xml-tokenizer 0.0.37 → 0.0.39
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
|
@@ -1 +1 @@
|
|
|
1
|
-
"use strict";var o=require("./ascii-constants.js"),c=require("./utils.js"),n=require("./XmlError.js"),C=Object.defineProperty,g=Object.getOwnPropertySymbols,x=Object.prototype.hasOwnProperty,E=Object.prototype.propertyIsEnumerable,m=(a,t,e)=>t in a?C(a,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):a[t]=e,f=(a,t)=>{for(var e in t||(t={}))x.call(t,e)&&m(a,e,t[e]);if(g)for(var e of g(t))E.call(t,e)&&m(a,e,t[e]);return a},d=(a,t,e)=>m(a,typeof t!="symbol"?t+"":t,e);class _{constructor(t,e={}){d(this,"_text"),d(this,"_pos"),d(this,"_end"),d(this,"config");const{pos:
|
|
1
|
+
"use strict";var o=require("./ascii-constants.js"),c=require("./utils.js"),n=require("./XmlError.js"),C=Object.defineProperty,g=Object.getOwnPropertySymbols,x=Object.prototype.hasOwnProperty,E=Object.prototype.propertyIsEnumerable,m=(a,t,e)=>t in a?C(a,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):a[t]=e,f=(a,t)=>{for(var e in t||(t={}))x.call(t,e)&&m(a,e,t[e]);if(g)for(var e of g(t))E.call(t,e)&&m(a,e,t[e]);return a},d=(a,t,e)=>m(a,typeof t!="symbol"?t+"":t,e);class _{constructor(t,e={}){d(this,"_text"),d(this,"_pos"),d(this,"_end"),d(this,"config");const{pos:s=0,strictDocument:r=!0,allowDtd:i=!0,rawTextElements:p=null,implicitSelfClosingElements:h=null,contextSliceSize:l=!1,debugLogger:u}=e;this._text=t,this._pos=s,this._end=this._text.length,this.config={strictDocument:r,allowDtd:i,rawTextElements:p,implicitSelfClosingElements:h,contextSliceSize:l,debugLogger:u},this.debug("XmlStream constructor",{textLength:t.length,initialPos:s,platform:process.platform,nodeVersion:process.version,textSample:t.slice(0,Math.min(100,t.length))+(t.length>100?"...":"")}),this.debug("XmlStream initialized",{textLength:this._text.length,initialPos:s,platform:process.platform,nodeVersion:process.version,textSample:this._text.slice(0,Math.min(100,this._text.length))+(this._text.length>100?"...":"")})}debug(t,...e){this.config.debugLogger&&this.config.debugLogger(t,...e)}clone(){return new _(this._text,f({pos:this._pos},this.config))}getPos(){return this._pos}atEnd(){return this._pos>=this._end}currCodeUnit(){if(this._pos>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos)}currCodeUnitUnchecked(){return this._text.charCodeAt(this._pos)}nextCodeUnit(){if(this._pos+1>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new n.XmlError({type:"InvalidChar",expected:t,actual:e},this.genTextPos());this._pos+=1}tryConsumeCodeUnit(t){return this.currCodeUnit()===t?(this._pos+=1,!0):!1}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked(),this);){if(!c.isXmlChar(this.currCodeUnitUnchecked()))throw new n.XmlError({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}goToEnd(){this._pos=this._end}startsWith(t){return this._text.startsWith(t,this._pos)}startsWithIgnoreCase(t){if(this._pos+t.length>this._end)return!1;for(let e=0;e<t.length;e++){const s=this._text.charCodeAt(this._pos+e),r=t.charCodeAt(e);if((s>=97&&s<=122?s-32:s)!==r)return!1}return!0}skipString(t){if(!this.startsWith(t))throw new n.XmlError({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}sliceBack(t){return this._text.slice(t,this._pos)}rangeFrom(t){return{start:t,end:this._pos}}skipSpaces(){for(;this.startsWithSpace();)this._pos+=1}startsWithSpace(){return this._pos<this._end&&c.isXmlSpaceByte(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new n.XmlError({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new n.XmlError({type:"InvalidChar",expected:"a whitespace",actual:this.currCodeUnitUnchecked()},this.genTextPos());this.skipSpaces()}tryConsumeReference(){const t=this._pos,e=this.clone(),s=e.consumeReference();return s!=null?(this._pos+=e.getPos()-t,s):null}consumeReference(){if(!this.tryConsumeCodeUnit(o.AMPERSAND))return null;let t;if(this.tryConsumeCodeUnit(o.HASH)){let e,s;this.tryConsumeCodeUnit(o.LOWERCASE_X)?