@mdgate/visio 0.1.0

package/README.md

@@ -0,0 +1,26 @@
+# @mdgate/visio
+
+Convert Microsoft Visio drawings to Markdown. Outputs GitHub-Flavored Markdown. Works in Node, Edge,
+and browsers. No native addons.
+
+Handles: `.vsd`, `.vsdx`, `.vss`, `.vst`, `.vssx`, `.vstx`
+
+## Usage
+
+```ts
+import { toMarkdown } from '@mdgate/visio';
+
+const markdown = await toMarkdown(bytes);
+```
+
+Compose with other converters:
+
+```ts
+import { create } from '@mdgate/core';
+import { visio } from '@mdgate/visio';
+
+const convert = create([visio()]);
+```
+
+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 { visio } from './visio.js';

package/dist/index.js

@@ -0,0 +1 @@
+import{create as Fn}from"@mdgate/core";import{CompoundFile as Tn,detectOleDoc as D,detectZipDoc as O,hasOleMagic as W,Package as Un}from"@mdgate/containers";import{ConvertError as x}from"@mdgate/core";import{documentToMarkdown as Ln}from"@mdgate/document";import{fileExtension as _n}from"@mdgate/utils";import{hasOleMagic as vn}from"@mdgate/containers";import{ConvertError as Sn}from"@mdgate/core";import{CompoundFile as Q}from"@mdgate/containers";import{ConvertError as d}from"@mdgate/core";import{emptyDocument as q,plain as X}from"@mdgate/document";import{cleanText as Y,collapseWs as H,isAlphanumeric as J}from"@mdgate/utils";var K=2,w=2;function A(n){let t;try{t=Q.open(n)}catch(i){if(i instanceof d)throw i;let a=i instanceof Error?i.message:String(i);throw d.malformed(`not an OLE2 compound file: ${a}`)}if(t.exists("EncryptionInfo")||t.exists("EncryptedPackage"))throw d.encrypted();let r=!1,o=[];for(let i of t.readRootStorage()){if(nn(i.name,"VisioDocument"))r=!0;if(i.type!==K)continue;try{Z(t.readStream(i.name),o)}catch(a){if(a instanceof d&&a.isFatal())throw a}}if(!r)throw d.malformed("not a Visio document");let s=new Set,e=q();for(let i of o){let a=H(Y(i)).trim();if(!b(a)||s.has(a))continue;s.add(a),e.blocks.push({type:"paragraph",inlines:[X(a)]})}if(e.blocks.length===0)throw d.malformed("no readable text in Visio document");return e}function Z(n,t){z(n,t),j(n,t)}function z(n,t){let r=[],o=()=>{if(r.length>=w)t.push(y(r));r.length=0};for(let s=0;s+1<n.length;s+=2){let e=n[s]|n[s+1]<<8;if(e!==0&&C(e))r.push(e);else o()}o()}function j(n,t){let r=new TextDecoder("utf-8",{fatal:!1}).decode(n),o=-1,s=(e)=>{if(o>=0&&e-o>=w)t.push(r.slice(o,e));o=-1};for(let e=0;e<r.length;e+=1)if(C(r.charCodeAt(e))){if(o<0)o=e}else s(e);s(r.length)}function y(n){if(n.length<=8192)return String.fromCharCode(...n);let r="";for(let o=0;o<n.length;o+=8192)r+=String.fromCharCode(...n.slice(o,o+8192));return r}function C(n){if(n===9||n===10||n===13||n===32)return!0;if(n<32||n>=127&&n<=159)return!1;return n!==65534&&n!==65535}function b(n){if(n.length<w)return!1;let t=0;for(let o of n)if(J(o))t+=1;if(t<2)return!1;let r=n.toLowerCase();if(r.includes("xmlns")||r.includes("http://schemas.")||r.includes("https://schemas.")