blob: 60f0e5ada81967d3743ef866438aaf9b8efdc497 [file] [log] [blame]
Bo Yang25ed3412013-03-28 17:47:30 +01001#include "git-compat-util.h"
2#include "line-range.h"
Thomas Rast13b8f682013-03-28 17:47:33 +01003#include "xdiff-interface.h"
Thomas Rast13b8f682013-03-28 17:47:33 +01004#include "userdiff.h"
Bo Yang25ed3412013-03-28 17:47:30 +01005
6/*
7 * Parse one item in the -L option
Eric Sunshine815834e2013-08-06 09:59:41 -04008 *
9 * 'begin' is applicable only to relative range anchors. Absolute anchors
10 * ignore this value.
11 *
12 * When parsing "-L A,B", parse_loc() is called once for A and once for B.
13 *
14 * When parsing A, 'begin' must be a negative number, the absolute value of
15 * which is the line at which relative start-of-range anchors should be
16 * based. Beginning of file is represented by -1.
17 *
18 * When parsing B, 'begin' must be the positive line number immediately
19 * following the line computed for 'A'.
Bo Yang25ed3412013-03-28 17:47:30 +010020 */
21static const char *parse_loc(const char *spec, nth_line_fn_t nth_line,
22 void *data, long lines, long begin, long *ret)
23{
24 char *term;
25 const char *line;
26 long num;
27 int reg_error;
28 regex_t regexp;
29 regmatch_t match[1];
30
31 /* Allow "-L <something>,+20" to mean starting at <something>
32 * for 20 lines, or "-L <something>,-5" for 5 lines ending at
33 * <something>.
34 */
Eric Sunshine5d57cac2013-07-31 04:15:45 -040035 if (1 <= begin && (spec[0] == '+' || spec[0] == '-')) {
Bo Yang25ed3412013-03-28 17:47:30 +010036 num = strtol(spec + 1, &term, 10);
37 if (term != spec + 1) {
Thomas Rast12da1d12013-03-28 17:47:32 +010038 if (!ret)
39 return term;
Eric Sunshineabba3532013-07-31 04:15:43 -040040 if (num == 0)
41 die("-L invalid empty range");
Bo Yang25ed3412013-03-28 17:47:30 +010042 if (spec[0] == '-')
43 num = 0 - num;
44 if (0 < num)
45 *ret = begin + num - 2;
46 else if (!num)
47 *ret = begin;
48 else
Isabella Stephens96cfa942018-06-15 16:29:27 +100049 *ret = begin + num > 0 ? begin + num : 1;
Bo Yang25ed3412013-03-28 17:47:30 +010050 return term;
51 }
52 return spec;
53 }
54 num = strtol(spec, &term, 10);
55 if (term != spec) {
Eric Sunshine5ce922a2013-08-06 09:59:49 -040056 if (ret) {
57 if (num <= 0)
58 die("-L invalid line number: %ld", num);
Thomas Rast12da1d12013-03-28 17:47:32 +010059 *ret = num;
Eric Sunshine5ce922a2013-08-06 09:59:49 -040060 }
Bo Yang25ed3412013-03-28 17:47:30 +010061 return term;
62 }
Eric Sunshine815834e2013-08-06 09:59:41 -040063
Eric Sunshinea6ac5f92013-08-06 09:59:45 -040064 if (begin < 0) {
65 if (spec[0] != '^')
66 begin = -begin;
67 else {
68 begin = 1;
69 spec++;
70 }
71 }
Eric Sunshine815834e2013-08-06 09:59:41 -040072
Bo Yang25ed3412013-03-28 17:47:30 +010073 if (spec[0] != '/')
74 return spec;
75
76 /* it could be a regexp of form /.../ */
77 for (term = (char *) spec + 1; *term && *term != '/'; term++) {
78 if (*term == '\\')
79 term++;
80 }
81 if (*term != '/')
82 return spec;
83
Thomas Rast12da1d12013-03-28 17:47:32 +010084 /* in the scan-only case we are not interested in the regex */
85 if (!ret)
86 return term+1;
87
Bo Yang25ed3412013-03-28 17:47:30 +010088 /* try [spec+1 .. term-1] as regexp */
89 *term = 0;
90 begin--; /* input is in human terms */
91 line = nth_line(data, begin);
92
93 if (!(reg_error = regcomp(&regexp, spec + 1, REG_NEWLINE)) &&
94 !(reg_error = regexec(&regexp, line, 1, match, 0))) {
95 const char *cp = line + match[0].rm_so;
