ws.c

   1 /*
   2  * Whitespace rules
   3  *
   4  * Copyright (c) 2007 Junio C Hamano
   5  */
   6 #include "cache.h"
   7 #include "attr.h"
   8
   9 static struct whitespace_rule {
  10         const char *rule_name;
  11         unsigned rule_bits;
  12         unsigned loosens_error:1,
  13                 exclude_default:1;
  14 } whitespace_rule_names[] = {
  15         { "trailing-space", WS_TRAILING_SPACE, 0 },
  16         { "space-before-tab", WS_SPACE_BEFORE_TAB, 0 },
  17         { "indent-with-non-tab", WS_INDENT_WITH_NON_TAB, 0 },
  18         { "cr-at-eol", WS_CR_AT_EOL, 1 },
  19         { "blank-at-eol", WS_BLANK_AT_EOL, 0 },
  20         { "blank-at-eof", WS_BLANK_AT_EOF, 0 },
  21         { "tab-in-indent", WS_TAB_IN_INDENT, 0, 1 },
  22 };
  23
  24 unsigned parse_whitespace_rule(const char *string)
  25 {
  26         unsigned rule = WS_DEFAULT_RULE;
  27
  28         while (string) {
  29                 int i;
  30                 size_t len;
  31                 const char *ep;
  32                 int negated = 0;
  33
  34                 string = string + strspn(string, ", \t\n\r");
  35                 ep = strchrnul(string, ',');
  36                 len = ep - string;
  37
  38                 if (*string == '-') {
  39                         negated = 1;
  40                         string++;
  41                         len--;
  42                 }
  43                 if (!len)
  44                         break;
  45                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
  46                         if (strncmp(whitespace_rule_names[i].rule_name,
  47                                     string, len))
  48                                 continue;
  49                         if (negated)
  50                                 rule &= ~whitespace_rule_names[i].rule_bits;
  51                         else
  52                                 rule |= whitespace_rule_names[i].rule_bits;
  53                         break;
  54                 }
  55                 if (strncmp(string, "tabwidth=", 9) == 0) {
  56                         unsigned tabwidth = atoi(string + 9);
  57                         if (0 < tabwidth && tabwidth < 0100) {
  58                                 rule &= ~WS_TAB_WIDTH_MASK;
  59                                 rule |= tabwidth;
  60                         }
  61                         else
  62                                 warning("tabwidth %.*s out of range",
  63                                         (int)(len - 9), string + 9);
  64                 }
  65                 string = ep;
  66         }
  67
  68         if (rule & WS_TAB_IN_INDENT && rule & WS_INDENT_WITH_NON_TAB)
  69                 die("cannot enforce both tab-in-indent and indent-with-non-tab");
  70         return rule;
  71 }
  72
  73 unsigned whitespace_rule(struct index_state *istate, const char *pathname)
  74 {
  75         static struct attr_check *attr_whitespace_rule;
  76         const char *value;
  77
  78         if (!attr_whitespace_rule)
  79                 attr_whitespace_rule = attr_check_initl("whitespace", NULL);
  80
  81         git_check_attr(istate, pathname, attr_whitespace_rule);
  82         value = attr_whitespace_rule->items[0].value;
  83         if (ATTR_TRUE(value)) {
  84                 /* true (whitespace) */
  85                 unsigned all_rule = ws_tab_width(whitespace_rule_cfg);
  86                 int i;
  87                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  88                         if (!whitespace_rule_names[i].loosens_error &&
  89                             !whitespace_rule_names[i].exclude_default)
  90                                 all_rule |= whitespace_rule_names[i].rule_bits;
  91                 return all_rule;
  92         } else if (ATTR_FALSE(value)) {
  93                 /* false (-whitespace) */
  94                 return ws_tab_width(whitespace_rule_cfg);
  95         } else if (ATTR_UNSET(value)) {
  96                 /* reset to default (!whitespace) */
  97                 return whitespace_rule_cfg;
  98         } else {
  99                 /* string */
 100                 return parse_whitespace_rule(value);
 101         }
 102 }
 103
 104 /* The returned string should be freed by the caller. */
 105 char *whitespace_error_string(unsigned ws)
 106 {
 107         struct strbuf err = STRBUF_INIT;
 108         if ((ws & WS_TRAILING_SPACE) == WS_TRAILING_SPACE)
 109                 strbuf_addstr(&err, "trailing whitespace");
 110         else {
 111                 if (ws & WS_BLANK_AT_EOL)
 112                         strbuf_addstr(&err, "trailing whitespace");
 113                 if (ws & WS_BLANK_AT_EOF) {
 114                         if (err.len)
 115                                 strbuf_addstr(&err, ", ");
 116                         strbuf_addstr(&err, "new blank line at EOF");
 117                 }
 118         }
 119         if (ws & WS_SPACE_BEFORE_TAB) {
 120                 if (err.len)
 121                         strbuf_addstr(&err, ", ");
 122                 strbuf_addstr(&err, "space before tab in indent");
 123         }
 124         if (ws & WS_INDENT_WITH_NON_TAB) {
 125                 if (err.len)
 126                         strbuf_addstr(&err, ", ");
 127                 strbuf_addstr(&err, "indent with spaces");
 128         }
 129         if (ws & WS_TAB_IN_INDENT) {
 130                 if (err.len)
