@mdgate/epub 0.1.0

package/README.md

@@ -0,0 +1,26 @@
+# @mdgate/epub
+
+Convert EPUB books to Markdown. Outputs GitHub-Flavored Markdown. Works in Node, Edge, and
+browsers. No native addons.
+
+Handles: `.epub`
+
+## Usage
+
+```ts
+import { toMarkdown } from '@mdgate/epub';
+
+const markdown = await toMarkdown(bytes);
+```
+
+Compose with other converters:
+
+```ts
+import { create } from '@mdgate/core';
+import { epub } from '@mdgate/epub';
+
+const convert = create([epub()]);
+```
+
+Part of [mdgate converters](https://github.com/mdgate/converters); install
+`@mdgate/converters` for every format at once.

package/dist/epub.d.ts

@@ -0,0 +1,2 @@
+import type { Converter } from '@mdgate/core';
+export declare function epub(): Converter;

package/dist/index.d.ts

@@ -0,0 +1,2 @@
+export { epub } from './epub.js';
+export { toMarkdown } from './to-markdown.js';

package/dist/index.js

@@ -0,0 +1 @@
+import{detectZipDoc as v8}from"@mdgate/containers";import{documentToMarkdown as T8}from"@mdgate/document";import{fileExtension as f8}from"@mdgate/utils";import{decodeFragment as M8,Package as Y8,resolve as _}from"@mdgate/containers";import{ConvertError as l}from"@mdgate/core";import{emptyDocument as _8,heading as N8,plain as D8}from"@mdgate/document";import{AssetSink as C8,mediaTypeFor as B8}from"@mdgate/office-common";import{isAbsoluteUri as t,trim as R8,warn as v}from"@mdgate/utils";import{cellSpanning as z8,GridBuilder as J8,inlinesAreEmpty as x,inlinesToPlainText as P,plain as K8,resolveHeaderRows as Q8}from"@mdgate/document";import{deltasEqual as V8,emptyDelta as o,mergeDelta as i,rebaseEmphasis as X8,resolveDelta as k}from"@mdgate/office-common";import{cleanText as y,collapseWs as g,trim as u}from"@mdgate/utils";function m(Z,z,J){let K=new E(z,J,!0);return K.walkChildren(Z,o()),K.finish()}var c={bold:void 0,italic:void 0,strike:void 0,code:void 0},R={delta:c,hidden:void 0};function w(){return{delta:o(),hidden:void 0}}function b(Z,z){return{delta:i(Z.delta,z.delta),hidden:z.hidden??Z.hidden}}function D(Z){return V8(Z.delta,c)&&Z.hidden===void 0}var h=1e5,n=1e6,$8=[],q8=[];class B{rules=[];get isEmpty(){return this.rules.length===0}addFrom(Z){let z=Z.rules;for(let J=0;J<z.length;J+=1)this.rules.push(z[J])}add(Z){if(Z.length===0)return;let z=F8(Z);if(z.length===0)return;for(let J of z.split("}")){let K=J.indexOf("{");if(K<0)continue;let V=J.slice(K+1);if(!W8(V))continue;let Q=r(V);if(D(Q.normal)&&D(Q.important))continue;let X=J.slice(0,K);for(let G of X.split(",")){let q=G.trim();if(q.length===0||q.includes(" ")||q.includes(":")||q.includes("["))continue;let $=q.indexOf("."),L,j;if($>=0){let F=q.slice(0,$);L=F.length===0?void 0:F.toLowerCase(),j=q.slice($+1)}else L=q.toLowerCase();let W=(j!==void 0?10:0)+(L!==void 0?1:0);if(!D(Q.normal))this.rules.push({tag:L,className:j,priority:W,props:Q.normal});if(!D(Q.important))this.rules.push({tag:L,className:j,priority:n+W,props:Q.important})}}}matchingRules(Z,z){let J=this.rules;if(J.length===0)return $8;let K=[];for(let V=0;V<J.length;V+=1){let Q=J[V];if(Q.tag!==void 0&&Q.tag!==Z)continue;if(Q.className!