xml-tokenizer 0.0.20 → 0.0.21

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

package/dist/esm/tokenizer/XmlStream.js

@@ -1 +1 @@
-import{AMPERSAND as C,HASH as f,LOWERCASE_X as U,ZERO as y,NINE as w,UPPERCASE_A as g,UPPERCASE_F as x,LOWERCASE_A as P,LOWERCASE_F as E,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 b,isAsciiDigit as T,isXmlNameStart as c,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},a=(r,t,e)=>p(r,typeof t!="symbol"?t+"":t,e);class l{constructor(t,e={}){a(this,"_text"),a(this,"_pos"),a(this,"_end"),a(this,"config");const{pos:i=0,strict:n=!0,allowDtd:s=!0}=e;this._text=t,this._pos=i,this._end=this._text.length,this.config={strict:n,allowDtd:s}}clone(){return new l(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&&b(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>=y&&s<=w||s>=g&&s<=x||s>=P&&s<=E),i=16):(e=this.consumeCodeUnitsWhile(s=>T(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(!_(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(_(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{l as XmlStream};
+import{AMPERSAND as C,HASH as f,LOWERCASE_X as U,ZERO as w,NINE as y,UPPERCASE_A as g,UPPERCASE_F as x,LOWERCASE_A as P,LOWERCASE_F as E,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 b,isAsciiDigit as A,isXmlNameStart as c,isXmlName as _}from"./utils.js";import{XmlError as o}from"./XmlError.js";var T=Object.defineProperty,m=Object.getOwnPropertySymbols,I=Object.prototype.hasOwnProperty,W=Object.prototype.propertyIsEnumerable,l=(r,t,e)=>t in r?T(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(m)for(var e of m(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,allowBooleanAttributes:h=!1}=e;this._text=t,this._pos=i,this._end=this._text.length,this.config={strictDocument:n,allowDtd:s,allowBooleanAttributes: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&&b(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>=g&&s<=x||s>=P&&s<=E),i=16):(e=this.consumeCodeUnitsWhile(s=>A(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(!_(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(_(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{LESS_THAN as d,HYPHEN as D,QUESTION_MARK as U,GREATER_THAN as r,OPEN_SQUARE_BRACKET as L,PERCENT as O,DOUBLE_QUOTE as R,SINGLE_QUOTE as b,UPPERCASE_S as B,UPPERCASE_P as H,SLASH as k,EQUALS as T,EXCLAMATION_MARK as M,CLOSE_SQUARE_BRACKET as K}from"./ascii-constants.js";import{XmlError as a}from"./XmlError.js";import{XmlStream as Y}from"./XmlStream.js";const X="\uFEFF",w="<?xml ",P="?>",z="<!ENTITY",G="<!ELEMENT",j="<!ATTLIST",q="<!NOTATION",J="<!DOCTYPE",W="<?",y="?>",f="<!--",x="-->",V="<![CDATA[",h="]]>",Z="NDATA",v="PUBLIC",$="SYSTEM",tt="version",nt="encoding",st="standalone";function et(t,n,s={}){A(new Y(t,s),n)}function A(t,n){if(t.startsWith(X)&&t.advance(1),t.startsWith(w)&&ot(t),u(t,n),t.skipSpaces(),t.startsWith(J)){if(!t.config.allowDtd)throw new a({type:"DtdDetected"});it(t,n),u(t,n)}if(t.skipSpaces(),t.config.strict){if(!t.atEnd()&&t.currCodeUnit()===d&&S(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?S(t,n):I(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(W))C(t,n);else break}function ot(t){function n(s){if(s.startsWithSpace())s.skipSpaces();else if(!s.startsWith(P)&&!s.atEnd())throw new a({type:"InvalidChar",expected:"a whitespace",actual:s.currCodeUnitUnchecked()},s.genTextPos())}if(t.advance(5),n(t),!t.startsWith(tt))throw new a({type:"InvalidString",expected:"version"},t.genTextPos());l(t),n(t),t.startsWith(nt)&&(l(t),n(t)),t.startsWith(st)&&l(t),t.skipSpaces(),t.skipString(P)}function g(t,n){const s=t.getPos();t.advance(4);const e=t.consumeCodeUnitsWhile((o,i)=>!(o===D&&i.startsWith(x)));if(t.skipString(x),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(y)));t.skipString(y),n({type:"ProcessingInstruction",target:e,content:o.length===0?void 0:o,range:t.rangeFrom(s)})}function it(t,n){const s=t.getPos();if(at(t),t.skipSpaces(),t.currCodeUnit()===r){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(z))ct(t,n);else if(t.startsWith(f))g(t,n);else if(t.startsWith(W))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(G)||t.startsWith(j)||t.startsWith(q))try{pt(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 at(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),N(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==L&&n!