⚡️ Speed up function cells_to_html by 8%

CHANGELOG.md

@@ -1,3 +1,7 @@
+## 1.0.8-dev1
+
+* Enhancement: Optimized cells_to_html (codeflash)
+
 ## 1.0.7
 
 * Fix a hardcoded file extension causing confusion in the logs

unstructured_inference/__version__.py

@@ -1 +1 @@
-__version__ = "1.0.7"  # pragma: no cover
+__version__ = "1.0.8-dev1"  # pragma: no cover

unstructured_inference/models/tables.py

@@ -688,25 +688,32 @@ def fill_cells(cells: List[dict]) -> List[dict]:
     if not cells:
         return []
 
-    table_rows_no = max({row for cell in cells for row in cell["row_nums"]})
-    table_cols_no = max({col for cell in cells for col in cell["column_nums"]})
-    filled = np.zeros((table_rows_no + 1, table_cols_no + 1), dtype=bool)
+    # Find max row and col indices
+    max_row = max(row for cell in cells for row in cell["row_nums"])
+    max_col = max(col for cell in cells for col in cell["column_nums"])
+    filled = set()
     for cell in cells:
         for row in cell["row_nums"]:
             for col in cell["column_nums"]:
-                filled[row, col] = True
-    # add cells for which filled is false
-    header_rows = {row for cell in cells if cell["column header"] for row in cell["row_nums"]}
+                filled.add((row, col))
+    header_rows = set()
+    for cell in cells:
+        if cell["column header"]:
+            header_rows.update(cell["row_nums"])
+
+    # Compose output list directly for speed
     new_cells = cells.copy()
-    not_filled_idx = np.where(filled == False)  # noqa: E712
-    for row, col in zip(not_filled_idx[0], not_filled_idx[1]):
-        new_cell = {
-            "row_nums": [row],
-            "column_nums": [col],
-            "cell text": "",
-            "column header": row in header_rows,
-        }
-        new_cells.append(new_cell)
+    for row in range(max_row + 1):
+        for col in range(max_col + 1):
+            if (row, col) not in filled:
+                new_cells.append(
+                    {
+                        "row_nums": [row],
+                        "column_nums": [col],
+                        "cell text": "",
+                        "column header": row in header_rows,
+                    }
+                )
     return new_cells
 
 
@@ -725,18 +732,20 @@ def cells_to_html(cells: List[dict]) -> str:
     Returns:
         str: HTML table string
     """
-    cells = sorted(fill_cells(cells), key=lambda k: (min(k["row_nums"]), min(k["column_nums"])))
+    # Pre-sort with tuple key, as per original
+    cells_filled = fill_cells(cells)
+    cells_sorted = sorted(cells_filled, key=lambda k: (min(k["row_nums"]), min(k["column_nums"])))
 
     table = ET.Element("table")
     current_row = -1
 
-    table_header = None
-    table_has_header = any(cell["column header"] for cell in cells)
-    if table_has_header:
-        table_header = ET.SubElement(table, "thead")
-
+    # Check if any column header exists
+    table_has_header = any(cell["column header"] for cell in cells_sorted)
+    table_header = ET.SubElement(table, "thead") if table_has_header else None
     table_body = ET.SubElement(table, "tbody")
-    for cell in cells:
+
+    row = None
+    for cell in cells_sorted:
         this_row = min(cell["row_nums"])
         attrib = {}
         colspan = len(cell["column_nums"])