||r.includes("visiodocument"))return!1;return!0}function nn(n,t){return n.length===t.length&&n.toLowerCase()===t.toLowerCase()}import{ns as h,Package as tn,probeOle as rn,readRels as E,relsPartFor as T,relType as on,resolve as en}from"@mdgate/containers";import{ConvertError as l}from"@mdgate/core";import{emptyDocument as an,heading as sn,plain as U}from"@mdgate/document";import{alternateBranch as cn}from"@mdgate/office-common";import{cleanText as L,collapseWs as _}from"@mdgate/utils";var fn="http://schemas.microsoft.com/office/visio/2012/main",un="http://schemas.microsoft.com/visio/2003/core",ln="http://schemas.microsoft.com/visio/2010/relationships/pages",mn="http://schemas.microsoft.com/visio/2010/relationships/page",dn="http://schemas.microsoft.com/visio/2010/relationships/masters",hn="http://schemas.microsoft.com/visio/2010/relationships/master",gn=[fn,un,h.R,h.MC];function I(n){let t;try{t=tn.open(n)}catch(e){throw Rn(e,n)}let r=pn(t);if(r.length===0)throw l.missingPart("visio/pages/pages.xml");let o=an(),s=!1;for(let e=0;e<r.length;e+=1){let i=r[e],a=t.optionalXmlPart(i.path);if(a===void 0)continue;let f=o.blocks.length,c=i.name.length>0?i.name:`Page ${e+1}`;if(o.blocks.push(sn(1,[U(c)])),P(a,o.blocks),o.blocks.length===f+1)o.blocks.pop();else s=!0}if(!s)throw l.malformed("no readable text in Visio document");return o}function pn(n){let t=xn(n)??"visio/document.xml",r=E(n,T(t)),o=N(r,t,ln)??"visio/pages/pages.xml",s=v(n,o,"Page",mn);if(s.length>0)return s;let e=S(n,"visio/pages/","pages.xml");if(e.length>0)return e;let i=N(r,t,dn)??"visio/masters/masters.xml",a=v(n,i,"Master",hn);if(a.length>0)return a;return S(n,"visio/masters/","masters.xml")}function xn(n){let t=E(n,"_rels/.rels"),r=V(t,on.OFFICE_DOCUMENT);if(r!==void 0){let o=p("",r.target);if(o!==void 0)return o}if(n.hasPart("visio/document.xml"))return"visio/document.xml";return}function v(n,t,r,o){let s=n.optionalXmlPart(t);if(s===void 0)return[];let e=E(n,T(t)),i=[],a=new Set;for(let f of s.descendantsAny(r)){let c=wn(f),m=c!==void 0?e.internalTarget(c):void 0,u=m!==void 0?p(t,m):void 0;if(u===void 0||a.has(u))continue;a.add(u),i.push({name:Pn(f),path:u})}if(i.length>0)return i;for(let[,f]of e.entries()){if(f.mode!=="internal"||!M(f.relType,o))continue;let c=p(t,f.target);if(c===void 0||a.has(c))continue;a.add(c),i.push({name:"",path:c})}return i}function S(n,t,r){let o=t.toLowerCase(),s=r.toLowerCase(),e=[];for(let i of n.partNames()){let a=i.toLowerCase();if(!a.startsWith(o)||!a.endsWith(".xml"))continue;if(a.endsWith(".rels")||a.endsWith(`/${s}`)||a===s)continue;e.push(i)}return e.sort((i,a)=>i.localeCompare(a,void 0,{numeric:!0,sensitivity:"base"})),e.map((i,a)=>({name:`Page ${a+1}`,path:i}))}function wn(n){let t=n.attrQualified(h.R,"id")??n.attrAny("id");if(t!==void 0&&t.length>0&&t.toLowerCase().startsWith("rid"))return t;for(let r of n.descendantsAny("Rel")){let o=r.attrQualified(h.R,"id")??r.attrAny("id");if(o!==void 0&&o.length>0)return o}return}function Pn(n){let t=n.attrAny("Name")??n.attrAny("NameU")??"";return _(L(t)).trim()}function N(n,t,r){let o=V(n,r);if(o===void 0)return;return p(t,o.target)}function V(n,t){let r=n.firstOfType(t);if(r!==void 0)return r;let o,s;for(let[e,i]of n.entries()){if(i.mode!=="internal"||!M(i.relType,t))continue;if(o===void 0||e<o)o=e,s=i}return s}function M(n,t){if(n===t)return!0;let r=t.slice(t.lastIndexOf("/"));return r.length>1&&n.endsWith(r)}function P(n,t){if(n.is(h.MC,"AlternateContent")){let r=cn(n,gn);if(r!==void 0)P(r,t);return}if(n.local==="Text"){g(t,n.text());return}if(n.local==="TextBlock"){g(t,n.text());let r=n.attrAny("V")??n.attrAny("Value");if(r!==void 0)g(t,r)}else if(n.local==="Cell"&&Cn(n.attrAny("N")??"","TextBlock")){let r=n.attrAny("V")??n.attrAny("Value");if(r!