xml-tokenizer 0.0.22 → 0.0.23

package/dist/cjs/tokenizer/XmlStream.js

@@ -1 +1 @@
-"use strict";var n=require("./ascii-constants.js"),a=require("./utils.js"),o=require("./XmlError.js"),_=Object.defineProperty,d=Object.getOwnPropertySymbols,C=Object.prototype.hasOwnProperty,f=Object.prototype.propertyIsEnumerable,u=(h,t,e)=>t in h?_(h,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):h[t]=e,E=(h,t)=>{for(var e in t||(t={}))C.call(t,e)&&u(h,e,t[e]);if(d)for(var e of d(t))f.call(t,e)&&u(h,e,t[e]);return h},p=(h,t,e)=>u(h,typeof t!="symbol"?t+"":t,e);class m{constructor(t,e={}){p(this,"_text"),p(this,"_pos"),p(this,"_end"),p(this,"config");const{pos:r=0,strictDocument:i=!0,allowDtd:s=!0,allowBooleanAttributes:l=!1,rawTextElements:c=null}=e;this._text=t,this._pos=r,this._end=this._text.length,this.config={strictDocument:i,allowDtd:s,allowBooleanAttributes:l,rawTextElements:c}}clone(){return new m(this._text,E({pos:this._pos},this.config))}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)}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}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(!a.isXmlChar(this.currCodeUnitUnchecked()))throw new o.XmlError({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}startsWith(t){return this._text.startsWith(t,this._pos)}skipString(t){if(!this.startsWith(t))throw new o.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&&a.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(n.AMPERSAND))return null;let t;if(this.tryConsumeCodeUnit(n.HASH)){let e,r;this.tryConsumeCodeUnit(n.LOWERCASE_X)?(e=this.consumeCodeUnitsWhile(s=>s>=n.ZERO&&s<=n.NINE||s>=n.UPPERCASE_A&&s<=n.UPPERCASE_F||s>=n.LOWERCASE_A&&s<=n.LOWERCASE_F),r=16):(e=this.consumeCodeUnitsWhile(s=>a.isAsciiDigit(s)),r=10);const i=parseInt(e,r);isNaN(i)||!a.isXmlChar(i)?t=null:t={type:"Char",value:String.fromCodePoint(i)}}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(n.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(!a.isXmlNameStart(e))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(t))}else if(!a.isXmlName(e))break;this._pos+=1}}consumeQName(){var t,e;const r=this._pos;let i=null;for(;this._pos<this._end;){const c=this.currCodeUnitUnchecked();if(c===n.COLON)if(i==null)i=this._pos,this._pos+=1;else throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));else if(a.isXmlName(c))this._pos+=1;else break}let s,l;if(i!=null?(s=this._text.slice(r,i),l=this.sliceBack(i+1)):(s="",l=this.sliceBack(r)),s.length>0&&!a.isXmlNameStart((t=s[0])==null?void 0:t.codePointAt(0)))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));if(l.length>0){if(!a.isXmlNameStart((e=l[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[s,l]}consumeQuote(){const t=this.currCodeUnit();if(t===n.SINGLE_QUOTE||t===n.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,i=1;for(let s=0;s<e;s++)this._text.charCodeAt(s)===n.LINE_FEED?(r++,i=1):i++;return{row:r,col:i}}}exports.XmlStream=m;
+"use strict";var n=require("./ascii-constants.js"),a=require("./utils.js"),o=require("./XmlError.js"),_=Object.defineProperty,d=Object.getOwnPropertySymbols,C=Object.prototype.hasOwnProperty,f=Object.prototype.propertyIsEnumerable,p=(h,t,e)=>t in h?_(h,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):h[t]=e,E=(h,t)=>{for(var e in t||(t={}))C.call(t,e)&&p(h,e,t[e]);if(d)for(var e of d(t))f.call(t,e)&&p(h,e,t[e]);return h},l=(h,t,e)=>p(h,typeof t!="symbol"?t+"":t,e);class u{constructor(t,e={}){l(this,"_text"),l(this,"_pos"),l(this,"_end"),l(this,"config");const{pos:r=0,strictDocument:i=!0,allowDtd:s=!0,rawTextElements:c=null}=e;this._text=t,this._pos=r,this._end=this._text.length,this.config={strictDocument:i,allowDtd:s,rawTextElements:c}}clone(){return new u(this._text,E({pos:this._pos},this.config))}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)}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}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(!a.isXmlChar(this.currCodeUnitUnchecked()))throw new o.XmlError({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}startsWith(t){return this._text.startsWith(t,this._pos)}skipString(t){if(!this.startsWith(t))throw new o.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&&a.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(n.AMPERSAND))return null;let t;if(this.tryConsumeCodeUnit(n.HASH)){let e,r;this.tryConsumeCodeUnit(n.LOWERCASE_X)?(e=this.consumeCodeUnitsWhile(s=>s>=n.ZERO&&s<=n.NINE||s>=n.UPPERCASE_A&&s<=n.UPPERCASE_F||s>=n.LOWERCASE_A&&s<=n.LOWERCASE_F),r=16):(e=this.consumeCodeUnitsWhile(s=>a.isAsciiDigit(s)),r=10);const i=parseInt(e,r);isNaN(i)||!a.isXmlChar(i)?t=null:t={type:"Char",value:String.fromCodePoint(i)}}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(n.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(!a.isXmlNameStart(e))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(t))}else if(!a.isXmlName(e))break;this._pos+=1}}consumeQName(){var t,e;const r=this._pos;let i=null;for(;this._pos<this._end;){const m=this.currCodeUnitUnchecked();if(m===n.COLON)if(i==null)i=this._pos,this._pos+=1;else throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));else if(a.isXmlName(m))this._pos+=1;else break}let s,c;if(i!=null?(s=this._text.slice(r,i),c=this.sliceBack(i+1)):(s="",c=this.sliceBack(r)),s.length>0&&!a.isXmlNameStart((t=s[0])==null?void 0:t.codePointAt(0)))throw new o.XmlError({type:"InvalidName"},this.genTextPosFrom(r));if(c.length>0){if(!a.isXmlNameStart((e=c[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[s,c]}consumeQuote(){const t=this.currCodeUnit();if(t===n.SINGLE_QUOTE||t===n.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,i=1;for(let s=0;s<e;s++)this._text.charCodeAt(s)===n.LINE_FEED?(r++,i=1):i++;return{row:r,col:i}}}exports.XmlStream=u;

