# PDF Validation Requirements

## Overview

Before documents are sent to AWS Textract for text extraction, the system performs several validation checks to ensure the PDF files meet Textract's requirements and can be processed successfully.

## Validation Checks

### 1. File Type

**Requirement**: Only PDF files are accepted.

**What we check**: The file must have the MIME type `application/pdf`.

### 2. File Readability

**Requirement**: The PDF must be valid and readable.

**What we check**: The system attempts to open and parse the PDF file. Corrupted or malformed PDFs will be rejected.

### 3. File Size

**Requirement**: Each page must be 5 MB or smaller.

**What we check**: Every page in the PDF is validated individually. If any single page exceeds 5 MB, the document is rejected.

**Why this matters**: This is a requirement imposed by AWS Textract for synchronous processing.

### 4. Page Dimensions

**Minimum**: 50 x 50 pixels
**Maximum**: 10,000 x 10,000 pixels

**What we check**: Each page's width and height must fall within this range.

**Why this matters**: Pages that are too small may not contain readable content. Pages that are too large exceed Textract's processing limits.

### 5. Image Resolution (DPI)

**Minimum**: 72 DPI (dots per inch)

**What we check**: Each page must have a resolution of at least 72 DPI for both width and height.

**Why this matters**: Low-resolution documents may not produce accurate text extraction results. Documents scanned or created at lower resolutions should be re-scanned or regenerated at a higher quality.

## Validation Failure Reasons

When a document fails validation, the system records one of the following failure reasons:

- `invalid_mimetype` - File is not a PDF
- `invalid_read` - PDF is corrupted or cannot be opened
- `invalid_read_pages` - One or more pages cannot be read
- `zero_page_count` - PDF contains no pages
- `large_file` - One or more pages exceeds 5 MB
- `small_dimensions` - Page dimensions are too small (below 50x50 pixels)
- `large_dimensions` - Page dimensions are too large (above 10,000x10,000 pixels)
- `small_dpi` - Page resolution is below 72 DPI
- `large_dpi` - Page resolution is too high (future use)