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 = strchr(string, ',');
  37                 if (!ep)
  38                         len = strlen(string);
  39                 else
  40                         len = ep - string;
  41
  42                 if (*string == '-') {
  43                         negated = 1;
  44                         string++;
  45                         len--;
  46                 }
  47                 if (!len)
  48                         break;
  49                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
  50                         if (strncmp(whitespace_rule_names[i].rule_name,
  51                                     string, len))
  52                                 continue;
  53                         if (negated)
  54                                 rule &= ~whitespace_rule_names[i].rule_bits;
  55                         else
  56                                 rule |= whitespace_rule_names[i].rule_bits;
  57                         break;
  58                 }
  59                 string = ep;
  60         }
  61
  62         if (rule & WS_TAB_IN_INDENT && rule & WS_INDENT_WITH_NON_TAB)
  63                 die("cannot enforce both tab-in-indent and indent-with-non-tab");
  64         return rule;
  65 }
  66
  67 static void setup_whitespace_attr_check(struct git_attr_check *check)
  68 {
  69         static struct git_attr *attr_whitespace;
  70
  71         if (!attr_whitespace)
  72                 attr_whitespace = git_attr("whitespace");
  73         check[0].attr = attr_whitespace;
  74 }
  75
  76 unsigned whitespace_rule(const char *pathname)
  77 {
  78         struct git_attr_check attr_whitespace_rule;
  79
  80         setup_whitespace_attr_check(&attr_whitespace_rule);
  81         if (!git_checkattr(pathname, 1, &attr_whitespace_rule)) {
  82                 const char *value;
  83
  84                 value = attr_whitespace_rule.value;
  85                 if (ATTR_TRUE(value)) {
  86                         /* true (whitespace) */
  87                         unsigned all_rule = 0;
  88                         int i;
  89                         for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  90                                 if (!whitespace_rule_names[i].loosens_error &&
  91                                     !whitespace_rule_names[i].exclude_default)
  92                                         all_rule |= whitespace_rule_names[i].rule_bits;
  93                         return all_rule;
  94                 } else if (ATTR_FALSE(value)) {
  95                         /* false (-whitespace) */
  96                         return 0;
  97                 } else if (ATTR_UNSET(value)) {
  98                         /* reset to default (!whitespace) */
  99                         return whitespace_rule_cfg;
 100                 } else {
 101                         /* string */
 102                         return parse_whitespace_rule(value);
 103                 }
 104         } else {
 105                 return whitespace_rule_cfg;
 106         }
 107 }
 108
 109 /* The returned string should be freed by the caller. */
 110 char *whitespace_error_string(unsigned ws)
 111 {
 112         struct strbuf err = STRBUF_INIT;
 113         if ((ws & WS_TRAILING_SPACE) == WS_TRAILING_SPACE)
 114                 strbuf_addstr(&err, "trailing whitespace");
 115         else {
 116                 if (ws & WS_BLANK_AT_EOL)
 117                         strbuf_addstr(&err, "trailing whitespace");
 118                 if (ws & WS_BLANK_AT_EOF) {
 119                         if (err.len)
 120                                 strbuf_addstr(&err, ", ");
 121                         strbuf_addstr(&err, "new blank line at EOF");
 122                 }
 123         }
 124         if (ws & WS_SPACE_BEFORE_TAB) {
 125                 if (err.len)
 126                         strbuf_addstr(&err, ", ");
 127                 strbuf_addstr(&err, "space before tab in indent");
 128         }
 129         if (ws & WS_INDENT_WITH_NON_TAB) {
 130                 if (err.len)
 131                         strbuf_addstr(&err, ", ");
 132                 strbuf_addstr(&err, "indent with spaces");
 133         }
 134         if (ws & WS_TAB_IN_INDENT) {
 135                 if (err.len)
 136                         strbuf_addstr(&err, ", ");
 137                 strbuf_addstr(&err, "tab in indent");
 138         }
 139         return strbuf_detach(&err, NULL);
 140 }
 141
 142 /* If stream is non-NULL, emits the line after checking. */
 143 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 144                                 FILE *stream, const char *set,
 145                                 const char *reset, const char *ws)
 146 {
 147         unsigned result = 0;
 148         int written = 0;
 149         int trailing_whitespace = -1;
 150         int trailing_newline = 0;
 151         int trailing_carriage_return = 0;
 152         int i;
 153
 154         /* Logic is simpler if we temporarily ignore the trailing newline. */
 155         if (len > 0 && line[len - 1] == '\n') {
 156                 trailing_newline = 1;
 157                 len--;
 158         }
 159         if ((ws_rule & WS_CR_AT_EOL) &&
 160             len > 0 && line[len - 1] == '\r') {
 161                 trailing_carriage_return = 1;
 162                 len--;
 163         }
 164
 165         /* Check for trailing whitespace. */
 166         if (ws_rule & WS_BLANK_AT_EOL) {
 167                 for (i = len - 1; i >= 0; i--) {
 168                         if (isspace(line[i])) {
 169                                 trailing_whitespace = i;
 170                                 result |= WS_BLANK_AT_EOL;
 171                         }
 172                         else
 173                                 break;
 174                 }
 175         }
 176
 177         /* Check indentation */
 178         for (i = 0; i < len; i++) {
 179                 if (line[i] == ' ')
 180                         continue;
 181                 if (line[i] != '\t')
 182                         break;
 183                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 184                         result |= WS_SPACE_BEFORE_TAB;
 185                         if (stream) {
 186                                 fputs(ws, stream);
 187                                 fwrite(line + written, i - written, 1, stream);
 188                                 fputs(reset, stream);
 189                                 fwrite(line + i, 1, 1, stream);
 190                         }
 191                 } else if (ws_rule & WS_TAB_IN_INDENT) {
 192                         result |= WS_TAB_IN_INDENT;
 193                         if (stream) {
 194                                 fwrite(line + written, i - written, 1, stream);
 195                                 fputs(ws, stream);
 196                                 fwrite(line + i, 1, 1, stream);
 197                                 fputs(reset, stream);
 198                         }
 199                 } else if (stream) {
 200                         fwrite(line + written, i - written + 1, 1, stream);
 201                 }
 202                 written = i + 1;
 203         }
 204
 205         /* Check for indent using non-tab. */
 206         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= 8) {
 207                 result |= WS_INDENT_WITH_NON_TAB;
 208                 if (stream) {
 209                         fputs(ws, stream);
 210                         fwrite(line + written, i - written, 1, stream);
 211                         fputs(reset, stream);
 212                 }
 213                 written = i;
 214         }
 215
 216         if (stream) {
 217                 /*
 218                  * Now the rest of the line starts at "written".
 219                  * The non-highlighted part ends at "trailing_whitespace".
 220                  */
 221                 if (trailing_whitespace == -1)
 222                         trailing_whitespace = len;
 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                             8 <= 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 == 8) {
 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 last = last_tab_in_indent + 1;
 366                 for (i = 0; i < last; i++) {
 367                         if (src[i] == '\t')
 368                                 do {
 369                                         strbuf_addch(dst, ' ');
 370                                 } while (dst->len % 8);
 371                         else
 372                                 strbuf_addch(dst, src[i]);
 373                 }
 374                 len -= last;
 375                 src += last;
 376                 fixed = 1;
 377         }
 378
 379         strbuf_add(dst, src, len);
 380         if (add_cr_to_tail)
 381                 strbuf_addch(dst, '\r');
 382         if (add_nl_to_tail)
 383                 strbuf_addch(dst, '\n');
 384         if (fixed && error_count)
 385                 (*error_count)++;
 386 }