 131                         strbuf_addstr(&err, ", ");
 132                 strbuf_addstr(&err, "tab in indent");
 133         }
 134         return strbuf_detach(&err, NULL);
 135 }
 136
 137 /* If stream is non-NULL, emits the line after checking. */
 138 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 139                                 FILE *stream, const char *set,
 140                                 const char *reset, const char *ws)
 141 {
 142         unsigned result = 0;
 143         int written = 0;
 144         int trailing_whitespace = -1;
 145         int trailing_newline = 0;
 146         int trailing_carriage_return = 0;
 147         int i;
 148
 149         /* Logic is simpler if we temporarily ignore the trailing newline. */
 150         if (len > 0 && line[len - 1] == '\n') {
 151                 trailing_newline = 1;
 152                 len--;
 153         }
 154         if ((ws_rule & WS_CR_AT_EOL) &&
 155             len > 0 && line[len - 1] == '\r') {
 156                 trailing_carriage_return = 1;
 157                 len--;
 158         }
 159
 160         /* Check for trailing whitespace. */
 161         if (ws_rule & WS_BLANK_AT_EOL) {
 162                 for (i = len - 1; i >= 0; i--) {
 163                         if (isspace(line[i])) {
 164                                 trailing_whitespace = i;
 165                                 result |= WS_BLANK_AT_EOL;
 166                         }
 167                         else
 168                                 break;
 169                 }
 170         }
 171
 172         if (trailing_whitespace == -1)
 173                 trailing_whitespace = len;
 174
 175         /* Check indentation */
 176         for (i = 0; i < trailing_whitespace; i++) {
 177                 if (line[i] == ' ')
 178                         continue;
 179                 if (line[i] != '\t')
 180                         break;
 181                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 182                         result |= WS_SPACE_BEFORE_TAB;
 183                         if (stream) {
 184                                 fputs(ws, stream);
 185                                 fwrite(line + written, i - written, 1, stream);
 186                                 fputs(reset, stream);
 187                                 fwrite(line + i, 1, 1, stream);
 188                         }
 189                 } else if (ws_rule & WS_TAB_IN_INDENT) {
 190                         result |= WS_TAB_IN_INDENT;
 191                         if (stream) {
 192                                 fwrite(line + written, i - written, 1, stream);
 193                                 fputs(ws, stream);
 194                                 fwrite(line + i, 1, 1, stream);
 195                                 fputs(reset, stream);
 196                         }
 197                 } else if (stream) {
 198                         fwrite(line + written, i - written + 1, 1, stream);
 199                 }
 200                 written = i + 1;
 201         }
 202
 203         /* Check for indent using non-tab. */
 204         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= ws_tab_width(ws_rule)) {
 205                 result |= WS_INDENT_WITH_NON_TAB;
 206                 if (stream) {
 207                         fputs(ws, stream);
 208                         fwrite(line + written, i - written, 1, stream);
 209                         fputs(reset, stream);
 210                 }
 211                 written = i;
 212         }
 213
 214         if (stream) {
 215                 /*
 216                  * Now the rest of the line starts at "written".
 217                  * The non-highlighted part ends at "trailing_whitespace".
 218                  */
 219
 220                 /* Emit non-highlighted (middle) segment. */
 221                 if (trailing_whitespace - written > 0) {
 222                         fputs(set, stream);
 223                         fwrite(line + written,
 224                             trailing_whitespace - written, 1, stream);
 225                         fputs(reset, stream);
 226                 }
 227
 228                 /* Highlight errors in trailing whitespace. */
 229                 if (trailing_whitespace != len) {
 230                         fputs(ws, stream);
 231                         fwrite(line + trailing_whitespace,
 232                             len - trailing_whitespace, 1, stream);
 233                         fputs(reset, stream);
 234                 }
 235                 if (trailing_carriage_return)
 236                         fputc('\r', stream);
 237                 if (trailing_newline)
 238                         fputc('\n', stream);
 239         }
 240         return result;
 241 }
 242
 243 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 244                    FILE *stream, const char *set,
 245                    const char *reset, const char *ws)
 246 {
 247         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 248 }
 249
 250 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 251 {
 252         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 253 }
 254
 255 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 256 {
 257         /*
 258          * We _might_ want to treat CR differently from other
 259          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 260          * for now we just use this stupid definition.
