mhtml-stream 2.0.0 → 2.0.2

package/dist/cli.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/cli.js

@@ -0,0 +1,45 @@
+/* eslint-disable no-console */
+import { mkdir } from "node:fs/promises";
+import { argv } from "bun";
+import yargs from "yargs";
+import { parseMhtml } from ".";
+async function* readFile(name) {
+    const file = Bun.file(name);
+    yield await file.bytes();
+}
+async function saveOutput(mkdir, path, content) {
+    await mkdir;
+    await Bun.write(path, content);
+}
+void (async () => {
+    const args = await yargs(argv.slice(2))
+        .scriptName("mhtml_stream")
+        .usage("$0 <mhtml>", "parse mhtml and extract component files")
+        .positional("mhtml", {
+        describe: "the mhtml file to parse",
+        type: "string",
+    })
+        .option("output", {
+        alias: "o",
+        describe: "output directory",
+        type: "string",
+    })
+        .help()
+        .alias("h", "help")
+        .demandOption("mhtml")
+        .strict().argv;
+    const makeOut = args.output ? mkdir(args.output, { recursive: true }) : null;
+    const proms = [];
+    for await (const file of parseMhtml(readFile(args.mhtml))) {
+        for (const [key, val] of file.headers) {
+            console.log(key, ":", val);
+        }
+        const loc = file.headers.get("Content-Location");
+        if (loc && args.output && makeOut) {
+            const path = `${args.output}/${loc.replaceAll(/[:/]/g, "_")}`;
+            proms.push(saveOutput(makeOut, path, file.content));
+        }
+        await Promise.all(proms);
+        console.log("[done]");
+    }
+})();

package/dist/index.d.ts

@@ -1,4 +1,4 @@
-import { MhtmlHeaders } from "./headers";
+import { type MhtmlHeaders } from "./headers";
 export type { MhtmlHeaders };
 /** a file that was encoded in MHTML format */
 export interface MhtmlFile {

package/dist/index.js

@@ -74,7 +74,7 @@ async function parseHeaders(iter) {
         const line = decoder.decode(value);
         if (/^\s/.test(line)) {
             // header folded
-            val += decodeLine(line.substr(1));
+            val += decodeLine(line.substring(1));
         }
         else {
             if (key) {
@@ -113,7 +113,7 @@ function getBoundary(headers) {
     if (contentType === null) {
         throw new Error(`first headers didn't contain a content type: ${JSON.stringify(Object.fromEntries(headers))}`);
     }
-    let bound = undefined;
+    let bound;
     let multipart = false;
     for (const field of contentType.split(/;\s*/)) {
         if (field.startsWith("multipart/")) {

