working prototype

cost_calculator.py

@@ -3,6 +3,7 @@ import os
 from PyPDF2 import PdfReader
 from pdf2image import convert_from_path
 import numpy as np
+from PIL import Image
 
 UPLOAD_FOLDER = "data/uploads"
 ALLOWED_EXTENSIONS = {"pdf"}
@@ -13,6 +14,7 @@ def allowed_file(filename: str) -> bool:
 
 
 def points_to_meters(points: float) -> float:
+    """Convert PDF points to meters (1 point = 1/72 inch)"""
     return (points / 72.0) * 0.0254
 
 
@@ -24,6 +26,21 @@ def get_rate_color() -> float:
     return float(os.environ.get("RATE_PER_M2_COLOR", "5.0"))
 
 
+def get_page_size(page):
+    """Returns width and height in meters, rotation-aware, prefers CropBox"""
+    box = getattr(page, "cropbox", None) or page.mediabox
+    width_pts = float(box.width)
+    height_pts = float(box.height)
+
+    rotation = page.get("/Rotate") or 0
+    if rotation in [90, 270]:
+        width_pts, height_pts = height_pts, width_pts
+
+    width_m = points_to_meters(width_pts)
+    height_m = points_to_meters(height_pts)
+    return width_m, height_m
+
+
 def analyze_pdf(path, compute_ink=True):
     reader = PdfReader(path)
     pages_info = []
@@ -31,9 +48,8 @@ def analyze_pdf(path, compute_ink=True):
     total_cost_black = total_cost_color = 0.0
 
     for i, page in enumerate(reader.pages):
-        box = page.mediabox
-        width_m = points_to_meters(float(box.width))
-        height_m = points_to_meters(float(box.height))
+        # Get page size robustly
+        width_m, height_m = get_page_size(page)
         area = width_m * height_m
 
         ink_pct = None
@@ -45,14 +61,22 @@ def analyze_pdf(path, compute_ink=True):
                 img = images[0].convert("RGB")
                 arr = np.array(img)
 
-                # ink pixels: any channel < 240
-                ink_mask = np.any(arr < 240, axis=2)
-                ink_pct = float(np.count_nonzero(ink_mask)) / (arr.shape[0] * arr.shape[1]) * 100.0
+                # Detect ink pixels (anything not near-white)
+                ink_mask = np.any(arr < 250, axis=2)
+                num_ink_pixels = np.count_nonzero(ink_mask)
+                total_pixels = arr.shape[0] * arr.shape[1]
+                ink_pct = (num_ink_pixels / total_pixels) * 100.0
+
+                if num_ink_pixels > 0:
+                    # Convert to HSV using Pillow
+                    hsv_img = img.convert("HSV")
+                    hsv_arr = np.array(hsv_img)
+                    saturation = hsv_arr[:, :, 1][ink_mask]
+
+                    # Color if even a tiny fraction of ink pixels have saturation > 10
+                    color_ratio = np.count_nonzero(saturation > 10) / len(saturation)
+                    is_color = color_ratio > 0.001  # 0.1% threshold
 
-                # simple color detection: if RGB channels differ significantly
-                avg_rgb = arr.mean(axis=(0, 1))
-                if np.ptp(avg_rgb) > 30:
-                    is_color = True
             except Exception as e:
                 print(f"Page {i+1} ink/color calc failed: {e}")
                 ink_pct = None