CLI, 100% TS, fixed various issues and added new models

.gitignore

@@ -1 +1,9 @@
 node_modules
+**/dist/
+**/examples/
+**/*.traineddata
+.env
+.env.local
+tmp/
+!tmp/.gitkeep
+*.tsbuildinfo

cli/README.md

@@ -0,0 +1,254 @@
+# Documind CLI
+
+Command-line interface for intelligent document processing and structured data extraction using Documind.
+
+## Installation
+
+```bash
+# Install dependencies
+npm install
+
+# Build the CLI
+npm run build
+```
+
+## Usage
+
+The CLI provides several commands to work with documents:
+
+### Extract Command
+
+Extract structured data from documents using schemas, templates, or auto-generation.
+
+```bash
+documind extract [options]
+```
+
+#### Options
+
+- `-f, --file <path>` - Path to the document file (required)
+- `-m, --model <model>` - LLM model to use (default: gpt-4o-mini)
+- `-s, --schema <path>` - Path to JSON schema file
+- `-t, --template <name>` - Name of a predefined template
+- `-a, --auto-schema` - Auto-generate schema from document
+- `-i, --instructions <text>` - Instructions for auto-schema generation
+- `-o, --output <path>` - Output file path for results (JSON)
+- `--base-url <url>` - Base URL for local LLM (for Ollama)
+
+#### Examples
+
+**Extract with auto-generated schema:**
+```bash
+documind extract -f invoice.pdf --auto-schema
+```
+
+**Extract with custom schema:**
+```bash
+documind extract -f document.pdf -s schema.json -o output.json
+```
+
+**Extract using a predefined template:**
+```bash
+documind extract -f invoice.pdf -t invoice -o result.json
+```
+
+**Extract with local LLM (Ollama):**
+```bash
+documind extract -f doc.pdf -m llama3.2-vision --base-url http://localhost:11434/v1 --auto-schema
+```
+
+**Extract with specific instructions:**
+```bash
+documind extract -f contract.pdf --auto-schema -i "Extract party names, dates, and monetary amounts"
+```
+
+### Convert Command
+
+Convert documents to markdown or plain text.
+
+```bash
+documind convert [options]
+```
+
+#### Options
+
+- `-f, --file <path>` - Path to the document file (required)
+- `-m, --model <model>` - LLM model to use (default: gpt-4o-mini)
+- `-t, --format <format>` - Output format: markdown or plaintext (default: markdown)
+- `-o, --output <path>` - Output file path
+- `--base-url <url>` - Base URL for local LLM (for Ollama)
+
+#### Examples
+
+**Convert to markdown:**
+```bash
+documind convert -f document.pdf -t markdown -o output.md
+```
+
+**Convert to plain text:**
+```bash
+documind convert -f document.pdf -t plaintext -o output.txt
+```
+
+### Templates Command
+
+List all available predefined templates.
+
+```bash
+documind templates
+```
+
+## Supported Models
+
+### OpenAI Models
+- `gpt-4o`
+- `gpt-4o-mini` (default)
+- `gpt-4.1`
+- `gpt-4.1-mini`
+
+### Google Models
+- `gemini-2.0-flash-001`
+- `gemini-2.0-flash-lite-preview-02-05`
+- `gemini-1.5-flash`
+- `gemini-1.5-flash-8b`
+- `gemini-1.5-pro`
+
+### Local Models (Ollama)
+- `llama3.2-vision`
+
+## Environment Variables
+
+Set these environment variables for API access:
+
+```bash
+# OpenAI API key for GPT models
+export OPENAI_API_KEY="your-openai-api-key"
+
+# Google Gemini API key
+export GEMINI_API_KEY="your-gemini-api-key"
+
+# Base URL for local LLM (Ollama)
+export BASE_URL="http://localhost:11434/v1"
+```
+
+Or create a `.env` file:
+
+```env
+OPENAI_API_KEY=your-openai-api-key
+GEMINI_API_KEY=your-gemini-api-key
+BASE_URL=http://localhost:11434/v1
+```
+
+## Schema Format
+
+Schemas should be in JSON format with the following structure:
+
+```json
+[
+  {
+    "name": "field_name",
+    "type": "string",
+    "description": "Field description"
+  },
+  {
+    "name": "nested_object",
+    "type": "object",
+    "description": "An object field",
+    "children": [
+      {
+        "name": "child_field",
+        "type": "number",
+        "description": "Child field description"
+      }
+    ]
+  },
+  {
+    "name": "items_list",
+    "type": "array",
+    "description": "A list of items",
+    "children": [
+      {
+        "name": "item_name",
+        "type": "string",
+        "description": "Item name"
+      }
+    ]
+  }
+]
+```
+
+### Supported Field Types
+
+- `string` - Text values
+- `number` - Numeric values
+- `boolean` - True/false values
+- `enum` - Predefined set of values (requires `values` array)
+- `object` - Nested objects (requires `children` array)
+- `array` - Lists (requires `children` array)
+
+## Using with Local LLMs
+
+To use Ollama or other local LLMs:
+
+1. Install and start Ollama:
+```bash
+# Install Ollama
+curl https://ollama.ai/install.sh | sh
+
+# Pull the vision model
+ollama pull llama3.2-vision
+
+# Start Ollama (if not already running)
+ollama serve
+```
+
+2. Use with the CLI:
+```bash
+documind extract \
+  -f document.pdf \
+  -m llama3.2-vision \
+  --base-url http://localhost:11434/v1 \
+  --auto-schema \
+  -o result.json
+```
+
+## Development
+
+```bash
+# Watch mode
+npm run dev
+
+# Build
+npm run build
+
+# Run
+npm start -- extract -f example.pdf --auto-schema
+```
+
+## Troubleshooting
+
+### "Cannot find module" errors
+
+Make sure you've built the project:
+```bash
+npm run build
+```
+
+### API Key errors
+
+Ensure your API keys are set:
+```bash
+echo $OPENAI_API_KEY
+echo $GEMINI_API_KEY
+```
+
+### Local LLM connection issues
+
+Check if Ollama is running:
+```bash
+curl http://localhost:11434/v1/models
+```
+
+## License
+
+AGPL-3.0

