From edf52ea0767467f205e91aeae1664f718f9b0860 Mon Sep 17 00:00:00 2001 From: Arlo Breault Date: Tue, 29 Dec 2015 16:30:43 -0800 Subject: [PATCH] Sync up with Parsoid parserTests. This now aligns with Parsoid commit cc0dfb3271495d764e471d18927b0b3ee8ce1cdb Change-Id: I3fd6f8cc31a5536471a04d2f6a52de92893f9911 --- tests/parser/parserTests.txt | 530 ++++++++++++++++++++++++++++------- 1 file changed, 423 insertions(+), 107 deletions(-) diff --git a/tests/parser/parserTests.txt b/tests/parser/parserTests.txt index 0ed8270e50..d0a3d08b62 100644 --- a/tests/parser/parserTests.txt +++ b/tests/parser/parserTests.txt @@ -14,6 +14,7 @@ # Plus any combination of these: # # cat add category links +# (ignored by Parsoid, since it emits s) # ill add inter-language links # (ignored by Parsoid, since it emits s) # subpage enable subpages (disabled by default) @@ -1335,6 +1336,8 @@ array ( ) +!! html/parsoid +

 !! end
 
 !! test
@@ -3329,14 +3332,18 @@ parsoid=wt2html,wt2wt
   
 !! end
 
+## We used to, but no longer wt2wt this test since the default serializer
+## will normalize all categories to serialize on their own line.
+## This wikitext usage is going to be fairly uncommon in production and
+## selser will take care of preserving formatting in those scenarios.
 !! test
 7b. Indent-pre and category links
 !! options
-parsoid=wt2html,wt2wt
+parsoid=wt2html
 !! wikitext
  [[Category:foo]] a
  [[Category:foo]] {{echo|b}}
-!! html
+!! html/parsoid
 
 a
   b

 !! end
@@ -7119,6 +7126,17 @@ Piped link with multiple pipe characters in link text
 
|The|Main|Page|

 !! end
 
+!! test
+Piped link with no link text
+!! wikitext
+[[Thomas Bek (bishop of St David's)|]]
+!! html/php
+
[[Thomas Bek (bishop of St David's)|]]
+

+!! html/parsoid
+
[[Thomas Bek (bishop of St David's)|]]

+!! end
+
 !! test
 Broken link
 !! wikitext
@@ -10860,8 +10878,14 @@ Un-closed 
 !! html
 !! end
 
+## We used to, but no longer wt2wt this test since the default serializer
+## will normalize the include directives to serialize on their own line.
+## Selser will take care of preserving formatting in scenarios where they
+## intermingled with other wikitext.
 !! test
 Includes and comments at SOL
+!! options
+parsoid=wt2html,html2html
 !! wikitext
 == hu ==
 
@@ -11042,10 +11066,14 @@ parsoid=wt2html,wt2wt
 
 !!end
 
+## We used to, but no longer wt2wt this test since the default serializer
+## will normalize the include directives to serialize on their own line.
+## Selser will take care of preserving formatting in scenarios where they
+## intermingled with other wikitext.
 !!test
 2. Table tag in SOL posn. should get reparsed correctly with valid TSR
 !!options
-parsoid=wt2html,wt2wt
+parsoid=wt2html
 !!wikitext
 a{| {{{b}}}
 |c
@@ -14245,7 +14273,7 @@ cat
 pst
 !! wikitext
 [[Category:MediaWiki User's Guide|]]
-!! html
+!! html/php
 [[Category:MediaWiki User's Guide|MediaWiki User's Guide]]
 !! end
 
@@ -14256,19 +14284,26 @@ cat
 pst
 !! wikitext
 [[Category:Foo (bar)|]]
-!! html
+!! html/php
 [[Category:Foo (bar)|Foo]]
 !! end
 
