xml-tokenizer 0.0.40 → 0.0.42
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.
- package/dist/cjs/html-to-md.js +8 -0
- package/dist/cjs/index.js +1 -1
- package/dist/cjs/xml-to-object.js +1 -1
- package/dist/cjs/xml-to-string.js +1 -0
- package/dist/esm/html-to-md.js +8 -0
- package/dist/esm/index.js +1 -1
- package/dist/esm/xml-to-object.js +1 -1
- package/dist/esm/xml-to-string.js +1 -0
- package/dist/types/html-to-md.d.ts +19 -0
- package/dist/types/index.d.ts +2 -0
- package/dist/types/xml-to-object.d.ts +1 -2
- package/dist/types/xml-to-string.d.ts +25 -0
- package/package.json +5 -5
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
"use strict";var b=require("./config.js"),u=require("./xml-to-string.js"),C=Object.defineProperty,c=Object.getOwnPropertySymbols,m=Object.prototype.hasOwnProperty,d=Object.prototype.propertyIsEnumerable,f=(e,l,o)=>l in e?C(e,l,{enumerable:!0,configurable:!0,writable:!0,value:o}):e[l]=o,$=(e,l)=>{for(var o in l||(l={}))m.call(l,o)&&f(e,o,l[o]);if(c)for(var o of c(l))d.call(l,o)&&f(e,o,l[o]);return e},X=(e,l)=>{var o={};for(var i in e)m.call(e,i)&&l.indexOf(i)<0&&(o[i]=e[i]);if(e!=null&&c)for(var i of c(e))l.indexOf(i)<0&&d.call(e,i)&&(o[i]=e[i]);return o};function v(e,l={}){const o=l,{xmlOptions:i=b.htmlConfig,transformers:h,skipNodes:p=y}=o,S=X(o,["xmlOptions","transformers","skipNodes"]);return u.xmlToString(e,$({xmlOptions:i,skipNodes:p,transformers:$({h1:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));return n===""?"":`${g(t)}# ${n}`},h2:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));return n===""?"":`${g(t)}## ${n}`},h3:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));return n===""?"":`${g(t)}### ${n}`},h4:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));return n===""?"":`${g(t)}#### ${n}`},h5:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));return n===""?"":`${g(t)}##### ${n}`},h6:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));return n===""?"":`${g(t)}###### ${n}`},p:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));return n===""?"":`${g(t)}${n}`},strong:r=>{const t=s(u.getXmlStringNodeContent(r));return t===""?"":`**${t}**`},b:r=>{const t=s(u.getXmlStringNodeContent(r));return t===""?"":`**${t}**`},em:r=>{const t=s(u.getXmlStringNodeContent(r));return t===""?"":`*${t}*`},i:r=>{const t=s(u.getXmlStringNodeContent(r));return t===""?"":`*${t}*`},a:r=>{var t;const n=(t=r.attributes.find(N=>N.local==="href"))==null?void 0:t.value,a=s(u.getXmlStringNodeContent(r));return a===""?"":n!=null?`[${a}](${n})`:a},ul:(r,t)=>{const n=u.getXmlStringNodeContent(r);return n===""?"":`${g(t)}${n}`},ol:(r,t)=>{const n=u.getXmlStringNodeContent(r);return n===""?"":`${g(t)}${n}`},li:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));if(n==="")return"";const a=O(t);return`${" ".repeat(Math.max(0,a-1))}- ${n}
|
|
2
|
+
`},blockquote:(r,t)=>{const n=s(u.getXmlStringNodeContent(r));return n===""?"":`${g(t)}> ${n}`},hr:(r,t)=>`${g(t)}---`,br:()=>`
|
|
3
|
+
`,code:r=>{const t=s(u.getXmlStringNodeContent(r));return t===""?"":`\`${t}\``},pre:(r,t)=>{const n=u.getXmlStringNodeContent(r);return n===""?"":`${g(t)}\`\`\`
|
|
4
|
+
${n}
|
|
5
|
+