cli/package.json

@@ -0,0 +1,37 @@
+{
+  "name": "cli",
+  "version": "1.0.0",
+  "description": "Command-line interface for Documind extractor",
+  "type": "module",
+  "main": "dist/index.js",
+  "types": "dist/index.d.ts",
+  "bin": {
+    "documind": "./dist/index.js"
+  },
+  "files": [
+    "dist"
+  ],
+  "scripts": {
+    "build": "tsc --build && tsc-alias -p tsconfig.json",
+    "start": "node dist/index.js",
+    "dev": "tsx src/index.ts"
+  },
+  "keywords": [
+    "cli",
+    "document",
+    "extraction",
+    "llm"
+  ],
+  "dependencies": {
+    "commander": "^12.0.0",
+    "chalk": "^5.3.0",
+    "ora": "^8.0.1",
+    "extractor": "*",
+    "dotenv": "^16.4.5"
+  },
+  "devDependencies": {
+    "@types/node": "^20.14.11",
+    "tsc-alias": "^1.8.8",
+    "typescript": "^5.6.3"
+  }
+}

cli/src/commands/convert.ts

@@ -0,0 +1,73 @@
+import { Command } from 'commander';
+import { formatter } from 'extractor';
+import ora from 'ora';
+import { logger } from '../utils/logger.js';
+import { fileExists, writeJsonFile, resolveFilePath } from '../utils/file-helper.js';
+import fs from 'fs';
+
+export function createConvertCommand(): Command {
+  const cmd = new Command('convert');
+
+  cmd
+    .description('Convert a document to markdown or plain text')
+    .requiredOption('-f, --file <path>', 'Path to the document file')
+    .option('-m, --model <model>', 'LLM model to use', 'gpt-4o-mini')
+    .option('-t, --format <format>', 'Output format (markdown or plaintext)', 'markdown')
+    .option('-o, --output <path>', 'Output file path')
+    .option('--base-url <url>', 'Base URL for local LLM (for Ollama)')
+    .action(async (options: ConvertOptions) => {
+      await handleConvert(options);
+    });
+
+  return cmd;
+}
+
+interface ConvertOptions {
+  file: string;
+  model: string;
+  format: 'markdown' | 'plaintext';
+  output?: string;
+  baseUrl?: string;
+}
+
+async function handleConvert(options: ConvertOptions): Promise<void> {
+  const spinner = ora('Starting conversion...').start();
+
+  try {
+    const filePath = resolveFilePath(options.file);
+
+    if (!fileExists(filePath)) {
+      spinner.fail(`File not found: ${filePath}`);
+      process.exit(1);
+    }
+
+    if (options.baseUrl) {
+      process.env.BASE_URL = options.baseUrl;
+    }
+
+    spinner.text = `Converting document to ${options.format}...`;
+
+    let result: string;
+
+    if (options.format === 'markdown') {
+      result = await formatter.markdown({ file: filePath, model: options.model });
+    } else {
+      result = await formatter.plaintext({ file: filePath, model: options.model });
+    }
+
+    spinner.succeed('Conversion completed!');
+
+    if (options.output) {
+      const outputPath = resolveFilePath(options.output);
+      fs.writeFileSync(outputPath, result, 'utf-8');
+      logger.success(`Output saved to: ${outputPath}`);
+    } else {
+      console.log('\n' + result);
+    }
+
+  } catch (error) {
+    spinner.fail('Conversion failed');
+    logger.error((error as Error).message);
+    process.exit(1);
+  }
+}