@mdgate/mobi 0.1.0

package/README.md

@@ -0,0 +1,35 @@
+# @mdgate/mobi
+
+Convert PalmDOC, MOBI, and KF8/AZW3 ebooks to Markdown. Outputs GitHub-Flavored Markdown. Works in
+Node, Edge, and browsers. No native addons.
+
+Handles: `.mobi`, `.azw`, `.azw3`, `.prc`
+
+## Usage
+
+```ts
+import { toMarkdown } from '@mdgate/mobi';
+
+const markdown = await toMarkdown(bytes);
+```
+
+Compose with other converters:
+
+```ts
+import { create } from '@mdgate/core';
+import { mobi } from '@mdgate/mobi';
+
+const convert = create([mobi()]);
+```
+
+Part of [mdgate converters](https://github.com/mdgate/converters); install
+`@mdgate/converters` for every format at once.
+
+## Limitations (v1)
+
+- DRM-encrypted books are rejected (`ConvertError.encrypted`).
+- Text records are extracted: uncompressed PalmDOC, PalmDOC LZ77, and HUFF/CDIC.
+- KF8/AZW3 XHTML is converted when it can be isolated from the text stream. Skeleton + fragment
+  reconstruction via INDX is not implemented, so some AZW3 books lose structure.
+- Images, fonts, audio, and NCX/toc indexes are not extracted.
+- Topaz (`TPZ`) Kindle files are not supported.

