natural-pdf 0.1.7__py3-none-any.whl → 0.1.9__py3-none-any.whl

docs/tutorials/10-form-field-extraction.md

@@ -1,201 +0,0 @@
-# Form Field Extraction
-
-Business documents like invoices, forms, and applications contain field-value pairs that need to be extracted. This tutorial shows how to identify and extract these form fields.
-
-```python
-#%pip install "natural-pdf[all]"
-```
-
-```python
-from natural_pdf import PDF
-
-# Load a PDF
-pdf = PDF("https://github.com/jsoma/natural-pdf/raw/refs/heads/main/pdfs/01-practice.pdf")
-page = pdf.pages[0]
-
-# Find fields with labels ending in colon
-labels = page.find_all('text:contains(":")')
-
-# Visualize the found labels
-labels.show(color="blue", label="Field Labels")
-
-# Count how many potential fields we found
-len(labels)
-```
-
-## Extracting Field Values
-
-```python
-# Extract the value for each field label
-form_data = {}
-
-for label in labels:
-    # Clean up the label text
-    field_name = label.text.strip().rstrip(':')
-    
-    # Find the value to the right of the label
-    value_region = label.right(width=200)
-    value = value_region.extract_text().strip()
-    
-    # Store in our dictionary
-    form_data[field_name] = value
-
-# Display the extracted data
-form_data
-```
-
-## Visualizing Labels and Values
-
-```python
-# Clear previous highlights
-page.clear_highlights()
-
-# Highlight both labels and their values
-for label in labels:
-    # Highlight the label in red
-    label.show(color="red", label="Label")
-    
-    # Highlight the value area in blue
-    label.right(width=200).show(color="blue", label="Value")
-
-# Show the page image with highlighted elements
-page.to_image()
-```
-
-## Handling Multi-line Values
-
-```python
-# Extract values that might span multiple lines
-multi_line_data = {}
-
-for label in labels:
-    # Get the field name
-    field_name = label.text.strip().rstrip(':')
-    
-    # Look both to the right and below
-    right_value = label.right(width=200).extract_text().strip()
-    below_value = label.below(height=50).extract_text().strip()
-    
-    # Combine the values if they're different
-    if right_value in below_value:
-        value = below_value
-    else:
-        value = f"{right_value} {below_value}".strip()
-    
-    # Add to results
-    multi_line_data[field_name] = value
-
-# Show fields with potential multi-line values
-multi_line_data
-```
-
-## Finding Pattern-Based Fields
-
-```python
-import re
-
-# Find dates in the format July 31, YYY
-date_pattern = r'\b\w+ \d+, \d\d\d\d\b'
-
-# Search all text elements for dates
-text_elements = page.find_all('text')
-print([elem.text for elem in text_elements])
-dates = text_elements.filter(lambda elem: re.search(date_pattern, elem.text))
-
-# Visualize the date fields
-dates.show(color="green", label="Date")
-
-# Extract just the date values
-date_texts = [re.search(date_pattern, elem.text).group(0) for elem in dates]
-date_texts
-```
-
-## Working with Form Tables
-
-```python
-# Run layout analysis to find table structures
-page.analyze_layout()
-
-# Find possible form tables
-tables = page.find_all('region[type=table]')
-
-if tables:
-    # Visualize the tables
-    tables.show(color="purple", label="Form Table")
-    
-    # Extract data from the first table
-    first_table = tables[0]
-    table_data = first_table.extract_table()
-    table_data
-else:
-    # Try to find form-like structure using text alignment
-    # Create a region where a form might be
-    form_region = page.create_region(50, 200, page.width - 50, 500)
-    
-    # Group text by vertical position
-    rows = {}
-    text_elements = form_region.find_all('text')
-    
-    for elem in text_elements:
-        # Round y-position to group elements in the same row
-        row_pos = round(elem.top / 5) * 5
-        if row_pos not in rows:
-            rows[row_pos] = []
-        rows[row_pos].append(elem)
-    
-    # Extract data from rows (first 5 rows)
-    row_data = []
-    for y in sorted(rows.keys())[:5]:
-        # Sort elements by x-position (left to right)
-        elements = sorted(rows[y], key=lambda e: e.x0)
-        
-        # Show the row
-        row_box = form_region.create_region(
-            min(e.x0 for e in elements), 
-            min(e.top for e in elements),
-            max(e.x1 for e in elements),
-            max(e.bottom for e in elements)
-        )
-        row_box.show(color=None, use_color_cycling=True)
-        
-        # Extract text from row
-        row_text = [e.text for e in elements]
-        row_data.append(row_text)
-    
-    # Show the extracted rows
-    row_data
-```
-
-## Combining Different Extraction Techniques
-
-```python
-# Combine label-based and pattern-based extraction
-all_fields = {}
-
-# 1. First get fields with explicit labels
-for label in labels:
-    field_name = label.text.strip().rstrip(':')
-    value = label.right(width=200).extract_text().strip()
-    all_fields[field_name] = value
-
-# 2. Add date fields that we found with pattern matching
-for date_elem in dates:
-    # Find the nearest label
-    nearby_label = date_elem.nearest('text:contains(":")')
-    
-    if nearby_label:
-        # Extract the label text
-        label_text = nearby_label.text.strip().rstrip(':')
-        
-        # Get the date value
-        date_value = re.search(date_pattern, date_elem.text).group(0)
-        
-        # Add to our results if not already present
-        if label_text not in all_fields:
-            all_fields[label_text] = date_value
-
-# Show all extracted fields
-all_fields
-```
-
-Form field extraction enables you to automate data entry and document processing. By combining different techniques like label detection, spatial navigation, and pattern matching, you can handle a wide variety of form layouts. 

docs/tutorials/11-enhanced-table-processing.ipynb

@@ -1,54 +0,0 @@
-{
- "cells": [
-  {
-   "cell_type": "markdown",
-   "id": "7674e123",
-   "metadata": {},
-   "source": [
-    "# Enhanced Table Processing\n",
-    "\n",
-    "Tables are a common way to present structured data in documents, but they can be challenging to extract correctly. This tutorial demonstrates advanced techniques for working with tables in natural-pdf.\n",
-    "\n",
-    "TK"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 1,
-   "id": "08c7c5f0",
-   "metadata": {
-    "execution": {
-     "iopub.execute_input": "2025-04-21T21:25:37.324499Z",
-     "iopub.status.busy": "2025-04-21T21:25:37.324337Z",
-     "iopub.status.idle": "2025-04-21T21:25:37.328739Z",
-     "shell.execute_reply": "2025-04-21T21:25:37.328344Z"
-    }
-   },
-   "outputs": [],
-   "source": [
-    "#%pip install \"natural-pdf[all]\""
-   ]
-  }
- ],
- "metadata": {
-  "jupytext": {
-   "cell_metadata_filter": "-all",
-   "main_language": "python",
-   "notebook_metadata_filter": "-all"
-  },
-  "language_info": {
-   "codemirror_mode": {
-    "name": "ipython",
-    "version": 3
-   },
-   "file_extension": ".py",
-   "mimetype": "text/x-python",
-   "name": "python",
-   "nbconvert_exporter": "python",
-   "pygments_lexer": "ipython3",
-   "version": "3.10.13"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 5
-}

docs/tutorials/11-enhanced-table-processing.md

@@ -1,9 +0,0 @@
-# Enhanced Table Processing
-
-Tables are a common way to present structured data in documents, but they can be challenging to extract correctly. This tutorial demonstrates advanced techniques for working with tables in natural-pdf.
-
-TK
-
-```python
-#%pip install "natural-pdf[all]"
-```