\`\`\``}},h)},S))}function s(e){return e.replace(/\s+/g," ").trim()}function g(e){if(e.length===0)return"";const l=e[e.length-1];if(l==null||l.content.length===0)return"";for(let o=l.content.length-1;o>=0;o--){const i=l.content[o];if(typeof i=="object"&&i.string!=null&&i.string!==""){if(k.has(i.local))return`
|
|
6
|
+
|
|
7
|
+
`;if(x.has(i.local))return`
|
|
8
|
+
`}}return""}function O(e){let l=0;for(let o=e.length-1;o>=0;o--){const i=e[o];i&&(i.local==="ul"||i.local==="ol")&&l++}return l}const k=new Set(["h1","h2","h3","h4","h5","h6","p","div","blockquote","hr","pre","ul","ol","li"]),x=new Set(["strong","b","em","i","a","code","br"]),y=["script","style","meta","link","noscript","iframe","object","embed","param","base","title","head"];exports.htmlToMarkdown=v;
|
package/dist/cjs/index.js
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
"use strict";var E=require("./config.js"),a=require("./get-q-name.js"),A=require("./processor/process.js"),T=require("./processor/processors/path-tracker.js"),
|
|
1
|
+
"use strict";var E=require("./config.js"),a=require("./get-q-name.js"),A=require("./html-to-md.js"),l=require("./processor/process.js"),T=require("./processor/processors/path-tracker.js"),m=require("./processor/validate-processor-order.js"),R=require("./selector/select.js"),C=require("./selector/TokenSelector.js"),t=require("./selector/TokenSelectState.js"),O=require("./selector/TokenSelectStateMachine.js"),n=require("./token-to-xml.js"),e=require("./tokenizer/ascii-constants.js"),i=require("./tokenizer/tokenize.js"),r=require("./tokenizer/utils.js"),N=require("./tokenizer/XmlError.js"),s=require("./tokenizer/XmlStream.js"),c=require("./tokens-to-xml.js"),_=require("./xml-to-object.js"),S=require("./xml-to-simplified-object.js"),o=require("./xml-to-string.js");exports.htmlConfig=E.htmlConfig,exports.svgConfig=E.svgConfig,exports.xmlConfig=E.xmlConfig,exports.getQName=a.getQName,exports.htmlToMarkdown=A.htmlToMarkdown,exports.process=l.process,exports.pathTracker=T.pathTracker,exports.validateProcessorOrder=m.validateProcessorOrder,exports.select=R.select,exports.TokenSelector=C.TokenSelector,exports.EToCacheNodeProps=t.EToCacheNodeProps,exports.ETokenMatchCriteria=t.ETokenMatchCriteria,exports.TokenSelectState=t.TokenSelectState,exports.TokenSelectStateMachine=O.TokenSelectStateMachine,exports.tokenToXml=n.tokenToXml,exports.AMPERSAND=e.AMPERSAND,exports.CARRIAGE_RETURN=e.CARRIAGE_RETURN,exports.CLOSE_SQUARE_BRACKET=e.CLOSE_SQUARE_BRACKET,exports.COLON=e.COLON,exports.DOUBLE_QUOTE=e.DOUBLE_QUOTE,exports.EQUALS=e.EQUALS,exports.EXCLAMATION_MARK=e.EXCLAMATION_MARK,exports.GREATER_THAN=e.GREATER_THAN,exports.HASH=e.HASH,exports.HORIZONTAL_TAB=e.HORIZONTAL_TAB,exports.HYPHEN=e.HYPHEN,exports.LESS_THAN=e.LESS_THAN,exports.LINE_FEED=e.LINE_FEED,exports.LOWERCASE_A=e.LOWERCASE_A,exports.LOWERCASE_F=e.LOWERCASE_F,exports.LOWERCASE_X=e.LOWERCASE_X,exports.LOWERCASE_Z=e.LOWERCASE_Z,exports.NINE=e.NINE,exports.OPEN_SQUARE_BRACKET=e.OPEN_SQUARE_BRACKET,exports.PERCENT=e.PERCENT,exports.PERIOD=e.PERIOD,exports.QUESTION_MARK=e.QUESTION_MARK,exports.SEMICOLON=e.SEMICOLON,exports.SINGLE_QUOTE=e.SINGLE_QUOTE,exports.SLASH=e.SLASH,exports.SPACE=e.SPACE,exports.UNDERSCORE=e.UNDERSCORE,exports