==void 0&&!z.includes(Q.className))continue;K.push([Q.priority,Q.props])}return K}}function W8(Z){return Z.includes("display")||Z.includes("font-weight")||Z.includes("font-style")||Z.includes("text-decoration")}function F8(Z){let z=Z.indexOf("/*");if(z<0)return Z;let J=[],K=0,V=z;while(V>=0){if(V>K)J.push(Z.slice(K,V));let Q=Z.indexOf("*/",V+2);if(Q<0)return J.join("");K=Q+2,V=Z.indexOf("/*",K)}if(K<Z.length)J.push(Z.slice(K));return J.join("")}function r(Z){let z={normal:w(),important:w()};for(let J of Z.split(";")){let K=J.indexOf(":");if(K<0)continue;let V=J.slice(0,K).trim().toLowerCase(),Q=J.slice(K+1).trim().toLowerCase(),X=!1,G=Q.indexOf("!");if(G>=0){if(Q.slice(G+1).trim()==="important")X=!0;Q=Q.slice(0,G).trimEnd()}let q=X?z.important:z.normal;switch(V){case"font-weight":{let $=A(Q);q.delta.bold=Q==="bold"||Q==="bolder"||$!==void 0&&$>=600;break}case"font-style":q.delta.italic=Q==="italic"||Q==="oblique";break;case"text-decoration":case"text-decoration-line":if(Q.includes("line-through"))q.delta.strike=!0;else if(Q==="none")q.delta.strike=!1;break;case"display":q.hidden=Q==="none";break;default:break}}return z}function A(Z){if(!/^\d+$/.test(Z))return;let z=Number(Z);if(!Number.isSafeInteger(z)||z<0||z>4294967295)return;return z}class E{blocks=[];inlines=[];css;ctx;startBoundary;constructor(Z,z,J){this.css=Z,this.ctx=z,this.startBoundary=J}finish(){return this.flushParagraph(),this.blocks}subBlocks(Z,z){return this.subBlocksAt(Z,z,!0)}subBlocksAt(Z,z,J){let K=new E(this.css,this.ctx,J);return K.walkChildren(Z,z),K.finish()}elementProps(Z){let z=Z.attrAny("style");if(this.css.isEmpty&&z===void 0)return R;let J=Z.attrAny("class"),K=J!==void 0&&J.length>0?J.split(/\s+/).filter((X)=>X.length>0):q8,V=this.css.matchingRules(Z.local,K);if(z!==void 0){let X=r(z);V.push([h,X.normal]),V.push([n+h,X.important])}if(V.length===0)return R;if(V.length===1)return b(R,V[0][1]);V.sort((X,G)=>X[0]-G[0]);let Q=w();for(let X=0;X<V.length;X+=1)Q=b(Q,V[X][1]);return Q}pushAnchor(Z){let z=Z.attrAny("id");if(z!==void 0&&z.length>0)this.inlines.push({type:"anchor",id:this.ctx.anchorId(z)});if(Z.local==="a"){let J=Z.attrAny("name");if(J!==void 0&&J.length>0)this.inlines.push({type:"anchor",id:this.ctx.anchorId(J)})}}walkChildren(Z,z){for(let J of Z.children)if(J.type==="text")this.pushText(J.text,z);else this.walkElem(J.elem,z)}pushText(Z,z){let J=g(y(Z));if(J.length===0)return;let K=J;if(S(this.inlines,this.startBoundary))K=O8(J," ");if(K.length===0)return;this.inlines.push({type:"text",text:K,style:k(z)})}walkElem(Z,z){let J=this.elementProps(Z);if(J.hidden===!0)return;z=i(G8(Z,z),J.delta);let K=Z.local;if(K==="h1"||K==="h2"||K==="h3"||K==="h4"||K==="h5"||K==="h6"){this.flushParagraph();let V=Number.parseInt(K.slice(1),10)||1,Q=this.inlineChildren(Z,z);X8(Q,k(z));let X=Z.attrAny("id"),G=X!==void 0?this.ctx.anchorId(X):void 0;if(!x(Q))this.blocks.push({type:"heading",level:V,anchor:G,content:Q});else{let q=[];if(G!==void 0)q.push({type:"anchor",id:G});for(let $ of Q)if($.type==="anchor")q.push($);if(q.length>0)this.blocks.push({type:"paragraph",inlines:q})}return}if(K==="p"){this.flushParagraph(),this.pushAnchor(Z);let V=this.inlines.splice(0,this.inlines.length);if(V.push(...this.inlineChildren(Z,z)),I(V))this.blocks.push({type:"paragraph",inlines:V});return}if(K==="ul"||K==="ol"){this.flushParagraph(),this.blocks.push(...this.parseList(Z,z));return}if(K==="table"){this.flushParagraph();let V=Z.childElems().find((X)=>X.local==="caption");if(V!