Merge d3b8848 into 002500a

jsvine · Jul 11, 2021 · 1f296af · 1f296af
2 parents 002500a + d3b8848
commit 1f296af
Show file tree

Hide file tree

Showing 3 changed files with 45 additions and 10 deletions.
diff --git a/pdfplumber/table.py b/pdfplumber/table.py
@@ -84,7 +84,7 @@ def words_to_edges_h(words, word_threshold=DEFAULT_MIN_WORDS_HORIZONTAL):
         return []
     min_x0 = min(map(itemgetter("x0"), rects))
     max_x1 = max(map(itemgetter("x1"), rects))
-    max_bottom = max(map(itemgetter("bottom"), rects))
+
     edges = [
         {
             "x0": min_x0,
@@ -95,17 +95,23 @@ def words_to_edges_h(words, word_threshold=DEFAULT_MIN_WORDS_HORIZONTAL):
             "orientation": "h",
         }
         for r in rects
-    ] + [
-        {
-            "x0": min_x0,
-            "x1": max_x1,
-            "top": max_bottom,
-            "bottom": max_bottom,
-            "width": max_x1 - min_x0,
-            "orientation": "h",
-        }
     ]
 
+    # For each detected row, we also add the 'bottom' line.
+    # This will generate extra edges, (some will be redundant with the next row
+    # 'top' line), but this catches the last row of every table.
+    for r in rects:
+        edges.append(
+            {
+                "x0": min_x0,
+                "x1": max_x1,
+                "top": r["bottom"],
+                "bottom": r["bottom"],
+                "width": max_x1 - min_x0,
+                "orientation": "h",
+            }
+        )
+
     return edges
 
 

diff --git a/tests/pdfs/issue-466-example.pdf b/tests/pdfs/issue-466-example.pdf
diff --git a/tests/test_table.py b/tests/test_table.py
@@ -141,3 +141,32 @@ def test_order(self):
             assert len(tables[0]) == 8
             assert len(tables[1]) == 11
             assert len(tables[2]) == 2
+
+    def test_issue_466_mixed_strategy(self):
+        """
+        See issue #466
+        """
+        path = os.path.join(HERE, "pdfs/issue-466-example.pdf")
+        with pdfplumber.open(path) as pdf:
+            tables = pdf.pages[0].extract_tables(
+                {
+                    "vertical_strategy": "lines",
+                    "horizontal_strategy": "text",
+                    "snap_tolerance": 8,
+                    "intersection_tolerance": 4,
+                }
+            )
+
+            # The engine only extracts the tables which have drawn horizontal
+            # lines.
+            # For the 3 extracted tables, some common properties are expected:
+            # - 4 rows
+            # - 3 columns
+            # - Data in last row contains the string 'last'
+            for t in tables:
+                assert len(t) == 4
+                assert len(t[0]) == 3
+
+                # Verify that all cell contain real data
+                for cell in t[3]:
+                    assert "last" in cell