(e=this.consumeCodeUnitsWhile(i=>i>=o.ZERO&&i<=o.NINE||i>=o.UPPERCASE_A&&i<=o.UPPERCASE_F||i>=o.LOWERCASE_A&&i<=o.LOWERCASE_F),s=16):(e=this.consumeCodeUnitsWhile(i=>c.isAsciiDigit(i)),s=10);const r=parseInt(e,s);isNaN(r)||!c.isXmlChar(r)?t=null:t={type:"Char",value:String.fromCodePoint(r)}}else{const e=this.consumeName();switch(e){case"quot":t={type:"Char",value:'"'};break;case"amp":t={type:"Char",value:"&"};break;case"apos":t={type:"Char",value:"'"};break;case"lt":t={type:"Char",value:"<"};break;case"gt":t={type:"Char",value:">"};break;default:t={type:"Entity",value:e}}}return this.tryConsumeCodeUnit(o.SEMICOLON)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new n.XmlError({type:"InvalidName"},this.genTextPosFrom(t));return e}skipName(){const t=this._pos;for(;this._pos<this._end;){const e=this.currCodeUnitUnchecked();if(this._pos===t){if(!c.isXmlNameStart(e))throw new n.XmlError({type:"InvalidName"},this.genTextPosFrom(t))}else if(!c.isXmlName(e))break;this._pos+=1}}consumeQName(){var t,e,s,r;const i=this._pos;let p=null;for(this.debug("Starting consumeQName",{position:this._pos,nextChars:this._text.slice(this._pos,this._pos+10)});this._pos<this._end;){const u=this.currCodeUnitUnchecked();if(u===o.COLON)if(p==null)p=this._pos,this._pos+=1,this.debug("Found colon splitter",{position:this._pos});else throw this.debug("Multiple colons found - error",{position:this._pos}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(i));else if(c.isXmlName(u))this._pos+=1;else break}let h,l;if(p!=null?(h=this._text.slice(i,p),l=this.sliceBack(p+1)):(h="",l=this.sliceBack(i)),this.debug("Finished consumeQName",{prefix:h,local:l,startPos:i,endPos:this._pos}),h.length>0&&!c.isXmlNameStart((t=h[0])==null?void 0:t.codePointAt(0)))throw this.debug("Invalid prefix start character",{prefix:h,firstChar:h[0],codePoint:(e=h[0])==null?void 0:e.codePointAt(0)}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(i));if(l.length>0){if(!c.isXmlNameStart((s=l[0])==null?void 0:s.codePointAt(0)))throw this.debug("Invalid local name start character",{local:l,firstChar:l[0],codePoint:(r=l[0])==null?void 0:r.codePointAt(0)}),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(i))}else throw this.debug("Empty local name - error"),new n.XmlError({type:"InvalidName"},this.genTextPosFrom(i));return[h,l]}consumeQuote(){const t=this.currCodeUnit();if(t===o.SINGLE_QUOTE||t===o.DOUBLE_QUOTE)return this._pos+=1,t;throw new n.XmlError({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}getTextAround(t,e){const s=Math.floor(e/2),r=Math.max(0,t-s),i=Math.min(this._end,t+s+e%2);return this._text.slice(r,i)}genTextPos(){return this.genTextPosFrom(this._pos)}genTextPosFrom(t){const e=Math.min(t,this._end);let s=1,r=1;for(let i=0;i<e;i++)this._text.charCodeAt(i)===o.LINE_FEED?(s++,r=1):r++;return f({row:s,col:r},this.config.contextSliceSize?{contextSlice:this.getTextAround(t,this.config.contextSliceSize)}:{})}getTextSnippet(t,e){return this._text.slice(t,t+e)}}exports.XmlStream=_;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
"use strict";var
|
|
1
|
+
"use strict";var p=require("../get-q-name.js"),o=require("./ascii-constants.js"),_=require("./utils.js"),c=require("./XmlError.js"),I=require("./XmlStream.js");const F="\uFEFF",P="<?xml ",w="?>",H="<!ENTITY",Q="<!ELEMENT",D="<!ATTLIST",O="<!NOTATION",E="<!DOCTYPE",x="<?",W="?>",f="<!