ereader module
Python CI / test (push) Failing after 6m25s

This commit is contained in:
2025-11-04 18:39:38 +01:00
parent 4fe5f8cf60
commit 25e7f1423e
7 changed files with 1658 additions and 3 deletions
@@ -0,0 +1,380 @@
#!/usr/bin/env python3
"""
EPUB page renderer using DocumentLayouter.
This tool uses pyWebLayout's DocumentLayouter to render EPUB content:
1. Load an EPUB file
2. Extract all blocks (paragraphs, images, etc.)
3. Use DocumentLayouter to layout blocks on pages
4. Save the pages as PNG images
Usage:
python epub_page_renderer_documentlayouter.py book.epub --pages 5 --output-dir rendered_pages
"""
import os
import sys
import argparse
from pathlib import Path
from typing import Optional, List, Tuple
# Add the parent directory to sys.path to import pyWebLayout
sys.path.insert(0, str(Path(__file__).parent.parent))
try:
from pyWebLayout.io.readers.epub_reader import read_epub
from pyWebLayout.layout.document_layouter import DocumentLayouter, paragraph_layouter, image_layouter
from pyWebLayout.concrete.page import Page
from pyWebLayout.style.page_style import PageStyle
from pyWebLayout.style.fonts import Font
from pyWebLayout.style.alignment import Alignment
from pyWebLayout.abstract.block import Block, Paragraph, Heading, HList, Table, Image as AbstractImage
from pyWebLayout.style.concrete_style import RenderingContext, StyleResolver
from PIL import Image, ImageDraw
except ImportError as e:
print(f"Error importing required modules: {e}")
print("Make sure pyWebLayout is properly installed and PIL is available")
sys.exit(1)
def get_all_blocks_from_book(book) -> List[Block]:
"""
Extract all blocks from all chapters in the book.
Args:
book: The Book object from epub_reader
Returns:
List of all Block objects
"""
all_blocks = []
# Iterate through all chapters
for chapter in book.chapters:
# Get blocks from the chapter
if hasattr(chapter, '_blocks'):
all_blocks.extend(chapter._blocks)
return all_blocks
def create_page(page_style: PageStyle, page_size: Tuple[int, int]) -> Page:
"""
Create a new Page with the given style and size.
Args:
page_style: Style configuration for the page
page_size: (width, height) tuple
Returns:
A new Page object
"""
page = Page(
size=page_size,
style=page_style
)
return page
def layout_blocks_on_pages(blocks: List[Block], page_style: PageStyle,
page_size: Tuple[int, int], max_pages: int,
alignment_override: Optional[Alignment] = None) -> List[Page]:
"""
Layout blocks across multiple pages using DocumentLayouter.
Args:
blocks: List of abstract blocks to layout
page_style: Style configuration for pages
page_size: (width, height) tuple for pages
max_pages: Maximum number of pages to generate
alignment_override: Optional alignment to override paragraph alignment
Returns:
List of rendered Page objects
"""
pages = []
current_block_index = 0
continuation_word_index = 0
continuation_pretext = None
# Create rendering context
rendering_context = RenderingContext(base_font_size=16)
while current_block_index < len(blocks) and len(pages) < max_pages:
# Create a new page
page = create_page(page_style, page_size)
page_has_content = False
# Try to layout blocks on this page
while current_block_index < len(blocks):
block = blocks[current_block_index]
if isinstance(block, (Paragraph, Heading)):
# Layout paragraph/heading
success, failed_word_index, remaining_pretext = paragraph_layouter(
block,
page,
start_word=continuation_word_index,
pretext=continuation_pretext,
alignment_override=alignment_override
)
if success:
# Block fully laid out, move to next block
page_has_content = True
current_block_index += 1
continuation_word_index = 0
continuation_pretext = None
else:
# Block partially laid out or page is full
if failed_word_index is not None:
# Partial layout - continue on next page
page_has_content = True
continuation_word_index = failed_word_index
continuation_pretext = remaining_pretext
# Break to create a new page
break
elif isinstance(block, AbstractImage):
# Layout image
success = image_layouter(block, page)
if success:
page_has_content = True
current_block_index += 1
continuation_word_index = 0
continuation_pretext = None
else:
# Image doesn't fit, try on next page
break
elif isinstance(block, HList):
# Layout list items as paragraphs
try:
list_items = list(block.items())
for item in list_items:
if isinstance(item, Paragraph):
success, failed_word_index, remaining_pretext = paragraph_layouter(
item,
page,
start_word=continuation_word_index,
pretext=continuation_pretext,
alignment_override=alignment_override
)
if not success:
# Can't fit more on this page
page_has_content = True
break
continuation_word_index = 0
continuation_pretext = None
# Move to next block after processing list
page_has_content = True
current_block_index += 1
except Exception as e:
print(f"Warning: Error processing list: {e}")
current_block_index += 1
elif isinstance(block, Table):
# Skip tables for now (not implemented)
print(f"Warning: Skipping table (not yet implemented)")
current_block_index += 1
else:
# Unknown block type, skip
print(f"Warning: Skipping unknown block type: {type(block).__name__}")
current_block_index += 1
# Add page if it has content
if page_has_content:
pages.append(page)
else:
# No content could be added to this page, stop
break
return pages
def main():
"""Main function to handle command line arguments and process the EPUB."""
