t5703: replace "grep -a" usage by perl
[git/debian.git] / line-range.c
blob9b50583dc05147fb6940c1d551ba274702704f07
1 #include "git-compat-util.h"
2 #include "line-range.h"
3 #include "xdiff-interface.h"
4 #include "strbuf.h"
5 #include "userdiff.h"
7 /*
8 * Parse one item in the -L option
10 * 'begin' is applicable only to relative range anchors. Absolute anchors
11 * ignore this value.
13 * When parsing "-L A,B", parse_loc() is called once for A and once for B.
15 * When parsing A, 'begin' must be a negative number, the absolute value of
16 * which is the line at which relative start-of-range anchors should be
17 * based. Beginning of file is represented by -1.
19 * When parsing B, 'begin' must be the positive line number immediately
20 * following the line computed for 'A'.
22 static const char *parse_loc(const char *spec, nth_line_fn_t nth_line,
23 void *data, long lines, long begin, long *ret)
25 char *term;
26 const char *line;
27 long num;
28 int reg_error;
29 regex_t regexp;
30 regmatch_t match[1];
32 /* Allow "-L <something>,+20" to mean starting at <something>
33 * for 20 lines, or "-L <something>,-5" for 5 lines ending at
34 * <something>.
36 if (1 <= begin && (spec[0] == '+' || spec[0] == '-')) {
37 num = strtol(spec + 1, &term, 10);
38 if (term != spec + 1) {
39 if (!ret)
40 return term;
41 if (num == 0)
42 die("-L invalid empty range");
43 if (spec[0] == '-')
44 num = 0 - num;
45 if (0 < num)
46 *ret = begin + num - 2;
47 else if (!num)
48 *ret = begin;
49 else
50 *ret = begin + num > 0 ? begin + num : 1;
51 return term;
53 return spec;
55 num = strtol(spec, &term, 10);
56 if (term != spec) {
57 if (ret) {
58 if (num <= 0)
59 die("-L invalid line number: %ld", num);
60 *ret = num;
62 return term;
65 if (begin < 0) {
66 if (spec[0] != '^')
67 begin = -begin;
68 else {
69 begin = 1;
70 spec++;
74 if (spec[0] != '/')
75 return spec;
77 /* it could be a regexp of form /.../ */
78 for (term = (char *) spec + 1; *term && *term != '/'; term++) {
79 if (*term == '\\')
80 term++;
82 if (*term != '/')
83 return spec;
85 /* in the scan-only case we are not interested in the regex */
86 if (!ret)
87 return term+1;
89 /* try [spec+1 .. term-1] as regexp */
90 *term = 0;
91 begin--; /* input is in human terms */
92 line = nth_line(data, begin);
94 if (!(reg_error = regcomp(&regexp, spec + 1, REG_NEWLINE)) &&
95 !(reg_error = regexec(&regexp, line, 1, match, 0))) {
96 const char *cp = line + match[0].rm_so;
97 const char *nline;
99 while (begin++ < lines) {
100 nline = nth_line(data, begin);
101 if (line <= cp && cp < nline)
102 break;
103 line = nline;
105 *ret = begin;
106 regfree(&regexp);
107 *term++ = '/';
108 return term;
110 else {
111 char errbuf[1024];
112 regerror(reg_error, &regexp, errbuf, 1024);
113 die("-L parameter '%s' starting at line %ld: %s",
114 spec + 1, begin + 1, errbuf);
118 static int match_funcname(xdemitconf_t *xecfg, const char *bol, const char *eol)
120 if (xecfg) {
121 char buf[1];
122 return xecfg->find_func(bol, eol - bol, buf, 1,
123 xecfg->find_func_priv) >= 0;
126 if (bol == eol)
127 return 0;
128 if (isalpha(*bol) || *bol == '_' || *bol == '$')
129 return 1;
130 return 0;
133 static const char *find_funcname_matching_regexp(xdemitconf_t *xecfg, const char *start,
