ws.c

   1 /*
   2  * Whitespace rules
   3  *
   4  * Copyright (c) 2007 Junio C Hamano
   5  */
   6
   7 #include "cache.h"
   8 #include "attr.h"
   9
  10 static struct whitespace_rule {
  11         const char *rule_name;
  12         unsigned rule_bits;
  13 } whitespace_rule_names[] = {
  14         { "trailing-space", WS_TRAILING_SPACE },
  15         { "space-before-tab", WS_SPACE_BEFORE_TAB },
  16         { "indent-with-non-tab", WS_INDENT_WITH_NON_TAB },
  17         { "cr-at-eol", WS_CR_AT_EOL },
  18 };
  19
  20 unsigned parse_whitespace_rule(const char *string)
  21 {
  22         unsigned rule = WS_DEFAULT_RULE;
  23
  24         while (string) {
  25                 int i;
  26                 size_t len;
  27                 const char *ep;
  28                 int negated = 0;
  29
  30                 string = string + strspn(string, ", \t\n\r");
  31                 ep = strchr(string, ',');
  32                 if (!ep)
  33                         len = strlen(string);
  34                 else
  35                         len = ep - string;
  36
  37                 if (*string == '-') {
  38                         negated = 1;
  39                         string++;
  40                         len--;
  41                 }
  42                 if (!len)
  43                         break;
  44                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
  45                         if (strncmp(whitespace_rule_names[i].rule_name,
  46                                     string, len))
  47                                 continue;
  48                         if (negated)
  49                                 rule &= ~whitespace_rule_names[i].rule_bits;
  50                         else
  51                                 rule |= whitespace_rule_names[i].rule_bits;
  52                         break;
  53                 }
  54                 string = ep;
  55         }
  56         return rule;
  57 }
  58
  59 static void setup_whitespace_attr_check(struct git_attr_check *check)
  60 {
  61         static struct git_attr *attr_whitespace;
  62
  63         if (!attr_whitespace)
  64                 attr_whitespace = git_attr("whitespace", 10);
  65         check[0].attr = attr_whitespace;
  66 }
  67
  68 unsigned whitespace_rule(const char *pathname)
  69 {
  70         struct git_attr_check attr_whitespace_rule;
  71
  72         setup_whitespace_attr_check(&attr_whitespace_rule);
  73         if (!git_checkattr(pathname, 1, &attr_whitespace_rule)) {
  74                 const char *value;
  75
  76                 value = attr_whitespace_rule.value;
  77                 if (ATTR_TRUE(value)) {
  78                         /* true (whitespace) */
  79                         unsigned all_rule = 0;
  80                         int i;
  81                         for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  82                                 all_rule |= whitespace_rule_names[i].rule_bits;
  83                         return all_rule;
  84                 } else if (ATTR_FALSE(value)) {
  85                         /* false (-whitespace) */
  86                         return 0;
  87                 } else if (ATTR_UNSET(value)) {
  88                         /* reset to default (!whitespace) */
  89                         return whitespace_rule_cfg;
  90                 } else {
  91                         /* string */
  92                         return parse_whitespace_rule(value);
  93                 }
  94         } else {
  95                 return whitespace_rule_cfg;
  96         }
  97 }
  98
  99 /* The returned string should be freed by the caller. */
 100 char *whitespace_error_string(unsigned ws)
 101 {
 102         struct strbuf err;
 103         strbuf_init(&err, 0);
 104         if (ws & WS_TRAILING_SPACE)
 105                 strbuf_addstr(&err, "trailing whitespace");
 106         if (ws & WS_SPACE_BEFORE_TAB) {
 107                 if (err.len)
 108                         strbuf_addstr(&err, ", ");
 109                 strbuf_addstr(&err, "space before tab in indent");
 110         }
 111         if (ws & WS_INDENT_WITH_NON_TAB) {
 112                 if (err.len)
 113                         strbuf_addstr(&err, ", ");
 114                 strbuf_addstr(&err, "indent with spaces");
 115         }
 116         return strbuf_detach(&err, NULL);
 117 }
 118
 119 /* If stream is non-NULL, emits the line after checking. */
 120 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 121                                 FILE *stream, const char *set,
 122                                 const char *reset, const char *ws)
 123 {
 124         unsigned result = 0;
 125         int written = 0;
 126         int trailing_whitespace = -1;
 127         int trailing_newline = 0;
 128         int trailing_carriage_return = 0;
 129         int i;
 130
 131         /* Logic is simpler if we temporarily ignore the trailing newline. */
 132         if (len > 0 && line[len - 1] == '\n') {
 133                 trailing_newline = 1;
 134                 len--;
 135         }
 136         if ((ws_rule & WS_CR_AT_EOL) &&
 137             len > 0 && line[len - 1] == '\r') {
 138                 trailing_carriage_return = 1;
 139                 len--;
 140         }
 141
 142         /* Check for trailing whitespace. */
 143         if (ws_rule & WS_TRAILING_SPACE) {
 144                 for (i = len - 1; i >= 0; i--) {
 145                         if (isspace(line[i])) {
 146                                 trailing_whitespace = i;
 147                                 result |= WS_TRAILING_SPACE;
 148                         }
 149                         else
 150                                 break;
 151                 }
 152         }
 153
 154         /* Check for space before tab in initial indent. */
 155         for (i = 0; i < len; i++) {
 156                 if (line[i] == ' ')
 157                         continue;
 158                 if (line[i] != '\t')
 159                         break;
 160                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 161                         result |= WS_SPACE_BEFORE_TAB;
 162                         if (stream) {
 163                                 fputs(ws, stream);
 164                                 fwrite(line + written, i - written, 1, stream);
 165                                 fputs(reset, stream);
 166                         }
 167                 } else if (stream)
 168                         fwrite(line + written, i - written, 1, stream);
 169                 if (stream)
 170                         fwrite(line + i, 1, 1, stream);
 171                 written = i + 1;
 172         }
 173
 174         /* Check for indent using non-tab. */
 175         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= 8) {
 176                 result |= WS_INDENT_WITH_NON_TAB;
 177                 if (stream) {
 178                         fputs(ws, stream);
 179                         fwrite(line + written, i - written, 1, stream);
 180                         fputs(reset, stream);
 181                 }
 182                 written = i;
 183         }
 184
 185         if (stream) {
 186                 /*
 187                  * Now the rest of the line starts at "written".
