked());)this._pos+=1}consumeCharsWhile(t){const e=this._pos;return this.skipCharsWhile(t),this.sliceBack(e)}skipCharsWhile(t){for(;this._pos<this._end;){const e=this._text[this._pos];if(e==null||!u(e.codePointAt(0)))throw new n({type:"NonXmlChar",char:e!=null?e:"\uFFFD"},this.genTextPos());if(t(this,e))this._pos+=e.length;else break}}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&&_(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(m))return null;let t;if(this.tryConsumeCodeUnit(C)){let e,i;this.tryConsumeCodeUnit(f)?(e=this.consumeCodeUnitsWhile(s=>s>=x&&s<=U||s>=w&&s<=y||s>=g&&s<=k),i=16):(e=this.consumeCodeUnitsWhile(s=>E(s)),i=10);const o=parseInt(e,i);isNaN(o)||!u(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(P)?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(!a(e))throw new n({type:"InvalidName"},this.genTextPosFrom(t))}else if(!d(e))break;this._pos+=1}}consumeQName(){var t,e;const i=this._pos;let o=null;for(;this._pos<this._end;){const p=this.currCodeUnitUnchecked();if(p===v)if(o==null)o=this._pos,this._pos+=1;else throw new n({type:"InvalidName"},this.genTextPosFrom(i));else if(d(p))this._pos+=1;else break}let s,r;if(o!=null?(s=this._text.slice(i,o),r=this.sliceBack(o+1)):(s="",r=this.sliceBack(i)),s.length>0&&!a((t=s[0])==null?void 0:t.codePointAt(0)))throw new n({type:"InvalidName"},this.genTextPosFrom(i));if(r.length>0){if(!a((e=r[0])==null?void 0:e.codePointAt(0)))throw new n({type:"InvalidName"},this.genTextPosFrom(i))}else throw new n({type:"InvalidName"},this.genTextPosFrom(i));return[s,r]}consumeQuote(){const t=this.currCodeUnit();if(t===N||t===S)return this._pos+=1,t;throw new n({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}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)===A?(i++,o=1):o++;return{row:i,col:o}}}export{l as XmlStream};

package/dist/esm/tokenizer/tokenizer.js

@@ -1 +1 @@
-import{LESS_THAN as C,GREATER_THAN as r,OPEN_BRACKET as w,PERCENT as W,DOUBLE_QUOTE as x,SINGLE_QUOTE as y,UPPERCASE_S as v,UPPERCASE_P as I,SLASH as S,EXCLAMATION_MARK as N,QUESTION_MARK as A}from"./utils.js";import{XmlError as i}from"./XmlError.js";function F(t,n,s){if(t.startsWith("\uFEFF")&&t.advance(1),t.startsWith("<?xml ")&&D(t),m(t,s),t.skipSpaces(),t.startsWith("<!DOCTYPE")){if(!n)throw new i({type:"DtdDetected"});L(t,s),m(t,s)}if(t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===C&&k(t,s),m(t,s),!t.atEnd())throw new i({type:"UnknownToken",message:"Not at end"},t.genTextPos())}function m(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith("<!--"))f(t,n);else if(t.startsWith("<?"))g(t,n);else break}function D(t){if(t.advance(5),h(t),!t.startsWith("version"))throw new i({type:"InvalidString",expected:"version"},t.genTextPos());l(t),h(t),t.startsWith("encoding")&&(l(t),h(t)),t.startsWith("standalone")&&l(t),t.skipSpaces(),t.skipString("?>")}function h(t){if(t.startsWithSpace())t.skipSpaces();else if(!t.startsWith("?>")&&!t.atEnd())throw new i({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos())}function f(t,n){const s=t.getPos();t.advance(4);const o=t.consumeCharsWhile((e,a)=>!(a==="-"&&e.startsWith("-->")));if(t.skipString("-->"),o.includes("--")||o.endsWith("-"))throw new i({type:"InvalidComment"},t.genTextPosFrom(s));n({type:"Comment",text:o,range:t.rangeFrom(s)})}function g(t,n){if(t.startsWith("<?xml "))throw new i({type:"UnexpectedDeclaration"},t.genTextPos());const s=t.getPos();t.advance(2);const o=t.consumeName();t.skipSpaces();const e=t.consumeCharsWhile((a,c)=>!