==void 0){let X=this.inlineChildren(V,z);if(I(X))this.blocks.push({type:"paragraph",inlines:X})}let Q=this.parseTable(Z,z);if(Q!==void 0)this.blocks.push(Q);return}if(K==="blockquote"){this.flushParagraph();let V=this.subBlocks(Z,z);if(V.length>0)this.blocks.push({type:"blockQuote",blocks:V});return}if(K==="pre"){this.flushParagraph();let V=Z.text();if(u(V).length>0)this.blocks.push({type:"codeBlock",lang:void 0,text:V});return}if(K==="hr"){this.flushParagraph(),this.blocks.push({type:"rule"});return}if(s(K)){if(this.pushAnchor(Z),L8(Z))this.flushParagraph(),this.walkChildren(Z,z),this.flushParagraph();else this.walkChildren(Z,z);return}if(K==="script"||K==="style"||K==="head"||K==="template"||K==="noscript")return;this.walkInline(Z,z)}walkInline(Z,z){this.pushAnchor(Z);let J=Z.local;if(J==="br"){this.inlines.push({type:"lineBreak"});return}if(J==="img"||J==="image"){let K=y(Z.attrAny("alt")??""),V=Z.attrAny("src")??Z.attrAny("href")??"",Q=this.ctx.imageSource(V);if(Q!==void 0||u(K).length>0)this.inlines.push({type:"image",alt:K,source:Q??{type:"unavailable"}});return}if(J==="a"){let K=Z.attrAny("href"),V=K!==void 0?this.ctx.linkTarget(K):void 0,Q=this.inlineChildrenAt(Z,z,S(this.inlines,this.startBoundary));if(V!==void 0)this.inlines.push({type:"link",content:Q,target:V});else this.inlines.push(...Q);return}this.walkChildren(Z,z)}inlineChildren(Z,z){return this.inlineChildrenAt(Z,z,!0)}inlineChildrenAt(Z,z,J){let K=this.subBlocksAt(Z,z,J);if(K.length===1&&K[0].type==="paragraph")return K[0].inlines;let V=[];for(let Q=0;Q<K.length;Q+=1){if(Q>0)V.push({type:"lineBreak"});let X=K[Q];if(X.type==="paragraph")V.push(...X.inlines);else if(X.type==="heading")V.push(...X.content);else V.push(K8(g(C(X))))}return V}parseList(Z,z){let J=Z.local==="ol",K=Z.childElems().filter((U)=>U.local==="li");if(K.length===0)return[];if(!J)return[{type:"list",list:{marker:"bullet",start:1,items:K.map((H)=>({blocks:this.subBlocks(H,z),checked:void 0,markerLabel:void 0}))}}];let V=Z.attrAny("type"),Q="decimal";if(V==="a")Q="lowerAlpha";else if(V==="A")Q="upperAlpha";else if(V==="i")Q="lowerRoman";else if(V==="I")Q="upperRoman";let X=Z.attrAny("reversed")!==void 0,G=Z.attrAny("start"),$=(G!==void 0?d(G):void 0)??(X?K.length:1),L=[];for(let U of K){let H=U.attrAny("value"),O=H!==void 0?d(H):void 0;if(O!==void 0)$=O;L.push($),$=X?$-1:$+1}if(L.some((U)=>U<1)){let U=K.map((H,O)=>({blocks:this.subBlocks(H,z),checked:void 0,markerLabel:`${L[O]}.`}));return[{type:"list",list:{marker:Q,start:1,items:U}}]}let j=[],W,F=0;for(let U=0;U<K.length;U+=1){let H=L[U],O={blocks:this.subBlocks(K[U],z),checked:void 0,markerLabel:void 0};if(!(W!==void 0&&F+1===H)){if(W!==void 0)j.push({type:"list",list:W});W={marker:Q,start:H,items:[]}}W.items.push(O),F=H}if(W!