ws.c

   1 /*
   2  * Whitespace rules
   3  *
   4  * Copyright (c) 2007 Junio C Hamano
   5  */
   6
   7 #define DISABLE_SIGN_COMPARE_WARNINGS
   8
   9 #include "git-compat-util.h"
  10 #include "attr.h"
  11 #include "strbuf.h"
  12 #include "ws.h"
  13
  14 unsigned whitespace_rule_cfg = WS_DEFAULT_RULE;
  15
  16 static struct whitespace_rule {
  17         const char *rule_name;
  18         unsigned rule_bits;
  19         unsigned loosens_error:1,
  20                 exclude_default:1;
  21 } whitespace_rule_names[] = {
  22         { "trailing-space", WS_TRAILING_SPACE, 0 },
  23         { "space-before-tab", WS_SPACE_BEFORE_TAB, 0 },
  24         { "indent-with-non-tab", WS_INDENT_WITH_NON_TAB, 0 },
  25         { "cr-at-eol", WS_CR_AT_EOL, 1 },
  26         { "blank-at-eol", WS_BLANK_AT_EOL, 0 },
  27         { "blank-at-eof", WS_BLANK_AT_EOF, 0 },
  28         { "tab-in-indent", WS_TAB_IN_INDENT, 0, 1 },
  29 };
  30
  31 unsigned parse_whitespace_rule(const char *string)
  32 {
  33         unsigned rule = WS_DEFAULT_RULE;
  34
  35         while (string) {
  36                 int i;
  37                 size_t len;
  38                 const char *ep;
  39                 const char *arg;
  40                 int negated = 0;
  41
  42                 string = string + strspn(string, ", \t\n\r");
  43                 ep = strchrnul(string, ',');
  44                 len = ep - string;
  45
  46                 if (*string == '-') {
  47                         negated = 1;
  48                         string++;
  49                         len--;
  50                 }
  51                 if (!len)
  52                         break;
  53                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
  54                         if (strncmp(whitespace_rule_names[i].rule_name,
  55                                     string, len))
  56                                 continue;
  57                         if (negated)
  58                                 rule &= ~whitespace_rule_names[i].rule_bits;
  59                         else
  60                                 rule |= whitespace_rule_names[i].rule_bits;
  61                         break;
  62                 }
  63                 if (skip_prefix(string, "tabwidth=", &arg)) {
  64                         unsigned tabwidth = atoi(arg);
  65                         if (0 < tabwidth && tabwidth < 0100) {
  66                                 rule &= ~WS_TAB_WIDTH_MASK;
  67                                 rule |= tabwidth;
  68                         }
  69                         else
  70                                 warning("tabwidth %.*s out of range",
  71                                         (int)(ep - arg), arg);
  72                 }
  73                 string = ep;
  74         }
  75
  76         if (rule & WS_TAB_IN_INDENT && rule & WS_INDENT_WITH_NON_TAB)
  77                 die("cannot enforce both tab-in-indent and indent-with-non-tab");
  78         return rule;
  79 }
  80
  81 unsigned whitespace_rule(struct index_state *istate, const char *pathname)
  82 {
  83         static struct attr_check *attr_whitespace_rule;
  84         const char *value;
  85
  86         if (!attr_whitespace_rule)
  87                 attr_whitespace_rule = attr_check_initl("whitespace", NULL);
  88
  89         git_check_attr(istate, pathname, attr_whitespace_rule);
  90         value = attr_whitespace_rule->items[0].value;
  91         if (ATTR_TRUE(value)) {
  92                 /* true (whitespace) */
  93                 unsigned all_rule = ws_tab_width(whitespace_rule_cfg);
  94                 int i;
  95                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  96                         if (!whitespace_rule_names[i].loosens_error &&
  97                             !whitespace_rule_names[i].exclude_default)
  98                                 all_rule |= whitespace_rule_names[i].rule_bits;
  99                 return all_rule;
 100         } else if (ATTR_FALSE(value)) {
 101                 /* false (-whitespace) */
 102                 return ws_tab_width(whitespace_rule_cfg);
 103         } else if (ATTR_UNSET(value)) {
 104                 /* reset to default (!whitespace) */
 105                 return whitespace_rule_cfg;
 106         } else {
 107                 /* string */
 108                 return parse_whitespace_rule(value);
 109         }
 110 }
 111
 112 /* The returned string should be freed by the caller. */
 113 char *whitespace_error_string(unsigned ws)
 114 {
 115         struct strbuf err = STRBUF_INIT;
 116         if ((ws & WS_TRAILING_SPACE) == WS_TRAILING_SPACE)
 117                 strbuf_addstr(&err, "trailing whitespace");