(c==="?"&&a.startsWith("?>")));t.skipString("?>"),n({type:"ProcessingInstruction",target:o,content:e.length===0?void 0:e,range:t.rangeFrom(s)})}function L(t,n){const s=t.getPos();if(O(t),t.skipSpaces(),t.currCodeUnit()===r){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith("<!ENTITY"))Q(t,n);else if(t.startsWith("<!--"))f(t,n);else if(t.startsWith("<?"))g(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===r){t.advance(1);break}else throw new i({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith("<!ELEMENT")||t.startsWith("<!ATTLIST")||t.startsWith("<!NOTATION"))try{B(t)}catch(o){throw new i({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(s))}else throw new i({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function O(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),E(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==w&&n!==r)throw new i({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function E(t){if(t.startsWith("SYSTEM")||t.startsWith("PUBLIC")){const n=t.getPos();t.advance(6);const s=t.sliceBack(n);t.consumeSpaces();const o=t.consumeQuote();if(t.consumeCodeUnitsWhile(e=>e!==o),t.consumeCodeUnit(o),s==="PUBLIC"){t.consumeSpaces();const e=t.consumeQuote();t.consumeCodeUnitsWhile(a=>a!==e),t.consumeCodeUnit(e)}return!0}return!1}function Q(t,n){t.advance(8),t.consumeSpaces();const s=!t.tryConsumeCodeUnit(W);s||t.consumeSpaces();const o=t.consumeName();t.consumeSpaces();const e=_(t,s);e!==null&&n({type:"EntityDeclaration",name:o,definition:e}),t.skipSpaces(),t.consumeCodeUnit(r)}function _(t,n){const s=t.currCodeUnit();if(s===x||s===y){const o=t.consumeQuote(),e=t.getPos();t.skipCodeUnitsWhile(c=>c!==o);const a=t.sliceBack(e);return t.consumeCodeUnit(o),a}else if(s===v||s===I){if(E(t))return n&&(t.skipSpaces(),t.startsWith("NDATA")&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new i({type:"InvalidExternalID"},t.genTextPos())}else throw new i({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:s},t.genTextPos())}function B(t){t.skipCodeUnitsWhile(n=>n!==r),t.consumeCodeUnit(r)}function k(t,n){const s=t.getPos();t.advance(1);const[o,e]=t.consumeQName();n({type:"ElementStart",prefix:o,local:e,start:s});let a=!1;for(;!t.atEnd();){const c=t.startsWithSpace();t.skipSpaces();const u=t.getPos(),p=t.currCodeUnit();if(p===S){t.advance(1),t.consumeCodeUnit(r);const d=t.rangeFrom(u);n({type:"ElementEnd",end:{type:"Empty"},range:d});break}else if(p===r){t.advance(1);const d=t.rangeFrom(u);n({type:"ElementEnd",end:{type:"Open"},range:d}),a=!0;break}else{if(!c)throw new i({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());const[d,U,T]=l(t),P=t.getPos();n({type:"Attribute",range:{start:u,end:P},prefix:d,local:U,value:T})}}a&&R(t,n)}function l(t){const[n,s]=t.consumeQName();t.consumeEq();const o=t.consumeQuote(),e=String.fromCharCode(o),a=t.getPos();t.skipCharsWhile((u,p)=>p!==e&&p!=="<");const c=t.sliceBack(a);return t.consumeCodeUnit(o),[n,s,c]}function R(t,n){for(;!t.atEnd();)if(t.currCodeUnit()===C){const s=t.nextCodeUnit();if(s===N)if(t.startsWith("<!--"))f(t,n);else if(t.startsWith("<![CDATA["))b(t,n);else throw new i({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===A)g(t,n);else if(s===S){M(t,n);break}else k(t,n)}else Y(t,n)}function b(t,n){const s=t.getPos();t.advance(9);const o=t.consumeCharsWhile((a,c)=>!(c==="]"&&a.startsWith("]]>")));t.skipString("]]>");const e=t.rangeFrom(s);n({type:"Cdata",text:o,range:e})}function M(t,n){const s=t.getPos();t.advance(2);const[o,e]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(r);const a=t.rangeFrom(s);n({type:"ElementEnd",end:{type:"Close",prefix:o,local:e},range:a})}function Y(t,n){const s=t.getPos(),o=t.consumeCharsWhile((e,a)=>a!=="<");if(o.includes(">")&&o.includes("]]>"))throw new i({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:o,range:t.rangeFrom(s)})}export{F as parseXmlStream};