--",y="-->",G="<![CDATA[",h="]]>",M="NDATA",N="PUBLIC",B="SYSTEM",z="version",q="encoding",Y="standalone";function K(t,n,e={}){var s,i;(s=e.debugLogger)==null||s.call(e,"Creating XmlStream",{stringLength:t.length});const r=new I.XmlStream(t,e);r.debug("Created XmlStream, starting tokenization",{stringLength:t.length}),(i=e.debugLogger)==null||i.call(e,"Created XmlStream, starting tokenization (2)",{stringLength:t.length}),v(r,n)}function v(t,n){if(t.debug("Starting XML stream tokenization"),t.startsWith(F)&&(t.debug("Skipping BOM"),t.advance(1)),t.startsWith(P)&&(t.debug("Found XML declaration"),$(t)),g(t,n),t.config.strictDocument){if(t.debug("Parsing in strict document mode"),t.skipSpaces(),t.startsWithIgnoreCase(E)&&(t.debug("Found DOCTYPE declaration"),T(t,n),g(t,n)),t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===o.LESS_THAN&&(t.debug("Starting root element parsing"),U(t,n)),g(t,n),!t.atEnd())throw t.debug("Unexpected content after root element",{position:t.getPos(),nextChars:t.getTextSnippet(t.getPos(),10)}),new c.XmlError({type:"UnknownToken",message:"Not at end"},t.genTextPos())}else for(t.debug("Parsing in non-strict mode");!t.atEnd();)t.currCodeUnit()===o.LESS_THAN?U(t,n):X(t,n),g(t,n);t.debug("Finished XML stream tokenization")}function g(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith(f))S(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(E))T(t,n);else if(t.startsWith(x))C(t,n);else break}function $(t){function n(e){if(e.startsWithSpace())e.skipSpaces();else if(!e.startsWith(w)&&!e.atEnd())throw new c.XmlError({type:"InvalidChar",expected:"a whitespace",actual:e.currCodeUnitUnchecked()},e.genTextPos())}if(t.advance(5),n(t),!t.startsWith(z))throw new c.XmlError({type:"InvalidString",expected:"version"},t.genTextPos());m(t),n(t),t.startsWith(q)&&(m(t),n(t)),t.startsWith(Y)&&m(t),t.skipSpaces(),t.skipString(w)}function S(t,n){const e=t.getPos();t.advance(4);const s=t.consumeCodeUnitsWhile((i,r)=>!(i===o.HYPHEN&&r.startsWith(y)));if(t.skipString(y),s.includes("--")||s.endsWith("-"))throw new c.XmlError({type:"InvalidComment"},t.genTextPosFrom(e));n({type:"Comment",text:s,range:t.rangeFrom(e)},t)}function C(t,n){if(t.startsWith(P))throw new c.XmlError({type:"UnexpectedDeclaration"},t.genTextPos());const e=t.getPos();t.advance(2);const s=t.consumeName();t.skipSpaces();const i=t.consumeCodeUnitsWhile((r,a)=>!(r===o.QUESTION_MARK&&a.startsWith(W)));t.skipString(W),n({type:"ProcessingInstruction",target:s,content:i.length===0?void 0:i,range:t.rangeFrom(e)},t)}function T(t,n){if(!t.config.allowDtd)throw new c.XmlError({type:"DtdDetected"});const e=t.getPos();if(j(t),t.skipSpaces(),t.currCodeUnit()===o.GREATER_THAN){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(H))J(t,n);else if(t.startsWith(f))S(t,n);else if(t.startsWith(x))C(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===o.GREATER_THAN){t.advance(1);break}else throw new c.XmlError({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith(Q)||t.startsWith(D)||t.startsWith(O))try{Z(t)}catch(s){throw new c.XmlError({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(e))}else throw new c.XmlError({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function j(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),b(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==o.OPEN_SQUARE_BRACKET&&n!==o.GREATER_THAN)throw new c.XmlError({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function b(t){if(t.startsWith(B)||t.startsWith(N)){const n=t.getPos();t.advance(6);const e=t.sliceBack(n);t.consumeSpaces();const s=t.consumeQuote();if(t.consumeCodeUnitsWhile(i=>i!==s),t.consumeCodeUnit(s),e===N){t.consumeSpaces();const i=t.consumeQuote();t.consumeCodeUnitsWhile(r=>r!==i),t.consumeCodeUnit(i)}return!0}return!1}function J(t,n){t.advance(8),t.consumeSpaces();const e=!t.tryConsumeCodeUnit(o.PERCENT);e||t.consumeSpaces();const s=t.consumeName();t.consumeSpaces();const i=V(t,e);i!==null&&n({type:"EntityDeclaration",name:s,definition:i},t),t.skipSpaces(),t.consumeCodeUnit(o.GREATER_THAN)}function V(t,n){const e=t.currCodeUnit();if(e===o.DOUBLE_QUOTE||e===o.SINGLE_QUOTE){const s=t.consumeQuote(),i=t.getPos();t.skipCodeUnitsWhile(a=>a!