+## We used to, but no longer wt2wt this test since the default serializer
+## will normalize all categories to serialize on their own line.
+## This wikitext usage is going to be fairly uncommon in production and
+## selser will take care of preserving formatting in those scenarios.
 !! test
 Category with link tail
 !! options
 cat
 pst
+parsoid=wt2html
 !! wikitext
 123[[Category:Foo]]456
-!! html
+!! html/php
 123[[Category:Foo]]456
+!! html/parsoid
+
123456

 !! end
 
 !! test
@@ -14278,7 +14313,7 @@ cat
 pst
 !! wikitext
 [[Category:{{echo|Foo}}]]
-!! html
+!! html/php
 [[Category:{{echo|Foo}}]]
 !! end
 
@@ -14289,7 +14324,7 @@ cat
 pst
 !! wikitext
 [[Category:Foo|{{echo|Bar}}]]
-!! html
+!! html/php
 [[Category:Foo|{{echo|Bar}}]]
 !! end
 
@@ -14300,12 +14335,18 @@ cat
 pst
 !! wikitext
 [[Category:{{echo|Foo}}|{{echo|Bar}}]]
-!! html
+!! html/php
 [[Category:{{echo|Foo}}|{{echo|Bar}}]]
 !! end
 
+## We used to, but no longer wt2wt this test since the default serializer
+## will normalize all categories to serialize on their own line.
+## This wikitext usage is going to be fairly uncommon in production and
+## selser will take care of preserving formatting in those scenarios.
 !! test
 Category / paragraph interactions
+!! options
+parsoid=wt2html
 !! wikitext
 Foo [[Category:Baz]] Bar
 
@@ -14332,7 +14373,7 @@ Bar
 [[Category:Baz]]
  {{echo|[[Category:Baz]]}}
 [[Category:Baz]]
-!! html
+!! html/php
 
Foo Bar
 
Foo
 Bar
@@ -14342,20 +14383,32 @@ Bar
 
Foo
 Bar
 

+!! html/parsoid
+
Foo  Bar

+
Foo  Bar

+
Foo  Bar

+
Foo  Bar

+
Foo    Bar     

+
 !! end
 
+## We used to, but no longer wt2wt this test since the default serializer
+## will normalize all categories to serialize on their own line.
+## This wikitext usage is going to be fairly uncommon in production and
+## selser will take care of preserving formatting in those scenarios.
+##
 ## The whitespace on the empty line is part of the test. Please do not delete
 !! test
 1. Categories and newlines: All preceding newlines should be suppressed (courtesy bug 87)
 !! options
-parsoid=wt2html,wt2wt
+parsoid=wt2html
 !! wikitext
 This
    
 [[Category:Foo]] and this should be part of same paragraph (not an indent-pre)
    
 {{echo|[[Category:Foo]] and so should this!}}
-!! html
+!! html/php
 
This and this should be part of same paragraph (not an indent-pre) and so should this!
 

 !! html/parsoid
@@ -14453,8 +14506,14 @@ parsoid=wt2html
 
 !! end
 
+## We used to, but no longer wt2wt this test since the default serializer
+## will normalize all categories to serialize on their own line.
+## This wikitext usage is going to be fairly uncommon in production and
+## selser will take care of preserving formatting in those scenarios.
 !! test
 6. Categories and newlines: migrateTrailingCategories dom pass should not migrate categories not preceded by newlines
+!! options
+parsoid=wt2html
 !! wikitext
 * a [[Category:Foo]]
 !! html/parsoid
@@ -14505,13 +14564,20 @@ parsoid
 

 !! end
 
-# html2wt localizes the "Category" namespace.
-# XXX the  element needs an empty data-parsoid attribute, or
-# else the html2html test fails because spaces are inserted.
+# We used to, but no longer wt2wt this test since the default serializer
+# will normalize all categories to serialize on their own line.
+# This wikitext usage is going to be fairly uncommon in production and
+# selser will take care of preventing whitespace insertion if this
+# occurs in an article.
+#
+# html2html disabled for the same reason (whitespace insertion between
+# x and y).
+#
+# html2wt disabled because it localizes the "Category" namespace.
 !! test
 Link prefix/suffixes aren't applied to category links
 !! options
-parsoid=wt2html,wt2wt,html2html
+parsoid=wt2html
 language=is
 !! wikitext
 x[[Category:Foo]]y
@@ -16152,10 +16218,15 @@ array (
 )
 
