@mdgate/fb2 0.1.0

package/README.md

@@ -0,0 +1,26 @@
+# @mdgate/fb2
+
+Convert FictionBook (FB2) files to Markdown. Outputs GitHub-Flavored Markdown. Works in Node, Edge,
+and browsers. No native addons.
+
+Handles: `.fb2`, `.fb2.zip`
+
+## Usage
+
+```ts
+import { toMarkdown } from '@mdgate/fb2';
+
+const markdown = await toMarkdown(bytes);
+```
+
+Compose with other converters:
+
+```ts
+import { create } from '@mdgate/core';
+import { fb2 } from '@mdgate/fb2';
+
+const convert = create([fb2()]);
+```
+
+Part of [mdgate converters](https://github.com/mdgate/converters); install
+`@mdgate/converters` for every format at once.

package/dist/fb2.d.ts

@@ -0,0 +1,2 @@
+import type { Converter } from '@mdgate/core';
+export declare function fb2(): Converter;

package/dist/index.d.ts

@@ -0,0 +1,2 @@
+export { fb2 } from './fb2.js';
+export { toMarkdown } from './to-markdown.js';

package/dist/index.js

@@ -0,0 +1 @@
+import{detectZipDoc as xn,hasOleMagic as R,Package as wn}from"@mdgate/containers";import{ConvertError as l}from"@mdgate/core";import{documentToMarkdown as Cn}from"@mdgate/document";import{fileExtension as Bn}from"@mdgate/utils";import{ns as $,parseXml as q}from"@mdgate/containers";import{ConvertError as W}from"@mdgate/core";import{cellSpanning as G,emptyDocument as K,GridBuilder as Y,heading as J,inlinesAreEmpty as u,inlinesToPlainText as V,MAX_ASSET_TOTAL_BYTES as nn,PLAIN as p,plain as U,resolveHeaderRows as en}from"@mdgate/document";import{cleanText as w,collapseWs as C,isAbsoluteUri as N,trim as B}from"@mdgate/utils";var v={bold:!1,italic:!0,strike:!1,code:!1},tn=new Set(["notes","comments","footnotes"]);function F(n){let e=q(n),i=rn(e);if(i===void 0)throw W.malformed("not a FictionBook document");let t={assets:[],assetTotal:0,byOrigin:new Map};on(i,t);let r=K(),o=sn(i,r,t),s=[];for(let a of k(i,"body")){let f=(a.attrAny("name")??"").toLowerCase();if(tn.has(f)){cn(a,t,s);continue}an(a,r,t,o)}return r.notes=s,r.assets=t.assets,r}function rn(n){if(n.local==="FictionBook")return n;for(let e of n.childElems())if(e.local==="FictionBook")return e;return}function on(n,e){for(let i of n.childElems()){if(i.local!=="binary")continue;let t=i.attrAny("id");if(t===void 0||t.length===0)continue;let r=bn(i.text());if(r===void 0)continue;let o=i.attrAny("content-type")??yn(r);kn(e,t,o,r)}}function sn(n,e,i){let t=g(n,"description"),r=t!==void 0?g(t,"title-info"):void 0;if(r===void 0)return;let o=y(r,"book-title");if(o.length>0)e.blocks.push(J(1,[U(o)]));let s=[];for(let d of k(r,"author")){let h=mn(d);if(h.length>0)s.push(h)}if(s.length>0)e.blocks.push({type:"paragraph",inlines:[U(s.join(", "))]});let a=g(r,"annotation");if(a!==void 0){let d=E(a,i,2);if(d.length>0)e.blocks.push({type:"blockQuote",blocks:d})}let f=g(r,"coverpage");if(f!==void 0)for(let d of k(f,"image")){let h=S(d,i);if(h!==void 0)e.blocks.push({type:"paragraph",inlines:[h]})}return o.length>0?o:void 0}function an(n,e,i,t){let r=g(n,"title");if(r!==void 0){let s=b(r,i),a=B(V(s));if(!u(s)&&!En(a,t))e.blocks.push({type:"heading",level:t!==void 0?2:1,anchor:void 0,content:s})}let o=t!==void 0?2:1;for(let s of n.childElems()){if(s.local==="title")continue;if(s.local==="section")e.blocks.push(...I(s,i,o));else e.blocks.push(...A(s,i,o))}}function cn(n,e,i){for(let t of k(n,"section"))M(t,e,i)}function M(n,e,i){let t=n.attrAny("id");if(t!