+import{LESS_THAN as C,GREATER_THAN as r,OPEN_BRACKET as w,PERCENT as W,DOUBLE_QUOTE as y,SINGLE_QUOTE as x,UPPERCASE_S as v,UPPERCASE_P as I,SLASH as S,EQUALS as N,EXCLAMATION_MARK as A,QUESTION_MARK as F}from"./utils.js";import{XmlError as i}from"./XmlError.js";function L(t,n,s){if(t.startsWith("\uFEFF")&&t.advance(1),t.startsWith("<?xml ")&&Q(t),m(t,s),t.skipSpaces(),t.startsWith("<!DOCTYPE")){if(!n)throw new i({type:"DtdDetected"});D(t,s),m(t,s)}if(t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===C&&E(t,s),m(t,s),!t.atEnd())throw new i({type:"UnknownToken",message:"Not at end"},t.genTextPos())}function m(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith("<!--"))f(t,n);else if(t.startsWith("<?"))g(t,n);else break}function Q(t){if(t.advance(5),h(t),!t.startsWith("version"))throw new i({type:"InvalidString",expected:"version"},t.genTextPos());u(t),h(t),t.startsWith("encoding")&&(u(t),h(t)),t.startsWith("standalone")&&u(t),t.skipSpaces(),t.skipString("?>")}function h(t){if(t.startsWithSpace())t.skipSpaces();else if(!t.startsWith("?>")&&!t.atEnd())throw new i({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos())}function f(t,n){const s=t.getPos();t.advance(4);const o=t.consumeCharsWhile((e,a)=>!(a==="-"&&e.startsWith("-->")));if(t.skipString("-->"),o.includes("--")||o.endsWith("-"))throw new i({type:"InvalidComment"},t.genTextPosFrom(s));n({type:"Comment",text:o,range:t.rangeFrom(s)})}function g(t,n){if(t.startsWith("<?xml "))throw new i({type:"UnexpectedDeclaration"},t.genTextPos());const s=t.getPos();t.advance(2);const o=t.consumeName();t.skipSpaces();const e=t.consumeCharsWhile((a,c)=>!(c==="?"&&a.startsWith("?>")));t.skipString("?>"),n({type:"ProcessingInstruction",target:o,content:e.length===0?void 0:e,range:t.rangeFrom(s)})}function D(t,n){const s=t.getPos();if(O(t),t.skipSpaces(),t.currCodeUnit()===r){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith("<!ENTITY"))_(t,n);else if(t.startsWith("<!--"))f(t,n);else if(t.startsWith("<?"))g(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===r){t.advance(1);break}else throw new i({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith("<!ELEMENT")||t.startsWith("<!ATTLIST")||t.startsWith("<!NOTATION"))try{R(t)}catch(o){throw new i({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(s))}else throw new i({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function O(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),k(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==w&&n!==r)throw new i({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function k(t){if(t.startsWith("SYSTEM")||t.startsWith("PUBLIC")){const n=t.getPos();t.advance(6);const s=t.sliceBack(n);t.consumeSpaces();const o=t.consumeQuote();if(t.consumeCodeUnitsWhile(e=>e!==o),t.consumeCodeUnit(o),s==="PUBLIC"){t.consumeSpaces();const e=t.consumeQuote();t.consumeCodeUnitsWhile(a=>a!