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         unsigned loosens_error:1,
  14                 exclude_default:1;
  15 } whitespace_rule_names[] = {
  16         { "trailing-space", WS_TRAILING_SPACE, 0 },
  17         { "space-before-tab", WS_SPACE_BEFORE_TAB, 0 },
  18         { "indent-with-non-tab", WS_INDENT_WITH_NON_TAB, 0 },
  19         { "cr-at-eol", WS_CR_AT_EOL, 1 },
  20         { "blank-at-eol", WS_BLANK_AT_EOL, 0 },
  21         { "blank-at-eof", WS_BLANK_AT_EOF, 0 },
  22         { "tab-in-indent", WS_TAB_IN_INDENT, 0, 1 },
  23 };
  24
  25 unsigned parse_whitespace_rule(const char *string)
  26 {
  27         unsigned rule = WS_DEFAULT_RULE;
  28
  29         while (string) {
  30                 int i;
  31                 size_t len;
  32                 const char *ep;
  33                 int negated = 0;
  34
  35                 string = string + strspn(string, ", \t\n\r");
  36                 ep = strchrnul(string, ',');
  37                 len = ep - string;
  38
  39                 if (*string == '-') {
  40                         negated = 1;
  41                         string++;
  42                         len--;
  43                 }
  44                 if (!len)
  45                         break;
  46                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
  47                         if (strncmp(whitespace_rule_names[i].rule_name,
  48                                     string, len))
  49                                 continue;
  50                         if (negated)
  51                                 rule &= ~whitespace_rule_names[i].rule_bits;
  52                         else
  53                                 rule |= whitespace_rule_names[i].rule_bits;
  54                         break;
  55                 }
  56                 if (strncmp(string, "tabwidth=", 9) == 0) {
  57                         unsigned tabwidth = atoi(string + 9);
  58                         if (0 < tabwidth && tabwidth < 0100) {
  59                                 rule &= ~WS_TAB_WIDTH_MASK;
  60                                 rule |= tabwidth;
  61                         }
  62                         else
  63                                 warning("tabwidth %.*s out of range",
  64                                         (int)(len - 9), string + 9);
  65                 }
  66                 string = ep;
  67         }
  68
  69         if (rule & WS_TAB_IN_INDENT && rule & WS_INDENT_WITH_NON_TAB)
  70                 die("cannot enforce both tab-in-indent and indent-with-non-tab");
  71         return rule;
  72 }
  73
  74 unsigned whitespace_rule(const char *pathname)
  75 {
  76         static struct attr_check *attr_whitespace_rule;
  77         const char *value;
  78
  79         if (!attr_whitespace_rule)
  80                 attr_whitespace_rule = attr_check_initl("whitespace", NULL);
  81
  82         git_check_attr(&the_index, pathname, attr_whitespace_rule);
  83         value = attr_whitespace_rule->items[0].value;
  84         if (ATTR_TRUE(value)) {
  85                 /* true (whitespace) */
  86                 unsigned all_rule = ws_tab_width(whitespace_rule_cfg);
  87                 int i;
  88                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  89                         if (!whitespace_rule_names[i].loosens_error &&
  90                             !whitespace_rule_names[i].exclude_default)
  91                                 all_rule |= whitespace_rule_names[i].rule_bits;
  92                 return all_rule;
  93         } else if (ATTR_FALSE(value)) {
  94                 /* false (-whitespace) */
  95                 return ws_tab_width(whitespace_rule_cfg);
  96         } else if (ATTR_UNSET(value)) {
  97                 /* reset to default (!whitespace) */
  98                 return whitespace_rule_cfg;
  99         } else {
 100                 /* string */
 101                 return parse_whitespace_rule(value);
 102         }
 103 }
 104
 105 /* The returned string should be freed by the caller. */
 106 char *whitespace_error_string(unsigned ws)
 107 {
 108         struct strbuf err = STRBUF_INIT;
 109         if ((ws & WS_TRAILING_SPACE) == WS_TRAILING_SPACE)
 110                 strbuf_addstr(&err, "trailing whitespace");
 111         else {
 112                 if (ws & WS_BLANK_AT_EOL)
 113                         strbuf_addstr(&err, "trailing whitespace");
 114                 if (ws & WS_BLANK_AT_EOF) {
 115                         if (err.len)
 116                                 strbuf_addstr(&err, ", ");
 117                         strbuf_addstr(&err, "new blank line at EOF");
 118                 }
 119         }
 120         if (ws & WS_SPACE_BEFORE_TAB) {
 121                 if (err.len)
 122                         strbuf_addstr(&err, ", ");
 123                 strbuf_addstr(&err, "space before tab in indent");
 124         }
 125         if (ws & WS_INDENT_WITH_NON_TAB) {
 126                 if (err.len)
 127                         strbuf_addstr(&err, ", ");
 128                 strbuf_addstr(&err, "indent with spaces");
 129         }
 130         if (ws & WS_TAB_IN_INDENT) {
 131                 if (err.len)
 132                         strbuf_addstr(&err, ", ");
 133                 strbuf_addstr(&err, "tab in indent");