.UPPERCASE_A=e.UPPERCASE_A,exports.UPPERCASE_F=e.UPPERCASE_F,exports.UPPERCASE_P=e.UPPERCASE_P,exports.UPPERCASE_S=e.UPPERCASE_S,exports.UPPERCASE_Z=e.UPPERCASE_Z,exports.ZERO=e.ZERO,exports.tokenize=i.tokenize,exports.tokenizeXmlStream=i.tokenizeXmlStream,exports.isAsciiDigit=r.isAsciiDigit,exports.isXmlChar=r.isXmlChar,exports.isXmlName=r.isXmlName,exports.isXmlNameByte=r.isXmlNameByte,exports.isXmlNameStart=r.isXmlNameStart,exports.isXmlSpaceByte=r.isXmlSpaceByte,exports.XmlError=N.XmlError,exports.XmlStream=s.XmlStream,exports.tokensToXml=c.tokensToXml,exports.xmlToObject=_.xmlToObject,exports.processTokenForSimplifiedObject=S.processTokenForSimplifiedObject,exports.xmlToSimplifiedObject=S.xmlToSimplifiedObject,exports.getXmlStringNodeContent=o.getXmlStringNodeContent,exports.xmlToString=o.xmlToString;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
"use strict";var
|
|
1
|
+
"use strict";var c=require("./tokenizer/tokenize.js");function r(t,l={}){const e={local:"root",attributes:[],content:[]},n=[e];return c.tokenize(t,o=>{a(o,n)},l),e.content[0]}function a(t,l){switch(t.type){case"ElementStart":{const e={local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,attributes:[],content:[]},n=l[l.length-1];n!=null&&n.content.push(e),l.push(e);break}case"ElementEnd":{(t.end.type==="Close"||t.end.type==="Empty")&&l.pop();break}case"Attribute":{const e=l[l.length-1];e!=null&&e.attributes.push({local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,value:t.value});break}case"Text":case"Cdata":{const e=l[l.length-1];e!=null&&t.text.trim().length>0&&e.content.push(t.text);break}}}exports.xmlToObject=r;
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
"use strict";var i=require("./get-q-name.js"),k=require("./tokenizer/tokenize.js"),g=Object.defineProperty,c=Object.getOwnPropertySymbols,m=Object.prototype.hasOwnProperty,b=Object.prototype.propertyIsEnumerable,p=(n,e,t)=>e in n?g(n,e,{enumerable:!0,configurable:!0,writable:!0,value:t}):n[e]=t,d=(n,e)=>{for(var t in e||(e={}))m.call(e,t)&&p(n,t,e[t]);if(c)for(var t of c(e))b.call(e,t)&&p(n,t,e[t]);return n};function v(n,e={}){const{transformers:t={},xmlOptions:s={},skipNodes:r=[],preserveTree:a=!1}=e,o={local:"root",attributes:[],content:[]},u={stack:[o],transformers:d({__default:l},t),skipNodes:new Set(r),preserveTree:a};return k.tokenize(n,f=>{x(f,u)},s),o.string=l(o),o}function x(n,e){switch(n.type){case"ElementStart":{const t=i.getQName(n.local,n.prefix.length>0?n.prefix:void 0).toLowerCase(),s=e.skipNodes.has(t),r={local:n.local,prefix:n.prefix.length>0?n.prefix:void 0,attributes:[],content:[],skip:s},a=e.stack[e.stack.length-1];a!=null&&!a.skip&&a.content.push(r),e.stack.push(r);break}case"ElementEnd":{if(n.end.type==="Close"||n.end.type==="Empty"){const t=e.stack.pop();if(t==null||t.skip)return;const s=i.getQName(t.local,t.prefix).toLowerCase(),r=e.transformers[s];r!=null?t.string=r(t,e.stack):e.transformers.__default!=null&&(t.string=e.transformers.__default(t,e.stack)),e.preserveTree||(t.content=[])}break}case"Attribute":{const t=e.stack[e.stack.length-1];if(t==null||t.skip)return;t.attributes.push({local:n.local,prefix:n.prefix.length>0?n.prefix:void 0,value:n.value});break}case"Text":case"Cdata":{const t=e.stack[e.stack.length-1];if(t==null||t.skip)return;n.text.trim().length>0&&t.content.push(n.text);break}}}function l(n){return n.content.map(e=>typeof e=="string"?e:e.string).join("")}exports.getXmlStringNodeContent=l,exports.xmlToString=v;
