Drop a redundant final 'l' operator to recognize rectangles

CHANGELOG.md

@@ -8,6 +8,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 ### Changed
 
 - Using absolute instead of relative imports ([[#995](https://github.com/pdfminer/pdfminer.six/pull/995)])
+- Remove redundant line to be able to recognize rectangles ([#1066](https://github.com/pdfminer/pdfminer.six/pull/1066))
 
 ### Deprecated
 

pdfminer/converter.py

@@ -162,6 +162,11 @@ def paint_path(
                 for o, p in zip(operators, transformed_points)
             ]
 
+            # Drop a redundant "l" on a path closed with "h"
+            if len(shape) > 3 and shape[-2:] == "lh" and pts[-2] == pts[0]:
+                shape = shape[:-2] + "h"
+                pts.pop()
+
             if shape in {"mlh", "ml"}:
                 # single line segment
                 #

tests/test_converter.py

@@ -81,6 +81,18 @@ def get_types(path):
             ],
         ) == [LTRect]
 
+        # Same but mllllh variation
+        assert get_types(
+            [
+                ("m", 10, 90),
+                ("l", 90, 90),
+                ("l", 90, 10),
+                ("l", 10, 10),
+                ("l", 10, 90),
+                ("h",),
+            ],
+        ) == [LTRect]
+
         # Bowtie shape
         assert get_types(
             [