parser = argparse.ArgumentParser(
description='Render EPUB pages using DocumentLayouter',
formatter_class=argparse.RawDescriptionHelpFormatter,
epilog="""
Examples:
python epub_page_renderer_documentlayouter.py book.epub --pages 5
python epub_page_renderer_documentlayouter.py book.epub --pages 10 --output-dir my_output --width 600 --height 800
"""
)
parser.add_argument(
'epub_file',
help='Path to the EPUB file to render'
)
parser.add_argument(
'--pages', '-p',
type=int,
default=5,
help='Number of pages to render (default: 5)'
)
parser.add_argument(
'--output-dir', '-o',
default='rendered_pages',
help='Output directory for rendered images (default: rendered_pages)'
)
parser.add_argument(
'--width', '-w',
type=int,
default=800,
help='Page width in pixels (default: 800)'
)
parser.add_argument(
'--height', '-t',
type=int,
default=1000,
help='Page height in pixels (default: 1000)'
)
parser.add_argument(
'--margin', '-m',
type=int,
default=40,
help='Page margin in pixels (default: 40)'
)
parser.add_argument(
'--align', '-a',
choices=['left', 'justify'],
default='left',
help='Text alignment: left or justify (default: left)'
)
args = parser.parse_args()
# Validate arguments
if not os.path.exists(args.epub_file):
print(f"Error: EPUB file '{args.epub_file}' not found")
return 1
if args.pages <= 0:
print("Error: Number of pages must be positive")
return 1
# Create output directory
try:
os.makedirs(args.output_dir, exist_ok=True)
except OSError as e:
print(f"Error creating output directory: {e}")
return 1
print(f"Loading EPUB file: {args.epub_file}")
# Load the EPUB file
try:
book = read_epub(args.epub_file)
print(f"Successfully loaded EPUB: {book.get_title() or 'Unknown Title'}")
# Print book information
author = book.get_metadata('AUTHOR')
if author:
print(f"Author: {author}")
print(f"Chapters: {len(book.chapters) if hasattr(book, 'chapters') else 'Unknown'}")
except Exception as e:
print(f"Error loading EPUB file: {e}")
import traceback
traceback.print_exc()
return 1
# Extract all blocks from the book
print("Extracting content blocks...")
try:
all_blocks = get_all_blocks_from_book(book)
print(f"Extracted {len(all_blocks)} content blocks")
if not all_blocks:
print("No content blocks found in EPUB. The book might be empty.")
return 1
except Exception as e:
print(f"Error extracting blocks: {e}")
import traceback
traceback.print_exc()
return 1
# Set up page style
page_size = (args.width, args.height)
page_style = PageStyle(
background_color=(255, 255, 255),
border_width=args.margin,
border_color=(200, 200, 200),
padding=(10, 10, 10, 10), # top, right, bottom, left
line_spacing=5,
inter_block_spacing=15
)
# Set alignment
alignment = Alignment.JUSTIFY if args.align == 'justify' else Alignment.LEFT
print(f"Setting up layouter with page size {page_size} and {args.align} alignment")
# Layout blocks on pages
print(f"Rendering up to {args.pages} pages using DocumentLayouter...")
try:
pages = layout_blocks_on_pages(
all_blocks,
page_style,
page_size,
args.pages,
alignment_override=alignment
)
if not pages:
print("No pages were generated.")