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
import{htmlConfig as O}from"./config.js";import{xmlToString as x,getXmlStringNodeContent as u}from"./xml-to-string.js";var y=Object.defineProperty,f=Object.getOwnPropertySymbols,$=Object.prototype.hasOwnProperty,p=Object.prototype.propertyIsEnumerable,h=(e,l,o)=>l in e?y(e,l,{enumerable:!0,configurable:!0,writable:!0,value:o}):e[l]=o,m=(e,l)=>{for(var o in l||(l={}))$.call(l,o)&&h(e,o,l[o]);if(f)for(var o of f(l))p.call(l,o)&&h(e,o,l[o]);return e},k=(e,l)=>{var o={};for(var s in e)$.call(e,s)&&l.indexOf(s)<0&&(o[s]=e[s]);if(e!=null&&f)for(var s of f(e))l.indexOf(s)<0&&p.call(e,s)&&(o[s]=e[s]);return o};function j(e,l={}){const o=l,{xmlOptions:s=O,transformers:b,skipNodes:g=P}=o,d=k(o,["xmlOptions","transformers","skipNodes"]);return x(e,m({xmlOptions:s,skipNodes:g,transformers:m({h1:(n,t)=>{const r=i(u(n));return r===""?"":`${a(t)}# ${r}`},h2:(n,t)=>{const r=i(u(n));return r===""?"":`${a(t)}## ${r}`},h3:(n,t)=>{const r=i(u(n));return r===""?"":`${a(t)}### ${r}`},h4:(n,t)=>{const r=i(u(n));return r===""?"":`${a(t)}#### ${r}`},h5:(n,t)=>{const r=i(u(n));return r===""?"":`${a(t)}##### ${r}`},h6:(n,t)=>{const r=i(u(n));return r===""?"":`${a(t)}###### ${r}`},p:(n,t)=>{const r=i(u(n));return r===""?"":`${a(t)}${r}`},strong:n=>{const t=i(u(n));return t===""?"":`**${t}**`},b:n=>{const t=i(u(n));return t===""?"":`**${t}**`},em:n=>{const t=i(u(n));return t===""?"":`*${t}*`},i:n=>{const t=i(u(n));return t===""?"":`*${t}*`},a:n=>{var t;const r=(t=n.attributes.find(v=>v.local==="href"))==null?void 0:t.value,c=i(u(n));return c===""?"":r!=null?`[${c}](${r})`:c},ul:(n,t)=>{const r=u(n);return r===""?"":`${a(t)}${r}`},ol:(n,t)=>{const r=u(n);return r===""?"":`${a(t)}${r}`},li:(n,t)=>{const r=i(u(n));if(r==="")return"";const c=w(t);return`${" ".repeat(Math.max(0,c-1))}- ${r}
|
|
2
|
+
`},blockquote:(n,t)=>{const r=i(u(n));return r===""?"":`${a(t)}> ${r}`},hr:(n,t)=>`${a(t)}---`,br:()=>`
|
|
3
|
+
`,code:n=>{const t=i(u(n));return t===""?"":`\`${t}\``},pre:(n,t)=>{const r=u(n);return r===""?"":`${a(t)}\`\`\`
|
|
4
|
+
${r}
|
|
5
|
+
\`\`\``}},b)},d))}function i(e){return e.replace(/\s+/g," ").trim()}function a(e){if(e.length===0)return"";const l=e[e.length-1];if(l==null||l.content.length===0)return"";for(let o=l.content.length-1;o>=0;o--){const s=l.content[o];if(typeof s=="object"&&s.string!=null&&s.string!==""){if(S.has(s.local))return`
|
|
6
|
+
|
|
7
|
+
`;if(N.has(s.local))return`
|
|
8
|
+
`}}return""}function w(e){let l=0;for(let o=e.length-1;o>=0;o--){const s=e[o];s&&(s.local==="ul"||s.local==="ol")&&l++}return l}const S=new Set(["h1","h2","h3","h4","h5","h6","p","div","blockquote","hr","pre","ul","ol","li"]),N=new Set(["strong","b","em","i","a","code","br"]),P=["script","style","meta","link","noscript","iframe","object","embed","param","base","title","head"];export{j as htmlToMarkdown};
|
package/dist/esm/index.js
