Merge pull request #3246 from masatake/rst--title

masatake · web-flow · commit e179627fd8d8 · 2022-01-06T02:00:04.000+09:00
ReStructuredText: extract titles and subtitles
diff --git a/Tmain/e-ctags-output.d/stdout-expected.txt b/Tmain/e-ctags-output.d/stdout-expected.txt
@@ -1,18 +1,18 @@
 #
 # with --sort=yes
 #
-A spaces B	input_space.rst	/^A spaces B$/;"	c
+A spaces B	input_space.rst	/^A spaces B$/;"	H
 test	input file.cc	/^int32 test(int32 a)$/;"	f	typeref:typename:int32	signature:(int32 a)
 test2	input file.cc	/^int32 test2(void)$/;"	f	typeref:typename:int32	signature:(void)
 test3	input file.cc	/^int32 test3(int32	a)$/;"	f	typeref:typename:int32	signature:(int32 a)
 test4	input file.cc	/^int32 test4(int32$/;"	f	typeref:typename:int32	signature:(int32 a)
 # WITH SCOPE
-X space Y	input_scope.rst	/^X space Y$/;"	c
-title	input_scope.rst	/^title$/;"	c
+X space Y	input_scope.rst	/^X space Y$/;"	c	title:title
+title	input_scope.rst	/^title$/;"	H
 topic in space	input_scope.rst	/^topic in space$/;"	s	chapter:X space Y
 # WITHOUT SCOPE
 X space Y	input_scope.rst	/^X space Y$/;"	c
-title	input_scope.rst	/^title$/;"	c
+title	input_scope.rst	/^title$/;"	H
 topic in space	input_scope.rst	/^topic in space$/;"	s
 topic in tab	input_scope.rst	/^topic in tab$/;"	s
 #
@@ -22,13 +22,13 @@ test	input file.cc	/^int32 test(int32 a)$/;"	f	typeref:typename:int32	signature:
 test2	input file.cc	/^int32 test2(void)$/;"	f	typeref:typename:int32	signature:(void)
 test3	input file.cc	/^int32 test3(int32	a)$/;"	f	typeref:typename:int32	signature:(int32 a)
 test4	input file.cc	/^int32 test4(int32$/;"	f	typeref:typename:int32	signature:(int32 a)
-A spaces B	input_space.rst	/^A spaces B$/;"	c
+A spaces B	input_space.rst	/^A spaces B$/;"	H
 # WITH SCOPE
-title	input_scope.rst	/^title$/;"	c
-X space Y	input_scope.rst	/^X space Y$/;"	c
+title	input_scope.rst	/^title$/;"	H
+X space Y	input_scope.rst	/^X space Y$/;"	c	title:title
 topic in space	input_scope.rst	/^topic in space$/;"	s	chapter:X space Y
 # WITHOUT SCOPE
-title	input_scope.rst	/^title$/;"	c
+title	input_scope.rst	/^title$/;"	H
 topic in tab	input_scope.rst	/^topic in tab$/;"	s
 X space Y	input_scope.rst	/^X space Y$/;"	c
 topic in space	input_scope.rst	/^topic in space$/;"	s
diff --git a/Tmain/list-fields-with-prefix.d/stdout-expected.txt b/Tmain/list-fields-with-prefix.d/stdout-expected.txt
@@ -36,6 +36,7 @@ x       UCTAGSxpath          no      NONE             s--    no    -- xpath for
 -       UCTAGSnameref        yes     Python           s--    no    -- the original name for the tag
 -       UCTAGSassignmentop   no      R                s--    no    -- operator for assignment
 -       UCTAGSconstructor    yes     R                s--    no    -- function used for making value assigned to the nameattr tag
+-       UCTAGSoverline       no      ReStructuredText --b    no    -- whether using overline & underline for declaring section
 -       UCTAGSsectionMarker  no      ReStructuredText s--    no    -- character used for declaring section
 -       UCTAGSmixin          yes     Ruby             s--    no    -- how the class or module is mixed in (mixin:HOW:MODULE)
 -       UCTAGSparameter      no      SystemVerilog    --b    no    -- parameter whose value can be overridden.