package/dist/cjs/tokenizer/tokenize.js

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

package/dist/esm/tokenizer/XmlStream.js

@@ -1 +1 @@
-import{AMPERSAND as C,HASH as f,LOWERCASE_X as w,ZERO as U,NINE as y,UPPERCASE_A as x,UPPERCASE_F as g,LOWERCASE_A as E,LOWERCASE_F as P,SEMICOLON as k,COLON as v,SINGLE_QUOTE as S,DOUBLE_QUOTE as N,LINE_FEED as O}from"./ascii-constants.js";import{isXmlChar as d,isXmlSpaceByte as T,isAsciiDigit as b,isXmlNameStart as l,isXmlName as _}from"./utils.js";import{XmlError as o}from"./XmlError.js";var A=Object.defineProperty,m=Object.getOwnPropertySymbols,I=Object.prototype.hasOwnProperty,W=Object.prototype.propertyIsEnumerable,p=(r,t,e)=>t in r?A(r,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):r[t]=e,F=(r,t)=>{for(var e in t||(t={}))I.call(t,e)&&p(r,e,t[e]);if(m)for(var e of m(t))W.call(t,e)&&p(r,e,t[e]);return r},c=(r,t,e)=>p(r,typeof t!="symbol"?t+"":t,e);class u{constructor(t,e={}){c(this,"_text"),c(this,"_pos"),c(this,"_end"),c(this,"config");const{pos:i=0,strictDocument:n=!0,allowDtd:s=!0,allowBooleanAttributes:h=!1,rawTextElements:a=null}=e;this._text=t,this._pos=i,this._end=this._text.length,this.config={strictDocument:n,allowDtd:s,allowBooleanAttributes:h,rawTextElements:a}}clone(){return new u(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 o({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({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new o({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(!d(this.currCodeUnitUnchecked()))throw new o({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}startsWith(t){return this._text.startsWith(t,this._pos)}skipString(t){if(!this.startsWith(t))throw new o({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&&T(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new o({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new o({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(C))return null;let t;if(this.tryConsumeCodeUnit(f)){let e,i;this.tryConsumeCodeUnit(w)?(e=this.consumeCodeUnitsWhile(s=>s>=U&&s<=y||s>=x&&s<=g||s>=E&&s<=P),i=16):(e=this.consumeCodeUnitsWhile(s=>b(s)),i=10);const n=parseInt(e,i);isNaN(n)||!d(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(k)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new o({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(!l(e))throw new o({type:"InvalidName"},this.genTextPosFrom(t))}else if(!_(e))break;this._pos+=1}}consumeQName(){var t,e;const i=this._pos;let n=null;for(;this._pos<this._end;){const a=this.currCodeUnitUnchecked();if(a===v)if(n==null)n=this._pos,this._pos+=1;else throw new o({type:"InvalidName"},this.genTextPosFrom(i));else if(_(a))this._pos+=1;else break}let s,h;if(n!=null?(s=this._text.slice(i,n),h=this.sliceBack(n+1)):(s="",h=this.sliceBack(i)),s.length>0&&!l((t=s[0])==null?void 0:t.codePointAt(0)))throw new o({type:"InvalidName"},this.genTextPosFrom(i));if(h.length>0){if(!l((e=h[0])==null?void 0:e.codePointAt(0)))throw new o({type:"InvalidName"},this.genTextPosFrom(i))}else throw new o({type:"InvalidName"},this.genTextPosFrom(i));return[s,h]}consumeQuote(){const t=this.currCodeUnit();if(t===S||t===N)return this._pos+=1,t;throw new o({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,n=1;for(let s=0;s<e;s++)this._text.charCodeAt(s)===O?(i++,n=1):n++;return{row:i,col:n}}}export{u as XmlStream};
+import{AMPERSAND as C,HASH as f,LOWERCASE_X as U,ZERO as w,NINE as y,UPPERCASE_A as x,UPPERCASE_F as g,LOWERCASE_A as E,LOWERCASE_F as P,SEMICOLON as k,COLON as v,SINGLE_QUOTE as S,DOUBLE_QUOTE as N,LINE_FEED as O}from"./ascii-constants.js";import{isXmlChar as d,isXmlSpaceByte as T,isAsciiDigit as b,isXmlNameStart as c,isXmlName as m}from"./utils.js";import{XmlError as o}from"./XmlError.js";var A=Object.defineProperty,_=Object.getOwnPropertySymbols,I=Object.prototype.hasOwnProperty,W=Object.prototype.propertyIsEnumerable,l=(r,t,e)=>t in r?A(r,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):r[t]=e,F=(r,t)=>{for(var e in t||(t={}))I.call(t,e)&&l(r,e,t[e]);if(_)for(var e of _(t))W.call(t,e)&&l(r,e,t[e]);return r},a=(r,t,e)=>l(r,typeof t!