@mdgate/email 0.1.0

package/README.md

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

package/dist/email.d.ts

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

package/dist/index.d.ts

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

package/dist/index.js

@@ -0,0 +1,6 @@
+import{hasOleMagic as X}from"@mdgate/containers";import{ConvertError as P}from"@mdgate/core";import{documentToMarkdown as $n}from"@mdgate/document";import{fileExtension as G}from"@mdgate/utils";import{hasOleMagic as Hn}from"@mdgate/containers";import{mimeAttachments as sn,mimeHeader as p,mimeTextHtml as un,mimeTextPlain as an,parseMime as fn}from"@mdgate/containers";import{emptyDocument as D}from"@mdgate/document";import{fileExtension as ln,trim as I}from"@mdgate/utils";import{mimeHeader as Z,parseXml as q}from"@mdgate/containers";import{ConvertError as K}from"@mdgate/core";import{heading as L,plain as m}from"@mdgate/document";import{Stylesheet as Q,toBlocks as nn}from"@mdgate/html";import{decode as tn,isAbsoluteUri as B,trim as k}from"@mdgate/utils";var en={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(B(n))return{type:"external",url:n};return{type:"relative",url:n}},imageSource(n){if(n.length===0)return;if(B(n))return{type:"external",url:n};return},anchorId(n){return n}};function E(n){let t=[],e=n.subject!==void 0?k(n.subject):"";if(e.length>0)t.push(L(1,[m(e)]));return _(t,"From",n.from),_(t,"To",n.to),_(t,"Cc",n.cc),_(t,"Date",n.date),t}function A(n){let t=n.map((e)=>k(e)).filter((e)=>e.length>0);if(t.length===0)return[];return[L(2,[m("Attachments")]),{type:"list",list:{marker:"bullet",start:1,items:t.map((e)=>({blocks:[{type:"paragraph",inlines:[m(e)]}],checked:void 0,markerLabel:void 0}))}}]}function T(n){let t=rn(n);if(t===void 0)return l(cn(n));b(t);let e=new Q;for(let r of t.descendantElems())if(r.local==="style")e.add(r.text());return nn(on(t),e,en)}function l(n){let e=n.replace(/\r\n/g,`
+`).replace(/\r/g,`
+`).split(/\n{2,}/),r=[];for(let o of e){let i=o.split(`
+`).map((s)=>k(s));while(i.length>0&&i[0].length===0)i.shift();while(i.length>0&&i[i.length-1].length===0)i.pop();if(i.length===0)continue;let a=[];for(let s=0;s<i.length;s+=1){if(s>0)a.push({type:"lineBreak"});if(i[s].length>0)a.push(m(i[s]))}if(a.length>0)r.push({type:"paragraph",inlines:a})}return r}function d(n){let t=Z(n,"content-type"),e="utf-8";if(t!==void 0){let r=/charset\s*=\s*("?)([^";\s]+)\1/i.exec(t);if(r!==null)e=r[2]}return w(n.bytes,e)}function w(n,t){try{return tn(n,t)}catch{return new TextDecoder("utf-8",{fatal:!1}).decode(n)}}function _(n,t,e){let r=e!==void 0?k(e):"";if(r.length===0)return;n.push({type:"paragraph",inlines:[m(t),{type:"lineBreak"},m(`: ${r}`)]})}function rn(n){let t=[n,`<div>${n}</div>`];for(let e of t)try{return q(new TextEncoder().encode(e))}catch(r){if(r instanceof K&&r.code==="resourceLimit")throw r}return}function on(n){if(n.local==="body")return n;let t=n.local==="html"?n:n.childElems().find((e)=>e.local==="html");if(t!==void 0)return t.childElems().find((r)=>r.local==="body")??t;return n.childElems().find((e)=>e.local==="body")??n}function b(n){n.local=n.local.toLowerCase();for(let t of n.attrs)t.local=t.local.toLowerCase();for(let t of n.childElems())b(t)}function cn(n){return n.replace(/<[^>]*>/g," ")}function v(n,t){let e=fn(mn(n,t));if(e.contentType==="application/mbox"){let r=D();for(let o=0;o<e.parts.length;o+=1){if(o>0)r.blocks.push({type:"rule"});let i=N(e.parts[o]);r.blocks.push(...i.blocks)}return r}return N(e)}function N(n){let t=D();t.blocks.push(...E({subject:p(n,"subject"),from:p(n,"from"),to:p(n,"to"),cc:p(n,"cc"),date:p(n,"date")}));let e=un(n),r=an(n);if(e!