diff --git a/Tmain/list-fields.d/stdout-expected.txt b/Tmain/list-fields.d/stdout-expected.txt
@@ -54,6 +54,7 @@ z	kind	no	NONE	s--	no	r-	[tags output] prepend "kind:" to k/ (or K/) field outpu
 -	nameref	yes	Python	s--	no	--	the original name for the tag
 -	assignmentop	no	R	s--	no	--	operator for assignment
 -	constructor	yes	R	s--	no	--	function used for making value assigned to the nameattr tag
+-	overline	no	ReStructuredText	--b	no	--	whether using overline & underline for declaring section
 -	sectionMarker	no	ReStructuredText	s--	no	--	character used for declaring section
 -	mixin	yes	Ruby	s--	no	--	how the class or module is mixed in (mixin:HOW:MODULE)
 -	parameter	no	SystemVerilog	--b	no	--	parameter whose value can be overridden.
diff --git a/Tmain/list-language-fields.d/stdout-expected.txt b/Tmain/list-language-fields.d/stdout-expected.txt
@@ -1,2 +1,3 @@
 #LETTER	NAME	ENABLED	LANGUAGE	JSTYPE	FIXED	OP	DESCRIPTION
+-	overline	no	ReStructuredText	--b	no	--	whether using overline & underline for declaring section
 -	sectionMarker	no	ReStructuredText	s--	no	--	character used for declaring section
diff --git a/Units/parser-html.r/class-and-id.d/expected.tags b/Units/parser-html.r/class-and-id.d/expected.tags
@@ -1,3 +1,4 @@
+Home &middot; Universal Ctags	input.html	/^  <\/title>$/;"	j	roles:def
 /public/css/poole.css	input.html	/^<link rel="stylesheet" href="\/public\/css\/poole.css">$/;"	C	roles:extFile
 /public/css/syntax.css	input.html	/^<link rel="stylesheet" href="\/public\/css\/syntax.css">$/;"	C	roles:extFile
 /public/css/hyde.css	input.html	/^<link rel="stylesheet" href="\/public\/css\/hyde.css">$/;"	C	roles:extFile
diff --git a/Units/parser-restructuredtext.r/citation.d/args.ctags b/Units/parser-restructuredtext.r/citation.d/args.ctags
@@ -1 +1,2 @@
 --sort=no
+--fields-ReStructuredText=+{overline}{sectionMarker}
diff --git a/Units/parser-restructuredtext.r/citation.d/expected.tags b/Units/parser-restructuredtext.r/citation.d/expected.tags
@@ -1,6 +1,6 @@
-KVM VCPU Requests	input.rst	/^KVM VCPU Requests$/;"	c
-VCPU Request Internals	input.rst	/^VCPU Request Internals$/;"	c
-References	input.rst	/^References$/;"	c
+KVM VCPU Requests	input.rst	/^KVM VCPU Requests$/;"	H	sectionMarker:=	overline:
+VCPU Request Internals	input.rst	/^VCPU Request Internals$/;"	c	title:KVM VCPU Requests	sectionMarker:=
+References	input.rst	/^References$/;"	c	title:KVM VCPU Requests	sectionMarker:=
 atomic-ops	input.rst	/^.. [atomic-ops] Documentation\/core-api\/atomic_ops.rst$/;"	C	chapter:References
 memory-barriers	input.rst	/^.. [memory-barriers] Documentation\/memory-barriers.txt$/;"	C	chapter:References
 lwn-mb	input.rst	/^.. [lwn-mb] https:\/\/lwn.net\/Articles\/573436\/$/;"	C	chapter:References
diff --git a/Units/parser-restructuredtext.r/iso8859-1-restructuredtext.d/expected.tags b/Units/parser-restructuredtext.r/iso8859-1-restructuredtext.d/expected.tags
@@ -1,4 +1,6 @@
-Ch�pt�r 1	input.rst	/^Ch�pt�r 1$/;"	c
+Ch�pt�r 1	input.rst	/^Ch�pt�r 1$/;"	c	subtitle:S�bit�tl�
 Subsubsection 1.1.1.1	input.rst	/^Subsubsection 1.1.1.1$/;"	t	subsection:S�bs
 S��ti�n 1.1	input.rst	/^S��ti�n 1.1$/;"	s	chapter:Ch�pt�r 1
 S�bs	input.rst	/^S�bs$/;"	S	section:S��ti�n 1.1
+S�bit�tl�	input.rst	/^S�bit�tl�$/;"	h	title:T�tl�
+T�tl�	input.rst	/^T�tl�$/;"	H
diff --git a/Units/parser-restructuredtext.r/iso8859-1-restructuredtext.d/input.rst b/Units/parser-restructuredtext.r/iso8859-1-restructuredtext.d/input.rst
@@ -1,3 +1,10 @@
+=======================================
+T�tl�
+=======================================
+---------------------------------------
+S�bit�tl�
+---------------------------------------
+
 Ch�pt�r 1
 =========
 
diff --git a/Units/parser-restructuredtext.r/substdef.d/expected.tags b/Units/parser-restructuredtext.r/substdef.d/expected.tags
@@ -1,3 +1,3 @@
 struct cpuidle_state	input.rst	/^.. |struct cpuidle_state| replace:: :c:type:`struct cpuidle_state <cpuidle_state>`$/;"	d
 cpufreq	input.rst	/^.. |cpufreq| replace:: :doc:`CPU Performance Scaling <cpufreq>`$/;"	d