 188                  * The non-highlighted part ends at "trailing_whitespace".
 189                  */
 190                 if (trailing_whitespace == -1)
 191                         trailing_whitespace = len;
 192
 193                 /* Emit non-highlighted (middle) segment. */
 194                 if (trailing_whitespace - written > 0) {
 195                         fputs(set, stream);
 196                         fwrite(line + written,
 197                             trailing_whitespace - written, 1, stream);
 198                         fputs(reset, stream);
 199                 }
 200
 201                 /* Highlight errors in trailing whitespace. */
 202                 if (trailing_whitespace != len) {
 203                         fputs(ws, stream);
 204                         fwrite(line + trailing_whitespace,
 205                             len - trailing_whitespace, 1, stream);
 206                         fputs(reset, stream);
 207                 }
 208                 if (trailing_carriage_return)
 209                         fputc('\r', stream);
 210                 if (trailing_newline)
 211                         fputc('\n', stream);
 212         }
 213         return result;
 214 }
 215
 216 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 217                    FILE *stream, const char *set,
 218                    const char *reset, const char *ws)
 219 {
 220         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 221 }
 222
 223 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 224 {
 225         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 226 }
 227
 228 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 229 {
 230         /*
 231          * We _might_ want to treat CR differently from other
 232          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 233          * for now we just use this stupid definition.
 234          */
 235         while (len-- > 0) {
 236                 if (!isspace(*line))
 237                         return 0;
 238                 line++;
 239         }
 240         return 1;
 241 }
 242
 243 /* Copy the line to the buffer while fixing whitespaces */
 244 int ws_fix_copy(char *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 245 {
 246         /*
 247          * len is number of bytes to be copied from src, starting
 248          * at src.  Typically src[len-1] is '\n', unless this is
 249          * the incomplete last line.
 250          */
 251         int i;
 252         int add_nl_to_tail = 0;
 253         int add_cr_to_tail = 0;
 254         int fixed = 0;
 255         int last_tab_in_indent = -1;
 256         int last_space_in_indent = -1;
 257         int need_fix_leading_space = 0;
 258         char *buf;
 259
 260         /*
 261          * Strip trailing whitespace
 262          */
 263         if ((ws_rule & WS_TRAILING_SPACE) &&
 264             (2 <= len && isspace(src[len-2]))) {
 265                 if (src[len - 1] == '\n') {
 266                         add_nl_to_tail = 1;
 267                         len--;
 268                         if (1 < len && src[len - 1] == '\r') {
 269                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 270                                 len--;
 271                         }
 272                 }
 273                 if (0 < len && isspace(src[len - 1])) {
 274                         while (0 < len && isspace(src[len-1]))
 275                                 len--;
 276                         fixed = 1;
 277                 }
 278         }
 279
 280         /*
 281          * Check leading whitespaces (indent)
 282          */
 283         for (i = 0; i < len; i++) {
 284                 char ch = src[i];
 285                 if (ch == '\t') {
 286                         last_tab_in_indent = i;
 287                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 288                             0 <= last_space_in_indent)
 289                             need_fix_leading_space = 1;
 290                 } else if (ch == ' ') {
 291                         last_space_in_indent = i;
 292                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 293                             8 <= i - last_tab_in_indent)
 294                                 need_fix_leading_space = 1;
 295                 } else
 296                         break;
 297         }
 298
 299         buf = dst;
 300         if (need_fix_leading_space) {
 301                 /* Process indent ourselves */
 302                 int consecutive_spaces = 0;
 303                 int last = last_tab_in_indent + 1;
 304
 305                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 306                         /* have "last" point at one past the indent */
 307                         if (last_tab_in_indent < last_space_in_indent)
 308                                 last = last_space_in_indent + 1;
 309                         else
 310                                 last = last_tab_in_indent + 1;
 311                 }
 312
 313                 /*
 314                  * between src[0..last-1], strip the funny spaces,
 315                  * updating them to tab as needed.
 316                  */
 317                 for (i = 0; i < last; i++) {
 318                         char ch = src[i];
 319                         if (ch != ' ') {
 320                                 consecutive_spaces = 0;
 321                                 *dst++ = ch;
 322                         } else {
 323                                 consecutive_spaces++;
 324                                 if (consecutive_spaces == 8) {
 325                                         *dst++ = '\t';
 326                                         consecutive_spaces = 0;
 327                                 }
 328                         }
 329                 }
 330                 while (0 < consecutive_spaces--)
 331                         *dst++ = ' ';
 332                 len -= last;
 333                 src += last;
 334                 fixed = 1;
 335         }
 336
 337         memcpy(dst, src, len);
 338         if (add_cr_to_tail)
 339                 dst[len++] = '\r';
 340         if (add_nl_to_tail)
 341                 dst[len++] = '\n';
 342         if (fixed && error_count)
 343                 (*error_count)++;
 344         return dst + len - buf;
 345 }