Pdf drawing extraction
[COLM'26] SkillLearnBench is the first benchmark for evaluating continual learning methods that automatically generate agent skills.
npx -y skills add cxcscmu/SkillLearnBench --skill pdf-drawing-extractionAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
What its author says it does
Copied from the file, not written here
Extracts vector graphics, colors, and text from PDF files using PyMuPDF to analyze calendars or visual schedules.
SKILL.md
1.4 KB, as published. Nobody here has run it
PDF Drawing Extraction with PyMuPDF
This skill demonstrates how to use PyMuPDF (fitz) to extract vector drawings and rectangles from a PDF. This is particularly useful for extracting calendar blocks and colors.
Setup
Ensure PyMuPDF is installed:
pip install PyMuPDF
Basic Usage
import fitz
def extract_colored_blocks(pdf_path):
doc = fitz.open(pdf_path)
page = doc[0] # first page
# Get all drawings
drawings = page.get_drawings()
blocks = []
for d in drawings:
rect = d.get('rect')
color = d.get('color')
fill = d.get('fill')
# Determine color (RGB values typically 0.0 to 1.0)
# Assuming fill color is used for block background
if fill:
r, g, b = fill
is_blue = (b > r and b > g) # basic blue check
blocks.append({
'rect': rect,
'is_blue': is_blue,
'fill': fill
})
return blocks
This extracts bounding boxes (fitz.Rect) and their associated fill colors. You can correlate these boxes with text in the same PDF by using page.get_text("dict") or checking overlaps between text rectangles and drawing rectangles.