==void 0)j.push({type:"list",list:W});return j}parseTable(Z,z){let J=[],K=0,V=!1;for(let $ of Z.childElems())if($.local==="thead"||$.local==="tbody"||$.local==="tfoot"){if(V)V=!1,K+=1;let L=$.local==="thead";for(let j of $.childElems())if(j.local==="tr")J.push([j,L,K]);K+=1}else if($.local==="tr")V=!0,J.push([$,!1,K]);if(J.length===0)return;let Q=new Map;for(let $=0;$<J.length;$+=1)Q.set(J[$][2],$);let X=new J8,G=0;for(let $=0;$<J.length;$+=1){let[L,j,W]=J[$];X.nextRow();let F=!0,U=!1;for(let H of L.childElems()){if(H.local!=="td"&&H.local!=="th")continue;if(U=!0,H.local!=="th")F=!1;let O=p(A(H.attrAny("colspan")??"")??1,1,1000),Y,N=H.attrAny("rowspan"),M=N!==void 0?A(N):void 0;if(M===0)Y=(Q.get(W)??$)-$+1;else if(M!==void 0)Y=p(M,1,65534);else Y=1;X.place(z8(this.subBlocks(H,z),O,Y))}if($===G&&(j||F&&U))G+=1}let q=X.finish("data");if(q.grid.length===0)return;return q.headerRows=Q8(q,G),{type:"table",table:q}}flushParagraph(){if(this.inlines.length>0){let Z=this.inlines.splice(0,this.inlines.length);if(I(Z))this.blocks.push({type:"paragraph",inlines:Z})}this.startBoundary=!0}}function G8(Z,z){let J={...z};switch(Z.local){case"b":case"strong":J.bold=!0;break;case"i":case"em":case"cite":case"dfn":case"var":J.italic=!0;break;case"s":case"del":case"strike":J.strike=!0;break;case"code":case"kbd":case"samp":case"tt":J.code=!0;break;default:break}return J}function C(Z){switch(Z.type){case"paragraph":return P(Z.inlines);case"heading":return P(Z.content);case"list":return Z.list.items.flatMap((z)=>z.blocks.map(C)).join(" ");case"blockQuote":return Z.blocks.map(C).join(" ");case"codeBlock":return Z.text;case"table":return Z.table.grid.flatMap((z)=>z.flatMap((J)=>J.type==="origin"?[J.cell.blocks.map(C).join(" ")]:[])).join(" ");case"rule":return""}}function s(Z){return Z==="div"||Z==="section"||Z==="article"||Z==="aside"||Z==="main"||Z==="nav"||Z==="header"||Z==="footer"||Z==="figure"||Z==="figcaption"||Z==="center"||Z==="details"||Z==="summary"||Z==="li"||Z==="dl"||Z==="dt"||Z==="dd"||Z==="body"}function U8(Z){return s(Z)||Z==="p"||Z==="ul"||Z==="ol"||Z==="table"||Z==="blockquote"||Z==="pre"||Z==="hr"||Z==="h1"||Z==="h2"||Z==="h3"||Z==="h4"||Z==="h5"||Z==="h6"}function L8(Z){return Z.childElems().some((z)=>U8(z.local))}function I(Z){return!x(Z)||Z.some((z)=>z.type==="anchor")}function S(Z,z){for(let J=Z.length-1;J>=0;J-=1){let K=Z[J];if(K.type==="anchor")continue;if(K.type==="text"){if(K.text.length===0)continue;return H8(j8(K.text))}if(K.type==="lineBreak")return!0;if(K.type==="link"){if(x(K.content))continue;return S(K.content,!1)}return!1}return z}function j8(Z){let z=Z.length;if(z===0)return 0;let J=Z.charCodeAt(z-1);if(J>=56320&&J<=57343&&z>=2){let K=Z.charCodeAt(z-2);if(K>=55296&&K<=56319)return(K-55296<<10)+(J-56320)+65536}return J}function H8(Z){return Z>=9&&Z<=13||Z===32||Z===133||Z===160||Z===5760||Z>=8192&&Z<=8202||Z===8232||Z===8233||Z===8239||Z===8287||Z===12288}function O8(Z,z){let J=0;while(J<Z.length&&Z[J]===z)J+=1;return Z.slice(J)}function d(Z){let z=Z.trim();if(!/^-?\d+$/.test(z))return;let J=Number(z);return Number.isFinite(J)?J:void 0}function p(Z,z,J){return Math.min(J,Math.max(z,Z))}function e(Z){let z=Y8.open(Z),J=z.requiredXmlPart("META-INF/container.xml"),K;for(let W of J.descendantsAny("rootfile")){K=W.attrAny("full-path");break}if(K===void 0)throw l.malformedPart("META-INF/container.xml","no rootfile entry");let V=z.requiredXmlPart(K),Q=_8();for(let W of V.descendantsAny("title")){let F=R8(W.text());if(F.length>0)Q.blocks.push(N8(1,[D8(F)]));break}let X=new Map;for(let W of V.descendantsAny("item")){let F=W.attrAny("id"),U=W.attrAny("href");if(F===void 0||U===void 0)continue;X.set(F,{href:U,media:W.attrAny("media-type")??""})}let G=[];for(let W of V.descendantsAny("itemref")){let F=W.attrAny("idref");if(F===void 0)continue;let U=X.get(F);if(U!==void 0)G.push(U.href)}let q=new Set;for(let W of G)try{q.add(_(K,W).path)}catch{}let $=new C8,L=new Map,j=0;for(let W of G){let F;try{F=_(K,W).path}catch(M){v(`skipping chapter with unresolvable href ${x8(W)}: ${E8(M)}`),j+=1;continue}let U=z.optionalXmlPart(F);if(U===void 0){v(`skipping unusable chapter ${F}`),j+=1;continue}let O=U.childElems().find((M)=>M.local==="html")?.childElems().find((M)=>M.local==="body");if(O===void 0){v(`skipping chapter ${F}: no body element`),j+=1;continue}let Y=I8(U,F,z,L),N=new Z8(z,$,F,q);Q.blocks.push({type:"paragraph",inlines:[{type:"anchor",id:F}]}),Q.blocks.push(...m(O,Y,N))}if(G.length>0&&j===G.length)throw l.malformed("no chapter in the book could be read");return Q.assets=$.assets,Q}function I8(Z,z,J,K){let V=new B,Q=Z.childElems();Q.reverse();while(Q.length>0){let X=Q.pop();switch(X.local){case"link":{let q=(X.attrAny("rel")??"").split(/\s+/).some((L)=>A8(L,"stylesheet")),$=X.attrAny("href");if(q&&$!==void 0){let L;try{L=_(z,$).path}catch{break}if(!K.has(L)){let W=J.optionalPart(L);if(W===void 0)K.set(L,void 0);else{let F=new B;F.add(w8(W)),K.set(L,F)}}let j=K.get(L);if(j!==void 0)V.addFrom(j)}break}case"style":V.add(X.text());break;default:{let G=Q.length;for(let q of X.childElems())Q.push(q);S8(Q,G);break}}}return V}class Z8{pkg;assets;chapterPath;spineParts;constructor(Z,z,J,K){this.pkg=Z;this.assets=z;this.chapterPath=J;this.spineParts=K}linkTarget(Z){if(Z.length===0)return;if(Z.startsWith("#")){let z=M8(Z.slice(1));return{type:"anchor",id:T(this.chapterPath,z)}}if(t(Z))return{type:"external",url:Z};try{let z=_(this.chapterPath,Z);if(this.spineParts.has(z.path))return{type:"anchor",id:T(z.path,z.fragment)}}catch{}return{type:"relative",url:Z}}imageSource(Z){if(Z.length===0)return;if(t(Z))return{type:"external",url:Z};let z;try{z=_(this.chapterPath,Z).path}catch{return}let J=this.pkg.optionalPart(z);if(J===void 0)return;return{type:"asset",id:this.assets.add(B8(z),z,J)}}anchorId(Z){return T(this.chapterPath,Z)}}function T(Z,z){if(z!==void 0&&z.length>0)return`${Z}#${z}`;return Z}function w8(Z){return new TextDecoder("utf-8",{fatal:!1}).decode(Z)}function A8(Z,z){if(Z.length!==z.length)return!1;for(let J=0;J<Z.length;J+=1){let K=Z.charCodeAt(J),V=z.charCodeAt(J);if(a(K)!==a(V))return!1}return!0}function a(Z){return Z>=65&&Z<=90?Z+32:Z}function S8(Z,z){let J=z,K=Z.length-1;while(J<K){let V=Z[J];Z[J]=Z[K],Z[K]=V,J+=1,K-=1}}function x8(Z){return JSON.stringify(Z)}function E8(Z){return Z instanceof Error?Z.message:String(Z)}function f(){return{id:"epub",sniff(Z,z){if(v8(Z)==="epub")return 2;if(z?.path!==void 0&&f8(z.path)==="epub")return 1;return 0},convert(Z){return{markdown:T8(e(Z))}}}}import{create as P8}from"@mdgate/core";var k8=P8([f()]);function y8(Z,z){return k8(Z,z)}export{y8 as toMarkdown,f as epub};