==r)throw new a({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function N(t){if(t.startsWith($)||t.startsWith(v)){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===v){t.consumeSpaces();const o=t.consumeQuote();t.consumeCodeUnitsWhile(i=>i!==o),t.consumeCodeUnit(o)}return!0}return!1}function ct(t,n){t.advance(8),t.consumeSpaces();const s=!t.tryConsumeCodeUnit(O);s||t.consumeSpaces();const e=t.consumeName();t.consumeSpaces();const o=rt(t,s);o!==null&&n({type:"EntityDeclaration",name:e,definition:o}),t.skipSpaces(),t.consumeCodeUnit(r)}function rt(t,n){const s=t.currCodeUnit();if(s===R||s===b){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===B||s===H){if(N(t))return n&&(t.skipSpaces(),t.startsWith(Z)&&(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 pt(t){t.skipCodeUnitsWhile(n=>n!==r),t.consumeCodeUnit(r)}function S(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(),E=t.currCodeUnit();if(E===k){t.advance(1),t.consumeCodeUnit(r);const p=t.rangeFrom(m);n({type:"ElementEnd",end:{type:"Empty"},range:p});break}else if(E===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,F,Q]=l(t),_=t.getPos();n({type:"Attribute",range:{start:m,end:_},prefix:p,local:F,value:Q})}}i&&dt(t,n)}function l(t){const[n,s]=t.consumeQName();let e;const o=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(T)){t.skipSpaces();const i=t.consumeQuote();e=t.consumeCodeUnitsWhile(c=>c!==i&&c!==d),t.consumeCodeUnit(i)}else{if(t.config.strict)throw new a({type:"InvalidChar",expected:T,actual:t.currCodeUnit()},t.genTextPos());t.goTo(o),e="true"}return[n,s,e]}function dt(t,n){for(;!t.atEnd();)if(t.currCodeUnit()===d){const s=t.nextCodeUnit();if(s===M)if(t.startsWith(f))g(t,n);else if(t.startsWith(V))ut(t,n);else throw new a({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===U)C(t,n);else if(s===k){lt(t,n);break}else S(t,n)}else I(t,n)}function ut(t,n){const s=t.getPos();t.advance(9);const e=t.consumeCodeUnitsWhile((i,c)=>!(i===K&&c.startsWith(h)));t.skipString(h);const o=t.rangeFrom(s);n({type:"Cdata",text:e,range:o})}function lt(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 I(t,n){const s=t.getPos(),e=t.consumeCodeUnitsWhile(o=>o!==d);if(e.includes(h))throw new a({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:e,range:t.rangeFrom(s)})}export{et as tokenize,A as tokenizeXmlStream};
+import{LESS_THAN as d,HYPHEN as _,QUESTION_MARK as U,GREATER_THAN as r,OPEN_SQUARE_BRACKET as L,PERCENT as O,DOUBLE_QUOTE as R,SINGLE_QUOTE as b,UPPERCASE_S as B,UPPERCASE_P as H,SLASH as k,EQUALS as T,EXCLAMATION_MARK as M,CLOSE_SQUARE_BRACKET as K}from"./ascii-constants.js";import{XmlError as a}from"./XmlError.js";import{XmlStream as Y}from"./XmlStream.js";const X="\uFEFF",w="<?xml ",P="?>",z="<!ENTITY",G="<!ELEMENT",j="<!ATTLIST",q="<!NOTATION",J="<!DOCTYPE",W="<?",y="?>",f="<!--",x="-->",V="<![CDATA[",h="]]>",Z="NDATA",v="PUBLIC",$="SYSTEM",tt="version",nt="encoding",st="standalone";function et(t,n,s={}){A(new Y(t,s),n)}function A(t,n){if(t.startsWith(X)&&t.advance(1),t.startsWith(w)&&ot(t),u(t,n),t.skipSpaces(),t.startsWith(J)){if(!t.config.allowDtd)throw new a({type:"DtdDetected"});it(t,n),u(t,n)}if(t.skipSpaces(),t.config.strictDocument){if(!t.atEnd()&&t.currCodeUnit()===d&&S(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?S(t,n):I(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(W))C(t,n);else break}function ot(t){function n(s){if(s.startsWithSpace())s.skipSpaces();else if(!s.startsWith(P)&&!s.atEnd())throw new a({type:"InvalidChar",expected:"a whitespace",actual:s.currCodeUnitUnchecked()},s.genTextPos())}if(t.advance(5),n(t),!t.startsWith(tt))throw new a({type:"InvalidString",expected:"version"},t.genTextPos());l(t),n(t),t.startsWith(nt)&&(l(t),n(t)),t.startsWith(st)&&l(t),t.skipSpaces(),t.skipString(P)}function g(t,n){const s=t.getPos();t.advance(4);const e=t.consumeCodeUnitsWhile((o,i)=>!