Merge pull request #497 from xxyzz/fr

Process forms table expanded from "fr-verbe-flexion" template
tatuylonen · Feb 8, 2024 · 0264748 · 0264748
2 parents 5c4c72d + fdaccad
commit 0264748
Show file tree

Hide file tree

Showing 2 changed files with 56 additions and 2 deletions.
diff --git a/src/wiktextract/extractor/fr/inflection.py b/src/wiktextract/extractor/fr/inflection.py
@@ -49,6 +49,21 @@ class ColspanHeader:
     span: int
 
 
+def table_data_cell_is_header(
+    wxr: WiktextractContext, cell_node: WikiNode, page_title: str
+) -> bool:
+    # first child is bold node
+    if cell_node.kind == NodeKind.TABLE_CELL:
+        for child in cell_node.filter_empty_str_child():
+            return (
+                isinstance(child, WikiNode)
+                and child.kind == NodeKind.BOLD
+                and clean_node(wxr, None, child) != page_title
+            )
+
+    return False
+
+
 def process_inflection_table(
     wxr: WiktextractContext,
     page_data: list[WordEntry],
@@ -83,7 +98,9 @@ def process_inflection_table(
             and "invisible" not in row_node_child.attrs.get("class", "")
         ]
         current_row_has_data_cell = any(
-            isinstance(cell, WikiNode) and cell.kind == NodeKind.TABLE_CELL
+            isinstance(cell, WikiNode)
+            and cell.kind == NodeKind.TABLE_CELL
+            and not table_data_cell_is_header(wxr, cell, page_data[-1].word)
             for cell in table_row_nodes
         )
         row_headers = []
@@ -98,7 +115,12 @@ def process_inflection_table(
         for column_num, table_cell in enumerate(table_row_nodes):
             form_data = Form()
             if isinstance(table_cell, WikiNode):
-                if table_cell.kind == NodeKind.TABLE_HEADER_CELL:
+                if (
+                    table_cell.kind == NodeKind.TABLE_HEADER_CELL
+                    or table_data_cell_is_header(
+                        wxr, table_cell, page_data[-1].word
+                    )
+                ):
                     if any(
                         table_cell.find_html(
                             "span",

diff --git a/tests/test_fr_inflection.py b/tests/test_fr_inflection.py
@@ -466,3 +466,35 @@ def test_en_adj(self):
                 },
             ],
         )
+
+    def test_fr_verbe_flexion(self):
+        page_data = [WordEntry(word="dièse", lang_code="fr", lang="Français")]
+        self.wxr.wtp.start_page("dièse")
+        root = self.wxr.wtp.parse("{{fr-verbe-flexion|diéser}}")
+        self.wxr.wtp.add_page(
+            "Modèle:fr-verbe-flexion",
+            10,
+            """{|
+!colspan="3"|Voir la conjugaison du verbe ''diéser''
+|-
+|rowspan="2" | '''Indicatif'''
+|rowspan="2" | '''Présent'''
+| je <nowiki />dièse
+|-
+| il/elle/on dièse
+|}""",
+        )
+        extract_inflection(self.wxr, page_data, root.children[0])
+        self.assertEqual(
+            [d.model_dump(exclude_defaults=True) for d in page_data[-1].forms],
+            [
+                {
+                    "form": "je dièse",
+                    "tags": ["Indicatif", "Présent"],
+                },
+                {
+                    "form": "il/elle/on dièse",
+                    "tags": ["Indicatif", "Présent"],
+                },
+            ],
+        )