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
  77         if (!attr_whitespace_rule)
  78                 attr_whitespace_rule = attr_check_initl("whitespace", NULL);
  79
  80         if (!git_check_attr(istate, pathname, attr_whitespace_rule)) {
  81                 const char *value;
  82
  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         } else {
 104                 return whitespace_rule_cfg;
 105         }
 106 }
 107
 108 /* The returned string should be freed by the caller. */
 109 char *whitespace_error_string(unsigned ws)
 110 {
 111         struct strbuf err = STRBUF_INIT;
 112         if ((ws & WS_TRAILING_SPACE) == WS_TRAILING_SPACE)
 113                 strbuf_addstr(&err, "trailing whitespace");
 114         else {
 115                 if (ws & WS_BLANK_AT_EOL)
 116                         strbuf_addstr(&err, "trailing whitespace");
 117                 if (ws & WS_BLANK_AT_EOF) {
 118                         if (err.len)
 119                                 strbuf_addstr(&err, ", ");
 120                         strbuf_addstr(&err, "new blank line at EOF");
 121                 }
 122         }
 123         if (ws & WS_SPACE_BEFORE_TAB) {
 124                 if (err.len)
 125                         strbuf_addstr(&err, ", ");
 126                 strbuf_addstr(&err, "space before tab in indent");
 127         }
 128         if (ws & WS_INDENT_WITH_NON_TAB) {
 129                 if (err.len)
 130                         strbuf_addstr(&err, ", ");
 131                 strbuf_addstr(&err, "indent with spaces");
 132         }
 133         if (ws & WS_TAB_IN_INDENT) {
 134                 if (err.len)
 135                         strbuf_addstr(&err, ", ");
 136                 strbuf_addstr(&err, "tab in indent");
 137         }
 138         return strbuf_detach(&err, NULL);
 139 }
 140
 141 /* If stream is non-NULL, emits the line after checking. */
 142 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 143                                 FILE *stream, const char *set,
 144                                 const char *reset, const char *ws)
 145 {
 146         unsigned result = 0;
 147         int written = 0;
 148         int trailing_whitespace = -1;
 149         int trailing_newline = 0;
 150         int trailing_carriage_return = 0;
 151         int i;
 152
 153         /* Logic is simpler if we temporarily ignore the trailing newline. */
 154         if (len > 0 && line[len - 1] == '\n') {
 155                 trailing_newline = 1;
 156                 len--;
 157         }
 158         if ((ws_rule & WS_CR_AT_EOL) &&
 159             len > 0 && line[len - 1] == '\r') {
 160                 trailing_carriage_return = 1;
 161                 len--;
 162         }
 163
 164         /* Check for trailing whitespace. */
 165         if (ws_rule & WS_BLANK_AT_EOL) {
 166                 for (i = len - 1; i >= 0; i--) {
 167                         if (isspace(line[i])) {
 168                                 trailing_whitespace = i;
 169                                 result |= WS_BLANK_AT_EOL;
 170                         }
 171                         else
 172                                 break;
 173                 }
 174         }
 175
 176         if (trailing_whitespace == -1)
 177                 trailing_whitespace = len;
 178
 179         /* Check indentation */
 180         for (i = 0; i < trailing_whitespace; i++) {
 181                 if (line[i] == ' ')
 182                         continue;
 183                 if (line[i] != '\t')
 184                         break;
 185                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 186                         result |= WS_SPACE_BEFORE_TAB;
 187                         if (stream) {
 188                                 fputs(ws, stream);
 189                                 fwrite(line + written, i - written, 1, stream);
 190                                 fputs(reset, stream);
 191                                 fwrite(line + i, 1, 1, stream);
 192                         }
 193                 } else if (ws_rule & WS_TAB_IN_INDENT) {
 194                         result |= WS_TAB_IN_INDENT;
 195                         if (stream) {
 196                                 fwrite(line + written, i - written, 1, stream);
 197                                 fputs(ws, stream);
 198                                 fwrite(line + i, 1, 1, stream);
 199                                 fputs(reset, stream);
 200                         }
 201                 } else if (stream) {
 202                         fwrite(line + written, i - written + 1, 1, stream);
 203                 }
 204                 written = i + 1;
 205         }
 206
 207         /* Check for indent using non-tab. */
 208         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= ws_tab_width(ws_rule)) {
 209                 result |= WS_INDENT_WITH_NON_TAB;
 210                 if (stream) {
 211                         fputs(ws, stream);
 212                         fwrite(line + written, i - written, 1, stream);
 213                         fputs(reset, stream);
 214                 }
 215                 written = i;
 216         }
 217
 218         if (stream) {
 219                 /*
 220                  * Now the rest of the line starts at "written".
 221                  * The non-highlighted part ends at "trailing_whitespace".