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
import{htmlConfig as r,svgConfig as
|
|
1
|
+
import{htmlConfig as r,svgConfig as t,xmlConfig as E}from"./config.js";import{getQName as S}from"./get-q-name.js";import{htmlToMarkdown as R}from"./html-to-md.js";import{process as p}from"./processor/process.js";import{pathTracker as T}from"./processor/processors/path-tracker.js";import{validateProcessorOrder as l}from"./processor/validate-processor-order.js";import{select as O}from"./selector/select.js";import{TokenSelector as _}from"./selector/TokenSelector.js";import{EToCacheNodeProps as P,ETokenMatchCriteria as n,TokenSelectState as s}from"./selector/TokenSelectState.js";import{TokenSelectStateMachine as L}from"./selector/TokenSelectStateMachine.js";import{tokenToXml as X}from"./token-to-xml.js";import{AMPERSAND as g,CARRIAGE_RETURN as I,CLOSE_SQUARE_BRACKET as H,COLON as M,DOUBLE_QUOTE as d,EQUALS as h,EXCLAMATION_MARK as Q,GREATER_THAN as B,HASH as D,HORIZONTAL_TAB as F,HYPHEN as K,LESS_THAN as W,LINE_FEED as Z,LOWERCASE_A as b,LOWERCASE_F as j,LOWERCASE_X as G,LOWERCASE_Z as v,NINE as y,OPEN_SQUARE_BRACKET as z,PERCENT as w,PERIOD as Y,QUESTION_MARK as q,SEMICOLON as u,SINGLE_QUOTE as J,SLASH as V,SPACE as $,UNDERSCORE as oo,UPPERCASE_A as eo,UPPERCASE_F as ro,UPPERCASE_P as to,UPPERCASE_S as Eo,UPPERCASE_Z as mo,ZERO as So}from"./tokenizer/ascii-constants.js";import{tokenize as Ro,tokenizeXmlStream as io}from"./tokenizer/tokenize.js";import{isAsciiDigit as Co,isXmlChar as To,isXmlName as fo,isXmlNameByte as lo,isXmlNameStart as xo,isXmlSpaceByte as Oo}from"./tokenizer/utils.js";import{XmlError as _o}from"./tokenizer/XmlError.js";import{XmlStream as Po}from"./tokenizer/XmlStream.js";import{tokensToXml as so}from"./tokens-to-xml.js";import{xmlToObject as Lo}from"./xml-to-object.js";import{processTokenForSimplifiedObject as Xo,xmlToSimplifiedObject as ko}from"./xml-to-simplified-object.js";import{getXmlStringNodeContent as Io,xmlToString as Ho}from"./xml-to-string.js";export{g as AMPERSAND,I as CARRIAGE_RETURN,H as CLOSE_SQUARE_BRACKET,M as COLON,d as DOUBLE_QUOTE,h as EQUALS,P as EToCacheNodeProps,n as ETokenMatchCriteria,Q as EXCLAMATION_MARK,B as GREATER_THAN,D as HASH,F as HORIZONTAL_TAB,K as HYPHEN,W as LESS_THAN,Z as LINE_FEED,b as LOWERCASE_A,j as LOWERCASE_F,G as LOWERCASE_X,v as LOWERCASE_Z,y as NINE,z as OPEN_SQUARE_BRACKET,w as PERCENT,Y as PERIOD,q as QUESTION_MARK,u as SEMICOLON,J as SINGLE_QUOTE,V as SLASH,$ as SPACE,s as TokenSelectState,L as TokenSelectStateMachine,_ as TokenSelector,oo as UNDERSCORE,eo as UPPERCASE_A,ro as UPPERCASE_F,to as UPPERCASE_P,Eo as UPPERCASE_S,mo as UPPERCASE_Z,_o as XmlError,Po as XmlStream,So as ZERO,S as getQName,Io as getXmlStringNodeContent,r as htmlConfig,R as htmlToMarkdown,Co as isAsciiDigit,To as isXmlChar,fo as isXmlName,lo as isXmlNameByte,xo as isXmlNameStart,Oo as isXmlSpaceByte,T as pathTracker,p as process,Xo as processTokenForSimplifiedObject,O as select,t as svgConfig,X as tokenToXml,Ro as tokenize,io as tokenizeXmlStream,so as tokensToXml,l as validateProcessorOrder,E as xmlConfig,Lo as xmlToObject,ko as xmlToSimplifiedObject,Ho as xmlToString};
|
|
@@ -1 +1 @@
|
|
|
1
|
-
import{tokenize as
|
|
1
|
+
