Commit ·
b874e9b
1
Parent(s): 4b2c414
Update
Browse files- app.py +174 -0
- requirements.txt +0 -0
app.py
ADDED
|
@@ -0,0 +1,174 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import fitz # PyMuPDF
|
| 2 |
+
import gradio as gr
|
| 3 |
+
import zipfile
|
| 4 |
+
import io
|
| 5 |
+
import os
|
| 6 |
+
import tempfile
|
| 7 |
+
from datetime import datetime # Import to get current date
|
| 8 |
+
|
| 9 |
+
# Helper function to find color areas
|
| 10 |
+
def find_color_areas(page, target_color, tolerance=30):
|
| 11 |
+
pix = page.get_pixmap()
|
| 12 |
+
width, height = pix.width, pix.height
|
| 13 |
+
visited = [[False for _ in range(width)] for _ in range(height)]
|
| 14 |
+
rectangles = []
|
| 15 |
+
|
| 16 |
+
def flood_fill(x, y):
|
| 17 |
+
stack = [(x, y)]
|
| 18 |
+
rects = []
|
| 19 |
+
while stack:
|
| 20 |
+
cx, cy = stack.pop()
|
| 21 |
+
if visited[cy][cx]:
|
| 22 |
+
continue
|
| 23 |
+
|
| 24 |
+
visited[cy][cx] = True
|
| 25 |
+
pixel_color = pix.pixel(cx, cy)
|
| 26 |
+
r, g, b = pixel_color[:3]
|
| 27 |
+
|
| 28 |
+
if (abs(r - target_color[0]) <= tolerance and
|
| 29 |
+
abs(g - target_color[1]) <= tolerance and
|
| 30 |
+
abs(b - target_color[2]) <= tolerance):
|
| 31 |
+
rects.append(fitz.Rect(cx, cy, cx + 1, cy + 1))
|
| 32 |
+
if cx > 0: stack.append((cx - 1, cy))
|
| 33 |
+
if cx < width - 1: stack.append((cx + 1, cy))
|
| 34 |
+
if cy > 0: stack.append((cx, cy - 1))
|
| 35 |
+
if cy < height - 1: stack.append((cx, cy + 1))
|
| 36 |
+
|
| 37 |
+
if rects:
|
| 38 |
+
bbox = fitz.Rect(min([r.x0 for r in rects]),
|
| 39 |
+
min([r.y0 for r in rects]),
|
| 40 |
+
max([r.x1 for r in rects]),
|
| 41 |
+
max([r.y1 for r in rects]))
|
| 42 |
+
return bbox
|
| 43 |
+
return None
|
| 44 |
+
|
| 45 |
+
for y in range(height):
|
| 46 |
+
for x in range(width):
|
| 47 |
+
if not visited[y][x]:
|
| 48 |
+
bbox = flood_fill(x, y)
|
| 49 |
+
if bbox:
|
| 50 |
+
rectangles.append(bbox)
|
| 51 |
+
return rectangles
|
| 52 |
+
|
| 53 |
+
def merge_overlapping_rectangles(rectangles):
|
| 54 |
+
merged_rects = []
|
| 55 |
+
while rectangles:
|
| 56 |
+
rect = rectangles.pop(0)
|
| 57 |
+
to_merge = [rect]
|
| 58 |
+
for other in rectangles[:]:
|
| 59 |
+
if rect.intersects(other):
|
| 60 |
+
to_merge.append(other)
|
| 61 |
+
rectangles.remove(other)
|
| 62 |
+
merged_rect = fitz.Rect(
|
| 63 |
+
min([r.x0 for r in to_merge]),
|
| 64 |
+
min([r.y0 for r in to_merge]),
|
| 65 |
+
max([r.x1 for r in to_merge]),
|
| 66 |
+
max([r.y1 for r in to_merge])
|
| 67 |
+
)
|
| 68 |
+
merged_rects.append(merged_rect)
|
| 69 |
+
return merged_rects
|
| 70 |
+
|
| 71 |
+
def markup_color_regions(doc, color_comment_pairs, tolerance=30):
|
| 72 |
+
for page_num in range(len(doc)):
|
| 73 |
+
page = doc[page_num]
|
| 74 |
+
for color_comment_pair in color_comment_pairs:
|
| 75 |
+
target_color = color_comment_pair['color']
|
| 76 |
+
comment = color_comment_pair['comment']
|
| 77 |
+
stroke_color = color_comment_pair['stroke_color']
|
| 78 |
+
|
| 79 |
+
rectangles = find_color_areas(page, target_color, tolerance)
|
| 80 |
+
if rectangles:
|
| 81 |
+
merged_rectangles = merge_overlapping_rectangles(rectangles)
|
| 82 |
+
for bbox in merged_rectangles:
|
| 83 |
+
annot = page.add_rect_annot(bbox)
|
| 84 |
+
annot.set_colors(stroke=stroke_color)
|
| 85 |
+
annot.set_border(width=2)
|
| 86 |
+
annot.set_info({"title": "Markup", "content": comment})
|
| 87 |
+
annot.update()
|
| 88 |
+
|
| 89 |
+
def process_pdf_files(input_pdfs, selected_color_comment_indices, tolerance, custom_color, custom_comment, custom_stroke_color):
|
| 90 |
+
color_comment_pairs = [
|
| 91 |
+
{
|
| 92 |
+
"color": (235, 128, 138),
|
| 93 |
+
"comment": "Structural Slab greater than architectural slab",
|
| 94 |
+
"stroke_color": (1, 0, 0)
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"color": (128, 253, 128),
