Skip to content

Commit a2073a8

Browse files
masatakemasatake
committed
Man: extract subsections
Signed-off-by: Masatake YAMATO <[email protected]>
1 parent 2b2165a commit a2073a8

File tree

4 files changed

+295
-32
lines changed

4 files changed

+295
-32
lines changed

‎Units/parser-man.r/simple.d/expected.tags‎

Lines changed: 16 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,20 @@
1-
CTAGS input.1 /^.TH CTAGS 1 "Version 0.0.0" "Darren Hiebert" "Universal Ctags"$/;" t end:37
1+
CTAGS input.1 /^.TH CTAGS 1 "Version 0.0.0" "Darren Hiebert" "Universal Ctags"$/;" t end:89
22
NAME input.1 /^.SH "NAME"$/;" s title:CTAGS end:6
33
ABSTRACT input.1 /^ABSTRACT$/;" s title:CTAGS end:9
44
SYNOPSIS input.1 /^.SH SYNOPSIS$/;" s title:CTAGS end:16
55
DESCRIPTION input.1 /^.SH "DESCRIPTION"$/;" s title:CTAGS end:25
6-
AUTHOR input.1 /^AUTHOR$/;" s title:CTAGS end:30
7-
ACKNOWLEDGMENT input.1 /^ACKNOWLEDGMENT$/;" s title:CTAGS end:34
8-
SEE ALSO input.1 /^.SH SEE ALSO$/;" s title:CTAGS end:37
6+
AUTHOR input.1 /^AUTHOR$/;" s title:CTAGS end:52
7+
subsection1 input.1 /^.SS subsection1$/;" S section:CTAGS""AUTHOR end:34
8+
subsection2 input.1 /^.SS "subsection2"$/;" S section:CTAGS""AUTHOR end:38
9+
subsection3 input.1 /^subsection3$/;" S section:CTAGS""AUTHOR end:43
10+
sub section 4 input.1 /^.SS sub section 4$/;" S section:CTAGS""AUTHOR end:47
11+
sub section 5 input.1 /^sub section 5$/;" S section:CTAGS""AUTHOR end:52
12+
ACKNOWLEDGMENT input.1 /^ACKNOWLEDGMENT$/;" s title:CTAGS end:83
13+
subsection6 input.1 /^subsection6$/;" S section:CTAGS""ACKNOWLEDGMENT end:61
14+
subsection7 input.1 /^subsection7$/;" S section:CTAGS""ACKNOWLEDGMENT end:66
15+
subsection8 input.1 /^.SS "subsection8"$/;" S section:CTAGS""ACKNOWLEDGMENT end:70
16+
subsection9 input.1 /^.SS subsection9$/;" S section:CTAGS""ACKNOWLEDGMENT end:74
17+
sub section 10 input.1 /^.SS sub section 10$/;" S section:CTAGS""ACKNOWLEDGMENT end:78
18+
sub section 11 input.1 /^sub section 11$/;" S section:CTAGS""ACKNOWLEDGMENT end:83
19+
SEE ALSO input.1 /^.SH SEE ALSO$/;" s title:CTAGS end:89
20+
sub section 12 input.1 /^sub section 12$/;" S section:CTAGS""SEE ALSO end:89

‎Units/parser-man.r/simple.d/input.1‎

Lines changed: 52 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -28,10 +28,62 @@ AUTHOR
2828
.sp
2929
The Developers.
3030

31+
.SS subsection1
32+
.sp
33+
This is the first subsection.
34+
35+
.SS "subsection2"
36+
.sp
37+
This is the second subsection.
38+
39+
.SS
40+
subsection3
41+
.sp
42+
This is the third subsection.
43+
44+
.SS sub section 4
45+
.sp
46+
This is the fourth subsection.
47+
48+
.SS
49+
sub section 5
50+
.sp
51+
This is the fifth subsection.
52+
3153
.SH
3254
ACKNOWLEDGMENT
3355
.sp
3456

57+
.SS
58+
subsection6
59+
.sp
60+
This is the sixth subsection.
61+
62+
.SS
63+
subsection7
64+
.sp
65+
This is the seventh subsection.
66+
67+
.SS "subsection8"
68+
.sp
69+
This is the eight subsection.
70+
71+
.SS subsection9
72+
.sp
73+
This is the ninth subsection.
74+
75+
.SS sub section 10
76+
.sp
77+
This is the tenth subsection.
78+
79+
.SS
80+
sub section 11
81+
.sp
82+
This is the eleventh subsection.
83+
3584
.SH SEE ALSO
3685
.sp
3786
See tags(5) for the details of tags file format.
87+
88+
.SS
89+
sub section 12