return 1
print(f"Generated {len(pages)} pages")
# Save each page to an image
for i, page in enumerate(pages):
print(f"Saving page {i + 1}/{len(pages)}...")
try:
# Render the page
image = page.render()
# Save the image
output_filename = f"page_{i + 1:03d}.png"
output_path = os.path.join(args.output_dir, output_filename)
image.save(output_path, 'PNG')
print(f"Saved: {output_path}")
except Exception as e:
print(f"Error saving page {i + 1}: {e}")
import traceback
traceback.print_exc()
continue
print(f"\nCompleted! Rendered {len(pages)} pages to {args.output_dir}")
except Exception as e:
print(f"Error during pagination/rendering: {e}")
import traceback
traceback.print_exc()
return 1
return 0
if __name__ == "__main__":
sys.exit(main())
+329
View File
@@ -0,0 +1,329 @@
#!/usr/bin/env python3
"""
Comprehensive demo of the EbookReader functionality.
This script demonstrates all features of the pyWebLayout EbookReader:
- Loading EPUB files
- Page navigation (forward/backward)
- Position saving/loading
- Chapter navigation
- Font size and spacing adjustments
- Getting book and position information
Usage:
python ereader_demo.py path/to/book.epub
"""
import sys
import os
from pathlib import Path
# Add parent directory to path to import pyWebLayout
sys.path.insert(0, str(Path(__file__).parent.parent))
from pyWebLayout.layout.ereader_application import EbookReader
def print_separator():
"""Print a visual separator."""
print("\n" + "="*70 + "\n")
def demo_basic_navigation(reader: EbookReader):
"""Demonstrate basic page navigation."""
print("DEMO: Basic Navigation")
print_separator()
# Get current page
print("Getting first page...")
page = reader.get_current_page()
if page:
print(f"✓ Current page rendered: {page.size}")
reader.render_to_file("demo_page_001.png")
print(" Saved to: demo_page_001.png")
# Navigate forward
print("\nNavigating to next page...")
page = reader.next_page()
if page:
print(f"✓ Next page rendered: {page.size}")
reader.render_to_file("demo_page_002.png")
print(" Saved to: demo_page_002.png")
# Navigate backward
print("\nNavigating to previous page...")
page = reader.previous_page()
if page:
print(f"✓ Previous page rendered: {page.size}")
print_separator()
def demo_position_management(reader: EbookReader):
"""Demonstrate position save/load functionality."""
print("DEMO: Position Management")
print_separator()
# Navigate a few pages forward
print("Navigating forward 3 pages...")
for i in range(3):
reader.next_page()
# Save position
print("Saving current position as 'demo_bookmark'...")
success = reader.save_position("demo_bookmark")
if success:
print("✓ Position saved successfully")
# Get position info
pos_info = reader.get_position_info()
print(f"\nCurrent position info:")
print(f" Chapter: {pos_info.get('chapter', {}).get('title', 'N/A')}")
print(f" Block index: {pos_info['position']['block_index']}")
print(f" Word index: {pos_info['position']['word_index']}")
print(f" Progress: {pos_info['progress']*100:.1f}%")
# Navigate away
print("\nNavigating forward 5 more pages...")
for i in range(5):
reader.next_page()
# Load saved position
print("Loading saved position 'demo_bookmark'...")
page = reader.load_position("demo_bookmark")
if page:
print("✓ Position restored successfully")
reader.render_to_file("demo_restored_position.png")
print(" Saved to: demo_restored_position.png")
# List all saved positions
positions = reader.list_saved_positions()
print(f"\nAll saved positions: {positions}")
print_separator()
def demo_chapter_navigation(reader: EbookReader):
"""Demonstrate chapter navigation."""
print("DEMO: Chapter Navigation")
print_separator()
# Get all chapters
chapters = reader.get_chapters()
print(f"Found {len(chapters)} chapters:")
for title, idx in chapters[:5]: # Show first 5
print(f" [{idx}] {title}")
if len(chapters) > 5:
print(f" ... and {len(chapters) - 5} more")
# Jump to a chapter by index
if len(chapters) > 1:
print(f"\nJumping to chapter 1...")
