diff --git a/mwparserfromhell/nodes/tag.py b/mwparserfromhell/nodes/tag.py index 7cbe78d..0fe580f 100644 --- a/mwparserfromhell/nodes/tag.py +++ b/mwparserfromhell/nodes/tag.py @@ -35,7 +35,7 @@ class Tag(Node): def __init__(self, tag, contents=None, attrs=None, wiki_markup=None, self_closing=False, invalid=False, implicit=False, padding="", - closing_tag=None): + closing_tag=None, closing_wiki_markup=None): super(Tag, self).__init__() self._tag = tag if contents is None and not self_closing: @@ -44,6 +44,13 @@ class Tag(Node): self._contents = contents self._attrs = attrs if attrs else [] self._wiki_markup = wiki_markup + if wiki_markup and not self_closing: + if closing_wiki_markup: + self._closing_wiki_markup = closing_wiki_markup + else: + self._closing_wiki_markup = wiki_markup + else: + self._closing_wiki_markup = None self._self_closing = self_closing self._invalid = invalid self._implicit = implicit @@ -55,10 +62,11 @@ class Tag(Node): def __unicode__(self): if self.wiki_markup: + attrs = "".join([str(attr) for attr in self.attributes]) if self.attributes else "" if self.self_closing: return self.wiki_markup else: - return self.wiki_markup + str(self.contents) + self.wiki_markup + return self.wiki_markup + attrs + str(self.contents) + self.closing_wiki_markup result = ("``).""" return self._self_closing @@ -185,10 +206,19 @@ class Tag(Node): @wiki_markup.setter def wiki_markup(self, value): self._wiki_markup = str(value) if value else None + if not value or not self.closing_wiki_markup: + self.closing_wiki_markup = str(value) if value else None + + + @closing_wiki_markup.setter + def closing_wiki_markup(self, value): + self._closing_wiki_markup = str(value) if value and not self.self_closing else None @self_closing.setter def self_closing(self, value): self._self_closing = bool(value) + if not bool(value): + self.closing_wiki_markup = None @invalid.setter def invalid(self, value): diff --git a/mwparserfromhell/parser/builder.py b/mwparserfromhell/parser/builder.py index 2d68036..8d1852e 100644 --- a/mwparserfromhell/parser/builder.py +++ b/mwparserfromhell/parser/builder.py @@ -248,6 +248,7 @@ class Builder(object): close_tokens = (tokens.TagCloseSelfclose, tokens.TagCloseClose) implicit, attrs, contents, closing_tag = False, [], None, None wiki_markup, invalid = token.wiki_markup, token.invalid or False + closing_wiki_markup = None self._push() while self._tokens: token = self._tokens.pop() @@ -258,6 +259,7 @@ class Builder(object): tag = self._pop() self._push() elif isinstance(token, tokens.TagOpenClose): + closing_wiki_markup = token.wiki_markup contents = self._pop() self._push() elif isinstance(token, close_tokens): @@ -270,7 +272,7 @@ class Builder(object): self_closing = False closing_tag = self._pop() return Tag(tag, contents, attrs, wiki_markup, self_closing, - invalid, implicit, padding, closing_tag) + invalid, implicit, padding, closing_tag, closing_wiki_markup) else: self._write(self._handle_token(token)) raise ParserError("_handle_tag() missed a close token") diff --git a/tests/test_tag.py b/tests/test_tag.py index 7577cce..950233f 100644 --- a/tests/test_tag.py +++ b/tests/test_tag.py @@ -171,6 +171,24 @@ class TestTag(TreeEqualityTestCase): self.assertFalse(node.wiki_markup) self.assertEqual("italic text", node) + def test_closing_wiki_markup(self): + """test getter/setter behavior for closing_wiki_markup attribute""" + node = Tag(wraptext("table"), wraptext("\n")) + self.assertIs(None, node.closing_wiki_markup) + node.wiki_markup = "{|" + self.assertEqual("{|", node.closing_wiki_markup) + node.closing_wiki_markup = "|}" + self.assertEqual("|}", node.closing_wiki_markup) + self.assertEqual("{|\n|}", node) + node.wiki_markup = False + self.assertFalse(node.closing_wiki_markup) + node.self_closing = True + node.wiki_markup = "{|" + self.assertIs(None, node.closing_wiki_markup) + node.wiki_markup = False + node.self_closing = False + self.assertEqual("\n
