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  * 'begin' is applicable only to relative range anchors. Absolute anchors
  11  * ignore this value.
  12  *
  13  * When parsing "-L A,B", parse_loc() is called once for A and once for B.
  14  *
  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.
  18  *
  19  * When parsing B, 'begin' must be the positive line number immediately
  20  * following the line computed for 'A'.
  21  */
  22 static const char *parse_loc(const char *spec, nth_line_fn_t nth_line,
  23                              void *data, long lines, long begin, long *ret)
  24 {
  25         char *term;
  26         const char *line;
  27         long num;
  28         int reg_error;
  29         regex_t regexp;
  30         regmatch_t match[1];
  31
  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>.
  35          */
  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;
  51                         return term;
  52                 }
  53                 return spec;
  54         }
  55         num = strtol(spec, &term, 10);
  56         if (term != spec) {
  57                 if (ret)
  58                         *ret = num;
  59                 return term;
  60         }
  61
  62         if (begin < 0)
  63                 begin = -begin;
  64
  65         if (spec[0] != '/')
  66                 return spec;
  67
  68         /* it could be a regexp of form /.../ */
  69         for (term = (char *) spec + 1; *term && *term != '/'; term++) {
  70                 if (*term == '\\')
  71                         term++;
  72         }
  73         if (*term != '/')
  74                 return spec;
  75
  76         /* in the scan-only case we are not interested in the regex */
  77         if (!ret)
  78                 return term+1;
  79
  80         /* try [spec+1 .. term-1] as regexp */
  81         *term = 0;
  82         begin--; /* input is in human terms */
  83         line = nth_line(data, begin);
  84
  85         if (!(reg_error = regcomp(&regexp, spec + 1, REG_NEWLINE)) &&
  86             !(reg_error = regexec(&regexp, line, 1, match, 0))) {
  87                 const char *cp = line + match[0].rm_so;
  88                 const char *nline;
  89
  90                 while (begin++ < lines) {
  91                         nline = nth_line(data, begin);
  92                         if (line <= cp && cp < nline)
  93                                 break;
  94                         line = nline;
  95                 }
  96                 *ret = begin;
  97                 regfree(&regexp);
  98                 *term++ = '/';
  99                 return term;
 100         }
 101         else {
 102                 char errbuf[1024];
 103                 regerror(reg_error, &regexp, errbuf, 1024);
 104                 die("-L parameter '%s' starting at line %ld: %s",
 105                     spec + 1, begin + 1, errbuf);
 106         }
 107 }
 108
 109 static int match_funcname(xdemitconf_t *xecfg, const char *bol, const char *eol)
 110 {
 111         if (xecfg) {
 112                 char buf[1];
 113                 return xecfg->find_func(bol, eol - bol, buf, 1,
 114                                         xecfg->find_func_priv) >= 0;
 115         }
 116
 117         if (bol == eol)
 118                 return 0;
 119         if (isalpha(*bol) || *bol == '_' || *bol == '$')
 120                 return 1;
 121         return 0;
 122 }
 123
 124 static const char *find_funcname_matching_regexp(xdemitconf_t *xecfg, const char *start,
 125                                                  regex_t *regexp)
 126 {
 127         int reg_error;
 128         regmatch_t match[1];
 129         while (1) {
 130                 const char *bol, *eol;
 131                 reg_error = regexec(regexp, start, 1, match, 0);
 132                 if (reg_error == REG_NOMATCH)
 133                         return NULL;
 134                 else if (reg_error) {
 135                         char errbuf[1024];
 136                         regerror(reg_error, regexp, errbuf, 1024);
 137                         die("-L parameter: regexec() failed: %s", errbuf);
 138                 }
 139                 /* determine extent of line matched */
 140                 bol = start+match[0].rm_so;
 141                 eol = start+match[0].rm_eo;
 142                 while (bol > start && *bol != '\n')
 143                         bol--;
 144                 if (*bol == '\n')