page = reader.jump_to_chapter(1)
if page:
print("✓ Jumped to chapter successfully")
reader.render_to_file("demo_chapter_1.png")
print(" Saved to: demo_chapter_1.png")
# Get current chapter info
chapter_info = reader.get_current_chapter_info()
if chapter_info:
print(f" Current chapter: {chapter_info['title']}")
# Jump to a chapter by title (if we have chapters)
if len(chapters) > 0:
first_chapter_title = chapters[0][0]
print(f"\nJumping to chapter by title: '{first_chapter_title}'...")
page = reader.jump_to_chapter(first_chapter_title)
if page:
print("✓ Jumped to chapter by title successfully")
print_separator()
def demo_font_size_adjustment(reader: EbookReader):
"""Demonstrate font size adjustments."""
print("DEMO: Font Size Adjustment")
print_separator()
# Save current page for comparison
print("Rendering page at normal font size (1.0x)...")
page = reader.get_current_page()
if page:
reader.render_to_file("demo_font_normal.png")
print("✓ Saved to: demo_font_normal.png")
# Increase font size
print("\nIncreasing font size...")
page = reader.increase_font_size()
if page:
print(f"✓ Font size increased to {reader.get_font_size():.1f}x")
reader.render_to_file("demo_font_larger.png")
print(" Saved to: demo_font_larger.png")
# Increase again
print("\nIncreasing font size again...")
page = reader.increase_font_size()
if page:
print(f"✓ Font size increased to {reader.get_font_size():.1f}x")
reader.render_to_file("demo_font_largest.png")
print(" Saved to: demo_font_largest.png")
# Decrease font size
print("\nDecreasing font size...")
page = reader.decrease_font_size()
if page:
print(f"✓ Font size decreased to {reader.get_font_size():.1f}x")
# Set specific font size
print("\nResetting to normal font size (1.0x)...")
page = reader.set_font_size(1.0)
if page:
print("✓ Font size reset to 1.0x")
print_separator()
def demo_spacing_adjustment(reader: EbookReader):
"""Demonstrate line and block spacing adjustments."""
print("DEMO: Spacing Adjustment")
print_separator()
# Save current page
print("Rendering page with default spacing...")
page = reader.get_current_page()
if page:
reader.render_to_file("demo_spacing_default.png")
print("✓ Saved to: demo_spacing_default.png")
# Increase line spacing
print("\nIncreasing line spacing to 10px...")
page = reader.set_line_spacing(10)
if page:
print("✓ Line spacing increased")
reader.render_to_file("demo_spacing_lines_10.png")
print(" Saved to: demo_spacing_lines_10.png")
# Increase inter-block spacing
print("\nIncreasing inter-block spacing to 25px...")
page = reader.set_inter_block_spacing(25)
if page:
print("✓ Inter-block spacing increased")
reader.render_to_file("demo_spacing_blocks_25.png")
print(" Saved to: demo_spacing_blocks_25.png")
# Reset to defaults
print("\nResetting spacing to defaults (line: 5px, block: 15px)...")
reader.set_line_spacing(5)
page = reader.set_inter_block_spacing(15)
if page:
print("✓ Spacing reset to defaults")
print_separator()
def demo_book_information(reader: EbookReader):
"""Demonstrate getting book information."""
print("DEMO: Book Information")
print_separator()
# Get book info
book_info = reader.get_book_info()
print("Book Information:")
print(f" Title: {book_info['title']}")
print(f" Author: {book_info['author']}")
print(f" Document ID: {book_info['document_id']}")
print(f" Total blocks: {book_info['total_blocks']}")
print(f" Total chapters: {book_info['total_chapters']}")
print(f" Page size: {book_info['page_size']}")
print(f" Font scale: {book_info['font_scale']}")
# Get reading progress
progress = reader.get_reading_progress()
print(f"\nReading Progress: {progress*100:.1f}%")
# Get detailed position info
pos_info = reader.get_position_info()
print("\nDetailed Position:")
print(f" Chapter index: {pos_info['position']['chapter_index']}")
print(f" Block index: {pos_info['position']['block_index']}")
print(f" Word index: {pos_info['position']['word_index']}")
chapter = pos_info.get('chapter', {})
if chapter.get('title'):
print(f" Current chapter: {chapter['title']}")
print_separator()
def main():
"""Main function to run all demos."""
