Merge pull request #410 from xxyzz/table_row

Ignore `|-` token before first row and not at line beginning

Author: kristian-clausal

src/wikitextprocessor/parser.py

@@ -1571,8 +1571,15 @@ def table_row_fn(ctx: "Wtp", token: str) -> None:
         return text_fn(ctx, token)
 
     if not (ctx.beginning_of_line or ctx.wsp_beginning_of_line):
-        vbar_fn(ctx, "|")
-        return text_fn(ctx, "-")
+        node = ctx.parser_stack[-1]
+        # ignore "|-" token before first row and not at line beginning
+        if node.kind == NodeKind.TABLE and not node.contain_node(
+            NodeKind.TABLE_ROW
+        ):
+            return
+        else:
+            vbar_fn(ctx, "|")
+            return text_fn(ctx, "-")
 
     close_begline_lists(ctx)
     table_check_attrs(ctx)

tests/test_parser.py

@@ -3394,6 +3394,22 @@ def test_newline_in_ref(self):
             url_node.largs, [["http://purl.uni-rostock.de/demel/d00649426"]]
         )
 
+    def test_incorrt_row_separator(self):
+        # tatuylonen/wiktextract#1509
+        self.ctx.start_page("αζωικός")
+        root = self.ctx.parse("""{| class="wikitable"|-
+! Header A
+|-
+| row 1 A
+|}""")
+        table = root.children[0]
+        first_row = table.children[0]
+        self.assertEqual(first_row.kind, NodeKind.TABLE_ROW)
+        self.assertEqual(len(first_row.children), 1)
+        header = first_row.children[0]
+        self.assertEqual(header.kind, NodeKind.TABLE_HEADER_CELL)
+        self.assertEqual(header.children, [" Header A\n"])
+
 
 # XXX implement <nowiki/> marking for links, templates
 #  - https://en.wikipedia.org/wiki/Help:Wikitext#Nowiki