 
+!! html/parsoid
+

 !! end
 
+## Don't expect parsoid to rt this form.
 !! test
 Parser hook: empty input using terminated empty elements
+!! options
+parsoid=wt2html,html2html
 !! wikitext
 
 !! html/php
@@ -16165,6 +16236,8 @@ array (
 )
 
 
+!! html/parsoid
+

 !! end
 
 !! test
@@ -16178,6 +16251,8 @@ array (
 )
 
 
+!! html/parsoid
+

 !! end
 
 !! test
@@ -16191,11 +16266,15 @@ array (
 )
 
 
+!! html/parsoid
+

 !! end
 
-
+## Don't expect parsoid to rt this form.
 !! test
 Parser hook: case insensitive
+!! options
+parsoid=wt2html,html2html
 !! wikitext
 input
 !! html/php
@@ -16205,11 +16284,15 @@ array (
 )
 
 
+!! html/parsoid
+

 !! end
 
-
+## Don't expect parsoid to rt this form.
 !! test
 Parser hook: case insensitive, redux
+!! options
+parsoid=wt2html,html2html
 !! wikitext
 input
 !! html/php
@@ -16219,6 +16302,8 @@ array (
 )
 
 
+!! html/parsoid
+

 !! end
 
 !! test
@@ -16234,11 +16319,35 @@ array (
 )
 </tag>
 
+!! html/parsoid
+
</tag>
 !! end
 
 !! test
 Parser hook: basic arguments
 !! wikitext
+
+!! html/php
+
+''
+array (
+  'width' => '200',
+  'height' => '100',
+  'depth' => '50',
+  'square' => '',
+)
+

+
+!! html/parsoid
+

+!! end
+
+## Don't expect parsoid to rt this form.
+!! test
+Parser hook: basic arguments, variations
+!! options
+parsoid=wt2html,html2html
+!! wikitext
 
 !! html/php
 
@@ -16251,12 +16360,14 @@ array (
 )
 

 
+!! html/parsoid
+

 !! end
 
 !! test
 Parser hook: argument containing a forward slash (bug 5344)
 !! wikitext
-
+
 !! html/php
 
 ''
@@ -16265,10 +16376,15 @@ array (
 )
 

 
+!! html/parsoid
+

 !! end
 
+## Don't expect parsoid to rt this form.
 !! test
 Parser hook: empty input using terminated empty elements (bug 2374)
+!! options
+parsoid=wt2html,html2html
 !! wikitext
 text
 !! html/php
@@ -16279,6 +16395,8 @@ array (
 )
 text
 
+!! html/parsoid
+
text
 !! end
 
 #  should be output literally since there is no matching tag that begins it
@@ -16311,21 +16429,28 @@ array (
 Parser hook: static parser hook not inside a comment
 !! wikitext
 hello, world
-
+
+
 !! html/php
-
hello, world
+


+hello, world
 

+!! html/parsoid
+

+
hello, world

 !! end
 
-
 !! test
 Parser hook: static parser hook inside a comment
 !! wikitext
 
-
+
 !! html/php
 


 

+!! html/parsoid
+
+

 !! end
 
 # Nested template calls; this case was broken by Parser.php rev 1.506,
@@ -19211,17 +19336,25 @@ Category:分類
 blah
 !! endarticle
 
+## We used to, but no longer wt2wt this test since the default serializer
+## will normalize all categories to serialize on their own line.
+## This wikitext usage is going to be fairly uncommon in production and
+## selser will take care of preserving formatting in those scenarios.
 !! test
 Don't convert blue categorylinks to another variant (bug 33210)
 !! options
-language=zh cat
+cat
+language=zh
+parsoid=wt2html
 !! wikitext
 [[A]][[Category:分类]]
-!! html
+!! html/php
 分类
+!! html/parsoid
+
A