(o===_&&i.startsWith(x)));if(t.skipString(x),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(y)));t.skipString(y),n({type:"ProcessingInstruction",target:e,content:o.length===0?void 0:o,range:t.rangeFrom(s)})}function it(t,n){const s=t.getPos();if(at(t),t.skipSpaces(),t.currCodeUnit()===r){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(z))ct(t,n);else if(t.startsWith(f))g(t,n);else if(t.startsWith(W))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(G)||t.startsWith(j)||t.startsWith(q))try{pt(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 at(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),N(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==L&&n!==r)throw new a({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function N(t){if(t.startsWith($)||t.startsWith(v)){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===v){t.consumeSpaces();const o=t.consumeQuote();t.consumeCodeUnitsWhile(i=>i!==o),t.consumeCodeUnit(o)}return!0}return!1}function ct(t,n){t.advance(8),t.consumeSpaces();const s=!t.tryConsumeCodeUnit(O);s||t.consumeSpaces();const e=t.consumeName();t.consumeSpaces();const o=rt(t,s);o!==null&&n({type:"EntityDeclaration",name:e,definition:o}),t.skipSpaces(),t.consumeCodeUnit(r)}function rt(t,n){const s=t.currCodeUnit();if(s===R||s===b){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===B||s===H){if(N(t))return n&&(t.skipSpaces(),t.startsWith(Z)&&(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 pt(t){t.skipCodeUnitsWhile(n=>n!==r),t.consumeCodeUnit(r)}function S(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(),E=t.currCodeUnit();if(E===k){t.advance(1),t.consumeCodeUnit(r);const p=t.rangeFrom(m);n({type:"ElementEnd",end:{type:"Empty"},range:p});break}else if(E===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,F,D]=l(t),Q=t.getPos();n({type:"Attribute",range:{start:m,end:Q},prefix:p,local:F,value:D})}}i&&dt(t,n)}function l(t){const[n,s]=t.consumeQName();let e;const o=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(T)){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:T,actual:t.currCodeUnit()},t.genTextPos());return[n,s,e]}function dt(t,n){for(;!t.atEnd();)if(t.currCodeUnit()===d){const s=t.nextCodeUnit();if(s===M)if(t.startsWith(f))g(t,n);else if(t.startsWith(V))ut(t,n);else throw new a({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===U)C(t,n);else if(s===k){lt(t,n);break}else S(t,n)}else I(t,n)}function ut(t,n){const s=t.getPos();t.advance(9);const e=t.consumeCodeUnitsWhile((i,c)=>!(i===K&&c.startsWith(h)));t.skipString(h);const o=t.rangeFrom(s);n({type:"Cdata",text:e,range:o})}function lt(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 I(t,n){const s=t.getPos(),e=t.consumeCodeUnitsWhile(o=>o!==d);if(e.includes(h))throw new a({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:e,range:t.rangeFrom(s)})}export{et as tokenize,A as tokenizeXmlStream};

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

@@ -194,7 +194,29 @@ export declare class XmlStream {
     genTextPosFrom(pos: number): TTextPos;
 }
 export interface TXmlStreamConfig {
-    strict: boolean;
+    /**
+     * Whether to enforce strict XML document structure rules.
+     * When true:
+     * - 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
+     * When false:
+     * - Multiple root elements allowed
+     * - Text nodes allowed at document level
+     * - More HTML-like permissive parsing
+     * @default true
+     */
+    strictDocument: boolean;
+    /**
+     * 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
+     */
     allowDtd: boolean;
 }
 export type TXmlStreamOptions = {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "xml-tokenizer",
-  "version": "0.0.20",
+  "version": "0.0.21",
   "private": false,
   "description": "Straightforward and typesafe XML tokenizer that streams tokens through a callback mechanism",
   "keywords": [],