96 const char *nline;
97
98 while (begin++ < lines) {
99 nline = nth_line(data, begin);
100 if (line <= cp && cp < nline)
101 break;
102 line = nline;
103 }
104 *ret = begin;
105 regfree(&regexp);
106 *term++ = '/';
107 return term;
108 }
109 else {
110 char errbuf[1024];
111 regerror(reg_error, &regexp, errbuf, 1024);
Eric Sunshine815834e2013-08-06 09:59:41 -0400112 die("-L parameter '%s' starting at line %ld: %s",
113 spec + 1, begin + 1, errbuf);
Bo Yang25ed3412013-03-28 17:47:30 +0100114 }
115}
116
Thomas Rast13b8f682013-03-28 17:47:33 +0100117static int match_funcname(xdemitconf_t *xecfg, const char *bol, const char *eol)
Bo Yang25ed3412013-03-28 17:47:30 +0100118{
Thomas Rast13b8f682013-03-28 17:47:33 +0100119 if (xecfg) {
120 char buf[1];
121 return xecfg->find_func(bol, eol - bol, buf, 1,
122 xecfg->find_func_priv) >= 0;
123 }
124
125 if (bol == eol)
126 return 0;
127 if (isalpha(*bol) || *bol == '_' || *bol == '$')
128 return 1;
129 return 0;
130}
131
132static const char *find_funcname_matching_regexp(xdemitconf_t *xecfg, const char *start,
133 regex_t *regexp)
134{
135 int reg_error;
136 regmatch_t match[1];
Lars Kellogg-Stedman4e57c882022-12-19 17:48:50 -0500137 while (*start) {
Thomas Rast13b8f682013-03-28 17:47:33 +0100138 const char *bol, *eol;
139 reg_error = regexec(regexp, start, 1, match, 0);
140 if (reg_error == REG_NOMATCH)
141 return NULL;
142 else if (reg_error) {
143 char errbuf[1024];
144 regerror(reg_error, regexp, errbuf, 1024);
145 die("-L parameter: regexec() failed: %s", errbuf);
146 }
147 /* determine extent of line matched */
148 bol = start+match[0].rm_so;
149 eol = start+match[0].rm_eo;
Lars Kellogg-Stedman4e57c882022-12-19 17:48:50 -0500150 while (bol > start && *--bol != '\n')
151 ; /* nothing */
Thomas Rast13b8f682013-03-28 17:47:33 +0100152 if (*bol == '\n')
153 bol++;
154 while (*eol && *eol != '\n')
155 eol++;
156 if (*eol == '\n')
157 eol++;
158 /* is it a funcname line? */
159 if (match_funcname(xecfg, (char*) bol, (char*) eol))
160 return bol;
161 start = eol;
162 }
Lars Kellogg-Stedman4e57c882022-12-19 17:48:50 -0500163 return NULL;
Thomas Rast13b8f682013-03-28 17:47:33 +0100164}
165
Nguyễn Thái Ngọc Duy80e03852018-09-21 17:57:34 +0200166static const char *parse_range_funcname(
167 const char *arg, nth_line_fn_t nth_line_cb,
168 void *cb_data, long lines, long anchor, long *begin, long *end,
169 const char *path, struct index_state *istate)
Thomas Rast13b8f682013-03-28 17:47:33 +0100170{
171 char *pattern;
172 const char *term;
173 struct userdiff_driver *drv;
174 xdemitconf_t *xecfg = NULL;
175 const char *start;
176 const char *p;
177 int reg_error;
178 regex_t regexp;
179
Eric Sunshine215e76c2013-08-06 09:59:47 -0400180 if (*arg == '^') {
181 anchor = 1;
182 arg++;
183 }
184
Thomas Rast13b8f682013-03-28 17:47:33 +0100185 assert(*arg == ':');
186 term = arg+1;
187 while (*term && *term != ':') {
188 if (*term == '\\' && *(term+1))
189 term++;
190 term++;
191 }
192 if (term == arg+1)
193 return NULL;
194 if (!begin) /* skip_range_arg case */
195 return term;
196
197 pattern = xstrndup(arg+1, term-(arg+1));
198
Eric Sunshine1ce761a2013-08-06 09:59:46 -0400199 anchor--; /* input is in human terms */
200 start = nth_line_cb(cb_data, anchor);
Thomas Rast13b8f682013-03-28 17:47:33 +0100201
Nguyễn Thái Ngọc Duy80e03852018-09-21 17:57:34 +0200202 drv = userdiff_find_by_path(istate, path);
Thomas Rast13b8f682013-03-28 17:47:33 +0100203 if (drv && drv->funcname.pattern) {
