Andrew's git - gitweb.git/blob

   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 */
  22static 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                if (spec[0] != '^')
  64                        begin = -begin;
  65                else {
  66                        begin = 1;
  67                        spec++;
  68                }
  69        }
  70
  71        if (spec[0] != '/')
  72                return spec;
  73
  74        /* it could be a regexp of form /.../ */
  75        for (term = (char *) spec + 1; *term && *term != '/'; term++) {
  76                if (*term == '\\')
  77                        term++;
  78        }
  79        if (*term != '/')
  80                return spec;
  81
  82        /* in the scan-only case we are not interested in the regex */
  83        if (!ret)
  84                return term+1;
  85
  86        /* try [spec+1 .. term-1] as regexp */
  87        *term = 0;
  88        begin--; /* input is in human terms */
  89        line = nth_line(data, begin);
  90
  91        if (!(reg_error = regcomp(&regexp, spec + 1, REG_NEWLINE)) &&
  92            !(reg_error = regexec(&regexp, line, 1, match, 0))) {
  93                const char *cp = line + match[0].rm_so;
  94                const char *nline;
  95
  96                while (begin++ < lines) {
  97                        nline = nth_line(data, begin);
  98                        if (line <= cp && cp < nline)
  99                                break;
 100                        line = nline;
 101                }
 102                *ret = begin;
 103                regfree(&regexp);
 104                *term++ = '/';
 105                return term;
 106        }
 107        else {
 108                char errbuf[1024];
 109                regerror(reg_error, &regexp, errbuf, 1024);
 110                die("-L parameter '%s' starting at line %ld: %s",
 111                    spec + 1, begin + 1, errbuf);
 112        }
 113}
 114
 115static int match_funcname(xdemitconf_t *xecfg, const char *bol, const char *eol)
 116{
 117        if (xecfg) {
 118                char buf[1];
 119                return xecfg->find_func(bol, eol - bol, buf, 1,
 120                                        xecfg->find_func_priv) >= 0;
 121        }
 122
 123        if (bol == eol)
 124                return 0;
 125        if (isalpha(*bol) || *bol == '_' || *bol == '$')
 126                return 1;
 127        return 0;
 128}
 129
 130static const char *find_funcname_matching_regexp(xdemitconf_t *xecfg, const char *start,
 131                                                 regex_t *regexp)
 132{
 133        int reg_error;
 134        regmatch_t match[1];
 135        while (1) {
 136                const char *bol, *eol;
 137                reg_error = regexec(regexp, start, 1, match, 0);
 138                if (reg_error == REG_NOMATCH)
 139                        return NULL;
 140                else if (reg_error) {
 141                        char errbuf[1024];
 142                        regerror(reg_error, regexp, errbuf, 1024);
 143                        die("-L parameter: regexec() failed: %s", errbuf);
 144                }
 145                /* determine extent of line matched */
 146                bol = start+match[0].rm_so;
 147                eol = start+match[0].rm_eo;
 148                while (bol > start && *bol != '\n')
 149                        bol--;
 150                if (*bol == '\n')
 151                        bol++;
 152                while (*eol && *eol != '\n')
 153                        eol++;
 154                if (*eol == '\n')
 155                        eol++;
 156                /* is it a funcname line? */
 157                if (match_funcname(xecfg, (char*) bol, (char*) eol))
 158                        return bol;
 159                start = eol;
 160        }
 161}
 162
 163static const char *parse_range_funcname(const char *arg, nth_line_fn_t nth_line_cb,
 164                                        void *cb_data, long lines, long *begin, long *end,
 165                                        const char *path)
 166{
 167        char *pattern;
 168        const char *term;
 169        struct userdiff_driver *drv;
 170        xdemitconf_t *xecfg = NULL;
 171        const char *start;
 172        const char *p;
 173        int reg_error;
 174        regex_t regexp;
 175
 176        assert(*arg == ':');
 177        term = arg+1;
 178        while (*term && *term != ':') {
 179                if (*term == '\\' && *(term+1))
 180                        term++;
 181                term++;
 182        }
 183        if (term == arg+1)
 184                return NULL;
 185        if (!begin) /* skip_range_arg case */
 186                return term;
 187
 188        pattern = xstrndup(arg+1, term-(arg+1));
 189
 190        start = nth_line_cb(cb_data, 0);
 191
 192        drv = userdiff_find_by_path(path);
 193        if (drv && drv->funcname.pattern) {
 194                const struct userdiff_funcname *pe = &drv->funcname;
 195                xecfg = xcalloc(1, sizeof(*xecfg));
 196                xdiff_set_find_func(xecfg, pe->pattern, pe->cflags);
 197        }
 198
 199        reg_error = regcomp(&regexp, pattern, REG_NEWLINE);
 200        if (reg_error) {
 201                char errbuf[1024];
 202                regerror(reg_error, &regexp, errbuf, 1024);
 203                die("-L parameter '%s': %s", pattern, errbuf);
 204        }
 205
 206        p = find_funcname_matching_regexp(xecfg, (char*) start, &regexp);
 207        if (!p)
 208                die("-L parameter '%s': no match", pattern);
 209        *begin = 0;
 210        while (p > nth_line_cb(cb_data, *begin))
 211                (*begin)++;
 212
 213        if (*begin >= lines)
 214                die("-L parameter '%s' matches at EOF", pattern);
 215
 216        *end = *begin+1;
 217        while (*end < lines) {
 218                const char *bol = nth_line_cb(cb_data, *end);
 219                const char *eol = nth_line_cb(cb_data, *end+1);
 220                if (match_funcname(xecfg, bol, eol))
 221                        break;
 222                (*end)++;
 223        }
 224
 225        regfree(&regexp);
 226        free(xecfg);
 227        free(pattern);
 228
 229        /* compensate for 1-based numbering */
 230        (*begin)++;
 231
 232        return term;
 233}
 234
 235int parse_range_arg(const char *arg, nth_line_fn_t nth_line_cb,
 236                    void *cb_data, long lines, long anchor,
 237                    long *begin, long *end, const char *path)
 238{
 239        *begin = *end = 0;
 240
 241        if (anchor < 1)
 242                anchor = 1;
 243        if (anchor > lines)
 244                anchor = lines + 1;
 245
 246        if (*arg == ':') {
 247                arg = parse_range_funcname(arg, nth_line_cb, cb_data, lines, begin, end, path);
 248                if (!arg || *arg)
 249                        return -1;
 250                return 0;
 251        }
 252
 253        arg = parse_loc(arg, nth_line_cb, cb_data, lines, -anchor, begin);
 254
 255        if (*arg == ',')
 256                arg = parse_loc(arg + 1, nth_line_cb, cb_data, lines, *begin + 1, end);
 257
 258        if (*arg)
 259                return -1;
 260
 261        if (*begin && *end && *end < *begin) {
 262                long tmp;
 263                tmp = *end; *end = *begin; *begin = tmp;
 264        }
 265
 266        return 0;
 267}
 268
 269const char *skip_range_arg(const char *arg)
 270{
 271        if (*arg == ':')
 272                return parse_range_funcname(arg, NULL, NULL, 0, NULL, NULL, NULL);
 273
 274        arg = parse_loc(arg, NULL, NULL, 0, -1, NULL);
 275
 276        if (*arg == ',')
 277                arg = parse_loc(arg+1, NULL, NULL, 0, 0, NULL);
 278
 279        return arg;
 280}