 134         }
 135         return strbuf_detach(&err, NULL);
 136 }
 137
 138 /* If stream is non-NULL, emits the line after checking. */
 139 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 140                                 FILE *stream, const char *set,
 141                                 const char *reset, const char *ws)
 142 {
 143         unsigned result = 0;
 144         int written = 0;
 145         int trailing_whitespace = -1;
 146         int trailing_newline = 0;
 147         int trailing_carriage_return = 0;
 148         int i;
 149
 150         /* Logic is simpler if we temporarily ignore the trailing newline. */
 151         if (len > 0 && line[len - 1] == '\n') {
 152                 trailing_newline = 1;
 153                 len--;
 154         }
 155         if ((ws_rule & WS_CR_AT_EOL) &&
 156             len > 0 && line[len - 1] == '\r') {
 157                 trailing_carriage_return = 1;
 158                 len--;
 159         }
 160
 161         /* Check for trailing whitespace. */
 162         if (ws_rule & WS_BLANK_AT_EOL) {
 163                 for (i = len - 1; i >= 0; i--) {
 164                         if (isspace(line[i])) {
 165                                 trailing_whitespace = i;
 166                                 result |= WS_BLANK_AT_EOL;
 167                         }
 168                         else
 169                                 break;
 170                 }
 171         }
 172
 173         if (trailing_whitespace == -1)
 174                 trailing_whitespace = len;
 175
 176         /* Check indentation */
 177         for (i = 0; i < trailing_whitespace; i++) {
 178                 if (line[i] == ' ')
 179                         continue;
 180                 if (line[i] != '\t')
 181                         break;
 182                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 183                         result |= WS_SPACE_BEFORE_TAB;
 184                         if (stream) {
 185                                 fputs(ws, stream);
 186                                 fwrite(line + written, i - written, 1, stream);
 187                                 fputs(reset, stream);
 188                                 fwrite(line + i, 1, 1, stream);
 189                         }
 190                 } else if (ws_rule & WS_TAB_IN_INDENT) {
 191                         result |= WS_TAB_IN_INDENT;
 192                         if (stream) {
 193                                 fwrite(line + written, i - written, 1, stream);
 194                                 fputs(ws, stream);
 195                                 fwrite(line + i, 1, 1, stream);
 196                                 fputs(reset, stream);
 197                         }
 198                 } else if (stream) {
 199                         fwrite(line + written, i - written + 1, 1, stream);
 200                 }
 201                 written = i + 1;
 202         }
 203
 204         /* Check for indent using non-tab. */
 205         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= ws_tab_width(ws_rule)) {
 206                 result |= WS_INDENT_WITH_NON_TAB;
 207                 if (stream) {
 208                         fputs(ws, stream);
 209                         fwrite(line + written, i - written, 1, stream);
 210                         fputs(reset, stream);
 211                 }
 212                 written = i;
 213         }
 214
 215         if (stream) {
 216                 /*
 217                  * Now the rest of the line starts at "written".
 218                  * The non-highlighted part ends at "trailing_whitespace".
 219                  */
 220
 221                 /* Emit non-highlighted (middle) segment. */
 222                 if (trailing_whitespace - written > 0) {
 223                         fputs(set, stream);
 224                         fwrite(line + written,
 225                             trailing_whitespace - written, 1, stream);
 226                         fputs(reset, stream);
 227                 }
 228
 229                 /* Highlight errors in trailing whitespace. */
 230                 if (trailing_whitespace != len) {
 231                         fputs(ws, stream);
 232                         fwrite(line + trailing_whitespace,
 233                             len - trailing_whitespace, 1, stream);
 234                         fputs(reset, stream);
 235                 }
 236                 if (trailing_carriage_return)
 237                         fputc('\r', stream);
 238                 if (trailing_newline)
 239                         fputc('\n', stream);
 240         }
 241         return result;
 242 }
 243
 244 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 245                    FILE *stream, const char *set,
 246                    const char *reset, const char *ws)
 247 {
 248         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 249 }
 250
 251 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 252 {
 253         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 254 }
 255
 256 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 257 {
 258         /*
 259          * We _might_ want to treat CR differently from other
 260          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 261          * for now we just use this stupid definition.