==void 0&&t.length>0){let r=un(n,e);if(r.length>0)i.push({id:t,kind:"footnote",blocks:r})}for(let r of k(n,"section"))M(r,e,i)}function un(n,e){let i=[];for(let t of n.childElems()){if(t.local==="section")continue;if(t.local==="title"){let r=b(t,e);if(!u(r))i.push({type:"paragraph",inlines:r});continue}i.push(...A(t,e,2))}return i}function I(n,e,i){let t=[],r=n.attrAny("id"),o=g(n,"title");if(o!==void 0){let a=b(o,e);if(!u(a))t.push({type:"heading",level:i,anchor:r,content:a});else if(r!==void 0)t.push({type:"paragraph",inlines:[{type:"anchor",id:r}]})}else if(r!==void 0)t.push({type:"paragraph",inlines:[{type:"anchor",id:r}]});let s=Math.min(i+1,6);for(let a of n.childElems()){if(a.local==="title")continue;if(a.local==="section")t.push(...I(a,e,s));else t.push(...A(a,e,i))}return t}function E(n,e,i){let t=[];for(let r of n.childElems())if(r.local==="section")t.push(...I(r,e,i));else t.push(...A(r,e,i));return t}function A(n,e,i){switch(n.local){case"p":{let t=c(n,e,p);return u(t)?[]:[{type:"paragraph",inlines:t}]}case"subtitle":{let t=c(n,e,p);if(u(t))return[];return[{type:"heading",level:Math.min(i+1,6),anchor:void 0,content:t}]}case"empty-line":return[];case"poem":return ln(n,e,i);case"cite":case"epigraph":case"annotation":{let t=E(n,e,i);return t.length===0?[]:[{type:"blockQuote",blocks:t}]}case"text-author":{let t=c(n,e,v);return u(t)?[]:[{type:"paragraph",inlines:t}]}case"table":{let t=fn(n,e,i);return t===void 0?[]:[t]}case"image":{let t=S(n,e);return t===void 0?[]:[{type:"paragraph",inlines:[t]}]}case"title":{let t=b(n,e);if(u(t))return[];return[{type:"heading",level:i,anchor:void 0,content:t}]}case"stanza":return O(n,e);case"v":{let t=c(n,e,p);return u(t)?[]:[{type:"paragraph",inlines:t}]}case"date":{let t=c(n,e,p);return u(t)?[]:[{type:"paragraph",inlines:t}]}default:return E(n,e,i)}}function ln(n,e,i){let t=[];for(let r of n.childElems())switch(r.local){case"title":{let o=b(r,e);if(!u(o))t.push({type:"heading",level:Math.min(i+1,6),anchor:void 0,content:o});break}case"epigraph":{let o=E(r,e,i);if(o.length>0)t.push({type:"blockQuote",blocks:o});break}case"stanza":t.push(...O(r,e));break;case"text-author":{let o=c(r,e,v);if(!u(o))t.push({type:"paragraph",inlines:o});break}case"date":{let o=c(r,e,p);if(!u(o))t.push({type:"paragraph",inlines:o});break}default:t.push(...A(r,e,i));break}return t.length===0?[]:[{type:"blockQuote",blocks:t}]}function O(n,e){let i=[],t=[];for(let r of n.childElems()){if(r.local==="title"){let o=b(r,e);if(!u(o))t.push({type:"paragraph",inlines:o});continue}if(r.local==="v"){let o=c(r,e,p);if(!u(o))i.push(o)}}if(i.length>0){let r=[];for(let o=0;o<i.length;o+=1){if(o>0)r.push({type:"lineBreak"});r.push(...i[o])}t.push({type:"paragraph",inlines:r})}return t}function fn(n,e,i){let t=k(n,"tr");if(t.length===0)return;let r=new Y,o=0;for(let a=0;a<t.length;a+=1){let f=t[a];r.nextRow();let d=!0,h=!1;for(let m of f.childElems()){if(m.local!=="td"&&m.local!=="th")continue;if(h=!0,m.local!=="th")d=!1;let Z=L(m.attrAny("colspan")),j=L(m.attrAny("rowspan"));r.place(G(E(m,e,i),Z,j))}if(a===o&&d&&h)o+=1}let s=r.finish("data");if(s.grid.length===0)return;return s.headerRows=en(s,o),{type:"table",table:s}}function c(n,e,i){let t=[];for(let r of n.children){if(r.type==="text"){let o=C(w(r.text));if(o.length>0)t.push({type:"text",text:o,style:i});continue}dn(r.elem,e,i,t)}return t}function dn(n,e,i,t){switch(n.local){case"emphasis":t.push(...c(n,e,{...i,italic:!0}));return;case"strong":t.push(...c(n,e,{...i,bold:!0}));return;case"strikethrough":t.push(...c(n,e,{...i,strike:!0}));return;case"code":t.push(...c(n,e,{...i,code:!0}));return;case"a":{let r=D(n),o=(n.attrAny("type")??"").toLowerCase();if(r!