134 regex_t *regexp)
136 int reg_error;
137 regmatch_t match[1];
138 while (1) {
139 const char *bol, *eol;
140 reg_error = regexec(regexp, start, 1, match, 0);
141 if (reg_error == REG_NOMATCH)
142 return NULL;
143 else if (reg_error) {
144 char errbuf[1024];
145 regerror(reg_error, regexp, errbuf, 1024);
146 die("-L parameter: regexec() failed: %s", errbuf);
148 /* determine extent of line matched */
149 bol = start+match[0].rm_so;
150 eol = start+match[0].rm_eo;
151 while (bol > start && *bol != '\n')
152 bol--;
153 if (*bol == '\n')
154 bol++;
155 while (*eol && *eol != '\n')
156 eol++;
157 if (*eol == '\n')
158 eol++;
159 /* is it a funcname line? */
160 if (match_funcname(xecfg, (char*) bol, (char*) eol))
161 return bol;
162 start = eol;
166 static 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)
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;
180 if (*arg == '^') {
181 anchor = 1;
182 arg++;
185 assert(*arg == ':');
186 term = arg+1;
187 while (*term && *term != ':') {
188 if (*term == '\\' && *(term+1))
189 term++;
190 term++;
192 if (term == arg+1)
193 return NULL;
194 if (!begin) /* skip_range_arg case */
195 return term;
197 pattern = xstrndup(arg+1, term-(arg+1));
199 anchor--; /* input is in human terms */
200 start = nth_line_cb(cb_data, anchor);
202 drv = userdiff_find_by_path(istate, path);
203 if (drv && drv->funcname.pattern) {
204 const struct userdiff_funcname *pe = &drv->funcname;
205 xecfg = xcalloc(1, sizeof(*xecfg));
206 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
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);
216 p = find_funcname_matching_regexp(xecfg, (char*) start, &regexp);
217 if (!p)
218 die("-L parameter '%s' starting at line %ld: no match",
219 pattern, anchor + 1);
220 *begin = 0;
221 while (p > nth_line_cb(cb_data, *begin))
222 (*begin)++;
224 if (*begin >= lines)
225 die("-L parameter '%s' matches at EOF", pattern);
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)++;
236 regfree(&regexp);
237 free(xecfg);
238 free(pattern);
240 /* compensate for 1-based numbering */
241 (*begin)++;
243 return term;
246 int parse_range_arg(const char *arg, nth_line_fn_t nth_line_cb,
247 void *cb_data, long lines, long anchor,
248 long *begin, long *end,
249 const char *path, struct index_state *istate)
251 *begin = *end = 0;
253 if (anchor < 1)
254 anchor = 1;
255 if (anchor > lines)
256 anchor = lines + 1;
258 if (*arg == ':' || (*arg == '^' && *(arg + 1) == ':')) {
259 arg = parse_range_funcname(arg, nth_line_cb, cb_data,
260 lines, anchor, begin, end,
261 path, istate);
262 if (!arg || *arg)
263 return -1;
264 return 0;
267 arg = parse_loc(arg, nth_line_cb, cb_data, lines, -anchor, begin);
269 if (*arg == ',')
270 arg = parse_loc(arg + 1, nth_line_cb, cb_data, lines, *begin + 1, end);
272 if (*arg)
273 return -1;
275 if (*begin && *end && *end < *begin) {
276 SWAP(*end, *begin);
279 return 0;
282 const char *skip_range_arg(const char *arg, struct index_state *istate)
284 if (*arg == ':' || (*arg == '^' && *(arg + 1) == ':'))
285 return parse_range_funcname(arg, NULL, NULL,
286 0, 0, NULL, NULL,
287 NULL, istate);
289 arg = parse_loc(arg, NULL, NULL, 0, -1, NULL);
291 if (*arg == ',')
292 arg = parse_loc(arg+1, NULL, NULL, 0, 0, NULL);
294 return arg;