", node) + def test_self_closing(self): """test getter/setter for the self_closing attribute""" node = Tag(wraptext("ref"), wraptext("foobar")) diff --git a/tests/tokenizer/tables.mwtest b/tests/tokenizer/tables.mwtest index fa068fd..bfdd83f 100644 --- a/tests/tokenizer/tables.mwtest +++ b/tests/tokenizer/tables.mwtest @@ -1,14 +1,14 @@ name: empty_table label: Parsing an empty table. input: "{|\n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: inline_table label: Correctly handle tables with close on the same line. input: "{||}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- @@ -29,7 +29,7 @@ output: [Text(text="{| | ")] name: leading_whitespace_table label: Handle leading whitespace for a table. input: "foo \n \t {|\n|}" -output: [Text(text="foo \n \t "), TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [Text(text="foo \n \t "), TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- @@ -43,112 +43,119 @@ output: [Text(text="foo \n foo \t {|\n|}")] name: table_row_simple label: Simple table row. input: "{|\n |- \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|-"), Text(text="tr"), TagCloseSelfclose(), Text(text=" \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|-"), Text(text="tr"), TagCloseSelfclose(), Text(text=" \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_cell_simple label: Simple table cell. input: "{|\n | foo \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), Text(text=" foo \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), Text(text=" foo \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_cell_inline label: Multiple inline table cells. input: "{|\n | foo || bar || test \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), Text(text=" foo "), TagOpenOpen(wiki_markup="||"), Text(text="td"), TagCloseSelfclose(), Text(text=" bar "),TagOpenOpen(wiki_markup="||"), Text(text="td"), TagCloseSelfclose(), Text(text=" test \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), Text(text=" foo "), TagOpenOpen(wiki_markup="||"), Text(text="td"), TagCloseSelfclose(), Text(text=" bar "),TagOpenOpen(wiki_markup="||"), Text(text="td"), TagCloseSelfclose(), Text(text=" test \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_header_simple label: Simple header cell. input: "{|\n ! foo \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="!"), Text(text="th"), TagCloseSelfclose(), Text(text=" foo \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="!"), Text(text="th"), TagCloseSelfclose(), Text(text=" foo \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_header_inline label: Multiple inline header cells. input: "{|\n ! foo || bar !! test \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="!"), Text(text="th"), TagCloseSelfclose(), Text(text=" foo "), TagOpenOpen(wiki_markup="||"), Text(text="th"), TagCloseSelfclose(), Text(text=" bar "),TagOpenOpen(wiki_markup="!!"), Text(text="th"), TagCloseSelfclose(), Text(text=" test \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="!"), Text(text="th"), TagCloseSelfclose(), Text(text=" foo "), TagOpenOpen(wiki_markup="||"), Text(text="th"), TagCloseSelfclose(), Text(text=" bar "),TagOpenOpen(wiki_markup="!!"), Text(text="th"), TagCloseSelfclose(), Text(text=" test \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: nowiki_inside_table label: Nowiki handles pipe characters in tables. input: "{|\n | foo | |- {| |} || ! !! bar \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), Text(text=" foo "), TagOpenOpen(), Text(text="nowiki"), TagCloseOpen(padding=""), Text(text="| |- {| |} || ! !!"), TagOpenClose(), Text(text="nowiki"), TagCloseClose(), Text(text=" bar \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), Text(text=" foo "), TagOpenOpen(), Text(text="nowiki"), TagCloseOpen(padding=""), Text(text="| |- {| |} || ! !!"), TagOpenClose(), Text(text="nowiki"), TagCloseClose(), Text(text=" bar \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_text_outside_cell label: Parse text inside table but outside of a cell. input: "{|\n bar \n | foo \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n bar \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), Text(text=" foo \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n bar \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), Text(text=" foo \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: no_table_cell_with_leading_characters label: Fail to create a table cell when there are leading non-whitespace characters. input: "{|\n bar | foo \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n bar | foo \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n bar | foo \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: no_table_row_with_leading_characters label: Fail to create a table row when there are leading non-whitespace characters. input: "{|\n bar |- foo \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n bar |- foo \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n bar |- foo \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: template_inside_table_cell label: Template within table cell. input: "{|\n |{{foo\n|bar=baz}} \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), TemplateOpen(), Text(text="foo\n"), TemplateParamSeparator(), Text(text="bar"), TemplateParamEquals(), Text(text="baz"), TemplateClose(), Text(text=" \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text="\n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagCloseSelfclose(), TemplateOpen(), Text(text="foo\n"), TemplateParamSeparator(), Text(text="bar"), TemplateParamEquals(), Text(text="baz"), TemplateClose(), Text(text=" \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_cell_attributes label: Parse table cell style attributes. input: "{| \n | name="foo bar"| test \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo bar"), TagCloseSelfclose(), Text(text=" test \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo bar"), TagCloseSelfclose(), Text(text=" test \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_cell_attributes_quote_with_pipe label: Pipe inside an attribute quote should still be used as a style separator. input: "{| \n | name="foo|bar"| test \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo"), TagCloseSelfclose(), Text(text="bar\"| test \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo"), TagCloseSelfclose(), Text(text="bar\"| test \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_cell_attributes_name_with_pipe label: Pipe inside an attribute name should still be used as a style separator. input: "{| \n | name|="foo bar"| test \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagCloseSelfclose(), Text(text="=\"foo bar\"| test \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagCloseSelfclose(), Text(text="=\"foo bar\"| test \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_cell_attributes_pipe_after_equals label: Pipe inside an attribute should still be used as a style separator after an equals. input: "{| \n | name=|"foo|bar"| test \n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagCloseSelfclose(), Text(text="\"foo|bar\"| test \n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagCloseSelfclose(), Text(text="\"foo|bar\"| test \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] + +--- + +name: table_cell_attributes_templates +label: Pipe inside attributes shouldn't be style separator. +input: "{| \n | {{comment|template=baz}} | test \n|}" +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|"), Text(text="td"), TagAttrStart(pad_after_eq="", pad_first=" ", pad_before_eq=" "), TemplateOpen(), Text(text="comment"), TemplateParamSeparator(), Text(text="template"), TemplateParamEquals(), Text(text="baz"), TemplateClose(), TagCloseSelfclose(), Text(text=" test \n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_row_attributes label: Parse table row style attributes. input: "{| \n |- name="foo bar"\n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|-"), Text(text="tr"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo bar"), TagCloseSelfclose(), Text(text="\n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \n "), TagOpenOpen(wiki_markup="|-"), Text(text="tr"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo bar"), TagCloseSelfclose(), Text(text="\n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- name: table_row_attributes_crazy_whitespace label: Parse table row style attributes with different whitespace. input: "{| \t \n |- \t name="foo bar"\n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \t \n "), TagOpenOpen(wiki_markup="|-"), Text(text="tr"), TagAttrStart(pad_first=" \t ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo bar"), TagCloseSelfclose(), Text(text="\n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text(text=" \t \n "), TagOpenOpen(wiki_markup="|-"), Text(text="tr"), TagAttrStart(pad_first=" \t ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo bar"), TagCloseSelfclose(), Text(text="\n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()] --- @@ -156,4 +163,4 @@ output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(), Text name: table_attributes label: Parse table style attributes. input: "{| name="foo bar"\n|}" -output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"),TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo bar"), TagCloseOpen(), Text(text="\n"), TagOpenClose(), Text(text="table"), TagCloseClose()] +output: [TagOpenOpen(wiki_markup="{|"), Text(text="table"),TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(char="\""), Text(text="foo bar"), TagCloseOpen(), Text(text="\n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose()]