==void 0)g(t,r);return}for(let r of n.childElems())P(r,t)}function g(n,t){let r=_(L(t)).trim();if(r.length===0||En(r))return;n.push({type:"paragraph",inlines:[U(r)]})}function En(n){return/^-?\d+(\.\d+)?([eE][+-]?\d+)?$/.test(n)}function p(n,t){try{return en(n,t).path}catch{return}}function Rn(n,t){if(n instanceof l){if(n.code==="encrypted"||An(n))return l.encrypted();return n}let r=rn(t);if(r!==void 0)return r;let o=n instanceof Error?n.message:String(n);if($(o))return l.encrypted();return l.malformed(`not a readable zip archive: ${o}`)}function An(n){if(n instanceof l&&n.code==="encrypted")return!0;let t=n instanceof l?n.detail??"":"",r=n instanceof Error?`${n.message} ${t}`:String(n);return $(r)}function $(n){let t=n.toLowerCase();return t.includes("encrypted")||t.includes("password")}function Cn(n,t){return n.length===t.length&&n.toLowerCase()===t.toLowerCase()}function k(n){if(vn(n))return A(n);if(Nn(n))return I(n);throw Sn.unsupported("visio")}function Nn(n){return n.length>=4&&n[0]===80&&n[1]===75&&n[2]===3&&n[3]===4}var In=new Set(["vsd","vsdx","vss","vst","vssx","vstx"]);function R(){return{id:"visio",sniff(n,t){let r=O(n);if(r==="vsdx"||r===void 0&&Mn(n))return 2;if(r!==void 0)return 0;let o=D(n);if(o==="vsd"||o===void 0&&$n(n))return 2;if(o!==void 0)return 0;if(t?.path!==void 0&&In.has(_n(t.path)??""))return 1;return 0},convert(n){return Vn(n),{markdown:Ln(k(n))}}}}function Vn(n){if(On(n))throw x.unsupported("pdf");let t=O(n);if(t!==void 0&&t!=="vsdx")throw x.unsupported(t);let r=D(n);if(r!==void 0&&r!=="vsd")throw x.unsupported(r);if(!B(n)&&!W(n))throw x.unsupported("visio")}function Mn(n){if(!B(n))return!1;try{let t=Un.open(n);if(t.hasPart("visio/document.xml"))return!0;for(let o of t.partNames())if(o.toLowerCase().startsWith("visio/"))return!0;let r=t.optionalPart("[Content_Types].xml");if(r!==void 0&&kn(r,"visio"))return!0;return!1}catch{return!1}}function $n(n){if(!W(n))return!1;try{for(let t of Tn.open(n).readRootStorage())if(Dn(t.name,"VisioDocument"))return!0}catch{return!1}return!1}function kn(n,t){if(t.length===0||n.length<t.length)return!1;let r=t.length,o=t.charCodeAt(0),s=o>=65&&o<=90?o+32:o;for(let e=0;e<=n.length-r;e+=1){let i=n[e];if((i>=65&&i<=90?i+32:i)!==s)continue;let f=!0;for(let c=1;c<r;c+=1){let m=n[e+c],u=t.charCodeAt(c),F=m>=65&&m<=90?m+32:m,G=u>=65&&u<=90?u+32:u;if(F!==G){f=!1;break}}if(f)return!0}return!1}function Dn(n,t){return n.length===t.length&&n.toLowerCase()===t.toLowerCase()}function On(n){let t=Wn(n);return Bn(n,"%PDF-",t)}function B(n){return n.length>=4&&n[0]===80&&n[1]===75&&n[2]===3&&n[3]===4}function Wn(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 Bn(n,t,r){if(r+t.length>n.length)return!1;for(let o=0;o<t.length;o+=1)if(n[r+o]!==t.charCodeAt(o))return!1;return!0}var Gn=Fn([R()]);function Qn(n,t){return Gn(n,t)}export{R as visio,Qn 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 visio 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/visio.d.ts

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

package/package.json

@@ -0,0 +1,44 @@
+{
+  "name": "@mdgate/visio",
+  "version": "0.1.0",
+  "description": "mdgate Visio converter (vsd/vsdx)",
+  "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/office-common": "0.1.0",
+    "@mdgate/utils": "0.1.2"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "keywords": [
+    "mdgate",
+    "markdown",
+    "visio",
+    "vsd",
+    "vsdx"
+  ]
+}