package/dist/mhtml.esm.min.js

@@ -1 +1 @@
-var f=function(M){var Y=M.length;if(Y%4>0)throw new Error("Invalid string. Length must be a multiple of 4");var W=M.indexOf("=");if(W===-1)W=Y;var z=W===Y?0:4-W%4;return[W,z]};var h=function(M,Y,W){return(Y+W)*3/4-W},v=function(M){var Y,W=f(M),z=W[0],G=W[1],J=new m(h(M,z,G)),K=0,R=G>0?z-4:z,U;for(U=0;U<R;U+=4)Y=V[M.charCodeAt(U)]<<18|V[M.charCodeAt(U+1)]<<12|V[M.charCodeAt(U+2)]<<6|V[M.charCodeAt(U+3)],J[K++]=Y>>16&255,J[K++]=Y>>8&255,J[K++]=Y&255;if(G===2)Y=V[M.charCodeAt(U)]<<2|V[M.charCodeAt(U+1)]>>4,J[K++]=Y&255;if(G===1)Y=V[M.charCodeAt(U)]<<10|V[M.charCodeAt(U+1)]<<4|V[M.charCodeAt(U+2)]>>2,J[K++]=Y>>8&255,J[K++]=Y&255;return J};var j=v;var g=[],V=[],m=typeof Uint8Array!=="undefined"?Uint8Array:Array,E="ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";for(X=0,_=E.length;X<_;++X)g[X]=E[X],V[E.charCodeAt(X)]=X;var X,_;V["-".charCodeAt(0)]=62;V["_".charCodeAt(0)]=63;class F{#M=new Map;[Symbol.iterator](){return this.entries()}append(M,Y){const W=this.#M.get(M);if(W===void 0)this.#M.set(M,[Y]);else W.push(Y)}*entries(M=", "){for(let[Y,W]of this.#M.entries())yield[Y,W.join(M)]}*entriesAll(){for(let[M,Y]of this.#M.entries())for(let W of Y)yield[M,W]}get(M,Y=", "){const W=this.#M.get(M);if(W===void 0)return null;else return W.join(Y)}getAll(M){return this.#M.get(M)??[]}has(M){return this.#M.has(M)}keys(){return this.#M.keys()}*values(M=", "){for(let Y of this.#M.values())yield Y.join(M)}*valuesAll(){for(let M of this.#M.values())yield*M}}function N(M,Y){if(M.length!==Y.length)return!1;else if(M.byteOffset%4===Y.byteOffset%4){const W=(4-M.byteOffset%4)%4,z=Math.floor((M.byteLength-W)/4),G=W+z*4,J=new Uint32Array(M.buffer,M.byteOffset+W,z),K=new Uint32Array(Y.buffer,Y.byteOffset+W,z);for(let R=0;R<W;++R)if(M[R]!==Y[R])return!1;for(let R=0;R<z;++R)if(J[R]!==K[R])return!1;for(let R=G;R<M.length;++R)if(M[R]!==Y[R])return!1;return!0}else{for(let W of M.keys())if(M[W]!==Y[W])return!1;return!0}}function y(M,Y){return M.findIndex((W,z)=>{if(W!==Y[0]||z+Y.length>M.length)return!1;else{for(let G=1;G<Y.length;++G)if(M[z+G]!==Y[G])return!1;return!0}})}async function*q(M,Y){let W=new Uint8Array(0);for await(let z of M){W=W.length?A([W,z]):z;let G;while((G=y(W,Y))!==-1)yield W.subarray(0,G),W=W.subarray(G+Y.length)}yield W}var A=function(M){const Y=M.reduce((G,J)=>G+J.length,0),W=new Uint8Array(Y);let z=0;for(let G of M)W.set(G,z),z+=G.length;return W};async function C(M){const Y=[];for await(let W of M)Y.push(W);return A(Y)}async function*O(M,Y=b){for await(let W of M){const z=new Uint8Array(W.length+Y.length);let G=!1,J=0;for(let K=0;K<W.length;++K){const R=W[K];if(R>=128)throw new Error(`got non-ascii character when decoding quoted printable: ${R}`);if(R!==61)z[J++]=R;else{const U=W[++K];if(U===void 0)G=!0;else{const $=W[++K];if($===void 0)throw new Error("quoted printable escape (=) was not followed by two bytes");let Z=parseInt(String.fromCharCode(U),16);Z*=16,Z+=parseInt(String.fromCharCode($),16),z[J++]=Z}}}if(!G)z.set(Y,J),J+=Y.length;yield z.subarray(0,J)}}async function*Q(M){for await(let Y of M)yield j(L.decode(Y))}async function*P(M){for await(let Y of M)yield Y}function T(){throw new Error("binary transfer-encoding is explicitly not supported and trying to add an implementation will likely result in unexpected results, but if you want to ignore anyway, set binary to `decode8bit`")}var b=new Uint8Array([10]),L=new TextDecoder;var k=function(M){const Y=new Uint8Array(M.length);let W=0;for(let z=0;z<M.length;++z){const G=M.charCodeAt(z);let J;if(G>=128)throw new Error(`got non-ascii character when decoding q-quoted word: "${M}"`);else if(G===95)J=32;else if(G===61)J=parseInt(M[++z],16),J*=16,J+=parseInt(M[++z],16);else