line-range.c

   1 #include "git-compat-util.h"
   2 #include "line-range.h"
   3 #include "xdiff-interface.h"
   4 #include "strbuf.h"
   5 #include "userdiff.h"
   6
   7 /*
   8  * Parse one item in the -L option
   9  */
  10 static const char *parse_loc(const char *spec, nth_line_fn_t nth_line,
  11                              void *data, long lines, long begin, long *ret)
  12 {
  13         char *term;
  14         const char *line;
  15         long num;
  16         int reg_error;
  17         regex_t regexp;
  18         regmatch_t match[1];
  19
  20         /* Allow "-L <something>,+20" to mean starting at <something>
  21          * for 20 lines, or "-L <something>,-5" for 5 lines ending at
  22          * <something>.
  23          */
  24         if (1 <= begin && (spec[0] == '+' || spec[0] == '-')) {
  25                 num = strtol(spec + 1, &term, 10);
  26                 if (term != spec + 1) {
  27                         if (!ret)
  28                                 return term;
  29                         if (num == 0)
  30                                 die("-L invalid empty range");
  31                         if (spec[0] == '-')
  32                                 num = 0 - num;
  33                         if (0 < num)
  34                                 *ret = begin + num - 2;
  35                         else if (!num)
  36                                 *ret = begin;
  37                         else
  38                                 *ret = begin + num;
  39                         return term;
  40                 }
  41                 return spec;
  42         }
  43         num = strtol(spec, &term, 10);
  44         if (term != spec) {
  45                 if (ret)
  46                         *ret = num;
  47                 return term;
  48         }
  49         if (spec[0] != '/')
  50                 return spec;
  51
  52         /* it could be a regexp of form /.../ */
  53         for (term = (char *) spec + 1; *term && *term != '/'; term++) {
  54                 if (*term == '\\')
  55                         term++;
  56         }
  57         if (*term != '/')
  58                 return spec;
  59
  60         /* in the scan-only case we are not interested in the regex */
  61         if (!ret)
  62                 return term+1;
  63
  64         /* try [spec+1 .. term-1] as regexp */
  65         *term = 0;
  66         begin--; /* input is in human terms */
  67         line = nth_line(data, begin);
  68
  69         if (!(reg_error = regcomp(&regexp, spec + 1, REG_NEWLINE)) &&
  70             !(reg_error = regexec(&regexp, line, 1, match, 0))) {
  71                 const char *cp = line + match[0].rm_so;
  72                 const char *nline;
  73
  74                 while (begin++ < lines) {
  75                         nline = nth_line(data, begin);
  76                         if (line <= cp && cp < nline)
  77                                 break;
  78                         line = nline;
  79                 }
  80                 *ret = begin;
  81                 regfree(&regexp);
  82                 *term++ = '/';
  83                 return term;
  84         }
  85         else {
  86                 char errbuf[1024];
  87                 regerror(reg_error, &regexp, errbuf, 1024);
  88                 die("-L parameter '%s': %s", spec + 1, errbuf);
  89         }
  90 }
  91
  92 static int match_funcname(xdemitconf_t *xecfg, const char *bol, const char *eol)
  93 {
  94         if (xecfg) {
  95                 char buf[1];
  96                 return xecfg->find_func(bol, eol - bol, buf, 1,
  97                                         xecfg->find_func_priv) >= 0;
  98         }
  99
 100         if (bol == eol)
 101                 return 0;
 102         if (isalpha(*bol) || *bol == '_' || *bol == '$')
 103                 return 1;
 104         return 0;
 105 }
 106
 107 static const char *find_funcname_matching_regexp(xdemitconf_t *xecfg, const char *start,
 108                                                  regex_t *regexp)
 109 {
 110         int reg_error;
 111         regmatch_t match[1];
 112         while (1) {
 113                 const char *bol, *eol;
 114                 reg_error = regexec(regexp, start, 1, match, 0);
 115                 if (reg_error == REG_NOMATCH)
 116                         return NULL;
 117                 else if (reg_error) {
 118                         char errbuf[1024];
 119                         regerror(reg_error, regexp, errbuf, 1024);
 120                         die("-L parameter: regexec() failed: %s", errbuf);
 121                 }