if len(sys.argv) < 2:
print("Usage: python ereader_demo.py path/to/book.epub")
print("\nExample EPUBs to try:")
print(" - test.epub (if available in project root)")
print(" - test2.epub (if available in project root)")
sys.exit(1)
epub_path = sys.argv[1]
if not os.path.exists(epub_path):
print(f"Error: File not found: {epub_path}")
sys.exit(1)
print("="*70)
print(" EbookReader Demo - pyWebLayout")
print("="*70)
print(f"\nLoading EPUB: {epub_path}")
# Create reader with context manager
with EbookReader(page_size=(800, 1000)) as reader:
# Load the EPUB
if not reader.load_epub(epub_path):
print("Error: Failed to load EPUB file")
sys.exit(1)
print("✓ EPUB loaded successfully")
# Run all demos
try:
demo_basic_navigation(reader)
demo_position_management(reader)
demo_chapter_navigation(reader)
demo_font_size_adjustment(reader)
demo_spacing_adjustment(reader)
demo_book_information(reader)
print("\n" + "="*70)
print(" Demo Complete!")
print("="*70)
print("\nGenerated demo images:")
demo_files = [
"demo_page_001.png",
"demo_page_002.png",
"demo_restored_position.png",
"demo_chapter_1.png",
"demo_font_normal.png",
"demo_font_larger.png",
"demo_font_largest.png",
"demo_spacing_default.png",
"demo_spacing_lines_10.png",
"demo_spacing_blocks_25.png"
]
for filename in demo_files:
if os.path.exists(filename):
print(f"{filename}")
print("\nAll features demonstrated successfully!")
except Exception as e:
print(f"\nError during demo: {e}")
import traceback
traceback.print_exc()
sys.exit(1)
if __name__ == "__main__":
main()
+96
View File
@@ -0,0 +1,96 @@
#!/usr/bin/env python3
"""
Simple example showing the most common EbookReader usage.
This script loads an EPUB and allows you to navigate through it,
saving each page as an image.
Usage:
python simple_ereader_example.py book.epub
"""
import sys
from pathlib import Path
# Add parent directory to path to import pyWebLayout
sys.path.insert(0, str(Path(__file__).parent.parent))
from pyWebLayout.layout.ereader_application import EbookReader
def main():
if len(sys.argv) < 2:
print("Usage: python simple_ereader_example.py book.epub")
sys.exit(1)
epub_path = sys.argv[1]
# Create reader and load EPUB
print(f"Loading: {epub_path}")
reader = EbookReader(page_size=(800, 1000))
if not reader.load_epub(epub_path):
print("Failed to load EPUB")
sys.exit(1)
# Get book information
info = reader.get_book_info()
print(f"\nBook: {info['title']}")
print(f"Author: {info['author']}")
print(f"Total blocks: {info['total_blocks']}")
# Get chapters
chapters = reader.get_chapters()
print(f"Chapters: {len(chapters)}")
if chapters:
print("\nChapter list:")
for title, idx in chapters[:10]: # Show first 10
print(f" {idx}: {title}")
if len(chapters) > 10:
print(f" ... and {len(chapters) - 10} more")
# Navigate through first 10 pages
print("\nRendering first 10 pages...")
for i in range(10):
page = reader.get_current_page()
if page:
filename = f"page_{i+1:03d}.png"
reader.render_to_file(filename)
# Show progress
progress = reader.get_reading_progress()
chapter_info = reader.get_current_chapter_info()
chapter_name = chapter_info['title'] if chapter_info else "N/A"
print(f" Page {i+1}: {filename} (Progress: {progress*100:.1f}%, Chapter: {chapter_name})")
# Move to next page
if not reader.next_page():
print(" Reached end of book")
break
# Save current position
reader.save_position("stopped_at_page_10")
print("\nSaved position as 'stopped_at_page_10'")
# Example: Jump to a chapter (if available)
if len(chapters) >= 2:
print(f"\nJumping to chapter: {chapters[1][0]}")
reader.jump_to_chapter(1)
reader.render_to_file("chapter_2_start.png")
print(" Saved to: chapter_2_start.png")
# Example: Increase font size
print("\nIncreasing font size...")
reader.increase_font_size()
reader.render_to_file("larger_font.png")
print(f" Font size now: {reader.get_font_size():.1f}x")
print(" Saved to: larger_font.png")
# Close reader (saves current position automatically)
reader.close()
print("\nDone! Current position saved automatically.")
if __name__ == "__main__":
main()