J=G;Y[W++]=J}return Y.subarray(0,W)},H=function(M){const Y=p.exec(M);if(Y){const[,W,z,G]=Y;let J;if(z==="Q")J=k(G);else J=j(G);return new TextDecoder(W).decode(J)}else return M};async function c(M){const Y=new F;let W="",z="";for(;;){const{done:G,value:J}=await M.next();if(G)throw new Error("didn't find an empty line to signify the end of header parsing");const K=I.decode(J);if(/^\s/.test(K))z+=H(K.substr(1));else{if(W)Y.append(W,z);if(K){const R=K.indexOf(": ");if(R===-1)throw new Error(`header line didn't have key-value delimiter: "${K}"`);W=K.slice(0,R),z=H(K.slice(R+2))}else return Y}}}var s=function(M){const Y=M.get("Content-Type");if(Y===null)throw new Error(`first headers didn't contain a content type: ${JSON.stringify(Object.fromEntries(M))}`);let W=void 0,z=!1;for(let K of Y.split(/;\s*/))if(K.startsWith("multipart/"))z=!0;else if(K.startsWith("boundary=")){if(W=K.slice(9),W.startsWith('"')&&W.endsWith('"'))W=W.slice(1,-1)}if(!z||W===void 0)throw new Error("first content type header didn't contain 'multipart/...' and a boundary string");const G=w.encode(`--${W}`),J=w.encode(`--${W}--`);return[G,J]};async function*e(M,{decoderOverrides:Y=new Map}={}){const W=new Map([...a.entries(),...Y.entries()]),z=new Uint8Array([13,10]),G=q(M,z);let J=null,K=!0;while(K){const R=await c(G),[U,$]=J??(J=s(R)),Z=R.get("Content-Transfer-Encoding")??"7bit",S=W.get(Z);if(S===void 0)throw new Error(`unhandled encoding type: ${Z}`);const B=await C(S({[Symbol.asyncIterator](){return{async next(){const{done:x,value:D}=await G.next();if(x)throw new Error(`stream didn't end with the appropriate termination boundary: ${I.decode($)}`);else if(N(D,U))return{done:!0,value:void 0};else if(N(D,$))return K=!1,{done:!0,value:void 0};else return{value:D}}}}}));yield{headers:R,content:B}}}var w=new TextEncoder,I=new TextDecoder,p=/^=\?([^?\s]+)\?([BQ])\?([^?\s]+)\?=$/,a=new Map([["7bit",P],["base64",Q],["quoted-printable",O],["8bit",P],["binary",T]]);export{e as parseMhtml};
+var D=L;var g=[],Y=[],f=typeof Uint8Array<"u"?Uint8Array:Array,P="ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";for(Z=0,S=P.length;Z<S;++Z)g[Z]=P[Z],Y[P.charCodeAt(Z)]=Z;var Z,S;Y[45]=62;Y[95]=63;function m(G){var K=G.length;if(K%4>0)throw Error("Invalid string. Length must be a multiple of 4");var J=G.indexOf("=");if(J===-1)J=K;var M=J===K?0:4-J%4;return[J,M]}function v(G,K,J){return(K+J)*3/4-J}function L(G){var K,J=m(G),M=J[0],R=J[1],U=new f(v(G,M,R)),V=0,W=R>0?M-4:M,X;for(X=0;X<W;X+=4)K=Y[G.charCodeAt(X)]<<18|Y[G.charCodeAt(X+1)]<<12|Y[G.charCodeAt(X+2)]<<6|Y[G.charCodeAt(X+3)],U[V++]=K>>16&255,U[V++]=K>>8&255,U[V++]=K&255;if(R===2)K=Y[G.charCodeAt(X)]<<2|Y[G.charCodeAt(X+1)]>>4,U[V++]=K&255;if(R===1)K=Y[G.charCodeAt(X)]<<10|Y[G.charCodeAt(X+1)]<<4|Y[G.charCodeAt(X+2)]>>2,U[V++]=K>>8&255,U[V++]=K&255;return U}class _{#G=new Map;[Symbol.iterator](){return this.entries()}append(G,K){let J=this.#G.get(G);if(J===void 0)this.#G.set(G,[K]);else J.push(K)}*entries(G=", "){for(let[K,J]of this.#G.entries())yield[K,J.join(G)]}*entriesAll(){for(let[G,K]of this.#G.entries())for(let J of K)yield[G,J]}get(G,K=", "){let J=this.#G.get(G);if(J===void 0)return null;else return J.join(K)}getAll(G){return this.#G.get(G)??[]}has(G){return this.#G.has(G)}keys(){return this.#G.keys()}*values(G=", "){for(let K of this.#G.values())yield K.join(G)}*valuesAll(){for(let G of this.#G.values())yield*G}}function j(G,K){if(G.length!==K.length)return!1;else if(G.byteOffset%4===K.byteOffset%4){let J=(4-G.byteOffset%4)%4,M=Math.floor((G.byteLength-J)/4),R=J+M*4,U=new Uint32Array(G.buffer,G.byteOffset+J,M),V=new Uint32Array(K.buffer,K.byteOffset+J,M);for(let W=0;W<J;++W)if(G[W]!==K[W])return!1;for(let W=0;W<M;++W)if(U[W]!==V[W])return!1;for(let W=R;W<G.length;++W)if(G[W]!