line-range.c

   1 #include "git-compat-util.h"
   2 #include "line-range.h"
   3 #include "xdiff-interface.h"
   4 #include "userdiff.h"
   5
   6 /*
   7  * Parse one item in the -L option
   8  *
   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'.
  20  */
  21 static 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          */
  35         if (1 <= begin && (spec[0] == '+' || spec[0] == '-')) {
  36                 num = strtol(spec + 1, &term, 10);
  37                 if (term != spec + 1) {
  38                         if (!ret)
  39                                 return term;
  40                         if (num == 0)
  41                                 die("-L invalid empty range");
  42                         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
  49                                 *ret = begin + num > 0 ? begin + num : 1;
  50                         return term;
  51                 }
  52                 return spec;
  53         }
  54         num = strtol(spec, &term, 10);
  55         if (term != spec) {
  56                 if (ret) {
  57                         if (num <= 0)
  58                                 die("-L invalid line number: %ld", num);
  59                         *ret = num;
  60                 }
  61                 return term;
  62         }
  63
  64         if (begin < 0) {
  65                 if (spec[0] != '^')
  66                         begin = -begin;
  67                 else {
  68                         begin = 1;
  69                         spec++;
  70                 }
  71         }
  72
  73         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
  84         /* in the scan-only case we are not interested in the regex */
  85         if (!ret)
  86                 return term+1;
  87
  88         /* 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);
 112                 die("-L parameter '%s' starting at line %ld: %s",
 113                     spec + 1, begin + 1, errbuf);
 114         }
 115 }
 116
 117 static int match_funcname(xdemitconf_t *xecfg, const char *bol, const char *eol)
 118 {
 119         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
 132 static 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];
 137         while (*start) {
 138                 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;
 150                 while (bol > start && *--bol != '\n')
 151                         ; /* nothing */
 152                 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         }
 163         return NULL;
 164 }
 165
 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)
 170 {
 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
 180         if (*arg == '^') {
 181                 anchor = 1;
 182                 arg++;
 183         }
 184
 185         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
 199         anchor--; /* input is in human terms */
 200         start = nth_line_cb(cb_data, anchor);
 201
 202         drv = userdiff_find_by_path(istate, path);
 203         if (drv && drv->funcname.pattern) {
 204                 const struct userdiff_funcname *pe = &drv->funcname;
 205                 CALLOC_ARRAY(xecfg, 1);
 206                 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)
 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)++;
 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         if (xecfg)
 238                 xdiff_clear_find_func(xecfg);
 239         free(xecfg);
 240         free(pattern);
 241
 242         /* compensate for 1-based numbering */
 243         (*begin)++;
 244
 245         return term;
 246 }
 247
 248 int parse_range_arg(const char *arg, nth_line_fn_t nth_line_cb,
 249                     void *cb_data, long lines, long anchor,
 250                     long *begin, long *end,
 251                     const char *path, struct index_state *istate)
 252 {
 253         *begin = *end = 0;
 254
 255         if (anchor < 1)
 256                 anchor = 1;
 257         if (anchor > lines)
 258                 anchor = lines + 1;
 259
 260         if (*arg == ':' || (*arg == '^' && *(arg + 1) == ':')) {
 261                 arg = parse_range_funcname(arg, nth_line_cb, cb_data,
 262                                            lines, anchor, begin, end,
 263                                            path, istate);
 264                 if (!arg || *arg)
 265                         return -1;
 266                 return 0;
 267         }
 268
 269         arg = parse_loc(arg, nth_line_cb, cb_data, lines, -anchor, begin);
 270
 271         if (*arg == ',')
 272                 arg = parse_loc(arg + 1, nth_line_cb, cb_data, lines, *begin + 1, end);
 273
 274         if (*arg)
 275                 return -1;
 276
 277         if (*begin && *end && *end < *begin) {
 278                 SWAP(*end, *begin);
 279         }
 280
 281         return 0;
 282 }
 283
 284 const char *skip_range_arg(const char *arg, struct index_state *istate)
 285 {
 286         if (*arg == ':' || (*arg == '^' && *(arg + 1) == ':'))
 287                 return parse_range_funcname(arg, NULL, NULL,
 288                                             0, 0, NULL, NULL,
 289                                             NULL, istate);
 290
 291         arg = parse_loc(arg, NULL, NULL, 0, -1, NULL);
 292
 293         if (*arg == ',')
 294                 arg = parse_loc(arg+1, NULL, NULL, 0, 0, NULL);
 295
 296         return arg;
 297 }