==s);const r=t.sliceBack(i);return t.consumeCodeUnit(s),r}else if(e===o.UPPERCASE_S||e===o.UPPERCASE_P){if(b(t))return n&&(t.skipSpaces(),t.startsWith(M)&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new c.XmlError({type:"InvalidExternalID"},t.genTextPos())}else throw new c.XmlError({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:e},t.genTextPos())}function Z(t){t.skipCodeUnitsWhile(n=>n!==o.GREATER_THAN),t.consumeCodeUnit(o.GREATER_THAN)}function U(t,n){const e=t.getPos();t.advance(1),t.debug("Starting element parsing",{position:e,nextChars:t.getTextSnippet(e,10)});const[s,i]=t.consumeQName();t.debug("Element name parsed",{prefix:s,local:i}),n({type:"ElementStart",prefix:s,local:i,start:e},t);let r=!1;for(;!t.atEnd();){const a=t.startsWithSpace();t.skipSpaces();const l=t.getPos(),d=t.currCodeUnit();if(t.debug("Processing element content",{position:l,currCodeUnit:d,hasSpace:a}),d===o.SLASH){t.debug("Found self-closing tag"),t.advance(1),t.consumeCodeUnit(o.GREATER_THAN);const u=t.rangeFrom(l);n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else if(d===o.GREATER_THAN){t.advance(1);const u=t.rangeFrom(l);if(t.config.implicitSelfClosingElements!=null&&t.config.implicitSelfClosingElements.includes(p.getQName(i,s))){t.debug("Found implicit self-closing element",{element:p.getQName(i,s)}),n({type:"ElementEnd",end:{type:"Empty"},range:u},t);break}else{t.debug("Found opening tag end"),n({type:"ElementEnd",end:{type:"Open"},range:u},t),r=!0;break}}else{if(!a)throw t.debug("Missing whitespace before attribute",{position:t.getPos(),currCodeUnit:String.fromCharCode(d)}),new c.XmlError({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());t.debug("Parsing attribute");const[u,k,A]=m(t),L=t.getPos();t.debug("Attribute parsed",{prefix:u,local:k,value:A}),n({type:"Attribute",range:{start:l,end:L},prefix:u,local:k,value:A},t)}}if(r){if(t.config.rawTextElements!=null){const a=p.getQName(i,s);if(t.config.rawTextElements.includes(a)){t.debug("Processing raw text element content",{element:a}),R(t,n,`</${a}>`);return}}t.debug("Processing element content"),R(t,n)}t.debug("Finished element parsing",{prefix:s,local:i})}function m(t){const[n,e]=t.consumeQName();let s;const i=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(o.EQUALS)){t.skipSpaces();const r=t.currCodeUnit();if(r===o.SINGLE_QUOTE||r===o.DOUBLE_QUOTE){const a=t.consumeQuote();s=t.consumeCodeUnitsWhile(l=>l!==a&&l!==o.LESS_THAN),t.consumeCodeUnit(a)}else if(!t.config.strictDocument)s=t.consumeCodeUnitsWhile(a=>!_.isXmlSpaceByte(a)&&a!==o.GREATER_THAN&&a!==o.SLASH&&a!==o.LESS_THAN);else throw new c.XmlError({type:"InvalidChar",expected:"a quote",actual:r},t.genTextPos())}else if(!t.config.strictDocument)t.goTo(i),s="true";else throw new c.XmlError({type:"InvalidChar",expected:o.EQUALS,actual:t.currCodeUnit()},t.genTextPos());return[n,e,s]}function R(t,n,e=o.LESS_THAN){for(;!t.atEnd();)if(typeof e=="number"?t.currCodeUnit()===e:t.startsWith(e)){const s=t.nextCodeUnit();if(s===o.EXCLAMATION_MARK)if(t.startsWith(f))S(t,n);else if(t.startsWith(G))tt(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(E))T(t,n);else throw new c.XmlError({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===o.QUESTION_MARK)C(t,n);else if(s===o.SLASH){nt(t,n);break}else U(t,n)}else X(t,n,e)}function tt(t,n){const e=t.getPos();t.advance(9);const s=t.consumeCodeUnitsWhile((r,a)=>!(r===o.CLOSE_SQUARE_BRACKET&&a.startsWith(h)));t.skipString(h);const i=t.rangeFrom(e);n({type:"Cdata",text:s,range:i},t)}function nt(t,n){const e=t.getPos();t.advance(2);const[s,i]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(o.GREATER_THAN);const r=t.rangeFrom(e);n({type:"ElementEnd",end:{type:"Close",prefix:s,local:i},range:r},t)}function X(t,n,e=o.LESS_THAN){const s=t.getPos(),i=typeof e=="number"?t.consumeCodeUnitsWhile(r=>r!==e):t.consumeCodeUnitsWhile((r,a)=>!a.startsWith(e));if(t.config.strictDocument&&i.includes(h))throw new c.XmlError({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:i,range:t.rangeFrom(s)},t)}exports.tokenize=K,exports.tokenizeXmlStream=v;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