==void 0){let i=T(d(e));if(i.length>0)t.blocks.push(...i);else if(r!==void 0)t.blocks.push(...l(d(r)))}else if(r!==void 0)t.blocks.push(...l(d(r)));else if(n.parts.length===0&&n.bytes.length>0)t.blocks.push(...l(d(n)));let o=sn(n).map((i)=>i.filename??i.contentType??"attachment");return t.blocks.push(...A(o)),t}function mn(n,t){if((t!==void 0?ln(t):void 0)!=="emlx"&&!dn(n))return n;return pn(n)}function dn(n){let t=F(n);if(!/^\d+$/.test(I(t.text)))return!1;let e=n.subarray(t.next),r=0;while(r<e.length&&(e[r]===32||e[r]===9))r+=1;return hn(e,r)}function pn(n){let t=F(n),e=Number(I(t.text));if(!Number.isSafeInteger(e)||e<0)return n;let r=t.next,o=Math.min(n.length,r+e);return n.subarray(r,o)}function F(n){let t=0;if(n.length>=3&&n[0]===239&&n[1]===187&&n[2]===191)t=3;let e=t;while(t<n.length&&n[t]!==10&&n[t]!==13)t+=1;let r=new TextDecoder("utf-8",{fatal:!1}).decode(n.subarray(e,t));if(n[t]===13)t+=1;if(n[t]===10)t+=1;return{text:r,next:t}}function hn(n,t){let e=t;while(e<n.length&&n[e]!==10&&n[e]!==13&&n[e]!==58)e+=1;if(e>=n.length||n[e]!==58||e===t)return!1;for(let r=t;r<e;r+=1){let o=n[r];if(!(o>=65&&o<=90||o>=97&&o<=122||o>=48&&o<=57||o===45))return!1}return!0}import{CompoundFile as j,hasOleMagic as gn,mimeHeader as h,parseMime as xn}from"@mdgate/containers";import{ConvertError as S}from"@mdgate/core";import{emptyDocument as _n}from"@mdgate/document";import{trim as g}from"@mdgate/utils";var kn=55,En=57,An=66,Tn=101,wn=125,Sn=3098,Cn=3103,Mn=3587,Pn=3588,Un=3590,H=4096,Rn=4115,Bn=12289,Ln=14084,bn=14087,Nn=11644473600000;function M(n){if(!gn(n))return!1;try{let t=j.open(n);for(let e of t.readRootStorage())if(y(e.name))return!0}catch{return!1}return!1}function W(n){let t;try{t=j.open(n)}catch(f){if(f instanceof S)throw f;throw S.malformed("not an Outlook message")}if(!Dn(t))throw S.unsupported("ole");let e=In(t),r=O(u(t,Sn),u(t,Cn))??O(u(t,An),u(t,Tn))??e.from,o=u(t,kn)??e.subject,i=u(t,Pn)??e.to,a=u(t,Mn)??e.cc,s=$(t,En)??$(t,Un)??e.date,c=_n();c.blocks.push(...E({subject:o,from:r,to:i,cc:a,date:s}));let x=u(t,Rn);if(x!==void 0&&g(x).length>0){let f=T(x);if(f.length>0)c.blocks.push(...f);else{let R=u(t,H);if(R!==void 0)c.blocks.push(...l(R))}}else{let f=u(t,H);if(f!==void 0)c.blocks.push(...l(f))}return c.blocks.push(...A(vn(t))),c}function Dn(n){for(let t of n.readRootStorage())if(y(t.name))return!0;return!1}function y(n){let t=n.toLowerCase();return t.startsWith("__substg1.0_")||t.startsWith("__properties_version1.0")||t.startsWith("__nameid_version1.0")||t.startsWith("__recip_version1.0_")||t.startsWith("__attach_version1.0_")}function In(n){let t=u(n,wn);if(t===void 0||g(t).length===0)return{};let e=xn(new TextEncoder().encode(`${t}\r
+\r
