feat: add @embedpdf/cli with doc search and reader commands

.changeset/cli-search-improvements.md

@@ -0,0 +1,5 @@
+---
+"@embedpdf/cli": patch
+---
+
+Improve search-docs output: show descriptions in human-readable mode, reduce default limit from 10 to 5, and add minimum score threshold (0.25) to filter irrelevant results.

.gitignore

@@ -15,3 +15,8 @@ config/tsup.frameworks.config.bundled*
 publish-summary.json
 **/.svelte-kit
 **/*/*.tgz
+.env
+
+# Generated data files
+packages/cli/src/data/manifest.json
+website/src/data/embeddings.json

packages/cli/README.md

@@ -0,0 +1,105 @@
+<div align="center">
+  <a href="https://www.embedpdf.com">
+    <img alt="EmbedPDF logo" src="https://www.embedpdf.com/logo-192.png" height="96">
+  </a>
+  <h1>EmbedPDF</h1>
+
+<a href="https://www.npmjs.com/package/@embedpdf/cli"><img alt="NPM version" src="https://img.shields.io/npm/v/@embedpdf/cli.svg?style=for-the-badge&labelColor=000000"></a> <a href="https://github.com/embedpdf/embed-pdf-viewer/blob/main/packages/cli/LICENSE"><img alt="License" src="https://img.shields.io/npm/l/@embedpdf/cli.svg?style=for-the-badge&labelColor=000000"></a> <a href="https://github.com/embedpdf/embed-pdf-viewer/discussions"><img alt="Join the community on GitHub" src="https://img.shields.io/badge/Join%20the%20community-blueviolet.svg?style=for-the-badge&labelColor=000000"></a>
+
+</div>
+
+# @embedpdf/cli
+
+CLI for searching and reading EmbedPDF documentation directly from your terminal.
+
+## Documentation
+
+For complete guides, examples, and full API reference, visit:
+
+**[Official Documentation](https://www.embedpdf.com/docs)**
+
+## Installation
+
+```bash
+# npm
+npm install -g @embedpdf/cli
+
+# pnpm
+pnpm add -g @embedpdf/cli
+
+# or run directly
+npx @embedpdf/cli
+```
+
+## Commands
+
+### `embedpdf search-docs <query>`
+
+Full-text search over the entire documentation corpus.
+
+```bash
+# Basic search
+embedpdf search-docs "zoom"
+
+# Filter by framework
+embedpdf search-docs "getting started" --framework react
+
+# Filter by section
+embedpdf search-docs "annotations" --section headless
+
+# JSON output (for LLM consumption)
+embedpdf search-docs "rendering" --json --limit 5
+```
+
+| Option | Description |
+|---|---|
+| `--json` | Output as JSON |
+| `--framework <framework>` | Filter by framework (`react`, `vue`, `svelte`) |
+| `--section <section>` | Filter by section (`headless`, `viewer`, `snippet`, `engines`, `pdfium`) |
+| `--limit <n>` | Maximum number of results (default: `10`) |
+
+### `embedpdf doc [path]`
+
+Fetch and display a documentation page.
+
+```bash
+# Fetch a specific doc
+embedpdf doc react/headless/getting-started
+
+# List all available pages
+embedpdf doc --list
+
+# List docs for a specific framework
+embedpdf doc --list --framework vue
+
+# Output as JSON
+embedpdf doc react/headless/plugins/plugin-zoom --json
+```
+
+| Option | Description |
+|---|---|
+| `--json` | Output as JSON |
+| `--list` | List all available doc pages |
+| `--framework <framework>` | Filter by framework (`react`, `vue`, `svelte`) |
+| `--section <section>` | Filter by section (`headless`, `viewer`, `snippet`, `engines`, `pdfium`) |
+
+## Programmatic Usage
+
+The package also exports functions for use in Node.js:
+
+```typescript
+import { searchDocs, fetchDocContent, getManifest } from '@embedpdf/cli';
+
+// Search docs
+const results = searchDocs('zoom', { framework: 'react', limit: 5 });
+
+// Fetch a doc page
+const content = await fetchDocContent('react/headless/getting-started');
+
+// Get the full manifest
+const manifest = getManifest();
+```
+
+## License
+
+MIT – see the [LICENSE](https://github.com/embedpdf/embed-pdf-viewer/blob/main/packages/cli/LICENSE) file.

