Browse Source

Make showtag=True implicit.

tags/v0.3
Ben Kurtovic 11 years ago
parent
commit
5018d9d323
5 changed files with 48 additions and 47 deletions
  1. +3
    -2
      mwparserfromhell/parser/builder.py
  2. +1
    -1
      mwparserfromhell/parser/tokenizer.py
  3. +1
    -1
      tests/test_builder.py
  4. +1
    -1
      tests/tokenizer/integration.mwtest
  5. +42
    -42
      tests/tokenizer/tags.mwtest

+ 3
- 2
mwparserfromhell/parser/builder.py View File

@@ -205,9 +205,10 @@ class Builder(object):

def _handle_tag(self, token):
"""Handle a case where a tag is at the head of the tokens."""
showtag, invalid = token.showtag, token.get("invalid", False)
implicit, attrs, contents, closing_tag = False, [], None, None
close_tokens = (tokens.TagCloseSelfclose, tokens.TagCloseClose)
implicit, attrs, contents, closing_tag = False, [], None, None
showtag = token.get("showtag", True)
invalid = token.get("invalid", False)
self._push()
while self._tokens:
token = self._tokens.pop()


+ 1
- 1
mwparserfromhell/parser/tokenizer.py View File

@@ -441,7 +441,7 @@ class Tokenizer(object):
"""Actually parse an HTML tag, starting with the open (``<foo>``)."""
data = _TagOpenData()
self._push(contexts.TAG_OPEN)
self._emit(tokens.TagOpenOpen(showtag=True))
self._emit(tokens.TagOpenOpen())
while True:
this, next = self._read(), self._read(1)
can_exit = (not data.context & (data.CX_QUOTED | data.CX_NAME) or


+ 1
- 1
tests/test_builder.py View File

@@ -193,7 +193,7 @@ class TestBuilder(TreeEqualityTestCase):
def test_tag(self):
"""tests for building Tag nodes"""
tests = [
([tokens.TagOpenOpen(showtag=True), tokens.Text(text="ref"),
([tokens.TagOpenOpen(), tokens.Text(text="ref"),
tokens.TagCloseOpen(padding=""), tokens.TagOpenClose(),
tokens.Text(text="ref"), tokens.TagCloseClose()],
wrap([Tag(wraptext("ref"), wrap([]), [], True, False, "",


+ 1
- 1
tests/tokenizer/integration.mwtest View File

@@ -36,7 +36,7 @@ output: [Text(text="&n"), CommentStart(), Text(text="foo"), CommentEnd(), Text(t
name: rich_tags
label: a HTML tag with tons of other things in it
input: "{{dubious claim}}<ref name={{abc}} foo="bar {{baz}}" abc={{de}}f ghi=j{{k}}{{l}} \n mno = "{{p}} [[q]] {{r}}">[[Source]]</ref>"
output: [TemplateOpen(), Text(text="dubious claim"), TemplateClose(), TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TemplateOpen(), Text(text="abc"), TemplateClose(), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="foo"), TagAttrEquals(), TagAttrQuote(), Text(text="bar "), TemplateOpen(), Text(text="baz"), TemplateClose(), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="abc"), TagAttrEquals(), TemplateOpen(), Text(text="de"), TemplateClose(), Text(text="f"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="ghi"), TagAttrEquals(), Text(text="j"), TemplateOpen(), Text(text="k"), TemplateClose(), TemplateOpen(), Text(text="l"), TemplateClose(), TagAttrStart(pad_first=" \n ", pad_before_eq=" ", pad_after_eq=" "), Text(text="mno"), TagAttrEquals(), TagAttrQuote(), TemplateOpen(), Text(text="p"), TemplateClose(), Text(text=" "), WikilinkOpen(), Text(text="q"), WikilinkClose(), Text(text=" "), TemplateOpen(), Text(text="r"), TemplateClose(), TagCloseOpen(padding=""), WikilinkOpen(), Text(text="Source"), WikilinkClose(), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TemplateOpen(), Text(text="dubious claim"), TemplateClose(), TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TemplateOpen(), Text(text="abc"), TemplateClose(), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="foo"), TagAttrEquals(), TagAttrQuote(), Text(text="bar "), TemplateOpen(), Text(text="baz"), TemplateClose(), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="abc"), TagAttrEquals(), TemplateOpen(), Text(text="de"), TemplateClose(), Text(text="f"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="ghi"), TagAttrEquals(), Text(text="j"), TemplateOpen(), Text(text="k"), TemplateClose(), TemplateOpen(), Text(text="l"), TemplateClose(), TagAttrStart(pad_first=" \n ", pad_before_eq=" ", pad_after_eq=" "), Text(text="mno"), TagAttrEquals(), TagAttrQuote(), TemplateOpen(), Text(text="p"), TemplateClose(), Text(text=" "), WikilinkOpen(), Text(text="q"), WikilinkClose(), Text(text=" "), TemplateOpen(), Text(text="r"), TemplateClose(), TagCloseOpen(padding=""), WikilinkOpen(), Text(text="Source"), WikilinkClose(), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---