+
 !! end
 
-
 !! test
 Stripping -{}- tags (language variants)
 !! options
@@ -19756,7 +19889,7 @@ Tildes in comments
 pst
 !! wikitext
 
-!! html
+!! html/php
 
 !! end
 
@@ -20078,7 +20211,7 @@ Edit comment with link
 comment
 !! wikitext
 I like the [[Main Page]] a lot
-!! html
+!! html/php
 I like the Main Page a lot
 !!end
 
@@ -20088,7 +20221,7 @@ Edit comment with link and link text
 comment
 !! wikitext
 I like the [[Main Page|best pages]] a lot
-!! html
+!! html/php
 I like the best pages a lot
 !!end
 
@@ -20098,7 +20231,7 @@ Edit comment with link and link text with suffix
 comment
 !! wikitext
 I like the [[Main Page|best page]]s a lot
-!! html
+!! html/php
 I like the best pages a lot
 !!end
 
@@ -20108,7 +20241,7 @@ Edit comment with section link (non-local, eg in history list)
 comment title=[[Main Page]]
 !! wikitext
 /* External links */ removed bogus entries
-!! html
+!! html/php
 →‎External links:  removed bogus entries
 !!end
 
@@ -20118,7 +20251,7 @@ Edit comment with section link and text before it (non-local, eg in history list
 comment title=[[Main Page]]
 !! wikitext
 pre-comment text /* External links */ removed bogus entries
-!! html
+!! html/php
 pre-comment text →‎External links:  removed bogus entries
 !!end
 
@@ -20128,7 +20261,7 @@ Edit comment with section link (local, eg in diff view)
 comment local title=[[Main Page]]
 !! wikitext
 /* External links */ removed bogus entries
-!! html
+!! html/php
 →‎External links:  removed bogus entries
 !!end
 
@@ -20140,7 +20273,7 @@ subpage
 title=[[Subpage test]]
 !! wikitext
 Poked at a [[/subpage]] here...
-!! html
+!! html/php
 Poked at a /subpage here...
 !!end
 
@@ -20152,7 +20285,7 @@ subpage
 title=[[Subpage test]]
 !! wikitext
 Poked at a [[/subpage|neat little page]] here...
-!! html
+!! html/php
 Poked at a neat little page here...
 !!end
 
@@ -20163,7 +20296,7 @@ comment
 title=[[Subpage test]]
 !! wikitext
 Poked at a [[/subpage]] here...
-!! html
+!! html/php
 Poked at a /subpage here...
 !!end
 
@@ -20175,7 +20308,7 @@ local
 title=[[Main Page]]
 !! wikitext
 [[#section]]
-!! html
+!! html/php
 #section
 !! end
 
@@ -20186,24 +20319,28 @@ comment
 title=[[Main Page]]
 !! wikitext
 [[#section]]
-!! html
+!! html/php
 #section
 !! end
 
 !! test
 Anchor starting with underscore
+!! options
+title=[[Foo]]
 !! wikitext
 [[#_ref|One]]
-!! html
+!! html/php
 
One
 

+!! html/parsoid
+
One

 !! end
 
 !! test
 Id starting with underscore
 !! wikitext
 

-!! html
+!! html/*
 

 
 !! end
@@ -20215,7 +20352,7 @@ comment
 title=[[Main Page]]
 !! wikitext
 /* __hello__world__ */
-!! html
+!! html/php
 →‎__hello__world__
 !! end
 
@@ -20534,11 +20671,14 @@ HTML5 data attributes
 !! wikitext
 Baz
 
Quuz

-!! html
+!! html/php
 
Baz
 

 
Quuz

 
+!! html/parsoid
+
Baz

+
Quuz

 !! end
 
 !! test
@@ -21386,14 +21526,12 @@ parsoid=wt2html,wt2wt
 
 !!test
 Ref: 1. ref-location should be replaced with an index span
-!!options
-parsoid
 !! wikitext
 A foo
 B foo
 C 
 