+`));return{subject:h(e,"subject"),from:h(e,"from"),to:h(e,"to"),cc:h(e,"cc"),date:h(e,"date")}}function vn(n){let t=[];for(let e of n.readRootStorage()){if(!e.name.toLowerCase().startsWith("__attach_version1.0_"))continue;let r=`${e.name}/`,o=u(n,bn,r)??u(n,Ln,r)??u(n,Bn,r);if(o!==void 0&&g(o).length>0)t.push(o);else t.push("attachment")}return t}function u(n,t,e=""){let r=`${e}__substg1.0_${z(t)}`;if(n.exists(`${r}001F`))return C(Fn(n.readStream(`${r}001F`)));if(n.exists(`${r}001E`))return C(w(n.readStream(`${r}001E`),"windows-1252"));if(n.exists(`${r}0102`))return C(w(n.readStream(`${r}0102`),"utf-8"));return}function $(n,t){let e=`__substg1.0_${z(t)}0040`;if(!n.exists(e))return;let r=n.readStream(e);if(r.length<8)return;let o=new DataView(r.buffer,r.byteOffset,r.byteLength),i=BigInt(o.getUint32(0,!0)),a=BigInt(o.getUint32(4,!0)),s=Number((i+(a<<32n))/10000n)-Nn,c=new Date(s);if(Number.isNaN(c.getTime()))return;return c.toUTCString()}function O(n,t){let e=n!==void 0?g(n):"",r=t!==void 0?g(t):"";if(e.length>0&&r.length>0&&e!==r)return`${e} <${r}>`;if(r.length>0)return r;if(e.length>0)return e;return}function Fn(n){let t=n.length;if(t>=2&&n[t-2]===0&&n[t-1]===0)t-=2;return new TextDecoder("utf-16le").decode(n.subarray(0,t))}function C(n){let t=n.indexOf("\x00");return t<0?n:n.slice(0,t)}function z(n){return n.toString(16).toUpperCase().padStart(4,"0")}function Y(n,t){if(Hn(n))return W(n);return v(n,t)}var On=new Set(["eml","msg","mbox","emlx"]);function U(){return{id:"email",sniff(n,t){if(X(n)){if(M(n))return 2;if(t?.path!==void 0&&G(t.path)==="msg")return 1;return 0}if(Wn(n))return 2;if(t?.path!==void 0&&On.has(G(t.path)??""))return 1;return 0},convert(n,t){return jn(n),{markdown:$n(Y(n,t?.path))}}}}function jn(n){if(Xn(n))throw P.unsupported("pdf");if(Vn(n))throw P.unsupported("zip");if(X(n)&&!M(n))throw P.unsupported("ole")}function Wn(n){let t=J(n);if(t=yn(n,t),zn(n,t))t=Gn(n,t);let e=Math.min(n.length,t+4096);while(t<e){let r=V(n,t,e);if(r.line===t)break;let o=Yn(n,t,r.line);if(o==="from"||o==="mime-version")return!0;t=r.next}return!1}function yn(n,t){let e=t;while(e<n.length&&n[e]>=48&&n[e]<=57)e+=1;if(e===t)return t;while(e<n.length&&(n[e]===32||n[e]===9))e+=1;if(n[e]===13)e+=1;if(n[e]===10)e+=1;else return t;return e}function zn(n,t){return t+5<=n.length&&n[t]===70&&n[t+1]===114&&n[t+2]===111&&n[t+3]===109&&n[t+4]===32}function Yn(n,t,e){let r=-1;for(let s=t;s<e;s+=1)if(n[s]===58){r=s;break}if(r<=t)return;let o=t,i=r;while(o<i&&(n[o]===32||n[o]===9))o+=1;while(i>o&&(n[i-1]===32||n[i-1]===9))i-=1;if(i<=o)return;let a="";for(let s=o;s<i;s+=1){let c=n[s];if(!(c>=65&&c<=90||c>=97&&c<=122||c>=48&&c<=57||c===45))return;a+=String.fromCharCode(c>=65&&c<=90?c+32:c)}return a}function V(n,t,e){let r=t;while(r<e&&n[r]!==10&&n[r]!==13)r+=1;let o=r;if(n[o]===13)o+=1;if(n[o]===10)o+=1;return{line:r,next:o}}function Gn(n,t){return V(n,t,n.length).next}function J(n){let t=0;if(n.length>=3&&n[0]===239&&n[1]===187&&n[2]===191)t=3;while(t<n.length){let e=n[t];if(e!==9&&e!==10&&e!==13&&e!==32)break;t+=1}return t}function Xn(n){return Jn(n,"%PDF-",J(n))}function Vn(n){return n.length>=4&&n[0]===80&&n[1]===75&&n[2]===3&&n[3]===4}function Jn(n,t,e){if(e+t.length>n.length)return!1;for(let r=0;r<t.length;r+=1)if(n[e+r]!==t.charCodeAt(r))return!1;return!0}import{create as Zn}from"@mdgate/core";var qn=Zn([U()]);function Kn(n,t){return qn(n,t)}export{Kn as toMarkdown,U as email};

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 email 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 +1,45 @@
+{
+  "name": "@mdgate/email",
+  "version": "0.1.0",
+  "description": "mdgate email converter (eml, msg, mbox)",
+  "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",
+    "email",
+    "eml",
+    "msg",
+    "mbox"
+  ]
+}