204 const struct userdiff_funcname *pe = &drv->funcname;
René Scharfeca56dad2021-03-13 17:17:22 +0100205 CALLOC_ARRAY(xecfg, 1);
Thomas Rast13b8f682013-03-28 17:47:33 +0100206 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
207 }
208
209 reg_error = regcomp(&regexp, pattern, REG_NEWLINE);
210 if (reg_error) {
211 char errbuf[1024];
212 regerror(reg_error, &regexp, errbuf, 1024);
213 die("-L parameter '%s': %s", pattern, errbuf);
214 }
215
216 p = find_funcname_matching_regexp(xecfg, (char*) start, &regexp);
217 if (!p)
Eric Sunshine1ce761a2013-08-06 09:59:46 -0400218 die("-L parameter '%s' starting at line %ld: no match",
219 pattern, anchor + 1);
Thomas Rast13b8f682013-03-28 17:47:33 +0100220 *begin = 0;
221 while (p > nth_line_cb(cb_data, *begin))
222 (*begin)++;
223
224 if (*begin >= lines)
225 die("-L parameter '%s' matches at EOF", pattern);
226
227 *end = *begin+1;
228 while (*end < lines) {
229 const char *bol = nth_line_cb(cb_data, *end);
230 const char *eol = nth_line_cb(cb_data, *end+1);
231 if (match_funcname(xecfg, bol, eol))
232 break;
233 (*end)++;
234 }
235
236 regfree(&regexp);
237 free(xecfg);
238 free(pattern);
239
240 /* compensate for 1-based numbering */
241 (*begin)++;
242
243 return term;
244}
245
246int parse_range_arg(const char *arg, nth_line_fn_t nth_line_cb,
Eric Sunshine815834e2013-08-06 09:59:41 -0400247 void *cb_data, long lines, long anchor,
Nguyễn Thái Ngọc Duy80e03852018-09-21 17:57:34 +0200248 long *begin, long *end,
249 const char *path, struct index_state *istate)
Thomas Rast13b8f682013-03-28 17:47:33 +0100250{
Eric Sunshine3bf65f92013-07-17 17:25:27 -0400251 *begin = *end = 0;
252
Eric Sunshine815834e2013-08-06 09:59:41 -0400253 if (anchor < 1)
254 anchor = 1;
255 if (anchor > lines)
256 anchor = lines + 1;
257
Eric Sunshine215e76c2013-08-06 09:59:47 -0400258 if (*arg == ':' || (*arg == '^' && *(arg + 1) == ':')) {
Nguyễn Thái Ngọc Duy80e03852018-09-21 17:57:34 +0200259 arg = parse_range_funcname(arg, nth_line_cb, cb_data,
260 lines, anchor, begin, end,
261 path, istate);
Thomas Rast13b8f682013-03-28 17:47:33 +0100262 if (!arg || *arg)
263 return -1;
264 return 0;
265 }
266
Eric Sunshine815834e2013-08-06 09:59:41 -0400267 arg = parse_loc(arg, nth_line_cb, cb_data, lines, -anchor, begin);
Bo Yang25ed3412013-03-28 17:47:30 +0100268
269 if (*arg == ',')
270 arg = parse_loc(arg + 1, nth_line_cb, cb_data, lines, *begin + 1, end);
271
272 if (*arg)
273 return -1;
274
Eric Sunshine3bf65f92013-07-17 17:25:27 -0400275 if (*begin && *end && *end < *begin) {
René Scharfe35d803b2017-01-28 22:40:58 +0100276 SWAP(*end, *begin);
Eric Sunshine3bf65f92013-07-17 17:25:27 -0400277 }
278
Bo Yang25ed3412013-03-28 17:47:30 +0100279 return 0;
280}
Thomas Rast12da1d12013-03-28 17:47:32 +0100281
Nguyễn Thái Ngọc Duy80e03852018-09-21 17:57:34 +0200282const char *skip_range_arg(const char *arg, struct index_state *istate)
Thomas Rast12da1d12013-03-28 17:47:32 +0100283{
Eric Sunshine215e76c2013-08-06 09:59:47 -0400284 if (*arg == ':' || (*arg == '^' && *(arg + 1) == ':'))
Nguyễn Thái Ngọc Duy80e03852018-09-21 17:57:34 +0200285 return parse_range_funcname(arg, NULL, NULL,
286 0, 0, NULL, NULL,
287 NULL, istate);
Thomas Rast13b8f682013-03-28 17:47:33 +0100288
Thomas Rast12da1d12013-03-28 17:47:32 +0100289 arg = parse_loc(arg, NULL, NULL, 0, -1, NULL);
290
291 if (*arg == ',')
292 arg = parse_loc(arg+1, NULL, NULL, 0, 0, NULL);
293
294 return arg;
295}