 222                  */
 223
 224                 /* Emit non-highlighted (middle) segment. */
 225                 if (trailing_whitespace - written > 0) {
 226                         fputs(set, stream);
 227                         fwrite(line + written,
 228                             trailing_whitespace - written, 1, stream);
 229                         fputs(reset, stream);
 230                 }
 231
 232                 /* Highlight errors in trailing whitespace. */
 233                 if (trailing_whitespace != len) {
 234                         fputs(ws, stream);
 235                         fwrite(line + trailing_whitespace,
 236                             len - trailing_whitespace, 1, stream);
 237                         fputs(reset, stream);
 238                 }
 239                 if (trailing_carriage_return)
 240                         fputc('\r', stream);
 241                 if (trailing_newline)
 242                         fputc('\n', stream);
 243         }
 244         return result;
 245 }
 246
 247 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 248                    FILE *stream, const char *set,
 249                    const char *reset, const char *ws)
 250 {
 251         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 252 }
 253
 254 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 255 {
 256         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 257 }
 258
 259 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 260 {
 261         /*
 262          * We _might_ want to treat CR differently from other
 263          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 264          * for now we just use this stupid definition.
 265          */
 266         while (len-- > 0) {
 267                 if (!isspace(*line))
 268                         return 0;
 269                 line++;
 270         }
 271         return 1;
 272 }
 273
 274 /* Copy the line onto the end of the strbuf while fixing whitespaces */
 275 void ws_fix_copy(struct strbuf *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 276 {
 277         /*
 278          * len is number of bytes to be copied from src, starting
 279          * at src.  Typically src[len-1] is '\n', unless this is
 280          * the incomplete last line.
 281          */
 282         int i;
 283         int add_nl_to_tail = 0;
 284         int add_cr_to_tail = 0;
 285         int fixed = 0;
 286         int last_tab_in_indent = -1;
 287         int last_space_in_indent = -1;
 288         int need_fix_leading_space = 0;
 289
 290         /*
 291          * Strip trailing whitespace
 292          */
 293         if (ws_rule & WS_BLANK_AT_EOL) {
 294                 if (0 < len && src[len - 1] == '\n') {
 295                         add_nl_to_tail = 1;
 296                         len--;
 297                         if (0 < len && src[len - 1] == '\r') {
 298                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 299                                 len--;
 300                         }
 301                 }
 302                 if (0 < len && isspace(src[len - 1])) {
 303                         while (0 < len && isspace(src[len-1]))
 304                                 len--;
 305                         fixed = 1;
 306                 }
 307         }
 308
 309         /*
 310          * Check leading whitespaces (indent)
 311          */
 312         for (i = 0; i < len; i++) {
 313                 char ch = src[i];
 314                 if (ch == '\t') {
 315                         last_tab_in_indent = i;
 316                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 317                             0 <= last_space_in_indent)
 318                             need_fix_leading_space = 1;
 319                 } else if (ch == ' ') {
 320                         last_space_in_indent = i;
 321                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 322                             ws_tab_width(ws_rule) <= i - last_tab_in_indent)
 323                                 need_fix_leading_space = 1;
 324                 } else
 325                         break;
 326         }
 327
 328         if (need_fix_leading_space) {
 329                 /* Process indent ourselves */
 330                 int consecutive_spaces = 0;
 331                 int last = last_tab_in_indent + 1;
 332
 333                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 334                         /* have "last" point at one past the indent */
 335                         if (last_tab_in_indent < last_space_in_indent)
 336                                 last = last_space_in_indent + 1;
 337                         else
 338                                 last = last_tab_in_indent + 1;
 339                 }
 340
 341                 /*
 342                  * between src[0..last-1], strip the funny spaces,
 343                  * updating them to tab as needed.
 344                  */
 345                 for (i = 0; i < last; i++) {
 346                         char ch = src[i];
 347                         if (ch != ' ') {
 348                                 consecutive_spaces = 0;
 349                                 strbuf_addch(dst, ch);
 350                         } else {
 351                                 consecutive_spaces++;
 352                                 if (consecutive_spaces == ws_tab_width(ws_rule)) {
 353                                         strbuf_addch(dst, '\t');
 354                                         consecutive_spaces = 0;
 355                                 }
 356                         }
 357                 }
 358                 while (0 < consecutive_spaces--)
 359                         strbuf_addch(dst, ' ');
 360                 len -= last;
 361                 src += last;
 362                 fixed = 1;
 363         } else if ((ws_rule & WS_TAB_IN_INDENT) && last_tab_in_indent >= 0) {
 364                 /* Expand tabs into spaces */
 365                 int start = dst->len;
 366                 int last = last_tab_in_indent + 1;
 367                 for (i = 0; i < last; i++) {
 368                         if (src[i] == '\t')
 369                                 do {
 370                                         strbuf_addch(dst, ' ');
 371                                 } while ((dst->len - start) % ws_tab_width(ws_rule));
 372                         else
 373                                 strbuf_addch(dst, src[i]);
 374                 }
 375                 len -= last;
 376                 src += last;
 377                 fixed = 1;
 378         }
 379
 380         strbuf_add(dst, src, len);
 381         if (add_cr_to_tail)
 382                 strbuf_addch(dst, '\r');
 383         if (add_nl_to_tail)
 384                 strbuf_addch(dst, '\n');
 385         if (fixed && error_count)
 386                 (*error_count)++;
 387 }