import{AMPERSAND as
|
|
1
|
+
import{AMPERSAND as C,HASH as S,LOWERCASE_X as U,ZERO as w,NINE as y,UPPERCASE_A as P,UPPERCASE_F as v,LOWERCASE_A as E,LOWERCASE_F as b,SEMICOLON as k,COLON as N,SINGLE_QUOTE as T,DOUBLE_QUOTE as A,LINE_FEED as O}from"./ascii-constants.js";import{isXmlChar as _,isXmlSpaceByte as I,isAsciiDigit as W,isXmlNameStart as u,isXmlName as g}from"./utils.js";import{XmlError as n}from"./XmlError.js";var F=Object.defineProperty,f=Object.getOwnPropertySymbols,L=Object.prototype.hasOwnProperty,R=Object.prototype.propertyIsEnumerable,d=(c,t,e)=>t in c?F(c,t,{enumerable:!0,configurable:!0,writable:!0,value:e}):c[t]=e,x=(c,t)=>{for(var e in t||(t={}))L.call(t,e)&&d(c,e,t[e]);if(f)for(var e of f(t))R.call(t,e)&&d(c,e,t[e]);return c},p=(c,t,e)=>d(c,typeof t!="symbol"?t+"":t,e);class m{constructor(t,e={}){p(this,"_text"),p(this,"_pos"),p(this,"_end"),p(this,"config");const{pos:s=0,strictDocument:o=!0,allowDtd:i=!0,rawTextElements:a=null,implicitSelfClosingElements:r=null,contextSliceSize:h=!1,debugLogger:l}=e;this._text=t,this._pos=s,this._end=this._text.length,this.config={strictDocument:o,allowDtd:i,rawTextElements:a,implicitSelfClosingElements:r,contextSliceSize:h,debugLogger:l},this.debug("XmlStream constructor",{textLength:t.length,initialPos:s,platform:process.platform,nodeVersion:process.version,textSample:t.slice(0,Math.min(100,t.length))+(t.length>100?"...":"")}),this.debug("XmlStream initialized",{textLength:this._text.length,initialPos:s,platform:process.platform,nodeVersion:process.version,textSample:this._text.slice(0,Math.min(100,this._text.length))+(this._text.length>100?"...":"")})}debug(t,...e){this.config.debugLogger&&this.config.debugLogger(t,...e)}clone(){return new m(this._text,x({pos:this._pos},this.config))}getPos(){return this._pos}atEnd(){return this._pos>=this._end}currCodeUnit(){if(this._pos>=this._end)throw new n({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos)}currCodeUnitUnchecked(){return this._text.charCodeAt(this._pos)}nextCodeUnit(){if(this._pos+1>=this._end)throw new n({type:"UnexpectedEndOfStream"});return this._text.charCodeAt(this._pos+1)}consumeCodeUnit(t){const e=this.currCodeUnit();if(e!==t)throw new n({type:"InvalidChar",expected:t,actual:e},this.genTextPos());this._pos+=1}tryConsumeCodeUnit(t){return this.currCodeUnit()===t?(this._pos+=1,!0):!1}consumeCodeUnitsWhile(t){const e=this._pos;return this.skipCodeUnitsWhile(t),this.sliceBack(e)}skipCodeUnitsWhile(t){for(;this._pos<this._end&&t(this.currCodeUnitUnchecked(),this);){if(!_(this.currCodeUnitUnchecked()))throw new n({type:"NonXmlChar",char:String.fromCodePoint(this.currCodeUnitUnchecked())},this.genTextPos());this._pos+=1}}advance(t){this._pos+=t}goTo(t){this._pos=t}goToEnd(){this._pos=this._end}startsWith(t){return this._text.startsWith(t,this._pos)}startsWithIgnoreCase(t){if(this._pos+t.length>this._end)return!1;for(let e=0;e<t.length;e++){const s=this._text.charCodeAt(this._pos+e),o=t.charCodeAt(e);if((s>=97&&s<=122?s-32:s)!==o)return!1}return!0}skipString(t){if(!this.startsWith(t))throw new n({type:"InvalidString",expected:t},this.genTextPos());this._pos+=t.length}sliceBack(t){return this._text.slice(t,this._pos)}rangeFrom(t){return{start:t,end:this._pos}}skipSpaces(){for(;this.startsWithSpace();)this._pos+=1}startsWithSpace(){return this._pos<this._end&&I(this.currCodeUnitUnchecked())}consumeSpaces(){if(this._pos>=this._end)throw new n({type:"UnexpectedEndOfStream"},this.genTextPos());if(!this.startsWithSpace())throw new n({type:"InvalidChar",expected:"a whitespace",actual:this.currCodeUnitUnchecked()},this.genTextPos());this.skipSpaces()}tryConsumeReference(){const t=this._pos,e=this.clone(),s=e.consumeReference();return s!