Add README.md

Author: RAOEUS

README.md

@@ -0,0 +1,89 @@
+# Drupal AI Alt Text Generator Module
+
+Provides AI‑generated alt text suggestions for images without alt text in the media library, with a human review workflow.
+
+## Requirements
+
+- Drupal 10  
+- PHP 7.4 or higher  
+- PHP extensions: cURL, GD or Imagick  
+- OpenAI API key  
+
+## Installation
+
+1. Copy the `alt_text_review` folder into your `modules/contrib` directory.  
+2. Enable the module:
+   ```bash
+   drush en alt_text_review -y
+
+3. Assign the **Access Alt Text Review UI** permission to the roles that should be able to review and save alt text.
+
+## Configuration
+
+Visit **Configuration » Media » Alt Text Review Settings** (`/admin/config/media/alt-text-review`) and configure:
+
+* **OpenAI API key**
+* **AI Prompt**
+
+  > A template for the AI, with the token `[max_length]` replaced by your max‑length setting.
+* **Alt text maximum character length**
+
+  > Integer value (default 128) that replaces `[max_length]` in your prompt.
+* **Enable debug logging**
+
+  > Logs full request/response payloads when checked. Disable in production.
+
+## Usage
+
+Go to **Content » Media » Review Alt Text** (`/admin/content/media/alt-text-review`):
+
+1. The module queries for the first media item of type **Image** where `field_media_image.alt` is empty or NULL.
+2. It **automatically downscales** the image to fit within an 800 × 800 px box (preserving aspect ratio) to limit payload size.
+3. The downscaled image is base64‑encoded and sent, along with your prompt, to the AI.
+4. You see the image plus the AI’s suggested alt text.
+5. Edit or accept the suggestion and **Save alt text**. You’re redirected to the next image without alt text.
+
+## How It Works
+
+1. **Discovery**
+   Runs an entity query for `media` bundle `image` where the alt field is empty.
+2. **Downscaling**
+   Uses Drupal’s ImageFactory to load the file and resize it so neither width nor height exceeds 800 px.
+3. **Encoding**
+   Reads the (downscaled) image file, base64‑encodes it, and wraps it in a `data:` URI.
+4. **Prompt Assembly**
+   Injects your max‑length setting into the prompt template.
+5. **API Call**
+   Sends a Chat Completion request to `https://api.openai.com/v1/chat/completions` with:
+
+   * **model**: `gpt-4o-mini`
+   * **messages**: `[ { role: "user", content: [ { type: "text", text: prompt }, { type: "image_url", image_url: { url: dataUri } } ] } ]`
+   * **max\_tokens**: calculated from your max length.
+6. **Error Handling**
+
+   * Missing API key → early error message.
+   * File not found → log and fallback message.
+   * Request failure → log details if debug enabled, show generic error.
+7. **Review & Save**
+   The form lets you tweak or accept the AI suggestion and saves it to the media entity’s `alt` attribute.
+
+## Model and Pricing
+
+This module uses **GPT‑4o mini**, priced at:
+
+* **\$0.15 per 1 M input tokens**
+* **\$0.60 per 1 M output tokens** ([OpenAI][1], [Reuters][2])
+
+Because images are downscaled to 800×800 px, a typical JPEG (\~50 KB) yields about 68 000 base64 characters (≈17 000 tokens). With a 32‑token response, the cost per image is:
+
+```text
+Input cost  = (17 000 / 1 000 000) × $0.15 ≈ $0.00255  
+Output cost = (   32 / 1 000 000) × $0.60 ≈ $0.00002  
+Total ≈ $0.00257 per image
+```
+
+> To further reduce cost, you can lower the downscale dimensions or adjust JPEG quality via your image styles.
+
+[1]: https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence/?utm_source=chatgpt.com "GPT-4o mini: advancing cost-efficient intelligence - OpenAI"
+[2]: https://www.reuters.com/technology/artificial-intelligence/openai-unveils-cheaper-small-ai-model-gpt-4o-mini-2024-07-18/?utm_source=chatgpt.com "OpenAI unveils cheaper small AI model GPT-4o mini"
+