==K[W])return!1;return!0}else{for(let J of G.keys())if(G[J]!==K[J])return!1;return!0}}function h(G,K){return G.findIndex((J,M)=>{if(J!==K[0]||M+K.length>G.length)return!1;else{for(let R=1;R<K.length;++R)if(G[M+R]!==K[R])return!1;return!0}})}async function*Q(G,K){let J=new Uint8Array(0);for await(let M of G){J=J.length?E([J,M]):M;let R;while((R=h(J,K))!==-1)yield J.subarray(0,R),J=J.subarray(R+K.length)}yield J}function E(G){let K=G.reduce((R,U)=>R+U.length,0),J=new Uint8Array(K),M=0;for(let R of G)J.set(R,M),M+=R.length;return J}async function O(G){let K=[];for await(let J of G)K.push(J);return E(K)}var k=new Uint8Array([10]);async function*T(G,K=k){for await(let J of G){let M=new Uint8Array(J.length+K.length),R=!1,U=0;for(let V=0;V<J.length;++V){let W=J[V];if(W>=128)throw Error(`got non-ascii character when decoding quoted printable: ${W}`);if(W!==61)M[U++]=W;else{let X=J[++V];if(X===void 0)R=!0;else{let z=J[++V];if(z===void 0)throw Error("quoted printable escape (=) was not followed by two bytes");let $=parseInt(String.fromCharCode(X),16);$*=16,$+=parseInt(String.fromCharCode(z),16),M[U++]=$}}}if(!R)M.set(K,U),U+=K.length;yield M.subarray(0,U)}}var y=new TextDecoder;async function*C(G){for await(let K of G)yield D(y.decode(K))}async function*q(G){for await(let K of G)yield K}function A(){throw Error("binary transfer-encoding is explicitly not supported and trying to add an implementation will likely result in unexpected results, but if you want to ignore anyway, set binary to `decode8bit`")}var I=new TextEncoder,B=new TextDecoder;function b(G){let K=new Uint8Array(G.length),J=0;for(let M=0;M<G.length;++M){let R=G.charCodeAt(M),U;if(R>=128)throw Error(`got non-ascii character when decoding q-quoted word: "${G}"`);else if(R===95)U=32;else if(R===61)U=parseInt(G[++M],16),U*=16,U+=parseInt(G[++M],16);else U=R;K[J++]=U}return K.subarray(0,J)}var p=/^=\?([^?\s]+)\?([BQ])\?([^?\s]+)\?=$/;function H(G){let K=p.exec(G);if(K){let[,J,M,R]=K,U;if(M==="Q")U=b(R);else U=D(R);return new TextDecoder(J).decode(U)}else return G}async function c(G){let K=new _,J="",M="";for(;;){let{done:R,value:U}=await G.next();if(R)throw Error("didn't find an empty line to signify the end of header parsing");let V=B.decode(U);if(/^\s/.test(V))M+=H(V.substring(1));else{if(J)K.append(J,M);if(V){let W=V.indexOf(": ");if(W===-1)throw Error(`header line didn't have key-value delimiter: "${V}"`);J=V.slice(0,W),M=H(V.slice(W+2))}else return K}}}var u=new Map([["7bit",q],["base64",C],["quoted-printable",T],["8bit",q],["binary",A]]);function s(G){let K=G.get("Content-Type");if(K===null)throw Error(`first headers didn't contain a content type: ${JSON.stringify(Object.fromEntries(G))}`);let J,M=!1;for(let V of K.split(/;\s*/))if(V.startsWith("multipart/"))M=!0;else if(V.startsWith("boundary=")){if(J=V.slice(9),J.startsWith('"')&&J.endsWith('"'))J=J.slice(1,-1)}if(!M||J===void 0)throw Error("first content type header didn't contain 'multipart/...' and a boundary string");let R=I.encode(`--${J}`),U=I.encode(`--${J}--`);return[R,U]}async function*t(G,{decoderOverrides:K=new Map}={}){let J=new Map([...u.entries(),...K.entries()]),M=new Uint8Array([13,10]),R=Q(G,M),U=null,V=!0;while(V){let W=await c(R),[X,z]=U??(U=s(W)),$=W.get("Content-Transfer-Encoding")??"7bit",F=J.get($);if(F===void 0)throw Error(`unhandled encoding type: ${$}`);let w=await O(F({[Symbol.asyncIterator](){return{async next(){let{done:x,value:N}=await R.next();if(x)throw Error(`stream didn't end with the appropriate termination boundary: ${B.decode(z)}`);else if(j(N,X))return{done:!0,value:void 0};else if(j(N,z))return V=!1,{done:!0,value:void 0};else return{value:N}}}}}));yield{headers:W,content:w}}}export{t as parseMhtml};

