@mdgate/data 0.1.0

package/README.md

@@ -0,0 +1,28 @@
+# @mdgate/data
+
+Convert JSON, JSONL, XML, and YAML files to Markdown. Outputs GitHub-Flavored Markdown. Works in
+Node, Edge, and browsers. No native addons.
+
+Handles: `.json`, `.jsonl`, `.xml`, `.yaml`, `.yml`
+
+XML is never claimed by content signature — flat ODF and many office parts start with `<?xml`.
+
+## Usage
+
+```ts
+import { toMarkdown } from '@mdgate/data';
+
+const markdown = await toMarkdown(bytes);
+```
+
+Compose with other converters:
+
+```ts
+import { create } from '@mdgate/core';
+import { data } from '@mdgate/data';
+
+const convert = create([data()]);
+```
+
+Part of [mdgate converters](https://github.com/mdgate/converters); install
+`@mdgate/converters` for every format at once.

package/dist/data.d.ts

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

package/dist/index.d.ts

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

package/dist/index.js

@@ -0,0 +1,5 @@
+import{ConvertError as m}from"@mdgate/core";import{documentToMarkdown as l}from"@mdgate/document";import{fileExtension as T}from"@mdgate/utils";import{ConvertError as s}from"@mdgate/core";var S=[208,207,17,224,161,177,26,225];function g(n){let c=0;if(n.length>=3&&n[0]===239&&n[1]===187&&n[2]===191)c=3;while(c<n.length){let r=n[c];if(r!==9&&r!==10&&r!==13&&r!==32)break;c+=1}return c}function $(n){let c=g(n);return A(n,"%PDF-",c)}function j(n){return n.length>=4&&n[0]===80&&n[1]===75&&n[2]===3&&n[3]===4}function L(n){if(n.length<S.length)return!1;for(let c=0;c<S.length;c+=1)if(n[c]!==S[c])return!1;return!0}function O(n){let c=g(n);return A(n,"{\\rtf",c)}function V(n){return $(n)||j(n)||L(n)||O(n)}function w(n){let c=g(n);if(c>=n.length)return!1;let r=n[c];return r===123||r===91}function J(n){let c=g(n);if(A(n,"<?xml",c))return!0;if(c>=n.length||n[c]!==60)return!1;let r=n[c+1];return r!==void 0&&v(r)}function W(n){if(n.length>=2&&n[0]===255&&n[1]===254)return U(new TextDecoder("utf-16le").decode(n));if(n.length>=2&&n[0]===254&&n[1]===255)return U(new TextDecoder("utf-16be").decode(n));let c=n.length>=3&&n[0]===239&&n[1]===187&&n[2]===191?n.subarray(3):n;try{return new TextDecoder("utf-8",{fatal:!0}).decode(c)}catch{throw s.malformed("invalid utf-8")}}function C(n){let c=n.split(/[/\\]/).pop()??"",r=c.lastIndexOf("."),o=r<=0?c:c.slice(0,r);return o.length>0?o:void 0}function U(n){return n.charCodeAt(0)===65279?n.slice(1):n}function A(n,c,r){if(r+c.length>n.length)return!1;for(let o=0;o<c.length;o+=1)if(n[r+o]!==c.charCodeAt(o))return!1;return!0}function v(n){return n>=48&&n<=57||n>=65&&n<=90||n>=97&&n<=122||n===45||n===95||n===58}import{ConvertError as R}from"@mdgate/core";import{cleanText as rn}from"@mdgate/utils";import{emptyDocument as I,heading as y,plain as N,tableFromRows as b}from"@mdgate/document";import{cleanText as nn}from"@mdgate/utils";function d(n,c){let r=I();return r.blocks.push({type:"codeBlock",lang:n,text:c}),r}function p(n,c){let r=I();if(n!==void 0&&n.length>0)r.blocks.push(y(1,[N(nn(n))]));return c(r),r}function Z(n){return n.map((c)=>({blocks:[{type:"paragraph",inlines:[N(c)]}],checked:void 0,markerLabel:void 0}))}function H(n){return{blocks:[{type:"paragraph",inlines:[N(n)]}],colSpan:1,rowSpan:1}}function _(n,c){let r=[n.map(H),...c.map((o)=>n.map((u,i)=>H(o[i]??"")))];return{type:"table",table:b(r,1,"data")}}var cn=24,G=50,P=12,on=200,un=80;function F(n,c){let r;try{r=JSON.parse(n)}catch(u){throw R.malformed(u instanceof Error?u.message:"invalid json")}return ln(r,c)??d("json",JSON.stringify(r,null,2))}function X(n,c){let r=en(n);if(r.length===0)throw R.malformed("empty jsonl");if(r.length<=G&&r.every(K)){let i=B(r);if(i!==void 0&&i.length>0&&i.length<=P)return p(c,(e)=>{e.blocks.push(_(i,r.map((t)=>i.map((f)=>z(t[f])))))})}let o=p(c,()=>{return}),u=Math.min(r.length,un);for(let i=0;i<u;i+=1)o.blocks.push({type:"codeBlock",lang:"json",text:JSON.stringify(r[i],null,2)});return o}function x(n,c){try{return F(n,c)}catch(r){if(fn(n))return X(n,c);throw r}}function en(n){let c=n.split(`