=null?(this._pos+=e.getPos()-t,s):null}consumeReference(){if(!this.tryConsumeCodeUnit(C))return null;let t;if(this.tryConsumeCodeUnit(S)){let e,s;this.tryConsumeCodeUnit(U)?(e=this.consumeCodeUnitsWhile(i=>i>=w&&i<=y||i>=P&&i<=v||i>=E&&i<=b),s=16):(e=this.consumeCodeUnitsWhile(i=>W(i)),s=10);const o=parseInt(e,s);isNaN(o)||!_(o)?t=null:t={type:"Char",value:String.fromCodePoint(o)}}else{const e=this.consumeName();switch(e){case"quot":t={type:"Char",value:'"'};break;case"amp":t={type:"Char",value:"&"};break;case"apos":t={type:"Char",value:"'"};break;case"lt":t={type:"Char",value:"<"};break;case"gt":t={type:"Char",value:">"};break;default:t={type:"Entity",value:e}}}return this.tryConsumeCodeUnit(k)?t:null}consumeName(){const t=this._pos;this.skipName();const e=this.sliceBack(t);if(e.length===0)throw new n({type:"InvalidName"},this.genTextPosFrom(t));return e}skipName(){const t=this._pos;for(;this._pos<this._end;){const e=this.currCodeUnitUnchecked();if(this._pos===t){if(!u(e))throw new n({type:"InvalidName"},this.genTextPosFrom(t))}else if(!g(e))break;this._pos+=1}}consumeQName(){var t,e,s,o;const i=this._pos;let a=null;for(this.debug("Starting consumeQName",{position:this._pos,nextChars:this._text.slice(this._pos,this._pos+10)});this._pos<this._end;){const l=this.currCodeUnitUnchecked();if(l===N)if(a==null)a=this._pos,this._pos+=1,this.debug("Found colon splitter",{position:this._pos});else throw this.debug("Multiple colons found - error",{position:this._pos}),new n({type:"InvalidName"},this.genTextPosFrom(i));else if(g(l))this._pos+=1;else break}let r,h;if(a!=null?(r=this._text.slice(i,a),h=this.sliceBack(a+1)):(r="",h=this.sliceBack(i)),this.debug("Finished consumeQName",{prefix:r,local:h,startPos:i,endPos:this._pos}),r.length>0&&!u((t=r[0])==null?void 0:t.codePointAt(0)))throw this.debug("Invalid prefix start character",{prefix:r,firstChar:r[0],codePoint:(e=r[0])==null?void 0:e.codePointAt(0)}),new n({type:"InvalidName"},this.genTextPosFrom(i));if(h.length>0){if(!u((s=h[0])==null?void 0:s.codePointAt(0)))throw this.debug("Invalid local name start character",{local:h,firstChar:h[0],codePoint:(o=h[0])==null?void 0:o.codePointAt(0)}),new n({type:"InvalidName"},this.genTextPosFrom(i))}else throw this.debug("Empty local name - error"),new n({type:"InvalidName"},this.genTextPosFrom(i));return[r,h]}consumeQuote(){const t=this.currCodeUnit();if(t===T||t===A)return this._pos+=1,t;throw new n({type:"InvalidChar",expected:"a quote",actual:t},this.genTextPos())}getTextAround(t,e){const s=Math.floor(e/2),o=Math.max(0,t-s),i=Math.min(this._end,t+s+e%2);return this._text.slice(o,i)}genTextPos(){return this.genTextPosFrom(this._pos)}genTextPosFrom(t){const e=Math.min(t,this._end);let s=1,o=1;for(let i=0;i<e;i++)this._text.charCodeAt(i)===O?(s++,o=1):o++;return x({row:s,col:o},this.config.contextSliceSize?{contextSlice:this.getTextAround(t,this.config.contextSliceSize)}:{})}getTextSnippet(t,e){return this._text.slice(t,t+e)}}export{m as XmlStream};
|
|
@@ -1 +1 @@
|
|
|
1
|
-
import{getQName as f}from"../get-q-name.js";import{LESS_THAN as
|
|
1
|
+
import{getQName as f}from"../get-q-name.js";import{LESS_THAN as u,GREATER_THAN as c,SLASH as h,EQUALS as W,SINGLE_QUOTE as y,DOUBLE_QUOTE as b,HYPHEN as M,QUESTION_MARK as v,OPEN_SQUARE_BRACKET as B,PERCENT as z,EXCLAMATION_MARK as Y,UPPERCASE_S as H,UPPERCASE_P as K,CLOSE_SQUARE_BRACKET as q}from"./ascii-constants.js";import{isXmlSpaceByte as G}from"./utils.js";import{XmlError as r}from"./XmlError.js";import{XmlStream as $}from"./XmlStream.js";const j="\uFEFF",I="<?xml ",A="?>",J="<!ENTITY",V="<!ELEMENT",Z="<!ATTLIST",tt="<!NOTATION",C="<!DOCTYPE",F="<?",N="?>",S="<!