packages/cli/package.json

@@ -0,0 +1,49 @@
+{
+  "name": "@embedpdf/cli",
+  "version": "2.14.3",
+  "private": false,
+  "description": "CLI for searching and reading EmbedPDF documentation.",
+  "type": "module",
+  "bin": {
+    "embedpdf": "./dist/bin.js"
+  },
+  "main": "./dist/bin.cjs",
+  "module": "./dist/bin.js",
+  "exports": {
+    ".": {
+      "types": "./dist/bin.d.ts",
+      "import": "./dist/bin.js",
+      "require": "./dist/bin.cjs"
+    }
+  },
+  "scripts": {
+    "generate-manifest": "node scripts/generate-manifest.mjs",
+    "build": "pnpm run generate-manifest && pnpm run clean && vite build",
+    "clean": "rimraf dist"
+  },
+  "files": [
+    "dist",
+    "README.md"
+  ],
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/embedpdf/embed-pdf-viewer",
+    "directory": "packages/cli"
+  },
+  "homepage": "https://www.embedpdf.com/docs",
+  "bugs": {
+    "url": "https://github.com/embedpdf/embed-pdf-viewer/issues"
+  },
+  "author": "EmbedPDF",
+  "license": "MIT",
+  "dependencies": {
+    "commander": "^13.1.0"
+  },
+  "devDependencies": {
+    "@embedpdf/build": "workspace:*",
+    "typescript": "^5.0.0"
+  },
+  "publishConfig": {
+    "access": "public"
+  }
+}

