markup: move functionality to library

markup/matching.py

@@ -1,7 +1,25 @@
 from markup.img import find_shapes, write_debug_image
+from markup.pdf import parse_pdf
 from markup.utils import overlaps
 
 
+def find_marked_products(pdf, subdir, catname, debug=0):
+    """Main entry point.  Give a pdf, get matches."""
+    (prods, scribbles) = parse_pdf(pdf, subdir, catname, debug)
+
+    if not prods or len(prods) < 1:
+        print('no product placement markers found')
+        return None
+
+    find_scribbles_shapes(scribbles)
+    matches = find_matches(prods, scribbles, 0.10)
+
+    for s in scribbles:
+        write_debug_image(subdir, catname, s['page'], prods, scribbles)
+
+    return matches
+
+
 def find_scribbles_shapes(scribbles):
     for scribble in scribbles:
         imgw, imgh, shapes = find_shapes(scribble['image'])
@@ -33,6 +51,7 @@ def find_matches(all_prods, scribbles, overlap_threshold):
     matches = []
     for s in scribbles:
         pagenum = s['page']
+        if not pagenum in page_prods: continue
         prods = page_prods[pagenum]
         for p in prods:
             for box in s['bboxes']: