Update mdr_pdf_parser.py

mdr_pdf_parser.py

@@ -222,7 +222,7 @@ class MDRTableLayoutParsedFormat(Enum):
     HTML = auto()
 
 
-@dataclass
+@dataclass(eq=False)
 class MDRBaseLayoutElement:
     """Base class for layout elements found on a page."""
     rect: MDRRectangle;
@@ -1852,7 +1852,7 @@ class MDROcrEngine:
                 # much lower thresholds so we actually get some candidate masks:
                 det_db_thresh=0.1,
                 det_db_box_thresh=0.3,
-                drop_score=0.1,
+                drop_score=0.0,
                 use_angle_cls=False,
             )
             try:
@@ -1906,7 +1906,7 @@ class MDROcrEngine:
             try:
                 pts = [(float(p[0]), float(p[1])) for p in box_pts]
                 if len(pts) == 4:
-                    r = MDRRectangle(lt=pts[0], rt=pts[1], rb=pts[2], lb=pts[3])
+                    r = MDRRectangle(lt=pts[0], rt=pts[1], lb=pts[2], rb=pts[3])
                     if r.is_valid and r.area > 1:
                         yield MDROcrFragment(order=-1, text=txt, rank=float(conf), rect=r)
                         fragments_generated_count += 1