package/dist/index.d.ts

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

package/dist/index.js

@@ -0,0 +1,3 @@
+import{ConvertError as _}from"@mdgate/core";import{fileExtension as Fn}from"@mdgate/utils";import{parseXml as gn}from"@mdgate/containers";import{ConvertError as a}from"@mdgate/core";import{documentToMarkdown as V,emptyDocument as q,heading as y,plain as N}from"@mdgate/document";import{Stylesheet as hn,toBlocks as an}from"@mdgate/html";import{decode as sn,isAbsoluteUri as G,trim as H}from"@mdgate/utils";function p(n,t){return(n[t]??0)<<8|(n[t+1]??0)}function h(n,t){return((n[t]??0)<<24|(n[t+1]??0)<<16|(n[t+2]??0)<<8|(n[t+3]??0))>>>0}function $(n,t,r){if(t<0||r<=0||t+r>n.length)return"";let e="";for(let o=0;o<r;o+=1)e+=String.fromCharCode(n[t+o]);return e}function k(n,t,r){if(t+r.length>n.length)return!1;for(let e=0;e<r.length;e+=1)if(n[t+e]!==r.charCodeAt(e))return!1;return!0}function M(n,t){let r=new Uint8Array(t),e=0;for(let o of n)r.set(o,e),e+=o.length;return r}import{ConvertError as s}from"@mdgate/core";var L=1,I=2,P=17480,rn=1048576;function Y(n,t,r,e){let o=Math.min(e,rn);if(t===L){if(n.length>o)throw s.resourceLimit("max_entry_bytes","text record exceeds the read cap");return n}if(t===I)return en(n,o);if(t===P){if(r===void 0)throw s.malformed("HUFF/CDIC dictionary is missing");return r.unpack(n,o)}throw s.unsupported(`compression ${t}`)}function en(n,t){let r=new Uint8Array(Math.min(Math.max(n.length*2,64),t)),e=0,o=0;while(o<n.length){let f=n[o];if(o+=1,f>=1&&f<=8){let i=Math.min(f,n.length-o);r=O(r,e,i,t),r.set(n.subarray(o,o+i),e),e+=i,o+=i;continue}if(f<128){r=O(r,e,1,t),r[e]=f,e+=1;continue}if(f>=192){r=O(r,e,2,t),r[e]=32,r[e+1]=f^128,e+=2;continue}if(o>=n.length)break;let l=f<<8|n[o];o+=1;let u=l>>3&2047,g=(l&7)+3;if(u===0||u>e)throw s.malformed("invalid PalmDOC backreference");r=O(r,e,g,t);for(let i=0;i<g;i+=1)r[e+i]=r[e-u+i];e+=g}return r.subarray(0,e)}class D{dict1;minCode;maxCode;dictionary;constructor(n){if(n.length===0)throw s.malformed("HUFF/CDIC dictionary is missing");let t=on(n[0]);this.dict1=t.dict1,this.minCode=t.minCode,this.maxCode=t.maxCode,this.dictionary=[];for(let r=1;r<n.length;r+=1)fn(n[r],this.dictionary);if(this.dictionary.length===0)throw s.malformed("HUFF/CDIC dictionary is empty")}unpack(n,t){return C(n,this.dict1,this.minCode,this.maxCode,this.dictionary,t)}}function on(n){if(n.length<16||!k(n,0,"HUFF")||h(n,4)!==24)throw s.malformed("invalid HUFF header");let t=h(n,8),r=h(n,12);if(t+1024>n.length||r+256>n.length)throw s.malformed("truncated HUFF tables");let e=[];for(let l=0;l<256;l+=1){let u=h(n,t+l*4),g=u&31;if(g===0)throw s.malformed("invalid HUFF dictionary");e.push({codeLen:g,term:(u&128)!==0,maxCode:Z(u>>>8,32-g)})}let o=[0],f=[0];for(let l=0;l<32;l+=1){let u=h(n,r+l*8),g=h(n,r+l*8+4),i=32-(l+1);o.push(ln(u,i)),f.push(Z(g,i))}return{dict1:e,minCode:o,maxCode:f}}function fn(n,t){if(n.length<16||!k(n,0,"CDIC")||h(n,4)!==16)throw s.malformed("invalid CDIC header");let r=h(n,8),e=h(n,12);if(e>16)throw s.malformed("invalid CDIC bit width");let o=Math.min(1<<e,Math.max(0,r-t.length));if(16+o*2>n.length)throw s.malformed("truncated CDIC index");for(let f=0;f<o;f+=1){let u=16+p(n,16+f*2);if(u+2>n.length)throw s.malformed("truncated CDIC phrase");let g=p(n,u),i=g&32767,m=u+2;if(m+i>n.length)throw s.malformed("truncated CDIC phrase");t.push({data:n.subarray(m,m+i),unpacked:(g&32768)!