@mdgate/wps 0.1.0

package/README.md

@@ -0,0 +1,30 @@
+# @mdgate/wps
+
+Convert Kingsoft WPS Writer, Spreadsheets, and Presentation files to Markdown.
+Outputs GitHub-Flavored Markdown. Works in Node, Edge, and browsers. No native
+addons.
+
+Handles: `.wps`, `.wpt`, `.et`, `.ett`, `.dps`, `.dpt`
+
+OOXML/OLE Office-compatible files are delegated to the matching official
+converter. Proprietary Kingsoft packages are best-effort text extraction.
+
+## Usage
+
+```ts
+import { toMarkdown } from '@mdgate/wps';
+
+const markdown = await toMarkdown(bytes);
+```
+
+Compose with other converters:
+
+```ts
+import { create } from '@mdgate/core';
+import { wps } from '@mdgate/wps';
+
+const convert = create([wps()]);
+```
+
+Part of [mdgate converters](https://github.com/mdgate/converters); install
+`@mdgate/converters` for every format at once.

package/dist/index.d.ts

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

package/dist/index.js

@@ -0,0 +1 @@
+import{create as Zq}from"@mdgate/core";import{CompoundFile as n,detectOleDoc as r,detectZipDoc as i,hasOleMagic as o,Package as b}from"@mdgate/containers";import{ConvertError as y}from"@mdgate/core";import{doc as s}from"@mdgate/doc";import{documentToMarkdown as t}from"@mdgate/document";import{docx as e}from"@mdgate/docx";import{ppt as qq}from"@mdgate/ppt";import{pptx as Gq}from"@mdgate/pptx";import{fileExtension as Hq}from"@mdgate/utils";import{xlsx as Jq}from"@mdgate/xlsx";import{CompoundFile as P,hasOleMagic as v,Package as C,parseXml as w}from"@mdgate/containers";import{ConvertError as W}from"@mdgate/core";import{emptyDocument as k,plain as h}from"@mdgate/document";import{cleanText as B,collapseWs as I,isAlphanumeric as g}from"@mdgate/utils";var u=2,R=2;function O(q){let G=[];try{if(a(q))x(q,G);else if(v(q))m(q,G);else S(q,G)}catch(Q){if(Q instanceof W&&(Q.code==="encrypted"||Q.code==="resourceLimit"))throw Q;S(q,G)}let H=new Set,J=k();for(let Q of G){let V=I(B(Q)).trim();if(!F(V)||H.has(V))continue;H.add(V),J.blocks.push({type:"paragraph",inlines:[h(V)]})}if(J.blocks.length===0)throw W.malformed("no readable text in WPS document");return J}function x(q,G){let H=C.open(q);for(let J of H.partNames()){if(l(J))continue;let Q=H.optionalPart(J);if(Q!==void 0)S(Q,G)}}function m(q,G){let H=P.open(q);if(H.exists("EncryptionInfo")||H.exists("EncryptedPackage"))throw W.encrypted();let J=!1;for(let Q of H.readRootStorage()){if(Q.type!==u)continue;try{S(H.readStream(Q.name),G),J=!0}catch(V){if(V instanceof W&&V.isFatal())throw V}}if(!J)S(q,G)}function S(q,G){if(c(q))try{f(w(q),G);return}catch(H){if(H instanceof W&&H.isFatal())throw H}d(q,G),p(q,G)}function f(q,G){let H=(J)=>{let Q="",V=!1;for(let Y of J.children)if(Y.type==="text")Q+=Y.text;else V=!0;let $=I(B(Q)).trim();if(F($))G.push($);if(V)for(let Y of J.childElems())H(Y)};H(q)}function d(q,G){let H=[],J=()=>{if(H.length>=R)G.push(String.fromCharCode(...H));H.length=0};for(let Q=0;Q+1<q.length;Q+=2){let V=q[Q]|q[Q+1]<<8;if(V!==0&&A(V))H.push(V);else J()}J()}function p(q,G){let J=new TextDecoder("utf-8",{fatal:!1}).decode(q),Q=-1,V=($)=>{if(Q>=0&&$-Q>=R)G.push(J.slice(Q,$));Q=-1};for(let $=0;$<J.length;$+=1){let Y=J.charCodeAt($);if(A(Y)){if(Q<0)Q=$}else V($)}V(J.length)}function A(q){if(q===9||q===10||q===13||q===32)return!0;if(q<32||q>=127&&q<=159)return!1;return q!==65534&&q!==65535}function F(q){if(q.length<R)return!1;let G=0;for(let J of q)if(g(J))G+=1;if(G<2)return!1;let H=q.toLowerCase();if(H.includes("xmlns")||H.includes("http://schemas.")||H.includes("https://schemas."))return!1;if(H.startsWith("application/"))return!1;return!0}function c(q){let G=0;if(q.length>=3&&q[0]===239&&q[1]===187&&q[2]===191)G=3;while(G<q.length){let H=q[G];if(H!