‎optlib/man.c‎

Lines changed: 124 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -8,19 +8,71 @@
88
#include "xtag.h"
99

1010

11+
typedef enum {
12+
K_TITLE,
13+
K_SECTION,
14+
K_SUBSECTION,
15+
} ManKind;
16+
17+
1118
static void initializeManParser (const langType language)
1219
{
1320
addLanguageOptscriptToHook (language, SCRIPT_HOOK_PRELUDE,
14-
"{{ % /replace SCOPE-ACTION -\n"
15-
" % /push SCOPE-ACTION -\n"
16-
" /scope-action {\n"
21+
"{{ % adjustment:int FILL-END-OF-SCOPE -\n"
22+
" %\n"
23+
" % Note: The group 1 of a regex matching is assumed.\n"
24+
" % An entry on the scope stack is assumed.\n"
25+
" /fill-end-of-scope {\n"
26+
" _scopetop pop exch\n"
27+
" % scope-top:int adjustment:int\n"
28+
" 1 /start _matchloc _matchloc2line exch\n"
29+
" % scope-top:int line adjustment:int\n"
30+
" 2 copy gt {\n"
31+
" sub end:\n"
32+
" } {\n"
33+
" pop pop pop\n"
34+
" } ifelse\n"
35+
" } def\n"
36+
" % adjustment:int /replace HEADING-ACTION -\n"
37+
" % /push HEADING-ACTION -\n"
38+
" /heading-action {\n"
1739
" /replace eq {\n"
18-
" _scopetop pop 1 /start _matchloc _matchloc2line dup 2 gt {\n"
19-
" 2 sub end:\n"
20-
" } {\n"
21-
" pop\n"
22-
" pop\n"
23-
" } ifelse\n"
40+
" %\n"
41+
" % Before removing the tag at the top of the scope stack,\n"
42+
" % fill the end field of the tag.\n"
43+
" %\n"
44+
" % input0.man\n"
45+
" % ----------------------------------------------------\n"
46+
" % .SH SEC1\n"
47+
" % ...\n"
48+
" %E: This is the end of the SEC1.\n"
49+
" % .SH\n"
50+
" %C: SEC2\n"
51+
" %\n"
52+
" % ----------------------------------------------------\n"
53+
" %\n"
54+
" % C represents the current input line. The parser must\n"
55+
" % fill the end field of \"SEC1\", the tag at the top,\n"
56+
" % with the line number of E.\n"
57+
" %\n"
58+
" % E = C - 2\n"
59+
" %\n"
60+
" % input1.man\n"
61+
" % ----------------------------------------------------\n"
62+
" % .SH SEC3\n"
63+
" % ...\n"
64+
" %E: This is the end of the SEC1.\n"
65+
" %C: .SH SEC4\n"
66+
" %\n"
67+
" % ----------------------------------------------------\n"
68+
" %\n"
69+
" % In this case\n"
70+
" %\n"
71+
" % E = C - 1\n"
72+
" %\n"
73+
" % The offset for the adjustment depends on the conctxt.\n"
74+
" %\n"
75+
" fill-end-of-scope\n"
2476
" _scopepop\n"
2577
" } if\n"
2678
"\n"
@@ -34,6 +86,8 @@ static void initializeManParser (const langType language)
3486
addLanguageRegexTable (language, "main");
3587
addLanguageRegexTable (language, "section");
3688
addLanguageRegexTable (language, "sectionheading");
89+
addLanguageRegexTable (language, "subsection");
90+
addLanguageRegexTable (language, "subsectionheading");
3791
addLanguageRegexTable (language, "EOF");
3892
addLanguageRegexTable (language, "SKIP");
3993
addLanguageRegexTable (language, "REST");
@@ -76,10 +130,22 @@ static void initializeManParser (const langType language)
76130
"^\\.SH[\t ]+([^\n]+)\n",
77131
"\\1", "s", "{icase}{scope=replace}", NULL);
78132
addLanguageTagMultiTableRegex (language, "section",
79-
"^\\.SH[\t ]*\n",
80-
"", "", "{icase}{tenter=sectionheading}"
133+
"^(\\.SH)[\t ]*\n",
134+
"", "", "{icase}{tjump=sectionheading}"
81135
"{{\n"
82-
" /replace\n"
136+
" 2 /replace\n"
137+
"}}", NULL);
138+
addLanguageTagMultiTableRegex (language, "section",
139+