==void 0&&(o==="note"||o==="footnote")){let f=r.startsWith("#")?r.slice(1):r;if(f.length>0)t.push({type:"noteRef",id:f});return}let s=c(n,e,i),a=gn(r);if(a!==void 0)t.push({type:"link",content:s,target:a});else t.push(...s);return}case"image":{let r=S(n,e);if(r!==void 0)t.push(r);return}case"style":case"sub":case"sup":t.push(...c(n,e,i));return;default:if(An(n.local)){let r=C(w(n.text()));if(r.length>0)t.push({type:"text",text:r,style:i});return}t.push(...c(n,e,i))}}function S(n,e){let i=D(n),t=B(n.attrAny("alt")??n.attrAny("title")??""),r=pn(i,e),o=t.length>0?t:hn(i);if(r===void 0&&o.length===0)return;return{type:"image",alt:o,source:r??{type:"unavailable"}}}function pn(n,e){if(n===void 0||n.length===0)return;if(N(n)&&!n.startsWith("#"))return{type:"external",url:n};let i=n.startsWith("#")?n.slice(1):n;if(i.length===0)return;let t=e.byOrigin.get(i);if(t===void 0)return{type:"unavailable"};return{type:"asset",id:t}}function hn(n){if(n===void 0||n.length===0)return"";return n.startsWith("#")?n.slice(1):n}function gn(n){if(n===void 0||n.length===0)return;if(n.startsWith("#")){let e=n.slice(1);return e.length>0?{type:"anchor",id:e}:void 0}if(N(n))return{type:"external",url:n};return{type:"relative",url:n}}function b(n,e){let i=[];for(let t of n.childElems()){if(t.local==="empty-line")continue;let r=c(t,e,p);if(u(r))continue;if(i.length>0)i.push({type:"text",text:" ",style:p});i.push(...r)}if(i.length>0)return i;return c(n,e,p)}function mn(n){let e=y(n,"first-name"),i=y(n,"middle-name"),t=y(n,"last-name"),r=y(n,"nickname"),o=[e,i,t].filter((s)=>s.length>0);if(o.length>0)return o.join(" ");return r}function kn(n,e,i,t){let r=n.byOrigin.get(e);if(r!==void 0)return r;if(n.assetTotal+=t.length,n.assetTotal>nn)throw W.resourceLimit("max_asset_total_bytes","embedded assets exceed the retained-bytes cap");let o=n.assets.length;return n.byOrigin.set(e,o),n.assets.push({id:o,mediaType:i,originPart:e,bytes:t.slice()}),o}function bn(n){let e="";for(let i=0;i<n.length;i+=1){let t=n.charCodeAt(i);if(t===9||t===10||t===13||t===32)continue;e+=n[i]}if(e.length===0)return;try{let i=atob(e),t=new Uint8Array(i.length);for(let r=0;r<i.length;r+=1)t[r]=i.charCodeAt(r);return t}catch{return}}function yn(n){if(n.length>=8&&n[0]===137&&n[1]===80&&n[2]===78&&n[3]===71)return"image/png";if(n.length>=3&&n[0]===255&&n[1]===216&&n[2]===255)return"image/jpeg";if(n.length>=3&&n[0]===71&&n[1]===73&&n[2]===70)return"image/gif";return"application/octet-stream"}function D(n){return n.attr($.XLINK,"href")??n.attrAny("href")}function g(n,e){return n.childElems().find((i)=>i.local===e)}function k(n,e){return n.childElems().filter((i)=>i.local===e)}function y(n,e){let i=g(n,e);return i===void 0?"":B(C(w(i.text())))}function En(n,e){if(e===void 0)return!1;return n.toLowerCase()===e.toLowerCase()}function An(n){return n==="p"||n==="poem"||n==="cite"||n==="table"||n==="subtitle"||n==="empty-line"||n==="section"||n==="epigraph"}function L(n){if(n===void 0)return 1;let e=Number.parseInt(n,10);if(!Number.isFinite(e)||e<1)return 1;return Math.min(e,1000)}var In="http://www.gribuser.ru/xml/fictionbook/";function T(){return{id:"fb2",sniff(n,e){if(z(n))return 2;if(e?.path!==void 0&&Un(e.path))return 1;return 0},convert(n){return{markdown:Cn(F(Sn(n)))}}}}function Sn(n){if(H(n))throw l.unsupported("pdf");if(R(n))throw l.unsupported("ole");if(X(n)){let e=xn(n);if(e!