import{tokenize as r}from"./tokenizer/tokenize.js";function c(t,l={}){const e={local:"root",attributes:[],content:[]},n=[e];return r(t,o=>{a(o,n)},l),e.content[0]}function a(t,l){switch(t.type){case"ElementStart":{const e={local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,attributes:[],content:[]},n=l[l.length-1];n!=null&&n.content.push(e),l.push(e);break}case"ElementEnd":{(t.end.type==="Close"||t.end.type==="Empty")&&l.pop();break}case"Attribute":{const e=l[l.length-1];e!=null&&e.attributes.push({local:t.local,prefix:t.prefix.length>0?t.prefix:void 0,value:t.value});break}case"Text":case"Cdata":{const e=l[l.length-1];e!=null&&t.text.trim().length>0&&e.content.push(t.text);break}}}export{c as xmlToObject};
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
import{getQName as i}from"./get-q-name.js";import{tokenize as k}from"./tokenizer/tokenize.js";var m=Object.defineProperty,c=Object.getOwnPropertySymbols,g=Object.prototype.hasOwnProperty,b=Object.prototype.propertyIsEnumerable,p=(n,e,t)=>e in n?m(n,e,{enumerable:!0,configurable:!0,writable:!0,value:t}):n[e]=t,d=(n,e)=>{for(var t in e||(e={}))g.call(e,t)&&p(n,t,e[t]);if(c)for(var t of c(e))b.call(e,t)&&p(n,t,e[t]);return n};function v(n,e={}){const{transformers:t={},xmlOptions:s={},skipNodes:r=[],preserveTree:o=!1}=e,a={local:"root",attributes:[],content:[]},f={stack:[a],transformers:d({__default:l},t),skipNodes:new Set(r),preserveTree:o};return k(n,u=>{x(u,f)},s),a.string=l(a),a}function x(n,e){switch(n.type){case"ElementStart":{const t=i(n.local,n.prefix.length>0?n.prefix:void 0).toLowerCase(),s=e.skipNodes.has(t),r={local:n.local,prefix:n.prefix.length>0?n.prefix:void 0,attributes:[],content:[],skip:s},o=e.stack[e.stack.length-1];o!=null&&!o.skip&&o.content.push(r),e.stack.push(r);break}case"ElementEnd":{if(n.end.type==="Close"||n.end.type==="Empty"){const t=e.stack.pop();if(t==null||t.skip)return;const s=i(t.local,t.prefix).toLowerCase(),r=e.transformers[s];r!=null?t.string=r(t,e.stack):e.transformers.__default!=null&&(t.string=e.transformers.__default(t,e.stack)),e.preserveTree||(t.content=[])}break}case"Attribute":{const t=e.stack[e.stack.length-1];if(t==null||t.skip)return;t.attributes.push({local:n.local,prefix:n.prefix.length>0?n.prefix:void 0,value:n.value});break}case"Text":case"Cdata":{const t=e.stack[e.stack.length-1];if(t==null||t.skip)return;n.text.trim().length>0&&t.content.push(n.text);break}}}function l(n){return n.content.map(e=>typeof e=="string"?e:e.string).join("")}export{l as getXmlStringNodeContent,v as xmlToString};
|
|
@@ -0,0 +1,19 @@
|
|
|
1
|
+
import { TXmlStringNode, TXmlToStringOptions } from './xml-to-string';
|
|
2
|
+
/**
|
|
3
|
+
* Converts HTML to Markdown using a streaming XML tokenizer approach.
|
|
4
|
+
*
|
|
5
|
+
* @param html - The HTML string to convert
|
|
6
|
+
* @param options - Configuration options for the conversion
|
|
7
|
+
* @returns The converted Markdown string with node tree
|
|
8
|
+
*
|
|
9
|
+
* @remarks
|
|
10
|
+
* This implementation uses a streaming approach where transformers can only see
|
|
11
|
+
* previously processed nodes. Spacing is determined by looking at the previous
|
|
12
|
+
* sibling element type (block vs inline).