--",L="-->",nt="<![CDATA[",E="]]>",et="NDATA",D="PUBLIC",st="SYSTEM",ot="version",it="encoding",at="standalone";function rt(t,n,e={}){var s,o;(s=e.debugLogger)==null||s.call(e,"Creating XmlStream",{stringLength:t.length});const i=new $(t,e);i.debug("Created XmlStream, starting tokenization",{stringLength:t.length}),(o=e.debugLogger)==null||o.call(e,"Created XmlStream, starting tokenization (2)",{stringLength:t.length}),O(i,n)}function O(t,n){if(t.debug("Starting XML stream tokenization"),t.startsWith(j)&&(t.debug("Skipping BOM"),t.advance(1)),t.startsWith(I)&&(t.debug("Found XML declaration"),ct(t)),p(t,n),t.config.strictDocument){if(t.debug("Parsing in strict document mode"),t.skipSpaces(),t.startsWithIgnoreCase(C)&&(t.debug("Found DOCTYPE declaration"),P(t,n),p(t,n)),t.skipSpaces(),!t.atEnd()&&t.currCodeUnit()===u&&(t.debug("Starting root element parsing"),T(t,n)),p(t,n),!t.atEnd())throw t.debug("Unexpected content after root element",{position:t.getPos(),nextChars:t.getTextSnippet(t.getPos(),10)}),new r({type:"UnknownToken",message:"Not at end"},t.genTextPos())}else for(t.debug("Parsing in non-strict mode");!t.atEnd();)t.currCodeUnit()===u?T(t,n):R(t,n),p(t,n);t.debug("Finished XML stream tokenization")}function p(t,n){for(;!t.atEnd();)if(t.skipSpaces(),t.startsWith(S))U(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(C))P(t,n);else if(t.startsWith(F))k(t,n);else break}function ct(t){function n(e){if(e.startsWithSpace())e.skipSpaces();else if(!e.startsWith(A)&&!e.atEnd())throw new r({type:"InvalidChar",expected:"a whitespace",actual:e.currCodeUnitUnchecked()},e.genTextPos())}if(t.advance(5),n(t),!t.startsWith(ot))throw new r({type:"InvalidString",expected:"version"},t.genTextPos());m(t),n(t),t.startsWith(it)&&(m(t),n(t)),t.startsWith(at)&&m(t),t.skipSpaces(),t.skipString(A)}function U(t,n){const e=t.getPos();t.advance(4);const s=t.consumeCodeUnitsWhile((o,i)=>!(o===M&&i.startsWith(L)));if(t.skipString(L),s.includes("--")||s.endsWith("-"))throw new r({type:"InvalidComment"},t.genTextPosFrom(e));n({type:"Comment",text:s,range:t.rangeFrom(e)},t)}function k(t,n){if(t.startsWith(I))throw new r({type:"UnexpectedDeclaration"},t.genTextPos());const e=t.getPos();t.advance(2);const s=t.consumeName();t.skipSpaces();const o=t.consumeCodeUnitsWhile((i,a)=>!(i===v&&a.startsWith(N)));t.skipString(N),n({type:"ProcessingInstruction",target:s,content:o.length===0?void 0:o,range:t.rangeFrom(e)},t)}function P(t,n){if(!t.config.allowDtd)throw new r({type:"DtdDetected"});const e=t.getPos();if(dt(t),t.skipSpaces(),t.currCodeUnit()===c){t.advance(1);return}for(t.advance(1);!t.atEnd();)if(t.skipSpaces(),t.startsWith(J))lt(t,n);else if(t.startsWith(S))U(t,n);else if(t.startsWith(F))k(t,n);else if(t.startsWith("]"))if(t.advance(1),t.skipSpaces(),t.currCodeUnit()===c){t.advance(1);break}else throw new r({type:"InvalidChar",expected:"'>'",actual:t.currCodeUnitUnchecked()},t.genTextPos());else if(t.startsWith(V)||t.startsWith(Z)||t.startsWith(tt))try{gt(t)}catch(s){throw new r({type:"UnknownToken",message:"Failed to consume declaration"},t.genTextPosFrom(e))}else throw new r({type:"UnknownToken",message:"Failed to parse doctype"},t.genTextPos())}function dt(t){t.advance(9),t.consumeSpaces(),t.skipName(),t.skipSpaces(),Q(t),t.skipSpaces();const n=t.currCodeUnit();if(n!==B&&n!==c)throw new r({type:"InvalidChar",expected:"'[' or '>'",actual:n},t.genTextPos())}function Q(t){if(t.startsWith(st)||t.startsWith(D)){const n=t.getPos();t.advance(6);const e=t.sliceBack(n);t.consumeSpaces();const s=t.consumeQuote();if(t.consumeCodeUnitsWhile(o=>o!==s),t.consumeCodeUnit(s),e===D){t.consumeSpaces();const o=t.consumeQuote();t.consumeCodeUnitsWhile(i=>i!==o),t.consumeCodeUnit(o)}return!0}return!1}function lt(t,n){t.advance(8),t.consumeSpaces();const e=!t.tryConsumeCodeUnit(z);e||t.consumeSpaces();const s=t.consumeName();t.consumeSpaces();const o=ut(t,e);o!==null&&n({type:"EntityDeclaration",name:s,definition:o},t),t.skipSpaces(),t.consumeCodeUnit(c)}function ut(t,n){const e=t.currCodeUnit();if(e===b||e===y){const s=t.consumeQuote(),o=t.getPos();t.skipCodeUnitsWhile(a=>a!