+`),r=[];for(let o=0;o<c.length;o+=1){let u=c[o];if(u.endsWith("\r"))u=u.slice(0,-1);let i=u.trim();if(i.length===0)continue;if(r.length>=on)break;try{r.push(JSON.parse(i))}catch(e){throw R.malformed(e instanceof Error?`jsonl line ${o+1}: ${e.message}`:`jsonl line ${o+1}`)}}return r}function fn(n){let c=0;for(let r of n.split(`
+`)){let o=(r.endsWith("\r")?r.slice(0,-1):r).trim();if(o.length===0)continue;try{JSON.parse(o)}catch{return!1}if(c+=1,c>=2)return!0}return!1}function ln(n,c){if(K(n)){let r=Object.keys(n);if(r.length===0||r.length>cn)return;return p(c,(o)=>{o.blocks.push({type:"list",list:{marker:"bullet",start:1,items:Z(r.map((u)=>`${u}: ${z(n[u])}`))}})})}if(Array.isArray(n)&&n.length>0&&n.length<=G){if(!n.every(K))return;let r=B(n);if(r===void 0||r.length===0||r.length>P)return;return p(c,(o)=>{o.blocks.push(_(r,n.map((u)=>r.map((i)=>z(u[i])))))})}return}function B(n){let c=Object.keys(n[0]),r=new Set(c);for(let o=1;o<n.length;o+=1){let u=Object.keys(n[o]);if(u.length!==r.size)return;for(let i of u)if(!r.has(i))return}return c}function K(n){if(n===null||typeof n!=="object"||Array.isArray(n))return!1;for(let c of Object.values(n))if(!tn(c))return!1;return!0}function tn(n){return n===null||typeof n==="string"||typeof n==="number"||typeof n==="boolean"}function z(n){if(n===null||n===void 0)return"";if(typeof n==="string")return rn(n);if(typeof n==="number"&&Number.isFinite(n))return String(n);if(typeof n==="boolean")return n?"true":"false";return""}function q(n){return d("xml",dn(n))}function E(n){return d("yaml",n.trimEnd())}function dn(n){let c=n.trim();if(c.length===0)return"";let r=pn(c),o=[],u=0;for(let i=0;i<r.length;i+=1){let e=r[i],t=Y(e);if(t==="text"){let f=e.trim();if(f.length===0)continue;let a=r[i+1],D=r[i-1];if(a!==void 0&&D!==void 0&&Y(D)==="open"&&Y(a)==="close"&&!f.includes(`
+`)){let k=o.pop();o.push(`${k??`${h(u)}${D}`}${f}${a}`),i+=1,u=Math.max(0,u-1);continue}for(let k of f.split(/\r?\n/)){let Q=k.trim();if(Q.length>0)o.push(`${h(u)}${Q}`)}continue}if(t==="close"){u=Math.max(0,u-1),o.push(`${h(u)}${e}`);continue}if(o.push(`${h(u)}${e}`),t==="open")u+=1}return o.join(`
+`)}function pn(n){let c=[],r=0;while(r<n.length){if(n.charCodeAt(r)!==60){let i=n.indexOf("<",r),e=i<0?n.length:i;c.push(n.slice(r,e)),r=e;continue}if(n.startsWith("<!--",r)){let i=n.indexOf("-->",r+4),e=i<0?n.length:i+3;c.push(n.slice(r,e)),r=e;continue}if(n.startsWith("<![CDATA[",r)){let i=n.indexOf("]]>",r+9),e=i<0?n.length:i+3;c.push(n.slice(r,e)),r=e;continue}let o=n.indexOf(">",r+1),u=o<0?n.length:o+1;c.push(n.slice(r,u)),r=u}return c}function Y(n){if(n.charCodeAt(0)!==60)return"text";if(n.startsWith("</"))return"close";if(n.startsWith("<?")||n.startsWith("<!"))return"misc";if(n.endsWith("/>"))return"empty";return"open"}function h(n){return"  ".repeat(n)}var mn=new Set(["json","jsonl"]),gn=new Set(["xml","yaml","yml"]);function M(){return{id:"data",sniff(n,c){if(V(n))return 0;let r=c?.path!==void 0?T(c.path):void 0;if(r!==void 0&&mn.has(r))return 2;if(!J(n)&&w(n))return 2;if(r!==void 0&&gn.has(r))return 1;return 0},convert(n,c){hn(n);let r=c?.path!==void 0?T(c.path):void 0,o=c?.path!==void 0?C(c.path):void 0,u=W(n);if(r==="jsonl")return{markdown:l(X(u,o))};if(r==="json")return{markdown:l(F(u,o))};if(r==="xml")return{markdown:l(q(u))};if(r==="yaml"||r==="yml")return{markdown:l(E(u))};if(w(n))return{markdown:l(x(u,o))};if(J(n))return{markdown:l(q(u))};throw m.unsupported("data")}}}function hn(n){if($(n))throw m.unsupported("pdf");if(L(n))throw m.unsupported("ole");if(j(n))throw m.unsupported("zip");if(O(n))throw m.unsupported("rtf")}import{create as an}from"@mdgate/core";var Dn=an([M()]);function kn(n,c){return Dn(n,c)}export{kn as toMarkdown,M as data};

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 data 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 +1,43 @@
+{
+  "name": "@mdgate/data",
+  "version": "0.1.0",
+  "description": "mdgate JSON, JSONL, XML, and YAML 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/core": "0.1.3",
+    "@mdgate/document": "0.1.0",
+    "@mdgate/utils": "0.1.2"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "engines": {
+    "node": ">=20"
+  },
+  "keywords": [
+    "mdgate",
+    "markdown",
+    "json",
+    "jsonl",
+    "xml",
+    "yaml"
+  ]
+}