|
|
13
|
+
*/
|
|
14
|
+
export declare function htmlToMarkdown(html: string, options?: THtmlToMarkdownOptions): TXmlStringNode & {
|
|
15
|
+
string: string;
|
|
16
|
+
};
|
|
17
|
+
interface THtmlToMarkdownOptions extends TXmlToStringOptions {
|
|
18
|
+
}
|
|
19
|
+
export {};
|
package/dist/types/index.d.ts
CHANGED
|
@@ -1,5 +1,6 @@
|
|
|
1
1
|
export * from './config';
|
|
2
2
|
export * from './get-q-name';
|
|
3
|
+
export * from './html-to-md';
|
|
3
4
|
export * from './processor';
|
|
4
5
|
export * from './selector';
|
|
5
6
|
export * from './token-to-xml';
|
|
@@ -7,3 +8,4 @@ export * from './tokenizer';
|
|
|
7
8
|
export * from './tokens-to-xml';
|
|
8
9
|
export * from './xml-to-object';
|
|
9
10
|
export * from './xml-to-simplified-object';
|
|
11
|
+
export * from './xml-to-string';
|
|
@@ -1,6 +1,5 @@
|
|
|
1
|
-
import { type TXmlStreamOptions
|
|
1
|
+
import { type TXmlStreamOptions } from './tokenizer';
|
|
2
2
|
export declare function xmlToObject(xmlString: string, options?: TXmlStreamOptions): TXmlNode;
|
|
3
|
-
export declare function processTokenForObject(token: TXmlToken, stack: TXmlNode[]): void;
|
|
4
3
|
export interface TXmlNode {
|
|
5
4
|
local: string;
|
|
6
5
|
prefix?: string;
|
|
@@ -0,0 +1,25 @@
|
|
|
1
|
+
import { type TXmlStreamOptions } from './tokenizer';
|
|
2
|
+
export declare function xmlToString(xmlString: string, options?: TXmlToStringOptions): TXmlStringNode & {
|
|
3
|
+
string: string;
|
|
4
|
+
};
|
|
5
|
+
export type TXmlToStringOptions = {
|
|
6
|
+
transformers?: TXmlStringTransformers;
|
|
7
|
+
xmlOptions?: TXmlStreamOptions;
|
|
8
|
+
skipNodes?: string[];
|
|
9
|
+
preserveTree?: boolean;
|
|
10
|
+
};
|
|
11
|
+
export type TXmlStringTransformer = (node: TXmlStringNode, stack: TXmlStringNode[]) => string;
|
|
12
|
+
export type TXmlStringTransformers = Record<string, TXmlStringTransformer>;
|
|
13
|
+
export interface TXmlStringNode {
|
|
14
|
+
local: string;
|
|
15
|
+
prefix?: string;
|
|
16
|
+
attributes: {
|
|
17
|
+
local: string;
|
|
18
|
+
prefix?: string;
|
|
19
|
+
value: string;
|
|
20
|
+
}[];
|
|
21
|
+
content: (TXmlStringNode | string)[];
|
|
22
|
+
skip?: boolean;
|
|
23
|
+
string?: string;
|
|
24
|
+
}
|
|
25
|
+
export declare function getXmlStringNodeContent(node: TXmlStringNode): string;
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "xml-tokenizer",
|
|
3
|
-
"version": "0.0.
|
|
3
|
+
"version": "0.0.42",
|
|
4
4
|
"private": false,
|
|
5
5
|
"description": "Straightforward and typesafe XML tokenizer that streams tokens through a callback mechanism",
|
|
6
6
|
"keywords": [],
|
|
@@ -23,17 +23,17 @@
|
|
|
23
23
|
"README.md"
|
|
24
24
|
],
|
|
25
25
|
"devDependencies": {
|
|
26
|
-
"@types/node": "^
|
|
26
|
+
"@types/node": "^24.3.0",
|
|
27
27
|
"@types/sax": "^1.2.7",
|
|
28
28
|
"@types/xml2js": "^0.4.14",
|
|
29
29
|
"camaro": "^6.2.3",
|
|
30
|
-
"fast-xml-parser": "^5.2.
|
|
30
|
+
"fast-xml-parser": "^5.2.5",
|
|
31
31
|
"sax": "^1.4.1",
|
|
32
32
|
"saxen": "^10.0.0",
|
|
33
33
|
"txml": "^5.1.1",
|
|
34
34
|
"xml2js": "^0.6.2",
|
|
35
|
-
"
|
|
36
|
-
"
|
|
35
|
+
"@blgc/config": "0.0.37",
|
|
36
|
+
"rollup-presets": "0.0.23"
|
|
37
37
|
},
|
|
38
38
|
"size-limit": [
|
|
39
39
|
{
|