="symbol"?t+"":t,e);class p{constructor(t,e={}){a(this,"_text"),a(this,"_pos"),a(this,"_end"),a(this,"config");const{pos:i=0,strictDocument:n=!0,allowDtd:s=!0,rawTextElements:h=null}=e;this._text=t,this._pos=i,this._end=this._text.length,this.config={strictDocument:n,allowDtd:s,rawTextElements:h}}clone(){return new p(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 o({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({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new o({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(!d(this.currCodeUnitUnchecked()))throw new o({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}startsWith(t){return this._text.startsWith(t,this._pos)}skipString(t){if(!this.startsWith(t))throw new o({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&&T(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new o({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new o({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(C))return null;let t;if(this.tryConsumeCodeUnit(f)){let e,i;this.tryConsumeCodeUnit(U)?(e=this.consumeCodeUnitsWhile(s=>s>=w&&s<=y||s>=x&&s<=g||s>=E&&s<=P),i=16):(e=this.consumeCodeUnitsWhile(s=>b(s)),i=10);const n=parseInt(e,i);isNaN(n)||!d(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(k)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new o({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(e))throw new o({type:"InvalidName"},this.genTextPosFrom(t))}else if(!m(e))break;this._pos+=1}}consumeQName(){var t,e;const i=this._pos;let n=null;for(;this._pos<this._end;){const u=this.currCodeUnitUnchecked();if(u===v)if(n==null)n=this._pos,this._pos+=1;else throw new o({type:"InvalidName"},this.genTextPosFrom(i));else if(m(u))this._pos+=1;else break}let s,h;if(n!=null?(s=this._text.slice(i,n),h=this.sliceBack(n+1)):(s="",h=this.sliceBack(i)),s.length>0&&!c((t=s[0])==null?void 0:t.codePointAt(0)))throw new o({type:"InvalidName"},this.genTextPosFrom(i));if(h.length>0){if(!c((e=h[0])==null?void 0:e.codePointAt(0)))throw new o({type:"InvalidName"},this.genTextPosFrom(i))}else throw new o({type:"InvalidName"},this.genTextPosFrom(i));return[s,h]}consumeQuote(){const t=this.currCodeUnit();if(t===S||t===N)return this._pos+=1,t;throw new o({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,n=1;for(let s=0;s<e;s++)this._text.charCodeAt(s)===O?(i++,n=1):n++;return{row:i,col:n}}}export{p as XmlStream};

package/dist/esm/tokenizer/tokenize.js

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

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

@@ -200,10 +200,11 @@ export interface TXmlStreamConfig {
      * - Document must have exactly one root element
      * - No text nodes at document level (outside root element)
      * - Only comments and processing instructions allowed outside root element
+     * - Attribute requires an explicit value and the value must be quoted
      * When false:
      * - Multiple root elements allowed
      * - Text nodes allowed at document level
-     * - More HTML-like permissive parsing
+     * - More HTML-like permissive parsing (unquoted attributes, boolean attributes)
      * @default true
      */
     strictDocument: boolean;
@@ -214,12 +215,6 @@ export interface TXmlStreamConfig {
      * @default null
      */
     rawTextElements: string[] | null;
-    /**
-     * Whether to allow HTML-style boolean attributes (e.g., <div hidden>).
-     * When true, attributes without values are treated as having their name as the value.
-     * @default false
-     */
-    allowBooleanAttributes: boolean;
     /**
      * Whether to allow DOCTYPE declarations.
      * @default true

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "xml-tokenizer",
-  "version": "0.0.22",
+  "version": "0.0.23",
   "private": false,
   "description": "Straightforward and typesafe XML tokenizer that streams tokens through a callback mechanism",
   "keywords": [],