==9&&H!==10&&H!==13&&H!==32)break;G+=1}return q[G]===60}function l(q){let G=q.toLowerCase();if(G==="mimetype"||G==="[content_types].xml")return!0;if(G.endsWith(".rels"))return!0;return/\.(png|jpe?g|gif|bmp|emf|wmf|bin|ttf|otf|woff2?)$/i.test(G)}function a(q){return q.length>=4&&q[0]===80&&q[1]===75&&q[2]===3&&q[3]===4}var Qq=new Set(["wps","wpt","et","ett","dps","dpt"]),Vq=e(),$q=s(),Yq=Jq(),Wq=Gq(),Sq=qq();function Z(){return{id:"wps",sniff(q,G){if(U(q)!==void 0)return 0;if(zq(q))return 2;if(G?.path!==void 0&&Qq.has(Hq(G.path)??""))return 1;return 0},convert(q,G){if(Dq(q))throw y.unsupported("pdf");let H=U(q);if(H==="docx")return Vq.convert(q,G);if(H==="doc")return $q.convert(q,G);if(H==="xlsx"||H==="xls")return Yq.convert(q,G);if(H==="pptx")return Wq.convert(q,G);if(H==="ppt")return Sq.convert(q,G);return{markdown:t(O(q))}}}}function U(q){let G=i(q);if(G==="docx"||G==="xlsx"||G==="pptx")return G;let H=r(q);if(H==="doc"||H==="xls"||H==="ppt")return H;return}function zq(q){if(Lq(q))try{let G=b.open(q),H=G.optionalPart("mimetype");if(H!==void 0&&X(K(H)))return!0;for(let Q of G.partNames())if(M(Q))return!0;let J=G.optionalPart("[Content_Types].xml");if(J!==void 0&&X(K(J)))return!0;return!1}catch{}else if(o(q))try{let G=n.open(q);for(let H of G.readRootStorage())if(M(H.name))return!0}catch{}return _q(q)}function M(q){let G=q.charCodeAt(0)===5?q.slice(1):q;for(let H of G.split(/[/\\]/)){if(H.length===0)continue;if(N(H,"ks")||N(H,"wps"))return!0}return!1}function X(q){let G=q.toLowerCase();return G.includes("kingsoft")||G.includes("application/wps")||G.includes("application/vnd.wps")||G.includes("application/vnd.kingsoft")}function _q(q){let G=q.subarray(0,Math.min(q.length,65536));return L(G,"kingsoft")||L(G,"application/wps")||L(G,"application/vnd.wps")||L(G,"application/vnd.kingsoft")}function Dq(q){let G=0;if(q.length>=3&&q[0]===239&&q[1]===187&&q[2]===191)G=3;while(G<q.length){let H=q[G];if(H!==9&&H!==10&&H!==13&&H!==32)break;G+=1}return Rq(q,"%PDF-",G)}function Lq(q){return q.length>=4&&q[0]===80&&q[1]===75&&q[2]===3&&q[3]===4}function K(q){return new TextDecoder("utf-8",{fatal:!1}).decode(q)}function N(q,G){if(q.length<G.length)return!1;return q.slice(0,G.length).toLowerCase()===G}function Rq(q,G,H){if(H+G.length>q.length)return!1;for(let J=0;J<G.length;J+=1)if(q[H+J]!==G.charCodeAt(J))return!1;return!0}function L(q,G){if(G.length===0||q.length<G.length)return!1;let H=G.length,J=G.charCodeAt(0),Q=J>=65&&J<=90?J+32:J;for(let V=0;V<=q.length-H;V+=1){let $=q[V];if(($>=65&&$<=90?$+32:$)!==Q)continue;let j=!0;for(let z=1;z<H;z+=1){let _=q[V+z],D=G.charCodeAt(z),T=_>=65&&_<=90?_+32:_,E=D>=65&&D<=90?D+32:D;if(T!==E){j=!1;break}}if(j)return!0}return!1}var jq=Zq([Z()]);function Bq(q,G){return jq(q,G)}export{Z as wps,Bq as toMarkdown};

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 wps 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/dist/wps.d.ts

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

package/package.json

@@ -0,0 +1,49 @@
+{
+  "name": "@mdgate/wps",
+  "version": "0.1.0",
+  "description": "mdgate Kingsoft WPS converter (wps/et/dps)",
+  "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/doc": "0.1.0",
+    "@mdgate/document": "0.1.0",
+    "@mdgate/docx": "0.1.0",
+    "@mdgate/ppt": "0.1.0",
+    "@mdgate/pptx": "0.1.0",
+    "@mdgate/utils": "0.1.2",
+    "@mdgate/xlsx": "0.1.0"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "keywords": [
+    "mdgate",
+    "markdown",
+    "wps",
+    "et",
+    "dps",
+    "kingsoft"
+  ]
+}