"^\\.SS[\t ]+\"([^\"\n]+)\"[^\n]*\n",
140+
"\\1", "S", "{icase}{scope=push}{tenter=subsection}", NULL);
141+
addLanguageTagMultiTableRegex (language, "section",
142+
"^\\.SS[\t ]+([^\n]+)\n",
143+
"\\1", "S", "{icase}{scope=push}{tenter=subsection}", NULL);
144+
addLanguageTagMultiTableRegex (language, "section",
145+
"^\\.SS[\t ]*\n",
146+
"", "", "{icase}{tenter=subsectionheading}"
147+
"{{\n"
148+
" /push\n"
83149
"}}", NULL);
84150
addLanguageTagMultiTableRegex (language, "section",
85151
"^[^\n]*\n|[^\n]+",
@@ -89,16 +155,56 @@ static void initializeManParser (const langType language)
89155
"", "", "{scope=clear}{tquit}", NULL);
90156
addLanguageTagMultiTableRegex (language, "sectionheading",
91157
"^[ \t]*([^\n]+)\n",
92-
"\\1", "s", "{tleave}"
158+
"\\1", "s", "{tjump=section}"
93159
"{{\n"
94-
" scope-action\n"
160+
" heading-action\n"
95161
"}}", NULL);
96162
addLanguageTagMultiTableRegex (language, "sectionheading",
97163
"^[^\n]*\n|[^\n]+",
98164
"", "", "", NULL);
99165
addLanguageTagMultiTableRegex (language, "sectionheading",
100166
"^",
101167
"", "", "{scope=clear}{tquit}", NULL);
168+
addLanguageTagMultiTableRegex (language, "subsection",
169+
"^([^\n.]|\\.[^\nst])[^\n]*\n",
170+
"", "", "{icase}", NULL);
171+
addLanguageTagMultiTableRegex (language, "subsection",
172+
"^\\.SS[\t ]+\"([^\"\n]+)\"[^\n]*\n",
173+
"\\1", "S", "{icase}{scope=replace}", NULL);
174+
addLanguageTagMultiTableRegex (language, "subsection",
175+
"^\\.SS[\t ]+([^\n]+)\n",
176+
"\\1", "S", "{icase}{scope=replace}", NULL);
177+
addLanguageTagMultiTableRegex (language, "subsection",
178+
"^\\.SS[\t ]*\n",
179+
"", "", "{icase}{tjump=subsectionheading}"
180+
"{{\n"
181+
" 2 /replace\n"
182+
"}}", NULL);
183+
addLanguageTagMultiTableRegex (language, "subsection",
184+
"^(\\.SH)",
185+
"", "", "{icase}{_advanceTo=0start}{tleave}"
186+
"{{\n"
187+
" 1 fill-end-of-scope\n"
188+
" _scopepop\n"
189+
"}}", NULL);
190+
addLanguageTagMultiTableRegex (language, "subsection",
191+
"^[^\n]*\n|[^\n]+",
192+
"", "", "", NULL);
193+
addLanguageTagMultiTableRegex (language, "subsection",
194+
"^",
195+
"", "", "{scope=clear}{tquit}", NULL);
196+
addLanguageTagMultiTableRegex (language, "subsectionheading",
197+
"^[ \t]*([^\n]+)\n",
198+
"\\1", "S", "{tjump=subsection}"
199+
"{{\n"
200+
" heading-action\n"
201+
"}}", NULL);
202+
addLanguageTagMultiTableRegex (language, "subsectionheading",
203+
"^[^\n]*\n|[^\n]+",
204+
"", "", "", NULL);
205+
addLanguageTagMultiTableRegex (language, "subsectionheading",
206+
"^",
207+
"", "", "{scope=clear}{tquit}", NULL);
102208
addLanguageTagMultiTableRegex (language, "EOF",
103209
"^",
104210
"", "", "{scope=clear}{tquit}", NULL);
@@ -149,6 +255,9 @@ extern parserDefinition* ManParser (void)
149255
{
150256
true, 's', "section", "sections",
151257
},
258+
{
259+
true, 'S', "subsection", "sub sections",
260+
},
152261
};
153262

154263
parserDefinition* const def = parserNew ("Man");
@@ -161,6 +270,7 @@ extern parserDefinition* ManParser (void)
161270
def->useCork = CORK_QUEUE;
162271
def->kindTable = ManKindTable;
163272
def->kindCount = ARRAY_SIZE(ManKindTable);
273+
def->defaultScopeSeparator = "\"\"";
164274
def->initialize = initializeManParser;
165275

166276
return def;

0 commit comments

Comments
 (0)