+ 42
- 42
tests/tokenizer/tags.mwtest View File

@@ -1,112 +1,112 @@
name: basic
label: a basic tag with an open and close
input: "<ref></ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: basic_selfclosing
label: a basic self-closing tag
input: "<ref/>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagCloseSelfclose(padding="")]
output: [TagOpenOpen(), Text(text="ref"), TagCloseSelfclose(padding="")]

---

name: content
label: a tag with some content in the middle
input: "<ref>this is a reference</ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagCloseOpen(padding=""), Text(text="this is a reference"), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagCloseOpen(padding=""), Text(text="this is a reference"), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: padded_open
label: a tag with some padding in the open tag
input: "<ref ></ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagCloseOpen(padding=" "), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagCloseOpen(padding=" "), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: padded_close
label: a tag with some padding in the close tag
input: "<ref></ref >"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref "), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref "), TagCloseClose()]

---

name: padded_selfclosing
label: a self-closing tag with padding
input: "<ref />"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagCloseSelfclose(padding=" ")]
output: [TagOpenOpen(), Text(text="ref"), TagCloseSelfclose(padding=" ")]

---

name: attribute
label: a tag with a single attribute
input: "<ref name></ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: attribute_value
label: a tag with a single attribute with a value
input: "<ref name=foo></ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), Text(text="foo"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), Text(text="foo"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: attribute_quoted
label: a tag with a single quoted attribute
input: "<ref name="foo bar"></ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), Text(text="foo bar"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), Text(text="foo bar"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: attribute_hyphen
label: a tag with a single attribute, containing a hyphen
input: "<ref name=foo-bar></ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), Text(text="foo-bar"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), Text(text="foo-bar"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: attribute_quoted_hyphen
label: a tag with a single quoted attribute, containing a hyphen
input: "<ref name="foo-bar"></ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), Text(text="foo-bar"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), Text(text="foo-bar"), TagCloseOpen(padding=""), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: attribute_selfclosing
label: a self-closing tag with a single attribute
input: "<ref name/>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagCloseSelfclose(padding="")]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagCloseSelfclose(padding="")]

---

name: attribute_selfclosing_value
label: a self-closing tag with a single attribute with a value
input: "<ref name=foo/>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), Text(text="foo"), TagCloseSelfclose(padding="")]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), Text(text="foo"), TagCloseSelfclose(padding="")]

---

name: attribute_selfclosing_value_quoted
label: a self-closing tag with a single quoted attribute
input: "<ref name="foo"/>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), Text(text="foo"), TagCloseSelfclose(padding="")]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), Text(text="foo"), TagCloseSelfclose(padding="")]