-CPU Idle Time Management	input.rst	/^CPU Idle Time Management$/;"	c
+CPU Idle Time Management	input.rst	/^CPU Idle Time Management$/;"	H
diff --git a/Units/parser-restructuredtext.r/target-restructuredtext.d/expected.tags b/Units/parser-restructuredtext.r/target-restructuredtext.d/expected.tags
@@ -1,3 +1,4 @@
+ANOTHER TITLE	input.rst	/^ANOTHER TITLE$/;"	c
 TITLE	input.rst	/^TITLE$/;"	c
 title-target	input.rst	/^.. _title-target:$/;"	T	chapter:TITLE
 section0	input.rst	/^section0$/;"	s	chapter:TITLE
diff --git a/Units/parser-restructuredtext.r/target-restructuredtext.d/input.rst b/Units/parser-restructuredtext.r/target-restructuredtext.d/input.rst
@@ -1,3 +1,7 @@
+===========================================
+ANOTHER TITLE
+===========================================
+
 ===========================================
 TITLE
 ===========================================
diff --git a/Units/parser-restructuredtext.r/title.d/args.ctags b/Units/parser-restructuredtext.r/title.d/args.ctags
@@ -0,0 +1,3 @@
+--sort=no
+--fields=+ZK
+--fields-ReStructuredText=+{sectionMarker}{overline}
diff --git a/Units/parser-restructuredtext.r/title.d/expected.tags b/Units/parser-restructuredtext.r/title.d/expected.tags
@@ -0,0 +1,30 @@
+TITLE	input.rst	/^TITLE$/;"	title	sectionMarker:=	overline:
+Subtitle	input.rst	/^Subtitle$/;"	subtitle	scope:title:TITLE	sectionMarker:-	overline:
+a	input.rst	/^.. [a] a.rst$/;"	citation	scope:subtitle:Subtitle
+Chapter	input.rst	/^Chapter$/;"	chapter	scope:subtitle:Subtitle	sectionMarker:=
+b	input.rst	/^.. [b] b.rst$/;"	citation	scope:chapter:Chapter
+Section	input.rst	/^Section$/;"	section	scope:chapter:Chapter	sectionMarker:-
+c	input.rst	/^.. [c] c.rst$/;"	citation	scope:section:Section
+Subsection	input.rst	/^Subsection$/;"	subsection	scope:section:Section	sectionMarker:|
+d	input.rst	/^.. [d] d.rst$/;"	citation	scope:subsection:Subsection
+SubSubsection	input.rst	/^SubSubsection$/;"	subsubsection	scope:subsection:Subsection	sectionMarker:.	overline:
+e	input.rst	/^.. [e] e.rst$/;"	citation	scope:subsubsection:SubSubsection
+f	input.rst	/^.. [f] f.rst$/;"	citation	scope:subsubsection:SubSubsection
+g	input.rst	/^.. [g] g.rst$/;"	citation	scope:subsubsection:SubSubsection
+Chapter2	input.rst	/^Chapter2$/;"	chapter	scope:subtitle:Subtitle	sectionMarker:=
+h	input.rst	/^.. [h] g.rst$/;"	citation	scope:chapter:Chapter2
+C0	input-0.rst	/^C0$/;"	chapter	sectionMarker:=	overline:
+S0	input-0.rst	/^S0$/;"	section	scope:chapter:C0	sectionMarker:-	overline:
+s0	input-0.rst	/^s0$/;"	subsection	scope:section:S0	sectionMarker:=
+C1	input-0.rst	/^C1$/;"	chapter	sectionMarker:=	overline:
+S1	input-0.rst	/^S1$/;"	section	scope:chapter:C1	sectionMarker:-	overline:
+s1	input-0.rst	/^s1$/;"	subsection	scope:section:S1	sectionMarker:=
+T0	input-1.rst	/^                     T0$/;"	title	sectionMarker:=	overline:
+xC0	input-1.rst	/^                     xC0$/;"	chapter	scope:title:T0	sectionMarker:-	overline:
+xs0	input-1.rst	/^xs0$/;"	section	scope:chapter:xC0	sectionMarker:=
+xC1	input-1.rst	/^xC1$/;"	chapter	scope:title:T0	sectionMarker:-	overline:
+xs1	input-1.rst	/^xs1$/;"	section	scope:chapter:xC1	sectionMarker:=
+yT0	input-2.rst	/^                   yT0$/;"	title	sectionMarker:=	overline:
+yST0	input-2.rst	/^yST0$/;"	subtitle	scope:title:yT0	sectionMarker:-
+yC0	input-2.rst	/^yC0$/;"	chapter	scope:subtitle:yST0	sectionMarker:=
+yC1	input-2.rst	/^yC1$/;"	chapter	scope:subtitle:yST0	sectionMarker:=
diff --git a/Units/parser-restructuredtext.r/title.d/input-0.rst b/Units/parser-restructuredtext.r/title.d/input-0.rst
@@ -0,0 +1,19 @@
+==============================================
+C0
+==============================================
+----------------------------------------------
+S0
+----------------------------------------------
+
+s0
+==============================================
+
+==============================================
+C1
+==============================================
+----------------------------------------------
+S1
