Include example PDFs with Git LFS and add helper scripts

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.pdf filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -68,7 +68,7 @@ htmlcov/
 # =========================
 uploads/
 data/
-*.pdf
+# *.pdf
 *.png
 *.jpg
 *.jpeg

18.1 Notes.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed3d2790d00abbd7d89fbb53b745de70b46a667613b2ad52db498be48ea4d1f1
+size 3392377

debug_rotation.py

@@ -0,0 +1,49 @@
+
+import pymupdf as fitz
+from PIL import Image, ImageDraw
+
+def create_rotated_pdf():
+    doc = fitz.open()
+    page = doc.new_page()
+    
+    # Insert text at specific coordinates (100, 100) -> (200, 150) roughly
+    page.insert_text((100, 100), "Hello World", fontsize=20)
+    
+    # Rotate page 90 degrees
+    page.set_rotation(90)
+    
+    doc.save("rotated_test.pdf")
+    return "rotated_test.pdf"
+
+def analyze_pdf(path):
+    doc = fitz.open(path)
+    page = doc[0]
+    
+    print(f"Page Rotation: {page.rotation}")
+    print(f"Page Rect (unrotated): {page.rect}")
+    
+    # Pixmap
+    pix = page.get_pixmap()
+    print(f"Pixmap WxH: {pix.width}x{pix.height}")
+    
+    # Text Dict
+    raw = page.get_text("dict")
+    block = raw["blocks"][0]
+    bbox = fitz.Rect(block["bbox"])
+    print(f"Text Bbox (raw): {bbox}")
+    
+    # Try transforming
+    # page.rotation_matrix might not exist as a property, let's check dir
+    if hasattr(page, "rotation_matrix"):
+        mat = page.rotation_matrix
+        print(f"Page.rotation_matrix: {mat}")
+        t_bbox = bbox * mat
+        print(f"Transformed Bbox: {t_bbox}")
+    else:
+        print("No page.rotation_matrix")
+        
+    doc.close()
+
+if __name__ == "__main__":
+    pdf_path = create_rotated_pdf()
+    analyze_pdf(pdf_path)

generate_test_pdf.py

@@ -0,0 +1,40 @@
+
+import pymupdf
+import os
+
+def create_test_pdf(filename="test_document.pdf"):
+    doc = pymupdf.open()
+    page = doc.new_page()
+    
+    # Title
+    page.insert_text((50, 50), "PDF Structure Inspector Test Document", fontsize=24)
+    
+    # Normal paragraph
+    text = "This is a normal paragraph of text to test basic extraction."
+    page.insert_text((50, 80), text, fontsize=12)
+    
+    # Math-like text to test math detection
+    math_text = "Here is some math: f(x) = sum(x_i) for i in N. Also x^2 + y^2 = r^2."
+    page.insert_text((50, 110), math_text, fontsize=12)
+    
+    # Text that might look like a header/column
+    page.insert_text((50, 150), "Column 1", fontsize=14)
+    page.insert_text((300, 150), "Column 2", fontsize=14)
+    
+    page.insert_text((50, 170), "Left side content.", fontsize=12)
+    page.insert_text((300, 170), "Right side content.", fontsize=12)
+    
+    # Add a drawing (vector)
+    page.draw_rect((50, 200, 150, 250), color=(0, 0, 1))
+    page.insert_text((55, 225), "Vector Box", fontsize=10, color=(1, 1, 1))
+
+    # Add a second page for batch testing
+    page2 = doc.new_page()
+    page2.insert_text((50, 50), "Page 2 - Batch Analysis Test", fontsize=24)
+    page2.insert_text((50, 80), "Just another page to verify multi-page processing.", fontsize=12)
+    
+    doc.save(filename)
+    print(f"Created {filename}")
+
+if __name__ == "__main__":
+    create_test_pdf()

logic.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:27322fa134c2fbce6dd38e1f24b2c0d65c40a50f7abd9bc154d71f6f91464e48
+size 635839

repro_error.py

@@ -0,0 +1,34 @@
+
+import pymupdf as fitz
+from layout_utils import extract_blocks_spans
+
+def test_extraction():
+    # Open the real PDF
+    doc = fitz.open("logic.pdf")
+    # doc.new_page() ... no
+    
+    # Analyze it using the function I modified
+    try:
+        blocks = extract_blocks_spans(doc, 0)
+        print(f"Blocks extracted: {len(blocks)}")
+        for i, b in enumerate(blocks):
+            bbox = b.bbox
+            print(f"Block {i} bbox type: {type(bbox)}")
+            if isinstance(bbox, tuple):
+                print(f"Block {i} bbox len: {len(bbox)}")
+                if len(bbox) == 4:
+                    print(f"Block {i} sample: {bbox}")
+                    # Try unpacking
+                    x0, y0, x1, y1 = bbox
+                    # Try explicit float cast
+                    _ = float(x0)
+            else:
+                print(f"Block {i} bbox is NOT tuple: {bbox}")
+
+    except Exception as e:
+        print(f"Error caught: {type(e).__name__}: {e}")
+        import traceback
+        traceback.print_exc()
+
+if __name__ == "__main__":
+    test_extraction()

rotated_test.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:276ce24070de6a5c2f034980be647556e73d8a9e25fd051089ef807c10ebbdaa
+size 746

test_app_analyze.py

@@ -0,0 +1,28 @@
+
+from app import analyze
+import traceback
+
+def test_analyze():
+    pdf_path = "test_document.pdf"
+    page_num = 1
+    dpi = 72
+    order_mode = "raw"
+    show_spans = False
+    highlight_math = False
+    
+    try:
+        print(f"Analyzing {pdf_path}...")
+        overlay, report, summary, preview = analyze(
+            pdf_path, page_num, dpi, order_mode, show_spans, highlight_math
+        )
+        print("Analysis successful!")
+        print(f"Summary length: {len(summary)}")
+        print(f"Preview length: {len(preview)}")
+        print(f"Report keys: {report.keys()}")
+        print(f"Overlay size: {overlay.size}")
+    except Exception as e:
+        print(f"CRASHED: {e}")
+        traceback.print_exc()
+
+if __name__ == "__main__":
+    test_analyze()

test_document.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f2648892600eb411eae617de92b54231946e51639c564331537886f55f94cd2
+size 2759