==0})}}function C(n,t,r,e,o,f){let l=n.length*8,u=0,g=W(n,0),i=32,m=[],d=0;for(;;){if(i<=0)u+=4,g=W(n,u),i+=32;let x=Number(g>>BigInt(i)&0xffffffffn)>>>0,{codeLen:c,term:U,maxCode:F}=t[x>>>24];if(!U){while(c<r.length&&x<r[c])c+=1;if(c>=e.length)throw s.malformed("invalid HUFF code");F=e[c]}if(i-=c,l-=c,l<0)break;let X=32-c,tn=X>=32?0:F-x>>>X,A=o[tn];if(A===void 0)throw s.malformed("invalid HUFF phrase");if(!A.unpacked)A.data=C(A.data,t,r,e,o,f),A.unpacked=!0;if(m.push(A.data),d+=A.data.length,d>f)throw s.resourceLimit("max_entry_bytes","decompressed text exceeds the read cap")}return M(m,d)}function W(n,t){let r=0n;for(let e=0;e<8;e+=1)r=r<<8n|BigInt(n[t+e]??0);return r}function ln(n,t){if(t<=0)return n>>>0;if(t>=32)return 0;return Number(BigInt(n)<<BigInt(t)&0xffffffffn)>>>0}function Z(n,t){if(t<=0)return n>>>0;if(t>=32)return 4294967295;return Number((BigInt(n)+1n<<BigInt(t))-1n&0xffffffffn)>>>0}function O(n,t,r,e){if(t+r>e)throw s.resourceLimit("max_entry_bytes","decompressed text exceeds the read cap");if(t+r<=n.length)return n;let o=n.length===0?64:n.length;while(o<t+r)o*=2;let f=new Uint8Array(Math.min(o,e));return f.set(n.subarray(0,t)),f}function K(n,t){let r=n,e=t>>>1;while(e!==0){if((e&1)!==0){let o=un(r);if(o<=0||o>r.length)break;r=r.subarray(0,r.length-o)}e>>>=1}if((t&1)!==0&&r.length>0){let o=(r[r.length-1]&3)+1;if(o<=r.length)r=r.subarray(0,r.length-o)}return r}function un(n){let t=0,r=Math.min(n.length,4);for(let e=0;e<r;e+=1){let o=n[n.length-1-e];if((o&128)!==0)t=0;t=t<<7|o&127}return t}var z=1e5,S=134217728,B=78,w=4294967295,cn=new TextEncoder,mn={linkTarget(n){if(n.length===0)return;if(n.startsWith("#")){let t=n.slice(1);return t.length>0?{type:"anchor",id:t}:void 0}if(G(n))return{type:"external",url:n};return{type:"relative",url:n}},imageSource(n){if(n.length===0)return;if(G(n))return{type:"external",url:n};return},anchorId(n){return n}};function b(n){if(Rn(n)||_n(n)||En(n))return!1;if(n.length>=68){let t=$(n,60,8);if(t==="BOOKMOBI"||t==="TEXtREAd")return!0}return Sn(n)}function v(n){let t=xn(n),r=Mn(t.text);if(r.length>0){let o=q();if(t.title!==void 0)o.blocks.push(y(1,[N(t.title)]));for(let f of r)o.blocks.push(...wn(f));if(o.blocks.length===0)throw a.malformed("empty text");return V(Tn(o,t.title))}let e=q();if(t.title!==void 0)e.blocks.push(y(1,[N(t.title)]));for(let o of Pn(t.text))e.blocks.push({type:"paragraph",inlines:[N(o)]});if(e.blocks.length===0)throw a.malformed("empty text");return V(e)}function xn(n){let t=dn(n);if(t.records.length===0)throw a.missingPart("text");let r=t.records[0],e=J(r);if(e.encryption!==0)throw a.encrypted();let o=Q(r);if(o!==void 0&&j(o))throw a.encrypted();let f=o!==void 0?pn(r,o):void 0,l=nn(o?.encoding??1252),u=Ln(r,t.name,o,f,l),g={records:t.records,header:r,palm:e,mobi:o},i=f?.kf8Boundary;if(i!==void 0&&i+1<t.records.length){let d=t.records[i];if(k(d,0,"BOUNDARY")){let x=t.records[i+1],c=J(x);if(c.encryption!==0)throw a.encrypted();let U=Q(x);if(U!==void 0&&j(U))throw a.encrypted();g={records:t.records,header:x,palm:c,mobi:U}}}let m=kn(g.records,g.palm,g.mobi,l);return{title:u,text:m}}function dn(n){if(n.length<B+8)throw a.malformed("truncated Palm database");let t=p(n,76);if(t===0)throw a.malformed("Palm database has no records");if(t>z)throw a.resourceLimit("max_entry_count",`Palm database has ${t} records`);if(n.length<B+t*8)throw a.malformed("truncated record list");let r=[];for(let o=0;o<t;o+=1)r.push(h(n,B+o*8));let e=[];for(let o=0;o<t;o+=1){let f=r[o],l=o+1<t?r[o+1]:n.length;if(f>n.length||l<f)throw a.malformed("invalid record offset");e.push(n.subarray(f,Math.min(l,n.length)))}return{name:In(n),records:e}}function J(n){if(n.length<16)throw a.malformed("truncated PalmDOC header");let t=p(n,0);if(t!