+----------------------------------------------
+
+s1
+==============================================
diff --git a/Units/parser-restructuredtext.r/title.d/input-1.rst b/Units/parser-restructuredtext.r/title.d/input-1.rst
@@ -0,0 +1,16 @@
+==============================================
+                     T0
+==============================================
+----------------------------------------------
+                     xC0
+----------------------------------------------
+
+xs0
+==============================================
+
+----------------------------------------------
+xC1
+----------------------------------------------
+
+xs1
+==============================================
diff --git a/Units/parser-restructuredtext.r/title.d/input-2.rst b/Units/parser-restructuredtext.r/title.d/input-2.rst
@@ -0,0 +1,13 @@
+==============================================
+                   yT0
+==============================================
+
+yST0
+----------------------------------------------
+
+yC0
+==============================================
+
+
+yC1
+==============================================
diff --git a/Units/parser-restructuredtext.r/title.d/input.rst b/Units/parser-restructuredtext.r/title.d/input.rst
@@ -0,0 +1,68 @@
+==============================================
+TITLE
+==============================================
+----------------------------------------------
+Subtitle
+----------------------------------------------
+
+[a]_
+
+.. [a] a.rst
+
+Chapter
+==============================================
+
+[b]_
+
+.. [b] b.rst
+
+Section
+---------------------------------------------
+
+[c]_
+
+.. [c] c.rst
+
+Subsection
+|||||||||||||||||||||||||||||||||||||||||||||
+
+[d]_
+
+.. [d] d.rst
+
+
+.............................................
+SubSubsection
+.............................................
+
+[e]_
+
+.. [e] e.rst
+
+
+,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
+SubSubSubsection will not be extracted.
+,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
+
+[f]_
+
+.. [f] f.rst
+
+@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
+SubSubSubSubsection will not be extracted.
+@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
+
+[g]_
+
+.. [g] g.rst
+
+Chapter2
+==============================================
+
+@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
+Title level inconsistent:
+@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
+
+[h]_
+
+.. [h] g.rst
diff --git a/Units/parser-restructuredtext.r/utf8-restructuredtext.d/expected.tags b/Units/parser-restructuredtext.r/utf8-restructuredtext.d/expected.tags
@@ -1,4 +1,6 @@
 @Ѐ–𐀀	input.rst	/^@Ѐ–𐀀$/;"	S	section:Šéçtiön 1.1
-Chàptěr 1	input.rst	/^Chàptěr 1$/;"	c
+Chàptěr 1	input.rst	/^Chàptěr 1$/;"	c	subtitle:Șůƀťițƚe
 Subsubsection 1.1.1.1	input.rst	/^Subsubsection 1.1.1.1$/;"	t	subsection:@Ѐ–𐀀
+Titlə	input.rst	/^Titlə$/;"	H
 Šéçtiön 1.1	input.rst	/^Šéçtiön 1.1$/;"	s	chapter:Chàptěr 1
+Șůƀťițƚe	input.rst	/^Șůƀťițƚe$/;"	h	title:Titlə
diff --git a/Units/parser-restructuredtext.r/utf8-restructuredtext.d/input.rst b/Units/parser-restructuredtext.r/utf8-restructuredtext.d/input.rst
@@ -1,3 +1,10 @@
+==========================================
+Titlə
+==========================================
+------------------------------------------
+Șůƀťițƚe
+------------------------------------------
+
 Chàptěr 1
 =========
 
diff --git a/parsers/html.c b/parsers/html.c
@@ -29,6 +29,7 @@
 typedef enum {
 	K_ANCHOR,
 	K_CLASS,
+	K_TITLE,
 	K_HEADING1,
 	K_HEADING2,
 	K_HEADING3,
@@ -66,6 +67,7 @@ static kindDefinition HtmlKinds [] = {
 	{ true, 'a', "anchor",		"named anchors" },
 	{ true, 'c', "class",		"classes",
 	  .referenceOnly = true, ATTACH_ROLES (ClassRoles)},
+	{ true, 't', "title",		"titles" },
 	{ true, 'h', "heading1",	"H1 headings" },
 	{ true, 'i', "heading2",	"H2 headings" },
 	{ true, 'j', "heading3",	"H3 headings" },
@@ -77,9 +79,16 @@ static kindDefinition HtmlKinds [] = {
 };
 
 typedef enum {
+	/* The order starting from "title" to "h3" should
+	 * not be changed.