package/dist/to-markdown.d.ts

@@ -0,0 +1,6 @@
+import { type ConvertHint } from '@mdgate/core';
+/**
+ * Convert a single file's bytes to Markdown with the epub converter.
+ * `hint.path` is only a sniff hint; it is never read from disk.
+ */
+export declare function toMarkdown(bytes: Uint8Array, hint?: ConvertHint): Promise<string>;

package/package.json

@@ -0,0 +1,43 @@
+{
+  "name": "@mdgate/epub",
+  "version": "0.1.0",
+  "description": "mdgate EPUB converter",
+  "type": "module",
+  "license": "MIT",
+  "sideEffects": false,
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js"
+    }
+  },
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "files": [
+    "dist/**/*.js",
+    "dist/**/*.d.ts"
+  ],
+  "scripts": {
+    "build": "bun ../../scripts/build-package.ts",
+    "prepublishOnly": "bun run build"
+  },
+  "dependencies": {
+    "@mdgate/containers": "0.1.0",
+    "@mdgate/core": "0.1.3",
+    "@mdgate/document": "0.1.0",
+    "@mdgate/office-common": "0.1.0",
+    "@mdgate/utils": "0.1.2"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "keywords": [
+    "mdgate",
+    "markdown",
+    "epub",
+    "ebook"
+  ]
+}