==e),t.consumeCodeUnit(e)}return!0}return!1}function _(t,n){t.advance(8),t.consumeSpaces();const s=!t.tryConsumeCodeUnit(W);s||t.consumeSpaces();const o=t.consumeName();t.consumeSpaces();const e=B(t,s);e!==null&&n({type:"EntityDeclaration",name:o,definition:e}),t.skipSpaces(),t.consumeCodeUnit(r)}function B(t,n){const s=t.currCodeUnit();if(s===y||s===x){const o=t.consumeQuote(),e=t.getPos();t.skipCodeUnitsWhile(c=>c!==o);const a=t.sliceBack(e);return t.consumeCodeUnit(o),a}else if(s===v||s===I){if(k(t))return n&&(t.skipSpaces(),t.startsWith("NDATA")&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new i({type:"InvalidExternalID"},t.genTextPos())}else throw new i({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:s},t.genTextPos())}function R(t){t.skipCodeUnitsWhile(n=>n!==r),t.consumeCodeUnit(r)}function E(t,n){const s=t.getPos();t.advance(1);const[o,e]=t.consumeQName();n({type:"ElementStart",prefix:o,local:e,start:s});let a=!1;for(;!t.atEnd();){const c=t.startsWithSpace();t.skipSpaces();const d=t.getPos(),l=t.currCodeUnit();if(l===S){t.advance(1),t.consumeCodeUnit(r);const p=t.rangeFrom(d);n({type:"ElementEnd",end:{type:"Empty"},range:p});break}else if(l===r){t.advance(1);const p=t.rangeFrom(d);n({type:"ElementEnd",end:{type:"Open"},range:p}),a=!0;break}else{if(!c)throw new i({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());const[p,U,T]=u(t),P=t.getPos();n({type:"Attribute",range:{start:d,end:P},prefix:p,local:U,value:T})}}a&&b(t,n)}function u(t){const[n,s]=t.consumeQName();let o;const e=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(N)){t.skipSpaces();const a=t.consumeQuote(),c=String.fromCharCode(a),d=t.getPos();t.skipCharsWhile((l,p)=>p!==c&&p!=="<"),o=t.sliceBack(d),t.consumeCodeUnit(a)}else t.goTo(e),o="true";return[n,s,o]}function b(t,n){for(;!t.atEnd();)if(t.currCodeUnit()===C){const s=t.nextCodeUnit();if(s===A)if(t.startsWith("<!--"))f(t,n);else if(t.startsWith("<![CDATA["))M(t,n);else throw new i({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===F)g(t,n);else if(s===S){Y(t,n);break}else E(t,n)}else H(t,n)}function M(t,n){const s=t.getPos();t.advance(9);const o=t.consumeCharsWhile((a,c)=>!(c==="]"&&a.startsWith("]]>")));t.skipString("]]>");const e=t.rangeFrom(s);n({type:"Cdata",text:o,range:e})}function Y(t,n){const s=t.getPos();t.advance(2);const[o,e]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(r);const a=t.rangeFrom(s);n({type:"ElementEnd",end:{type:"Close",prefix:o,local:e},range:a})}function H(t,n){const s=t.getPos(),o=t.consumeCharsWhile((e,a)=>a!=="<");if(o.includes(">")&&o.includes("]]>"))throw new i({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:o,range:t.rangeFrom(s)})}export{L as parseXmlStream};

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

@@ -55,6 +55,12 @@ export declare class XmlStream {
      * @param n - The number of characters to advance.
      */
     advance(n: number): void;
+    /**
+     * Go to a new position in the stream.
+     *
+     *  @param pos - The new position.
+     */
+    goTo(pos: number): void;
     /**
      * Checks if the stream starts with the given text.
      *
@@ -178,10 +184,6 @@ export declare class XmlStream {
      * @throws XmlError if an invalid qualified XML name is encountered.
      */
     consumeQName(): [string, string];
-    /**
-     * Consumes an equal sign, surrounded by optional whitespace.
-     */
-    consumeEq(): void;
     /**
      * Consumes a quote character.