==s);const i=t.sliceBack(o);return t.consumeCodeUnit(s),i}else if(e===H||e===K){if(Q(t))return n&&(t.skipSpaces(),t.startsWith(et)&&(t.advance(5),t.consumeSpaces(),t.skipName())),null;throw new r({type:"InvalidExternalID"},t.genTextPos())}else throw new r({type:"InvalidChar",expected:"a quote, SYSTEM or PUBLIC",actual:e},t.genTextPos())}function gt(t){t.skipCodeUnitsWhile(n=>n!==c),t.consumeCodeUnit(c)}function T(t,n){const e=t.getPos();t.advance(1),t.debug("Starting element parsing",{position:e,nextChars:t.getTextSnippet(e,10)});const[s,o]=t.consumeQName();t.debug("Element name parsed",{prefix:s,local:o}),n({type:"ElementStart",prefix:s,local:o,start:e},t);let i=!1;for(;!t.atEnd();){const a=t.startsWithSpace();t.skipSpaces();const d=t.getPos(),g=t.currCodeUnit();if(t.debug("Processing element content",{position:d,currCodeUnit:g,hasSpace:a}),g===h){t.debug("Found self-closing tag"),t.advance(1),t.consumeCodeUnit(c);const l=t.rangeFrom(d);n({type:"ElementEnd",end:{type:"Empty"},range:l},t);break}else if(g===c){t.advance(1);const l=t.rangeFrom(d);if(t.config.implicitSelfClosingElements!=null&&t.config.implicitSelfClosingElements.includes(f(o,s))){t.debug("Found implicit self-closing element",{element:f(o,s)}),n({type:"ElementEnd",end:{type:"Empty"},range:l},t);break}else{t.debug("Found opening tag end"),n({type:"ElementEnd",end:{type:"Open"},range:l},t),i=!0;break}}else{if(!a)throw t.debug("Missing whitespace before attribute",{position:t.getPos(),currCodeUnit:String.fromCharCode(g)}),new r({type:"InvalidChar",expected:"a whitespace",actual:t.currCodeUnitUnchecked()},t.genTextPos());t.debug("Parsing attribute");const[l,w,x]=m(t),X=t.getPos();t.debug("Attribute parsed",{prefix:l,local:w,value:x}),n({type:"Attribute",range:{start:d,end:X},prefix:l,local:w,value:x},t)}}if(i){if(t.config.rawTextElements!=null){const a=f(o,s);if(t.config.rawTextElements.includes(a)){t.debug("Processing raw text element content",{element:a}),_(t,n,`</${a}>`);return}}t.debug("Processing element content"),_(t,n)}t.debug("Finished element parsing",{prefix:s,local:o})}function m(t){const[n,e]=t.consumeQName();let s;const o=t.getPos();if(t.skipSpaces(),t.tryConsumeCodeUnit(W)){t.skipSpaces();const i=t.currCodeUnit();if(i===y||i===b){const a=t.consumeQuote();s=t.consumeCodeUnitsWhile(d=>d!==a&&d!==u),t.consumeCodeUnit(a)}else if(!t.config.strictDocument)s=t.consumeCodeUnitsWhile(a=>!G(a)&&a!==c&&a!==h&&a!==u);else throw new r({type:"InvalidChar",expected:"a quote",actual:i},t.genTextPos())}else if(!t.config.strictDocument)t.goTo(o),s="true";else throw new r({type:"InvalidChar",expected:W,actual:t.currCodeUnit()},t.genTextPos());return[n,e,s]}function _(t,n,e=u){for(;!t.atEnd();)if(typeof e=="number"?t.currCodeUnit()===e:t.startsWith(e)){const s=t.nextCodeUnit();if(s===Y)if(t.startsWith(S))U(t,n);else if(t.startsWith(nt))pt(t,n);else if(!t.config.strictDocument&&t.startsWithIgnoreCase(C))P(t,n);else throw new r({type:"UnknownToken",message:"Failed to parse content"},t.genTextPos());else if(s===v)k(t,n);else if(s===h){mt(t,n);break}else T(t,n)}else R(t,n,e)}function pt(t,n){const e=t.getPos();t.advance(9);const s=t.consumeCodeUnitsWhile((i,a)=>!(i===q&&a.startsWith(E)));t.skipString(E);const o=t.rangeFrom(e);n({type:"Cdata",text:s,range:o},t)}function mt(t,n){const e=t.getPos();t.advance(2);const[s,o]=t.consumeQName();t.skipSpaces(),t.consumeCodeUnit(c);const i=t.rangeFrom(e);n({type:"ElementEnd",end:{type:"Close",prefix:s,local:o},range:i},t)}function R(t,n,e=u){const s=t.getPos(),o=typeof e=="number"?t.consumeCodeUnitsWhile(i=>i!==e):t.consumeCodeUnitsWhile((i,a)=>!a.startsWith(e));if(t.config.strictDocument&&o.includes(E))throw new r({type:"InvalidCharacterData"},t.genTextPos());n({type:"Text",text:o,range:t.rangeFrom(s)},t)}export{rt as tokenize,O as tokenizeXmlStream};
|