+	 *
+	 */
+	KEYWORD_heading_start,
+	KEYWORD_title = KEYWORD_heading_start,
 	KEYWORD_h1,
 	KEYWORD_h2,
 	KEYWORD_h3,
+	KEYWORD_heading_end = KEYWORD_h3,
 	KEYWORD_a,
 	KEYWORD_script,
 	KEYWORD_style,
@@ -110,6 +119,7 @@ typedef enum {
 } keywordId;
 
 static const keywordTable HtmlKeywordTable[] = {
+	{"title", KEYWORD_title},
 	{"h1", KEYWORD_h1},
 	{"h2", KEYWORD_h2},
 	{"h3", KEYWORD_h3},
@@ -465,7 +475,7 @@ static void readTag (tokenInfo *token, vString *text, int depth)
 		bool stylesheet_expectation = false;
 
 		startTag = lookupKeyword (vStringValue (token->string), Lang_html);
-		isHeading = (startTag == KEYWORD_h1 || startTag == KEYWORD_h2 || startTag == KEYWORD_h3);
+		isHeading = (KEYWORD_heading_start <= startTag && startTag <= KEYWORD_heading_end);
 		isVoid = (startTag >= KEYWORD_area && startTag <= KEYWORD_wbr);
 		if (text == NULL && isHeading)
 		{
@@ -600,6 +610,8 @@ static void readTag (tokenInfo *token, vString *text, int depth)
 					{
 						htmlKind headingKind;
 
+						if (startTag == KEYWORD_title)
+							headingKind = K_TITLE;
 						if (startTag == KEYWORD_h1)
 							headingKind = K_HEADING1;
 						else if (startTag == KEYWORD_h2)
diff --git a/parsers/rst.c b/parsers/rst.c

Original file line number	Diff line number	Diff line change
`@@ -1,2 +1,3 @@`
`1`	`1`	`#LETTER NAME ENABLED LANGUAGE JSTYPE FIXED OP DESCRIPTION`
	`2`	`+- overline no ReStructuredText --b no -- whether using overline & underline for declaring section`
`2`	`3`	`- sectionMarker no ReStructuredText s-- no -- character used for declaring section`
Original file line number	Diff line number	Diff line change
`@@ -1,3 +1,4 @@`
	`1`	`+Home · Universal Ctags input.html /^ <\/title>$/;" j roles:def`
`1`	`2`	`/public/css/poole.css input.html /^<link rel="stylesheet" href="\/public\/css\/poole.css">$/;" C roles:extFile`
`2`	`3`	`/public/css/syntax.css input.html /^<link rel="stylesheet" href="\/public\/css\/syntax.css">$/;" C roles:extFile`
`3`	`4`	`/public/css/hyde.css input.html /^<link rel="stylesheet" href="\/public\/css\/hyde.css">$/;" C roles:extFile`
Original file line number	Diff line number	Diff line change
`@@ -1 +1,2 @@`
`1`	`1`	`--sort=no`
	`2`	`+--fields-ReStructuredText=+{overline}{sectionMarker}`
-Original file line number
+Diff line change
@@ @@ -1,3 +1,10 @@ @@
 +=======================================
 +Tîtlë
 +=======================================
 +---------------------------------------
 +Sûbitïtlê
 +---------------------------------------
++
 Chàptèr 1
 =========