---

name: nested_tag
label: a tag nested within the attributes of another
input: "<ref name=<span style="color: red;">foo</span>>citation</ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagOpenOpen(showtag=True), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), TagAttrQuote(), Text(text="color: red;"), TagCloseOpen(padding=""), Text(text="foo"), TagOpenClose(), Text(text="span"), TagCloseClose(), TagCloseOpen(padding=""), Text(text="citation"), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagOpenOpen(), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), TagAttrQuote(), Text(text="color: red;"), TagCloseOpen(padding=""), Text(text="foo"), TagOpenClose(), Text(text="span"), TagCloseClose(), TagCloseOpen(padding=""), Text(text="citation"), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: nested_tag_quoted
label: a tag nested within the attributes of another, quoted
input: "<ref name="<span style="color: red;">foo</span>">citation</ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), TagOpenOpen(showtag=True), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), TagAttrQuote(), Text(text="color: red;"), TagCloseOpen(padding=""), Text(text="foo"), TagOpenClose(), Text(text="span"), TagCloseClose(), TagCloseOpen(padding=""), Text(text="citation"), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), TagOpenOpen(), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), TagAttrQuote(), Text(text="color: red;"), TagCloseOpen(padding=""), Text(text="foo"), TagOpenClose(), Text(text="span"), TagCloseClose(), TagCloseOpen(padding=""), Text(text="citation"), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

@@ -120,7 +120,7 @@ output: [Text(text="<ref name=</ ><//>>citation</ref>")]
name: nested_troll_tag_quoted
label: a bogus tag that appears to be nested within the attributes of another, quoted
input: "<ref name="</ ><//>">citation</ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), Text(text="</ ><//>"), TagCloseOpen(padding=""), Text(text="citation"), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="name"), TagAttrEquals(), TagAttrQuote(), Text(text="</ ><//>"), TagCloseOpen(padding=""), Text(text="citation"), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

@@ -141,7 +141,7 @@ output: [Text(text="<ref>test</ ref>")]
name: valid_space_end
label: valid tag: spaces at the ends of both the open and close tags
input: "<ref >test</ref >"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagCloseOpen(padding=" "), Text(text="test"), TagOpenClose(), Text(text="ref "), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagCloseOpen(padding=" "), Text(text="test"), TagOpenClose(), Text(text="ref "), TagCloseClose()]

---

@@ -162,14 +162,14 @@ output: [Text(text="<ref "), TemplateOpen(), Text(text="foo"), TemplateClose(),
name: valid_template_end_open
label: valid tag: a template at the end of the open tag
input: "<ref {{foo}}>test</ref>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), TemplateOpen(), Text(text="foo"), TemplateClose(), TagCloseOpen(padding=""), Text(text="test"), TagOpenClose(), Text(text="ref"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), TemplateOpen(), Text(text="foo"), TemplateClose(), TagCloseOpen(padding=""), Text(text="test"), TagOpenClose(), Text(text="ref"), TagCloseClose()]

---

name: valid_template_end_open_space_end_close
label: valid tag: a template at the end of the open tag; whitespace at the end of the close tag
input: "<ref {{foo}}>test</ref\n>"
output: [TagOpenOpen(showtag=True), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), TemplateOpen(), Text(text="foo"), TemplateClose(), TagCloseOpen(padding=""), Text(text="test"), TagOpenClose(), Text(text="ref\n"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="ref"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), TemplateOpen(), Text(text="foo"), TemplateClose(), TagCloseOpen(padding=""), Text(text="test"), TagOpenClose(), Text(text="ref\n"), TagCloseClose()]

---

@@ -197,21 +197,21 @@ output: [Text(text="<"), TemplateOpen(), Text(text="foo"), TemplateClose(), Text
name: unclosed_quote
label: a quoted attribute that is never closed
input: "<span style="foobar>stuff</span>"
output: [TagOpenOpen(showtag=True), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), Text(text="\"foobar"), TagCloseOpen(padding=""), Text(text="stuff"), TagOpenClose(), Text(text="span"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), Text(text="\"foobar"), TagCloseOpen(padding=""), Text(text="stuff"), TagOpenClose(), Text(text="span"), TagCloseClose()]