==void 0)throw l.unsupported(e);return Tn(n)}return n}function Tn(n){let e;try{e=wn.open(n)}catch(t){if(t instanceof l){if(P(t))throw l.encrypted();throw t}throw l.malformed("not a readable zip archive")}let i;for(let t of e.partNames()){if(t.endsWith("/")||Ln(t))continue;let r=t.toLowerCase();if(!r.endsWith(".fb2")&&!r.endsWith(".xml"))continue;let o;try{o=e.part(t)}catch(s){if(s instanceof l&&s.isFatal())throw s;if(P(s))throw l.encrypted();continue}if(o===void 0)continue;if(r.endsWith(".fb2"))return o;if(i===void 0&&z(o))i=o}if(i!==void 0)return i;throw l.missingPart("*.fb2")}function z(n){if(H(n)||R(n)||X(n))return!1;let e=Q(n);if(e>=n.length||n[e]!==60)return!1;let i=Fn(n),t=Wn(i,0),r=i.slice(t);if(/^<([A-Za-z_][\w.-]*:)?FictionBook\b/i.test(r))return!0;return r.includes(In)}function Un(n){let i=(n.split(/[/\\]/).pop()??"").toLowerCase();return Bn(n)==="fb2"||i.endsWith(".fb2.zip")}function P(n){if(n instanceof l&&n.code==="encrypted")return!0;let e=n instanceof l?n.detail??"":"",t=(n instanceof Error?`${n.message} ${e}`:String(n)).toLowerCase();return t.includes("encrypted")||t.includes("password")}function Ln(n){for(let e of n.split("/"))if(e==="__MACOSX")return!0;return!1}function X(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)}function H(n){let e=Q(n);return vn(n,"%PDF-",e)}function Q(n){let e=0;if(n.length>=3&&n[0]===239&&n[1]===187&&n[2]===191)e=3;while(e<n.length){let i=n[e];if(i!==9&&i!==10&&i!==13&&i!==32)break;e+=1}return e}function Wn(n,e){let i=x(n,e);if(n.charCodeAt(i)===65279)i=x(n,i+1);if(_(n,"<?xml",i)){let t=n.indexOf("?>",i);if(t<0)return i;i=x(n,t+2)}for(;;){if(n.startsWith("<!--",i)){let t=n.indexOf("-->",i+4);if(t<0)return i;i=x(n,t+3);continue}if(_(n,"<!doctype",i)){i=Nn(n,i),i=x(n,i);continue}break}return i}function Nn(n,e){let i=0;e+=2;while(e<n.length){let t=n[e];if(t==='"'||t==="'"){let r=t;e+=1;while(e<n.length&&n[e]!==r)e+=1;if(e<n.length)e+=1;continue}if(t==="<")i+=1;else if(t===">"){if(i===0)return e+1;i-=1}e+=1}return e}function x(n,e){while(e<n.length){let i=n.charCodeAt(e);if(i!==9&&i!==10&&i!==13&&i!==32)break;e+=1}return e}function _(n,e,i){if(i+e.length>n.length)return!1;for(let t=0;t<e.length;t+=1){let r=n.charCodeAt(i+t),o=e.charCodeAt(t);if(r===o)continue;let s=r>=65&&r<=90?r+32:r,a=o>=65&&o<=90?o+32:o;if(s!==a)return!1}return!0}function vn(n,e,i){if(i+e.length>n.length)return!1;for(let t=0;t<e.length;t+=1)if(n[i+t]!==e.charCodeAt(t))return!1;return!0}function Fn(n){let e=Math.min(n.length,8192);if(n.length>=2&&n[0]===255&&n[1]===254)return new TextDecoder("utf-16le").decode(n.subarray(0,e));if(n.length>=2&&n[0]===254&&n[1]===255)return new TextDecoder("utf-16be").decode(n.subarray(0,e));let i=0;if(n.length>=3&&n[0]===239&&n[1]===187&&n[2]===191)i=3;return new TextDecoder("utf-8",{fatal:!1}).decode(n.subarray(i,i+8192))}import{create as Mn}from"@mdgate/core";var On=Mn([T()]);function Dn(n,e){return On(n,e)}export{Dn as toMarkdown,T as fb2};

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 fb2 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 +1,42 @@
+{
+  "name": "@mdgate/fb2",
+  "version": "0.1.0",
+  "description": "mdgate FictionBook (FB2) 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/utils": "0.1.2"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "keywords": [
+    "mdgate",
+    "markdown",
+    "fb2",
+    "fictionbook"
+  ]
+}