packages/cli/scripts/generate-manifest.mjs

@@ -0,0 +1,189 @@
+/**
+ * Scans website/src/content/docs/ for all MDX files,
+ * extracts frontmatter (title, description, searchable) and full text content,
+ * and writes src/lib/manifest.json.
+ *
+ * Run: node scripts/generate-manifest.mjs
+ */
+
+import fs from 'node:fs';
+import path from 'node:path';
+import { fileURLToPath } from 'node:url';
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+const CONTENT_DIR = path.resolve(__dirname, '../../../website/src/content/docs');
+const TOOLS_DIR = path.resolve(__dirname, '../../../website/src/content/tools');
+const OUTPUT_PATH = path.resolve(__dirname, '../src/data/manifest.json');
+
+function parseFrontmatter(content) {
+  const match = content.match(/^---\r?\n([\s\S]*?)\r?\n---/);
+  if (!match) return { fm: {}, body: content };
+
+  const fm = {};
+  for (const line of match[1].split('\n')) {
+    const colonIdx = line.indexOf(':');
+    if (colonIdx === -1) continue;
+    const key = line.slice(0, colonIdx).trim();
+    let value = line.slice(colonIdx + 1).trim();
+
+    // Strip surrounding quotes
+    if (
+      (value.startsWith('"') && value.endsWith('"')) ||
+      (value.startsWith("'") && value.endsWith("'"))
+    ) {
+      value = value.slice(1, -1);
+    }
+
+    // Parse booleans
+    if (value === 'true') value = true;
+    else if (value === 'false') value = false;
+
+    fm[key] = value;
+  }
+
+  // Everything after the frontmatter block
+  const body = content.slice(match[0].length);
+  return { fm, body };
+}
+
+/**
+ * Strip MDX/JSX syntax from doc content, returning plain searchable text.
+ */
+function stripMdx(content) {
+  let text = content;
+
+  // Remove import statements
+  text = text.replace(/^import\s+.*?;\s*$/gm, '');
+
+  // Remove JSX component wrappers but keep text children
+  // e.g. <ExampleWrapper>...</ExampleWrapper>
+  text = text.replace(/<ExampleWrapper[^>]*>[\s\S]*?<\/ExampleWrapper>/g, '');
+
+  // Remove Callout tags but keep content
+  text = text.replace(/<\/?Callout[^>]*>/g, '');
+
+  // Remove self-closing JSX tags
+  text = text.replace(/<[A-Z][a-zA-Z]*\s*[^>]*\/>/g, '');
+
+  // Remove remaining JSX open/close tags (keep text content)
+  text = text.replace(/<\/?[A-Z][a-zA-Z]*[^>]*>/g, '');
+
+  // Remove fenced code blocks (they're code examples, not prose)
+  text = text.replace(/```[\s\S]*?```/g, '');
+
+  // Remove inline code backticks but keep the text
+  text = text.replace(/`([^`]+)`/g, '$1');
+
+  // Remove markdown link syntax, keep text
+  text = text.replace(/\[([^\]]+)\]\([^)]+\)/g, '$1');
+
+  // Remove markdown image syntax
+  text = text.replace(/!\[([^\]]*)\]\([^)]+\)/g, '');
+
+  // Remove HTML comments
+  text = text.replace(/<!--[\s\S]*?-->/g, '');
+
+  // Remove markdown heading markers but keep text
+  text = text.replace(/^#{1,6}\s+/gm, '');
+
+  // Remove markdown bold/italic markers
+  text = text.replace(/\*{1,3}([^*]+)\*{1,3}/g, '$1');
+  text = text.replace(/_{1,3}([^_]+)_{1,3}/g, '$1');
+
+  // Remove horizontal rules
+  text = text.replace(/^[-*_]{3,}\s*$/gm, '');
+
+  // Remove markdown list markers
+  text = text.replace(/^\s*[-*+]\s+/gm, '');
+  text = text.replace(/^\s*\d+\.\s+/gm, '');
+
+  // Collapse multiple blank lines
+  text = text.replace(/\n{3,}/g, '\n\n');
+
+  return text.trim();
+}
+
+function inferMeta(relativePath, collection = 'docs') {
+  const parts = relativePath.replace(/\.mdx$/, '').split('/');
+
+  let framework = null;
+  let section = null;
+
+  if (collection === 'tools') {
+    section = 'tools';
+  } else if (['react', 'vue', 'svelte'].includes(parts[0])) {
+    framework = parts[0];
+    if (['headless', 'viewer'].includes(parts[1])) {
+      section = parts[1];
+    }
+  } else if (parts[0] === 'snippet') {
+    section = 'snippet';
+  } else if (parts[0] === 'engines') {
+    section = 'engines';
+  } else if (parts[0] === 'pdfium') {
+    section = 'pdfium';
+  }
+
+  const docPath = relativePath.replace(/\.mdx$/, '');
+  const urlPrefix = collection === 'tools' ? 'tools' : 'docs';
+  const url = `https://www.embedpdf.com/${urlPrefix}/${docPath}`;
+
+  return { docPath, url, framework, section };
+}
+
+function scanCollection(dir, collection) {
+  if (!fs.existsSync(dir)) return [];
+  const entries = fs.readdirSync(dir, { recursive: true });
+  const docs = [];
+
+  for (const entry of entries) {
+    const entryStr = String(entry);
+    if (!entryStr.endsWith('.mdx')) continue;
+
+    const fullPath = path.join(dir, entryStr);
+    const raw = fs.readFileSync(fullPath, 'utf-8');
+    const { fm, body } = parseFrontmatter(raw);
+
+    // Skip non-searchable pages
+    if (fm.searchable === false) continue;
+
+    // Skip index pages (they're usually just navigation wrappers)
+    const basename = path.basename(entryStr, '.mdx');
+    if (basename === 'index') continue;
+
+    const meta = inferMeta(entryStr, collection);
+    const content = stripMdx(body);
+
+    docs.push({
+      path: meta.docPath,
+      title: fm.title || basename,
+      description: fm.description || '',
+      url: meta.url,
+      framework: meta.framework,
+      section: meta.section,
+      content,
+    });
+  }
+
+  return docs;
+}
+
+function scanDocs() {
+  const docs = [
+    ...scanCollection(CONTENT_DIR, 'docs'),
+    ...scanCollection(TOOLS_DIR, 'tools'),
+  ];
+
+  // Sort by path for deterministic output
+  docs.sort((a, b) => a.path.localeCompare(b.path));
+  return docs;
+}
+
+const docs = scanDocs();
+
+fs.writeFileSync(OUTPUT_PATH, JSON.stringify({ docs }, null, 2) + '\n');
+
+const totalContentChars = docs.reduce((sum, d) => sum + d.content.length, 0);
+console.log(
+  `[generate-manifest] Wrote ${docs.length} docs (${(totalContentChars / 1024).toFixed(0)}KB of content) to ${path.relative(process.cwd(), OUTPUT_PATH)}`,
+);

packages/cli/src/bin.ts

@@ -0,0 +1,12 @@
+import { Command } from 'commander';
+import { registerDocCommand } from './commands/doc';
+import { registerSearchCommand } from './commands/search';
+
+const program = new Command();
+
+program.name('embedpdf').description('The official CLI for the EmbedPDF project').version('2.14.3');
+
+registerDocCommand(program);
+registerSearchCommand(program);
+
+program.parse();