---

name: fake_quote
label: a fake quoted attribute
input: "<span style="foo"bar>stuff</span>"
output: [TagOpenOpen(showtag=True), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), Text(text="\"foo\"bar"), TagCloseOpen(padding=""), Text(text="stuff"), TagOpenClose(), Text(text="span"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), Text(text="\"foo\"bar"), TagCloseOpen(padding=""), Text(text="stuff"), TagOpenClose(), Text(text="span"), TagCloseClose()]

---

name: fake_quote_complex
label: a fake quoted attribute, with spaces and templates and links
input: "<span style="foo {{bar}}\n[[baz]]"buzz >stuff</span>"
output: [TagOpenOpen(showtag=True), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), Text(text="\"foo"), TagAttrStart(pad_first=" ", pad_before_eq="\n", pad_after_eq=""), TemplateOpen(), Text(text="bar"), TemplateClose(), TagAttrStart(pad_first="", pad_before_eq=" ", pad_after_eq=""), WikilinkOpen(), Text(text="baz"), WikilinkClose(), Text(text="\"buzz"), TagCloseOpen(padding=""), Text(text="stuff"), TagOpenClose(), Text(text="span"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="span"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="style"), TagAttrEquals(), Text(text="\"foo"), TagAttrStart(pad_first=" ", pad_before_eq="\n", pad_after_eq=""), TemplateOpen(), Text(text="bar"), TemplateClose(), TagAttrStart(pad_first="", pad_before_eq=" ", pad_after_eq=""), WikilinkOpen(), Text(text="baz"), WikilinkClose(), Text(text="\"buzz"), TagCloseOpen(padding=""), Text(text="stuff"), TagOpenClose(), Text(text="span"), TagCloseClose()]

---

@@ -386,77 +386,77 @@ output: [Text(text="junk <></>")]
name: backslash_premature_before
label: a backslash before a quote before a space
input: "<foo attribute="this is\\" quoted">blah</foo>"
output: [TagOpenOpen(showtag=True), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this is\\\" quoted"), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this is\\\" quoted"), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]

---

name: backslash_premature_after
label: a backslash before a quote after a space
input: "<foo attribute="this is \\"quoted">blah</foo>"
output: [TagOpenOpen(showtag=True), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this is \\\"quoted"), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this is \\\"quoted"), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]

---

name: backslash_premature_middle
label: a backslash before a quote in the middle of a word
input: "<foo attribute="this i\\"s quoted">blah</foo>"
output: [TagOpenOpen(showtag=True), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this i\\\"s quoted"), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this i\\\"s quoted"), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]

---

name: backslash_adjacent
label: escaped quotes next to unescaped quotes
input: "<foo attribute="\\"this is quoted\\"">blah</foo>"
output: [TagOpenOpen(showtag=True), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="\\\"this is quoted\\\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="\\\"this is quoted\\\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]

---

name: backslash_endquote
label: backslashes before the end quote, causing the attribute to become unquoted
input: "<foo attribute="this_is quoted\\">blah</foo>"
output: [TagOpenOpen(showtag=True), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), Text(text="\"this_is"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="quoted\\\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), Text(text="\"this_is"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="quoted\\\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]

---

name: backslash_double
label: two adjacent backslashes, which do *not* affect the quote
input: "<foo attribute="this is\\\\" quoted">blah</foo>"
output: [TagOpenOpen(showtag=True), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this is\\\\"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="quoted\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this is\\\\"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="quoted\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]

---

name: backslash_triple
label: three adjacent backslashes, which do *not* affect the quote
input: "<foo attribute="this is\\\\\\" quoted">blah</foo>"
output: [TagOpenOpen(showtag=True), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this is\\\\\\"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="quoted\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="this is\\\\\\"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="quoted\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]