 122                 /* determine extent of line matched */
 123                 bol = start+match[0].rm_so;
 124                 eol = start+match[0].rm_eo;
 125                 while (bol > start && *bol != '\n')
 126                         bol--;
 127                 if (*bol == '\n')
 128                         bol++;
 129                 while (*eol && *eol != '\n')
 130                         eol++;
 131                 if (*eol == '\n')
 132                         eol++;
 133                 /* is it a funcname line? */
 134                 if (match_funcname(xecfg, (char*) bol, (char*) eol))
 135                         return bol;
 136                 start = eol;
 137         }
 138 }
 139
 140 static const char *parse_range_funcname(const char *arg, nth_line_fn_t nth_line_cb,
 141                                         void *cb_data, long lines, long *begin, long *end,
 142                                         const char *path)
 143 {
 144         char *pattern;
 145         const char *term;
 146         struct userdiff_driver *drv;
 147         xdemitconf_t *xecfg = NULL;
 148         const char *start;
 149         const char *p;
 150         int reg_error;
 151         regex_t regexp;
 152
 153         assert(*arg == ':');
 154         term = arg+1;
 155         while (*term && *term != ':') {
 156                 if (*term == '\\' && *(term+1))
 157                         term++;
 158                 term++;
 159         }
 160         if (term == arg+1)
 161                 return NULL;
 162         if (!begin) /* skip_range_arg case */
 163                 return term;
 164
 165         pattern = xstrndup(arg+1, term-(arg+1));
 166
 167         start = nth_line_cb(cb_data, 0);
 168
 169         drv = userdiff_find_by_path(path);
 170         if (drv && drv->funcname.pattern) {
 171                 const struct userdiff_funcname *pe = &drv->funcname;
 172                 xecfg = xcalloc(1, sizeof(*xecfg));
 173                 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
 174         }
 175
 176         reg_error = regcomp(&regexp, pattern, REG_NEWLINE);
 177         if (reg_error) {
 178                 char errbuf[1024];
 179                 regerror(reg_error, &regexp, errbuf, 1024);
 180                 die("-L parameter '%s': %s", pattern, errbuf);
 181         }
 182
 183         p = find_funcname_matching_regexp(xecfg, (char*) start, &regexp);
 184         if (!p)
 185                 die("-L parameter '%s': no match", pattern);
 186         *begin = 0;
 187         while (p > nth_line_cb(cb_data, *begin))
 188                 (*begin)++;
 189
 190         if (*begin >= lines)
 191                 die("-L parameter '%s' matches at EOF", pattern);
 192
 193         *end = *begin+1;
 194         while (*end < lines) {
 195                 const char *bol = nth_line_cb(cb_data, *end);
 196                 const char *eol = nth_line_cb(cb_data, *end+1);
 197                 if (match_funcname(xecfg, bol, eol))
 198                         break;
 199                 (*end)++;
 200         }
 201
 202         regfree(&regexp);
 203         free(xecfg);
 204         free(pattern);
 205
 206         /* compensate for 1-based numbering */
 207         (*begin)++;
 208
 209         return term;
 210 }
 211
 212 int parse_range_arg(const char *arg, nth_line_fn_t nth_line_cb,
 213                     void *cb_data, long lines, long *begin, long *end,
 214                     const char *path)
 215 {
 216         *begin = *end = 0;
 217
 218         if (*arg == ':') {
 219                 arg = parse_range_funcname(arg, nth_line_cb, cb_data, lines, begin, end, path);
 220                 if (!arg || *arg)
 221                         return -1;
 222                 return 0;
 223         }
 224
 225         arg = parse_loc(arg, nth_line_cb, cb_data, lines, 1, begin);
 226
 227         if (*arg == ',')
 228                 arg = parse_loc(arg + 1, nth_line_cb, cb_data, lines, *begin + 1, end);
 229
 230         if (*arg)
 231                 return -1;
 232
 233         if (*begin && *end && *end < *begin) {
 234                 long tmp;
 235                 tmp = *end; *end = *begin; *begin = tmp;
 236         }
 237
 238         return 0;
 239 }
 240
 241 const char *skip_range_arg(const char *arg)
 242 {
 243         if (*arg == ':')
 244                 return parse_range_funcname(arg, NULL, NULL, 0, NULL, NULL, NULL);
 245
 246         arg = parse_loc(arg, NULL, NULL, 0, -1, NULL);
 247
 248         if (*arg == ',')
 249                 arg = parse_loc(arg+1, NULL, NULL, 0, 0, NULL);
 250
 251         return arg;
 252 }