      *

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

@@ -1 +1 @@
-{"version":3,"file":"XmlStream.d.ts","sourceRoot":"","sources":["../../../src/tokenizer/XmlStream.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,MAAM,EAAE,KAAK,UAAU,EAAE,KAAK,QAAQ,EAAE,MAAM,SAAS,CAAC;AAyBtE;;;;;;GAMG;AACH,qBAAa,SAAS;IACrB,OAAO,CAAC,KAAK,CAAS;IACtB,OAAO,CAAC,IAAI,CAAS;IACrB,OAAO,CAAC,IAAI,CAAS;gBAEF,IAAI,EAAE,MAAM,EAAE,GAAG,SAAI;IAMxC;;;;OAIG;IACI,KAAK,IAAI,SAAS;IAIzB;;;;OAIG;IACI,MAAM,IAAI,MAAM;IAIvB;;;;OAIG;IACI,KAAK,IAAI,OAAO;IAIvB;;;;;OAKG;IACI,YAAY,IAAI,MAAM;IAO7B;;;;OAIG;IACI,qBAAqB,IAAI,MAAM;IAItC;;;;;OAKG;IACI,YAAY,IAAI,MAAM;IAO7B;;;;OAIG;IACI,OAAO,CAAC,CAAC,EAAE,MAAM,GAAG,IAAI;IAI/B;;;;;OAKG;IACI,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO;IAIxC;;;;;OAKG;IACI,eAAe,CAAC,QAAQ,EAAE,MAAM,GAAG,IAAI;IAW9C;;;;;;OAMG;IACI,kBAAkB,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO;IAQpD;;;;;OAKG;IACI,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI;IAOrC;;;;;OAKG;IACI,qBAAqB,CAAC,SAAS,EAAE,CAAC,QAAQ,EAAE,MAAM,KAAK,OAAO,GAAG,MAAM;IAM9E;;;;OAIG;IACI,kBAAkB,CAAC,SAAS,EAAE,CAAC,QAAQ,EAAE,MAAM,KAAK,OAAO,GAAG,IAAI;IAMzE;;;;;;OAMG;IACI,iBAAiB,CAAC,SAAS,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,KAAK,OAAO,GAAG,MAAM;IAMzF;;;;;OAKG;IACI,cAAc,CAAC,SAAS,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,KAAK,OAAO,GAAG,IAAI;IAapF;;;;;OAKG;IACI,SAAS,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM;IAIrC;;;;;OAKG;IACI,SAAS,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM;IAIvC;;OAEG;IACI,UAAU,IAAI,IAAI;IAMzB;;;;OAIG;IACI,eAAe,IAAI,OAAO;IAIjC;;;;;OAKG;IACI,aAAa,IAAI,IAAI;IAe5B;;;;;OAKG;IACI,mBAAmB,IAAI,UAAU,GAAG,IAAI;IAiB/C;;;;OAIG;IACI,gBAAgB,IAAI,UAAU,GAAG,IAAI;IA4D5C;;;;;;OAMG;IACI,WAAW,IAAI,MAAM;IAY5B;;;;OAIG;IACI,QAAQ,IAAI,IAAI;IAgBvB;;;;;;OAMG;IACI,YAAY,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC;IAiDvC;;OAEG;IACI,SAAS,IAAI,IAAI;IAMxB;;;;;OAKG;IACI,YAAY,IAAI,MAAM;IAY7B;;;;;OAKG;IACI,UAAU,IAAI,QAAQ;IAI7B;;;;;;OAMG;IACI,cAAc,CAAC,GAAG,EAAE,MAAM,GAAG,QAAQ;CAc5C"}
+{"version":3,"file":"XmlStream.d.ts","sourceRoot":"","sources":["../../../src/tokenizer/XmlStream.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,MAAM,EAAE,KAAK,UAAU,EAAE,KAAK,QAAQ,EAAE,MAAM,SAAS,CAAC;AAwBtE;;;;;;GAMG;AACH,qBAAa,SAAS;IACrB,OAAO,CAAC,KAAK,CAAS;IACtB,OAAO,CAAC,IAAI,CAAS;IACrB,OAAO,CAAC,IAAI,CAAS;gBAEF,IAAI,EAAE,MAAM,EAAE,GAAG,SAAI;IAMxC;;;;OAIG;IACI,KAAK,IAAI,SAAS;IAIzB;;;;OAIG;IACI,MAAM,IAAI,MAAM;IAIvB;;;;OAIG;IACI,KAAK,IAAI,OAAO;IAIvB;;;;;OAKG;IACI,YAAY,IAAI,MAAM;IAO7B;;;;OAIG;IACI,qBAAqB,IAAI,MAAM;IAItC;;;;;OAKG;IACI,YAAY,IAAI,MAAM;IAO7B;;;;OAIG;IACI,OAAO,CAAC,CAAC,EAAE,MAAM,GAAG,IAAI;IAI/B;;;;OAIG;IACI,IAAI,CAAC,GAAG,EAAE,MAAM,GAAG,IAAI;IAI9B;;;;;OAKG;IACI,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO;IAIxC;;;;;OAKG;IACI,eAAe,CAAC,QAAQ,EAAE,MAAM,GAAG,IAAI;IAW9C;;;;;;OAMG;IACI,kBAAkB,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO;IAQpD;;;;;OAKG;IACI,UAAU,CAAC,IAAI,EAAE,MAAM,GAAG,IAAI;IAOrC;;;;;OAKG;IACI,qBAAqB,CAAC,SAAS,EAAE,CAAC,QAAQ,EAAE,MAAM,KAAK,OAAO,GAAG,MAAM;IAM9E;;;;OAIG;IACI,kBAAkB,CAAC,SAAS,EAAE,CAAC,QAAQ,EAAE,MAAM,KAAK,OAAO,GAAG,IAAI;IAMzE;;;;;;OAMG;IACI,iBAAiB,CAAC,SAAS,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,KAAK,OAAO,GAAG,MAAM;IAMzF;;;;;OAKG;IACI,cAAc,CAAC,SAAS,EAAE,CAAC,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,KAAK,OAAO,GAAG,IAAI;IAapF;;;;;OAKG;IACI,SAAS,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM;IAIrC;;;;;OAKG;IACI,SAAS,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM;IAIvC;;OAEG;IACI,UAAU,IAAI,IAAI;IAMzB;;;;OAIG;IACI,eAAe,IAAI,OAAO;IAIjC;;;;;OAKG;IACI,aAAa,IAAI,IAAI;IAe5B;;;;;OAKG;IACI,mBAAmB,IAAI,UAAU,GAAG,IAAI;IAiB/C;;;;OAIG;IACI,gBAAgB,IAAI,UAAU,GAAG,IAAI;IA4D5C;;;;;;OAMG;IACI,WAAW,IAAI,MAAM;IAY5B;;;;OAIG;IACI,QAAQ,IAAI,IAAI;IAgBvB;;;;;;OAMG;IACI,YAAY,IAAI,CAAC,MAAM,EAAE,MAAM,CAAC;IAiDvC;;;;;OAKG;IACI,YAAY,IAAI,MAAM;IAY7B;;;;;OAKG;IACI,UAAU,IAAI,QAAQ;IAI7B;;;;;;OAMG;IACI,cAAc,CAAC,GAAG,EAAE,MAAM,GAAG,QAAQ;CAc5C"}

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

@@ -1 +1 @@
-{"version":3,"file":"tokenizer.d.ts","sourceRoot":"","sources":["../../../src/tokenizer/tokenizer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,cAAc,EAAE,MAAM,SAAS,CAAC;AAe9C,OAAO,EAAE,KAAK,SAAS,EAAE,MAAM,aAAa,CAAC;AAE7C;;;;;;GAMG;AACH,wBAAgB,cAAc,CAC7B,CAAC,EAAE,SAAS,EACZ,QAAQ,EAAE,OAAO,EACjB,aAAa,EAAE,cAAc,GAC3B,IAAI,CAgCN"}
+{"version":3,"file":"tokenizer.d.ts","sourceRoot":"","sources":["../../../src/tokenizer/tokenizer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,KAAK,cAAc,EAAE,MAAM,SAAS,CAAC;AAgB9C,OAAO,EAAE,KAAK,SAAS,EAAE,MAAM,aAAa,CAAC;AAE7C;;;;;;GAMG;AACH,wBAAgB,cAAc,CAC7B,CAAC,EAAE,SAAS,EACZ,QAAQ,EAAE,OAAO,EACjB,aAAa,EAAE,cAAc,GAC3B,IAAI,CAgCN"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "xml-tokenizer",
-  "version": "0.0.3",
+  "version": "0.0.5",
   "description": "Straightforward and typesafe XML tokenizer that streams tokens through a callback mechanism",
   "private": false,
   "source": "./src/index.ts",
@@ -26,9 +26,8 @@
     "sax": "^1.4.1",
     "saxen": "^10.0.0",
     "txml": "^5.1.1",
-    "xml-tokenizer": "^0.0.2",
     "xml2js": "^0.6.2",
-    "@blgc/config": "0.0.16"
+    "@blgc/config": "0.0.17"
   },
   "files": [
     "dist",