 261          */
 262         while (len-- > 0) {
 263                 if (!isspace(*line))
 264                         return 0;
 265                 line++;
 266         }
 267         return 1;
 268 }
 269
 270 /* Copy the line onto the end of the strbuf while fixing whitespaces */
 271 void ws_fix_copy(struct strbuf *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 272 {
 273         /*
 274          * len is number of bytes to be copied from src, starting
 275          * at src.  Typically src[len-1] is '\n', unless this is
 276          * the incomplete last line.
 277          */
 278         int i;
 279         int add_nl_to_tail = 0;
 280         int add_cr_to_tail = 0;
 281         int fixed = 0;
 282         int last_tab_in_indent = -1;
 283         int last_space_in_indent = -1;
 284         int need_fix_leading_space = 0;
 285
 286         /*
 287          * Strip trailing whitespace
 288          */
 289         if (ws_rule & WS_BLANK_AT_EOL) {
 290                 if (0 < len && src[len - 1] == '\n') {
 291                         add_nl_to_tail = 1;
 292                         len--;
 293                         if (0 < len && src[len - 1] == '\r') {
 294                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 295                                 len--;
 296                         }
 297                 }
 298                 if (0 < len && isspace(src[len - 1])) {
 299                         while (0 < len && isspace(src[len-1]))
 300                                 len--;
 301                         fixed = 1;
 302                 }
 303         }
 304
 305         /*
 306          * Check leading whitespaces (indent)
 307          */
 308         for (i = 0; i < len; i++) {
 309                 char ch = src[i];
 310                 if (ch == '\t') {
 311                         last_tab_in_indent = i;
 312                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 313                             0 <= last_space_in_indent)
 314                             need_fix_leading_space = 1;
 315                 } else if (ch == ' ') {
 316                         last_space_in_indent = i;
 317                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 318                             ws_tab_width(ws_rule) <= i - last_tab_in_indent)
 319                                 need_fix_leading_space = 1;
 320                 } else
 321                         break;
 322         }
 323
 324         if (need_fix_leading_space) {
 325                 /* Process indent ourselves */
 326                 int consecutive_spaces = 0;
 327                 int last = last_tab_in_indent + 1;
 328
 329                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 330                         /* have "last" point at one past the indent */
 331                         if (last_tab_in_indent < last_space_in_indent)
 332                                 last = last_space_in_indent + 1;
 333                         else
 334                                 last = last_tab_in_indent + 1;
 335                 }
 336
 337                 /*
 338                  * between src[0..last-1], strip the funny spaces,
 339                  * updating them to tab as needed.
 340                  */
 341                 for (i = 0; i < last; i++) {
 342                         char ch = src[i];
 343                         if (ch != ' ') {
 344                                 consecutive_spaces = 0;
 345                                 strbuf_addch(dst, ch);
 346                         } else {
 347                                 consecutive_spaces++;
 348                                 if (consecutive_spaces == ws_tab_width(ws_rule)) {
 349                                         strbuf_addch(dst, '\t');
 350                                         consecutive_spaces = 0;
 351                                 }
 352                         }
 353                 }
 354                 while (0 < consecutive_spaces--)
 355                         strbuf_addch(dst, ' ');
 356                 len -= last;
 357                 src += last;
 358                 fixed = 1;
 359         } else if ((ws_rule & WS_TAB_IN_INDENT) && last_tab_in_indent >= 0) {
 360                 /* Expand tabs into spaces */
 361                 int start = dst->len;
 362                 int last = last_tab_in_indent + 1;
 363                 for (i = 0; i < last; i++) {
 364                         if (src[i] == '\t')
 365                                 do {
 366                                         strbuf_addch(dst, ' ');
 367                                 } while ((dst->len - start) % ws_tab_width(ws_rule));
 368                         else
 369                                 strbuf_addch(dst, src[i]);
 370                 }
 371                 len -= last;
 372                 src += last;
 373                 fixed = 1;
 374         }
 375
 376         strbuf_add(dst, src, len);
 377         if (add_cr_to_tail)
 378                 strbuf_addch(dst, '\r');
 379         if (add_nl_to_tail)
 380                 strbuf_addch(dst, '\n');
 381         if (fixed && error_count)
 382                 (*error_count)++;
 383 }