---

name: backslash_unaffecting
label: backslashes near quotes, but not immediately adjacent, thus having no effect
input: "<foo attribute="\\quote\\d" also="quote\\d\\">blah</foo>"
output: [TagOpenOpen(showtag=True), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="\\quote\\d"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="also"), TagAttrEquals(), Text(text="\"quote\\d\\\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]
output: [TagOpenOpen(), Text(text="foo"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attribute"), TagAttrEquals(), TagAttrQuote(), Text(text="\\quote\\d"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="also"), TagAttrEquals(), Text(text="\"quote\\d\\\""), TagCloseOpen(padding=""), Text(text="blah"), TagOpenClose(), Text(text="foo"), TagCloseClose()]

---

name: unparsable
label: a tag that should not be put through the normal parser
input: "{{t1}}<nowiki>{{t2}}</nowiki>{{t3}}"
output: [TemplateOpen(), Text(text="t1"), TemplateClose(), TagOpenOpen(showtag=True), Text(text="nowiki"), TagCloseOpen(padding=""), Text(text="{{t2}}"), TagOpenClose(), Text(text="nowiki"), TagCloseClose(), TemplateOpen(), Text(text="t3"), TemplateClose()]
output: [TemplateOpen(), Text(text="t1"), TemplateClose(), TagOpenOpen(), Text(text="nowiki"), TagCloseOpen(padding=""), Text(text="{{t2}}"), TagOpenClose(), Text(text="nowiki"), TagCloseClose(), TemplateOpen(), Text(text="t3"), TemplateClose()]

---

name: unparsable_complex
label: a tag that should not be put through the normal parser; lots of stuff inside
input: "{{t1}}<pre>{{t2}}\n==Heading==\nThis is some text with a [[page|link]].</pre>{{t3}}"
output: [TemplateOpen(), Text(text="t1"), TemplateClose(), TagOpenOpen(showtag=True), Text(text="pre"), TagCloseOpen(padding=""), Text(text="{{t2}}\n==Heading==\nThis is some text with a [[page|link]]."), TagOpenClose(), Text(text="pre"), TagCloseClose(), TemplateOpen(), Text(text="t3"), TemplateClose()]
output: [TemplateOpen(), Text(text="t1"), TemplateClose(), TagOpenOpen(), Text(text="pre"), TagCloseOpen(padding=""), Text(text="{{t2}}\n==Heading==\nThis is some text with a [[page|link]]."), TagOpenClose(), Text(text="pre"), TagCloseClose(), TemplateOpen(), Text(text="t3"), TemplateClose()]

---

name: unparsable_attributed
label: a tag that should not be put through the normal parser; parsed attributes
input: "{{t1}}<nowiki attr=val attr2="{{val2}}">{{t2}}</nowiki>{{t3}}"
output: [TemplateOpen(), Text(text=u't1'), TemplateClose(), TagOpenOpen(showtag=True), Text(text="nowiki"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attr"), TagAttrEquals(), Text(text="val"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attr2"), TagAttrEquals(), TagAttrQuote(), TemplateOpen(), Text(text="val2"), TemplateClose(), TagCloseOpen(padding=""), Text(text="{{t2}}"), TagOpenClose(), Text(text="nowiki"), TagCloseClose(), TemplateOpen(), Text(text="t3"), TemplateClose()]
output: [TemplateOpen(), Text(text=u't1'), TemplateClose(), TagOpenOpen(), Text(text="nowiki"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attr"), TagAttrEquals(), Text(text="val"), TagAttrStart(pad_first=" ", pad_before_eq="", pad_after_eq=""), Text(text="attr2"), TagAttrEquals(), TagAttrQuote(), TemplateOpen(), Text(text="val2"), TemplateClose(), TagCloseOpen(padding=""), Text(text="{{t2}}"), TagOpenClose(), Text(text="nowiki"), TagCloseClose(), TemplateOpen(), Text(text="t3"), TemplateClose()]