-!! html
+!! html/parsoid
 
A [1]
 B [2]
 C [3]

@@ -21406,13 +21544,11 @@ C foo
 B 
 
-!! html
+!! html/parsoid
 
A [1]
 B [1]

 
    
@@ -21422,14 +21558,12 @@ B foo
 B 
 C 
 
-!! html
+!! html/parsoid
 
    A [1]
 B [1]
 C [1]
    
@@ -21441,12 +21575,10 @@ C foo    
 
-!! html
+!! html/parsoid
 
    A [1]
    
 
      
 
    1. ↑  foo
      2. 
@@ -21455,15 +21587,13 @@ A foo
 
 !!test
 Ref: 5. body should accept generic wikitext
-!!options
-parsoid
 !! wikitext
 A 
  This is a '''[[bolded link]]''' and this is a {{echo|transclusion}}
 
 
 
-!! html
+!! html/parsoid
 
      A [1]
      
 
 
        
@@ -21474,8 +21604,6 @@ A 
 
 !!test
 Ref: 6. indent-pres should not be output in ref-body
-!!options
-parsoid
 !! wikitext
 A 
  foo
@@ -21484,7 +21612,7 @@ A 
 
 
 
-!! html
+!! html/parsoid
 
        A [1]
        
 
 
          
@@ -21497,8 +21625,6 @@ A 
 
 !!test
 Ref: 7. No p-wrapping in ref-body
-!!options
-parsoid
 !! wikitext
 A 
 foo
@@ -21514,7 +21640,7 @@ booz
 
 
 
-!! html
+!! html/parsoid
 
          A [1]
          
 
 
            
@@ -21534,27 +21660,23 @@ booz
 
 !!test
 Ref: 8. transclusion wikitext has lower precedence
-!!options
-parsoid
 !! wikitext
 A  foo {{echo| B C}}
 
 
-!! html
+!! html/parsoid
 
            A [1] B C}}
            
 
              
-
            1. ↑  foo {{echo|
              2. 
+
            2. ↑  foo {{echo|
              3. 
 
            
 !!end
 
 !!test
 Ref: 9. unclosed comments should not leak out of ref-body
-!!options
-parsoid
 !! wikitext
 A  foo 
@@ -21563,13 +21685,11 @@ A  foo   [[Category:A3]]    how goes it ==
+==   [[Category:A3]]    how goes it ==
 
-== it goes well    [[Category:A4]]    ==
+== it goes well    [[Category:A4]]   ==
 
-==howdy [[Category:A5]] ==
+==howdy [[Category:A5]]==
 
 ==  __TOC__  ok ==
 !! end
@@ -25620,7 +25754,7 @@ parsoid={ "modes": ["html2wt"], "suppressErrors": true }
 # shown to sneak through on occasion. See T101768.
 # The original wikitext here is: [http://test.com [[one]] two three]
 !! test
-Strip span tags added to mark as misnested
+Strip span tags added to mark misnested links
 !! options
 parsoid=html2wt
 !! html/parsoid
@@ -25629,10 +25763,112 @@ parsoid=html2wt
 [http://test.com][[one]] two three
 !! end
 
+!! test
+Use data-parsoid.firstWikitextNode to compute newline constraints for template content
+!! options
+parsoid=html2wt
+!! html/parsoid
+a
+
+
            d
+
            
+!! wikitext
+{{echo|a}}
+{|{{echo|c
+{{!}}d
+}}
+|}
+!! end
+
+## This test verifies the presence and computation of this attribute indirectly
+## by making an edit and ensuring that the serialization is correct (which it would be
+## only if firstWikitextNode is properly set).
+!! test
+data-parsoid.firstWikitextNode should be computed properly in the presence of fostered content
+!! options
+parsoid= {
+  "modes": ["wt2wt"],
+  "changes": [
+    [ "div#x", "remove" ],
+    [ "div", "before", "
            new
            " ]
+  ]
+}
+!! wikitext
+
            foo
            