|
| 98 |
+
"comment": "Arch Slab greater than Structure",
|
| 99 |
+
"stroke_color": (0, 1, 0)
|
| 100 |
+
}
|
| 101 |
+
]
|
| 102 |
+
|
| 103 |
+
# Add custom color-comment pair if provided
|
| 104 |
+
if custom_color and custom_comment and custom_stroke_color:
|
| 105 |
+
custom_color_tuple = tuple(map(int, custom_color.split(','))) # Convert color to tuple
|
| 106 |
+
custom_stroke_tuple = tuple(map(int, custom_stroke_color.split(','))) # Convert stroke to tuple
|
| 107 |
+
color_comment_pairs.append({
|
| 108 |
+
"color": custom_color_tuple,
|
| 109 |
+
"comment": custom_comment,
|
| 110 |
+
"stroke_color": custom_stroke_tuple
|
| 111 |
+
})
|
| 112 |
+
|
| 113 |
+
# Get the selected color-comment pairs
|
| 114 |
+
selected_color_comment_pairs = [color_comment_pairs[i] for i in selected_color_comment_indices]
|
| 115 |
+
|
| 116 |
+
# List to keep track of all modified PDFs (in-memory)
|
| 117 |
+
modified_pdfs = []
|
| 118 |
+
|
| 119 |
+
# Process each input PDF file
|
| 120 |
+
for pdf_file in input_pdfs:
|
| 121 |
+
with open(pdf_file.name, "rb") as file_stream:
|
| 122 |
+
doc = fitz.open(stream=file_stream.read(), filetype="pdf")
|
| 123 |
+
markup_color_regions(doc, selected_color_comment_pairs, tolerance)
|
| 124 |
+
|
| 125 |
+
# Save the modified PDF in memory
|
| 126 |
+
pdf_in_memory = io.BytesIO()
|
| 127 |
+
doc.save(pdf_in_memory)
|
| 128 |
+
doc.close()
|
| 129 |
+
|
| 130 |
+
# Move pointer to the beginning of the BytesIO object
|
| 131 |
+
pdf_in_memory.seek(0)
|
| 132 |
+
modified_pdfs.append((os.path.basename(pdf_file.name), pdf_in_memory))
|
| 133 |
+
|
| 134 |
+
# Create an in-memory ZIP file containing all modified PDFs
|
| 135 |
+
zip_in_memory = io.BytesIO()
|
| 136 |
+
with zipfile.ZipFile(zip_in_memory, 'w') as zipf:
|
| 137 |
+
for pdf_name, pdf_bytes in modified_pdfs:
|
| 138 |
+
zipf.writestr(pdf_name, pdf_bytes.read())
|
| 139 |
+
|
| 140 |
+
# Move pointer to the beginning of the ZIP file BytesIO object
|
| 141 |
+
zip_in_memory.seek(0)
|
| 142 |
+
|
| 143 |
+
# Generate the current date string
|
| 144 |
+
current_date = datetime.now().strftime("%Y-%m-%d")
|
| 145 |
+
|
| 146 |
+
# Create a temporary file with a custom name including the date
|
| 147 |
+
temp_filename = f"CoordinationPDFS_{current_date}.zip"
|
| 148 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix=".zip", prefix="CoordinationPDFS_", dir=".") as tmp_file:
|
| 149 |
+
tmp_file.write(zip_in_memory.read())
|
| 150 |
+
temp_file_name = tmp_file.name # Store the temp file name
|
| 151 |
+
|
| 152 |
+
return temp_file_name # Return the path to the temp file
|
| 153 |
+
|
| 154 |
+
# Define the Gradio interface
|
| 155 |
+
interface = gr.Interface(
|
| 156 |
+
fn=process_pdf_files,
|
| 157 |
+
inputs=[
|
| 158 |
+
gr.Files(label="Input PDF Files", file_types=[".pdf"]),
|
| 159 |
+
gr.CheckboxGroup( # Use CheckboxGroup for multi-select
|
| 160 |
+
label="Select Color-Comment Pairs",
|
| 161 |
+
choices=["Structural Slab vs Arch Slab", "Arch Slab vs Structural Slab", "Custom Option"],
|
| 162 |
+
type="index"
|
| 163 |
+
),
|
| 164 |
+
gr.Slider(label="Tolerance", minimum=0, maximum=100, step=1, value=30),
|
| 165 |
+
gr.Textbox(label="Custom Color (R,G,B)", placeholder="Enter custom color in RGB format, e.g., 255,0,0"),
|
| 166 |
+
gr.Textbox(label="Custom Comment", placeholder="Enter custom comment for this color"),
|
| 167 |
+
gr.Textbox(label="Custom Stroke Color (R,G,B)", placeholder="Enter stroke color in RGB format, e.g., 0,0,255")
|
| 168 |
+
],
|
| 169 |
+
outputs=gr.File(label="Download Modified PDFs as ZIP"),
|
| 170 |
+
title="PDF Color Region Markup"
|
| 171 |
+
)
|
| 172 |
+
|
| 173 |
+
# Launch the Gradio app
|
| 174 |
+
interface.launch()
|
requirements.txt
ADDED
|
Binary file (1.91 kB). View file
|
|
|