 145                         bol++;
 146                 while (*eol && *eol != '\n')
 147                         eol++;
 148                 if (*eol == '\n')
 149                         eol++;
 150                 /* is it a funcname line? */
 151                 if (match_funcname(xecfg, (char*) bol, (char*) eol))
 152                         return bol;
 153                 start = eol;
 154         }
 155 }
 156
 157 static const char *parse_range_funcname(const char *arg, nth_line_fn_t nth_line_cb,
 158                                         void *cb_data, long lines, long *begin, long *end,
 159                                         const char *path)
 160 {
 161         char *pattern;
 162         const char *term;
 163         struct userdiff_driver *drv;
 164         xdemitconf_t *xecfg = NULL;
 165         const char *start;
 166         const char *p;
 167         int reg_error;
 168         regex_t regexp;
 169
 170         assert(*arg == ':');
 171         term = arg+1;
 172         while (*term && *term != ':') {
 173                 if (*term == '\\' && *(term+1))
 174                         term++;
 175                 term++;
 176         }
 177         if (term == arg+1)
 178                 return NULL;
 179         if (!begin) /* skip_range_arg case */
 180                 return term;
 181
 182         pattern = xstrndup(arg+1, term-(arg+1));
 183
 184         start = nth_line_cb(cb_data, 0);
 185
 186         drv = userdiff_find_by_path(path);
 187         if (drv && drv->funcname.pattern) {
 188                 const struct userdiff_funcname *pe = &drv->funcname;
 189                 xecfg = xcalloc(1, sizeof(*xecfg));
 190                 xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
 191         }
 192
 193         reg_error = regcomp(&regexp, pattern, REG_NEWLINE);
 194         if (reg_error) {
 195                 char errbuf[1024];
 196                 regerror(reg_error, &regexp, errbuf, 1024);
 197                 die("-L parameter '%s': %s", pattern, errbuf);
 198         }
 199
 200         p = find_funcname_matching_regexp(xecfg, (char*) start, &regexp);
 201         if (!p)
 202                 die("-L parameter '%s': no match", pattern);
 203         *begin = 0;
 204         while (p > nth_line_cb(cb_data, *begin))
 205                 (*begin)++;
 206
 207         if (*begin >= lines)
 208                 die("-L parameter '%s' matches at EOF", pattern);
 209
 210         *end = *begin+1;
 211         while (*end < lines) {
 212                 const char *bol = nth_line_cb(cb_data, *end);
 213                 const char *eol = nth_line_cb(cb_data, *end+1);
 214                 if (match_funcname(xecfg, bol, eol))
 215                         break;
 216                 (*end)++;
 217         }
 218
 219         regfree(&regexp);
 220         free(xecfg);
 221         free(pattern);
 222
 223         /* compensate for 1-based numbering */
 224         (*begin)++;
 225
 226         return term;
 227 }
 228
 229 int parse_range_arg(const char *arg, nth_line_fn_t nth_line_cb,
 230                     void *cb_data, long lines, long anchor,
 231                     long *begin, long *end, const char *path)
 232 {
 233         *begin = *end = 0;
 234
 235         if (anchor < 1)
 236                 anchor = 1;
 237         if (anchor > lines)
 238                 anchor = lines + 1;
 239
 240         if (*arg == ':') {
 241                 arg = parse_range_funcname(arg, nth_line_cb, cb_data, lines, begin, end, path);
 242                 if (!arg || *arg)
 243                         return -1;
 244                 return 0;
 245         }
 246
 247         arg = parse_loc(arg, nth_line_cb, cb_data, lines, -anchor, begin);
 248
 249         if (*arg == ',')
 250                 arg = parse_loc(arg + 1, nth_line_cb, cb_data, lines, *begin + 1, end);
 251
 252         if (*arg)
 253                 return -1;
 254
 255         if (*begin && *end && *end < *begin) {
 256                 long tmp;
 257                 tmp = *end; *end = *begin; *begin = tmp;
 258         }
 259
 260         return 0;
 261 }
 262
 263 const char *skip_range_arg(const char *arg)
 264 {
 265         if (*arg == ':')
 266                 return parse_range_funcname(arg, NULL, NULL, 0, NULL, NULL, NULL);
 267
 268         arg = parse_loc(arg, NULL, NULL, 0, -1, NULL);
 269
 270         if (*arg == ',')
 271                 arg = parse_loc(arg+1, NULL, NULL, 0, 0, NULL);
 272
 273         return arg;
 274 }