 262          */
 263         while (len-- > 0) {
 264                 if (!isspace(*line))
 265                         return 0;
 266                 line++;
 267         }
 268         return 1;
 269 }
 270
 271 /* Copy the line onto the end of the strbuf while fixing whitespaces */
 272 void ws_fix_copy(struct strbuf *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 273 {
 274         /*
 275          * len is number of bytes to be copied from src, starting
 276          * at src.  Typically src[len-1] is '\n', unless this is
 277          * the incomplete last line.
 278          */
 279         int i;
 280         int add_nl_to_tail = 0;
 281         int add_cr_to_tail = 0;
 282         int fixed = 0;
 283         int last_tab_in_indent = -1;
 284         int last_space_in_indent = -1;
 285         int need_fix_leading_space = 0;
 286
 287         /*
 288          * Strip trailing whitespace
 289          */
 290         if (ws_rule & WS_BLANK_AT_EOL) {
 291                 if (0 < len && src[len - 1] == '\n') {
 292                         add_nl_to_tail = 1;
 293                         len--;
 294                         if (0 < len && src[len - 1] == '\r') {
 295                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 296                                 len--;
 297                         }
 298                 }
 299                 if (0 < len && isspace(src[len - 1])) {
 300                         while (0 < len && isspace(src[len-1]))
 301                                 len--;
 302                         fixed = 1;
 303                 }
 304         }
 305
 306         /*
 307          * Check leading whitespaces (indent)
 308          */
 309         for (i = 0; i < len; i++) {
 310                 char ch = src[i];
 311                 if (ch == '\t') {
 312                         last_tab_in_indent = i;
 313                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 314                             0 <= last_space_in_indent)
 315                             need_fix_leading_space = 1;
 316                 } else if (ch == ' ') {
 317                         last_space_in_indent = i;
 318                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 319                             ws_tab_width(ws_rule) <= i - last_tab_in_indent)
 320                                 need_fix_leading_space = 1;
 321                 } else
 322                         break;
 323         }
 324
 325         if (need_fix_leading_space) {
 326                 /* Process indent ourselves */
 327                 int consecutive_spaces = 0;
 328                 int last = last_tab_in_indent + 1;
 329
 330                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 331                         /* have "last" point at one past the indent */
 332                         if (last_tab_in_indent < last_space_in_indent)
 333                                 last = last_space_in_indent + 1;
 334                         else
 335                                 last = last_tab_in_indent + 1;
 336                 }
 337
 338                 /*
 339                  * between src[0..last-1], strip the funny spaces,
 340                  * updating them to tab as needed.
 341                  */
 342                 for (i = 0; i < last; i++) {
 343                         char ch = src[i];
 344                         if (ch != ' ') {
 345                                 consecutive_spaces = 0;
 346                                 strbuf_addch(dst, ch);
 347                         } else {
 348                                 consecutive_spaces++;
 349                                 if (consecutive_spaces == ws_tab_width(ws_rule)) {
 350                                         strbuf_addch(dst, '\t');
 351                                         consecutive_spaces = 0;
 352                                 }
 353                         }
 354                 }
 355                 while (0 < consecutive_spaces--)
 356                         strbuf_addch(dst, ' ');
 357                 len -= last;
 358                 src += last;
 359                 fixed = 1;
 360         } else if ((ws_rule & WS_TAB_IN_INDENT) && last_tab_in_indent >= 0) {
 361                 /* Expand tabs into spaces */
 362                 int start = dst->len;
 363                 int last = last_tab_in_indent + 1;
 364                 for (i = 0; i < last; i++) {
 365                         if (src[i] == '\t')
 366                                 do {
 367                                         strbuf_addch(dst, ' ');
 368                                 } while ((dst->len - start) % ws_tab_width(ws_rule));
 369                         else
 370                                 strbuf_addch(dst, src[i]);
 371                 }
 372                 len -= last;
 373                 src += last;
 374                 fixed = 1;
 375         }
 376
 377         strbuf_add(dst, src, len);
 378         if (add_cr_to_tail)
 379                 strbuf_addch(dst, '\r');
 380         if (add_nl_to_tail)
 381                 strbuf_addch(dst, '\n');
 382         if (fixed && error_count)
 383                 (*error_count)++;
 384 }