+{|
+{{echo|
            boo
            
+{{!}}b}}
+|c
+|}
+!! wikitext/edited
+
+
            new
            
+{|
+{{echo|
            boo
            
+{{!}}b}}
+|c
+|}
+!! end
+
 # --------------------------------------------
 # Tests spec'ing wikitext serialization norms |
 # --------------------------------------------
 
+!! test
+1. Categories should always be serialized on their own line
+!! options
+parsoid=html2wt
+!! html/parsoid
+foobar
+!! wikitext
+foo
+[[Category:Foo]]
+bar
+!! end
+
+!! test
+2. Categories that are part of templates should not introduce a line break
+!! wikitext
+foo {{echo|bar [[Category:baz]]}} bar
+!! html/parsoid
+
            foo bar  bar
            
+!! end
+
+# Careful while editing these next 2 tests. There are \u200f characters
+# before and after the  tags in the HTML and following some
+# of the categories in wikitext
+# Do not remove these characters in edits.
+#
+# As part of the serialization, these bidi characters will get stripped.
+!! test
+RTL (\u200f) and LTR (\u200e) markers around category tags should be stripped
+!! options
+parsoid={
+  "modes": ["html2wt"],
+  "scrubWikitext": true
+}
+!! html/parsoid
+
            ‏‏
+‏‏
            
+!! wikitext
+[[קטגוריה:טקסים]]
+[[קטגוריה: שיטות משפט]]
+!! end
+
+!! test
+RTL (\u200f) and LTR (\u200e) markers should not be stripped if followed by a text node
+!! options
+parsoid={
+  "modes": ["html2wt"],
+  "scrubWikitext": true
+}
+!! html/parsoid
+
            ‏‏y
            
+!! wikitext
+[[קטגוריה:טקסים]]
+‏y
+!! end
+
 !! test
 Lists: Add space after bullets
 !! options
@@ -25720,6 +25956,35 @@ parsoid={
 !! wikitext/edited
 !! end
 
+!! test
+Headings: Replace 
             with a single whitespace char (when scrubWikitext = true)
+!! options
+parsoid={
+  "modes": ["html2wt"],
+  "scrubWikitext": true
+}
+!! html/parsoid
+
            foo
            bar
            
+
            foo 
            bar             baz
            
+!! wikitext
+== foo bar ==
+
+== foo  bar baz ==
+!! end
+
+!! test
+Headings: Replace 
             with a single whitespace char (when scrubWikitext = false)
+!! options
+parsoid={
+  "modes": ["html2wt"],
+  "scrubWikitext": false
+}
+!! html/parsoid
+
            foo
            bar
            
+!! wikitext
+== foo
             bar ==
+!! end
+
 !! test
 1. WT Quote Tags: suppress newly created empty style tags
 !! options
@@ -26225,10 +26490,61 @@ parsoid=html2wt
 <nowiki>''foo''</nowiki>
 !! end
 
+# This is meant to be an interim fix while we go about figuring out
+# how to not introduce these trailing s in the first place.
+!! test
+T115717: Strip trailing s (without affecting valid uses)
+!! options
+parsoid=html2wt
+!! html/parsoid
+
            x
+y
            
+
            
+
            
+!! wikitext
+x
+y
+
+{{echo|
+1 = }}
+
+{{echo|
+1 = 
+}}
+!! end
+
 # ---------------------------------------------------
 # End of tests spec'ing wikitext serialization norms |
 # ---------------------------------------------------
 
+# T104032
+!! test
+Bare inline nodes not wrapped inside p-tags should be treated as p-wrapped
+!! options
+parsoid=html2wt
+!! html/parsoid
+a
            b
            
+c
            d
            
+
+
+
+
            a
            b
            		c
            d
            
+!! wikitext
+a
+
+b
+
+'''c'''
+
+d
+{|
+|a
+b
+|'''c'''
+d
+|}
+!! end
+
 # -----------------------------------------------------------------
 # End of section for Parsoid-only html2wt tests for serialization
 # of new content
-- 
2.20.1