package/dist/utils.js

@@ -77,7 +77,9 @@ export function indexOf(haystack, needle) {
 export async function* splitStream(iter, split) {
     let current = new Uint8Array(0);
     for await (const chunk of iter) {
-        current = current.length ? concat([current, chunk]) : chunk;
+        current = current.length
+            ? concat([current, chunk])
+            : chunk;
         let nextInd;
         while ((nextInd = indexOf(current, split)) !== -1) {
             yield current.subarray(0, nextInd);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mhtml-stream",
-  "version": "2.0.0",
+  "version": "2.0.2",
   "description": "Zero dependency stream MHTML parser",
   "keywords": [
     "mhtml",
@@ -8,88 +8,43 @@
   ],
   "repository": {
     "type": "git",
-    "url": "https://github.com/erikbrinkman/mhtml-stream.git"
+    "url": "git+https://github.com/erikbrinkman/mhtml-stream.git"
   },
   "author": "Erik Brinkman <erik.brinkman@gmail.com>",
   "license": "MIT",
   "type": "module",
+  "sideEffects": false,
   "types": "dist/index.d.ts",
   "module": "dist/mhtml.esm.min.js",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "default": "./dist/mhtml.esm.min.js"
+    }
+  },
   "files": [
     "/dist/**/*.js",
     "/dist/**/*.d.ts"
   ],
   "scripts": {
-    "fmt": "prettier --cache --write 'src/*.ts' '*.{js,json}'",
-    "lint": "tsc && eslint --cache 'src/*.ts' && typedoc --emit none",
+    "cli": "bun src/cli.ts",
+    "fmt": "biome format --write",
+    "lint": "tsc && biome check && typedoc --emit none",
     "doc": "typedoc",
     "export": "tsc -p tsconfig.build.json && bun build src/index.ts --minify --outfile dist/mhtml.esm.min.js",
-    "prepack": "bun lint && bun test --coverage && bun export"
+    "prepack": "bun lint && bun test --coverage && bun export && publint"
   },
   "dependencies": {
     "base64-js": "^1.5.1"
   },
   "devDependencies": {
-    "@types/base64-js": "^1.3.2",
-    "@types/bun": "^1.1.0",
-    "eslint": "^9.0.0",
-    "eslint-config-prettier": "^9.1.0",
-    "eslint-plugin-spellcheck": "^0.0.20",
-    "eslint-plugin-tsdoc": "^0.2.17",
-    "prettier": "^3.2.5",
-    "prettier-plugin-organize-imports": "^3.2.4",
-    "typedoc": "^0.25.8",
-    "typescript": "^5.3.3",
-    "typescript-eslint": "^7.7.0"
-  },
-  "eslintConfig": {
-    "rules": {
-      "@typescript-eslint/no-inferrable-types": "off",
-      "@typescript-eslint/no-non-null-assertion": "off",
-      "@typescript-eslint/no-unused-vars": [
-        "error",
-        {
-          "varsIgnorePattern": "^_+$"
-        }
-      ],
-      "prefer-const": [
-        "error",
-        {
-          "destructuring": "all"
-        }
-      ],
-      "no-warning-comments": [
-        "error",
-        {
-          "terms": [
-            "fixme"
-          ],
-          "location": "anywhere"
-        }
-      ],
-      "spellcheck/spell-checker": [
-        "error",
-        {
-          "identifiers": false,
-          "skipIfMatch": [
-            "[^\\s]{20,}"
-          ],
-          "skipWords": [
-            "0x7fffffff",
-            "Chr",
-            "Dvalue",
-            "Hola",
-            "Mhtml",
-            "Today’s",
-            "Uint",
-            "Yorker",
-            "ascii",
-            "delim",
-            "multipart",
-            "versa"
-          ]
-        }
-      ]
-    }
+    "@biomejs/biome": "^2.4.15",
+    "@types/base64-js": "^1.5.0",
+    "@types/bun": "^1.3.14",
+    "@types/yargs": "^17.0.35",
+    "publint": "^0.3.21",
+    "typedoc": "^0.28.19",
+    "typescript": "~6.0.3",
+    "yargs": "^18.0.0"
   }
 }