---

@@ -470,21 +470,21 @@ output: [TemplateOpen(), Text(text="t1"), TemplateClose(), Text(text="<nowiki>")
name: single_open_close
label: a tag that supports being single; both an open and a close tag
input: "foo<li>bar{{baz}}</li>"
output: [Text(text="foo"), TagOpenOpen(showtag=True), Text(text="li"), TagCloseOpen(padding=""), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose(), TagOpenClose(), Text(text="li"), TagCloseClose()]
output: [Text(text="foo"), TagOpenOpen(), Text(text="li"), TagCloseOpen(padding=""), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose(), TagOpenClose(), Text(text="li"), TagCloseClose()]

---

name: single_open
label: a tag that supports being single; just an open tag
input: "foo<li>bar{{baz}}"
output: [Text(text="foo"), TagOpenOpen(showtag=True), Text(text="li"), TagCloseSelfclose(padding="", implicit=True), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]
output: [Text(text="foo"), TagOpenOpen(), Text(text="li"), TagCloseSelfclose(padding="", implicit=True), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]

---

name: single_selfclose
label: a tag that supports being single; a self-closing tag
input: "foo<li/>bar{{baz}}"
output: [Text(text="foo"), TagOpenOpen(showtag=True), Text(text="li"), TagCloseSelfclose(padding=""), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]
output: [Text(text="foo"), TagOpenOpen(), Text(text="li"), TagCloseSelfclose(padding=""), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]

---

@@ -498,32 +498,32 @@ output: [Text(text="foo</li>bar"), TemplateOpen(), Text(text="baz"), TemplateClo
name: single_only_open_close
label: a tag that can only be single; both an open and a close tag
input: "foo<br>bar{{baz}}</br>"
output: [Text(text="foo"), TagOpenOpen(showtag=True), Text(text="br"), TagCloseSelfclose(padding="", implicit=True), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose(), TagOpenOpen(showtag=True, invalid=True), Text(text="br"), TagCloseSelfclose(padding="", implicit=True)]
output: [Text(text="foo"), TagOpenOpen(), Text(text="br"), TagCloseSelfclose(padding="", implicit=True), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose(), TagOpenOpen(invalid=True), Text(text="br"), TagCloseSelfclose(padding="", implicit=True)]

---

name: single_only_open
label: a tag that can only be single; just an open tag
input: "foo<br>bar{{baz}}"
output: [Text(text="foo"), TagOpenOpen(showtag=True), Text(text="br"), TagCloseSelfclose(padding="", implicit=True), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]
output: [Text(text="foo"), TagOpenOpen(), Text(text="br"), TagCloseSelfclose(padding="", implicit=True), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]

---

name: single_only_selfclose
label: a tag that can only be single; a self-closing tag
input: "foo<br/>bar{{baz}}"
output: [Text(text="foo"), TagOpenOpen(showtag=True), Text(text="br"), TagCloseSelfclose(padding=""), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]
output: [Text(text="foo"), TagOpenOpen(), Text(text="br"), TagCloseSelfclose(padding=""), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]

---

name: single_only_close
label: a tag that can only be single; just a close tag
input: "foo</br>bar{{baz}}"
output: [Text(text="foo"), TagOpenOpen(showtag=True, invalid=True), Text(text="br"), TagCloseSelfclose(padding="", implicit=True), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]
output: [Text(text="foo"), TagOpenOpen(invalid=True), Text(text="br"), TagCloseSelfclose(padding="", implicit=True), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]

---

name: single_only_double
label: a tag that can only be single; a tag with backslashes at the beginning and end
input: "foo</br/>bar{{baz}}"
output: [Text(text="foo"), TagOpenOpen(showtag=True, invalid=True), Text(text="br"), TagCloseSelfclose(padding=""), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]
output: [Text(text="foo"), TagOpenOpen(invalid=True), Text(text="br"), TagCloseSelfclose(padding=""), Text(text="bar"), TemplateOpen(), Text(text="baz"), TemplateClose()]

Loading…
Cancel
Save