==L&&t!==I&&t!==P)throw a.unsupported(`compression ${t}`);return{compression:t,textLength:h(n,4),recordCount:p(n,8),encryption:p(n,12)}}function Q(n){if(n.length<24||!k(n,16,"MOBI"))return;let t=h(n,20);if(t<8)throw a.malformed("truncated MOBI header");let r=(f,l)=>n.length>=f+4?h(n,f):l,e=t>=228&&n.length>=244?h(n,240):0,o=t>=192&&n.length>=194?p(n,192):1;return{headerLength:t,encoding:n.length>=32?h(n,28):1252,firstNonBook:r(80,w),fullNameOffset:r(84,0),fullNameLength:r(88,0),firstImage:r(108,w),huffOffset:r(112,0),huffCount:r(116,0),exthFlags:r(128,0),drmOffset:r(168,w),drmCount:r(172,0),firstContent:o===0?1:o,extraFlags:e}}function pn(n,t){if((t.exthFlags&64)===0)return;let r=16+t.headerLength;if(r+12>n.length||!k(n,r,"EXTH"))return;let e=h(n,r+4),o=h(n,r+8);if(o>1e4)throw a.malformed("truncated EXTH header");let f=r+12,l=Math.min(n.length,r+e),u,g;for(let i=0;i<o&&f+8<=l;i+=1){let m=h(n,f),d=h(n,f+4);if(d<8||f+d>n.length)break;let x=n.subarray(f+8,f+d);if(m===503){let c=H(T(x,nn(t.encoding)));if(c.length>0)u=c}if(m===121&&x.length>=4)g=h(x,0);f+=d}return{title:u,kf8Boundary:g}}function kn(n,t,r,e){let o=r!==void 0&&r.firstContent>0?r.firstContent:1,f=o+Math.max(t.recordCount,1)-1;if(r!==void 0&&r.firstNonBook!==0&&r.firstNonBook!==w)f=Math.min(f,r.firstNonBook-1);if(r!==void 0&&r.firstImage!==0&&r.firstImage!==w)f=Math.min(f,r.firstImage-1);if(o>=n.length||f<o)throw a.missingPart("text");let l=t.textLength>0?Math.min(t.textLength,S):S,u=An(n,t,r),g=[],i=0;for(let d=o;d<=f&&d<n.length;d+=1){let x=n[d];if(x.length===0)continue;if(r!==void 0&&r.extraFlags!==0)x=K(x,r.extraFlags);let c=Y(x,t.compression,u,l-i);if(g.push(c),i+=c.length,i>S)throw a.resourceLimit("max_entry_bytes","book text exceeds the read cap");if(t.textLength>0&&i>=t.textLength)break}if(i===0)throw a.missingPart("text");let m=M(g,i);if(t.textLength>0&&m.length>t.textLength)m=m.subarray(0,t.textLength);return Dn(T(m,e))}function An(n,t,r){if(t.compression!==P)return;if(r===void 0||r.huffCount===0||r.huffOffset>=n.length)throw a.malformed("HUFF/CDIC dictionary is missing");let e=Math.min(n.length,r.huffOffset+r.huffCount);return new D(n.slice(r.huffOffset,e))}function wn(n){let t=On(n);try{let e=gn(cn.encode(Un(t))),o=Bn(e),f=Hn(e),l=an(o,f,mn);if(l.length>0)return l}catch(e){if(e instanceof a&&e.isFatal())throw e}let r=Nn(n);return r.length===0?[]:[{type:"paragraph",inlines:[N(r)]}]}function Un(n){return n.replace(/<(br|hr|img|meta|link|mbp:pagebreak)(\s[^>/]*)?\s*\/?>/gi,"<$1$2/>")}function Bn(n){if(n.local==="body")return n;let t=n.local==="html"?n:R(n,"html");if(t!==void 0)return R(t,"body")??t;return R(n,"body")??n}function R(n,t){for(let r of n.childElems())if(r.local===t)return r;return}function Hn(n){let t=new hn;for(let r of n.descendantElems())if(r.local==="style")t.add(r.text());return t}function Mn(n){let t=[],r=/<html\b[\s\S]*?<\/html>/gi,e=r.exec(n);while(e!==null)t.push(e[0]),e=r.exec(n);if(t.length>0)return t;if(/<[a-zA-Z][^>]*>/.test(n))return[n];return[]}function On(n){if(/<html[\s>]/i.test(n))return n;return`<html><body>${n}</body></html>`}function Pn(n){let t=n.replace(/\r\n/g,`
+`).replace(/\r/g,`
+`),r=[];for(let e of t.split(/\n(?:[ \t]*\n)+/)){let o=H(e.replace(/[ \t]*\n[ \t]*/g," "));if(o.length>0)r.push(o)}return r}function Nn(n){return H(n.replace(/<[^>]+>/g," ").replace(/\s+/g," "))}function Tn(n,t){if(t===void 0||n.blocks.length<2)return n;let r=n.blocks[0],e=n.blocks[1];if(r?.type!=="heading"||e?.type!=="heading")return n;let o=r.content[0],f=e.content[0];if(o?.type!=="text"||f?.type!=="text")return n;if(o.text.toLowerCase()!==f.text.toLowerCase())return n;return{...n,blocks:n.blocks.slice(1)}}function Ln(n,t,r,e,o){if(e?.title!==void 0&&e.title.length>0)return e.title;if(r!==void 0&&r.fullNameLength>0&&r.fullNameOffset>0){let f=r.fullNameOffset,l=f+r.fullNameLength;if(l<=n.length){let u=H(T(n.subarray(f,l),o));if(u.length>0)return u}}return t.length>0?t:void 0}function In(n){let t=Math.min(32,n.length),r=0;while(r<t&&n[r]!==0)r+=1;if(r===0)return"";return H(T(n.subarray(0,r),"windows-1252"))}function nn(n){if(n===65001||n===65000)return"utf-8";if(n===932)return"shift_jis";if(n===950)return"big5";if(n===949)return"euc-kr";if(n===936)return"gbk";return"windows-1252"}function T(n,t){try{return sn(n,t)}catch{return new TextDecoder("utf-8",{fatal:!1}).decode(n)}}function Dn(n){return n.includes("\x00")?n.replace(/\0+/g,""):n}function j(n){if(n.drmOffset===0||n.drmOffset===w)return!1;return n.drmCount!==0&&n.drmCount!==w}function Sn(n){if(n.length<B+8)return!1;let t=p(n,76);if(t<1||t>z)return!1;let r=h(n,B);return r+20<=n.length&&k(n,r+16,"MOBI")}function Rn(n){let t=0;if(n.length>=3&&n[0]===239&&n[1]===187&&n[2]===191)t=3;while(t<n.length){let r=n[t];if(r!==9&&r!==10&&r!==13&&r!==32)break;t+=1}return t+5<=n.length&&n[t]===37&&n[t+1]===80&&n[t+2]===68&&n[t+3]===70&&n[t+4]===45}function _n(n){return n.length>=8&&n[0]===208&&n[1]===207&&n[2]===17&&n[3]===224&&n[4]===161&&n[5]===177&&n[6]===26&&n[7]===225}function En(n){return n.length>=4&&n[0]===80&&n[1]===75&&(n[2]===3||n[2]===5||n[2]===7)&&(n[3]===4||n[3]===6||n[3]===8)}var Xn=new Set(["mobi","azw","azw3","prc"]);function E(){return{id:"mobi",sniff(n,t){if(b(n))return 2;if(t?.path!==void 0&&Xn.has(Fn(t.path)??""))return 1;return 0},convert(n){return $n(n),{markdown:v(n)}}}}function $n(n){if(Wn(n))throw _.unsupported("pdf");if(Zn(n))throw _.unsupported("ole");if(Yn(n))throw _.unsupported("zip")}function Wn(n){let t=Cn(n);return Kn(n,"%PDF-",t)}function Zn(n){return n.length>=8&&n[0]===208&&n[1]===207&&n[2]===17&&n[3]===224&&n[4]===161&&n[5]===177&&n[6]===26&&n[7]===225}function Yn(n){if(n.length<4||n[0]!==80||n[1]!==75)return!1;return n[2]===3&&n[3]===4||n[2]===5&&n[3]===6||n[2]===7&&n[3]===8}function Cn(n){let t=0;if(n.length>=3&&n[0]===239&&n[1]===187&&n[2]===191)t=3;while(t<n.length){let r=n[t];if(r!==9&&r!==10&&r!==13&&r!==32)break;t+=1}return t}function Kn(n,t,r){if(r+t.length>n.length)return!1;for(let e=0;e<t.length;e+=1)if(n[r+e]!==t.charCodeAt(e))return!1;return!0}import{create as Vn}from"@mdgate/core";var qn=Vn([E()]);function yn(n,t){return qn(n,t)}export{yn as toMarkdown,E as mobi};

package/dist/mobi.d.ts

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

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 mobi 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 +1,46 @@
+{
+  "name": "@mdgate/mobi",
+  "version": "0.1.0",
+  "description": "mdgate MOBI / AZW3 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/html": "0.1.0",
+    "@mdgate/utils": "0.1.2"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "keywords": [
+    "mdgate",
+    "markdown",
+    "mobi",
+    "azw",
+    "azw3",
+    "kindle",
+    "palmdoc"
+  ]
+}