 118         else {
 119                 if (ws & WS_BLANK_AT_EOL)
 120                         strbuf_addstr(&err, "trailing whitespace");
 121                 if (ws & WS_BLANK_AT_EOF) {
 122                         if (err.len)
 123                                 strbuf_addstr(&err, ", ");
 124                         strbuf_addstr(&err, "new blank line at EOF");
 125                 }
 126         }
 127         if (ws & WS_SPACE_BEFORE_TAB) {
 128                 if (err.len)
 129                         strbuf_addstr(&err, ", ");
 130                 strbuf_addstr(&err, "space before tab in indent");
 131         }
 132         if (ws & WS_INDENT_WITH_NON_TAB) {
 133                 if (err.len)
 134                         strbuf_addstr(&err, ", ");
 135                 strbuf_addstr(&err, "indent with spaces");
 136         }
 137         if (ws & WS_TAB_IN_INDENT) {
 138                 if (err.len)
 139                         strbuf_addstr(&err, ", ");
 140                 strbuf_addstr(&err, "tab in indent");
 141         }
 142         return strbuf_detach(&err, NULL);
 143 }
 144
 145 /* If stream is non-NULL, emits the line after checking. */
 146 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 147                                 FILE *stream, const char *set,
 148                                 const char *reset, const char *ws)
 149 {
 150         unsigned result = 0;
 151         int written = 0;
 152         int trailing_whitespace = -1;
 153         int trailing_newline = 0;
 154         int trailing_carriage_return = 0;
 155         int i;
 156
 157         /* Logic is simpler if we temporarily ignore the trailing newline. */
 158         if (len > 0 && line[len - 1] == '\n') {
 159                 trailing_newline = 1;
 160                 len--;
 161         }
 162         if ((ws_rule & WS_CR_AT_EOL) &&
 163             len > 0 && line[len - 1] == '\r') {
 164                 trailing_carriage_return = 1;
 165                 len--;
 166         }
 167
 168         /* Check for trailing whitespace. */
 169         if (ws_rule & WS_BLANK_AT_EOL) {
 170                 for (i = len - 1; i >= 0; i--) {
 171                         if (isspace(line[i])) {
 172                                 trailing_whitespace = i;
 173                                 result |= WS_BLANK_AT_EOL;
 174                         }
 175                         else
 176                                 break;
 177                 }
 178         }
 179
 180         if (trailing_whitespace == -1)
 181                 trailing_whitespace = len;
 182
 183         /* Check indentation */
 184         for (i = 0; i < trailing_whitespace; i++) {
 185                 if (line[i] == ' ')
 186                         continue;
 187                 if (line[i] != '\t')
 188                         break;
 189                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 190                         result |= WS_SPACE_BEFORE_TAB;
 191                         if (stream) {
 192                                 fputs(ws, stream);
 193                                 fwrite(line + written, i - written, 1, stream);
 194                                 fputs(reset, stream);
 195                                 fwrite(line + i, 1, 1, stream);
 196                         }
 197                 } else if (ws_rule & WS_TAB_IN_INDENT) {
 198                         result |= WS_TAB_IN_INDENT;
 199                         if (stream) {
 200                                 fwrite(line + written, i - written, 1, stream);
 201                                 fputs(ws, stream);
 202                                 fwrite(line + i, 1, 1, stream);
 203                                 fputs(reset, stream);
 204                         }
 205                 } else if (stream) {
 206                         fwrite(line + written, i - written + 1, 1, stream);
 207                 }
 208                 written = i + 1;
 209         }
 210
 211         /* Check for indent using non-tab. */
 212         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= ws_tab_width(ws_rule)) {
 213                 result |= WS_INDENT_WITH_NON_TAB;
 214                 if (stream) {
 215                         fputs(ws, stream);
 216                         fwrite(line + written, i - written, 1, stream);
 217                         fputs(reset, stream);
 218                 }
 219                 written = i;
 220         }
 221
 222         if (stream) {
 223                 /*
 224                  * Now the rest of the line starts at "written".
 225                  * The non-highlighted part ends at "trailing_whitespace".
 226                  */
 227
 228                 /* Emit non-highlighted (middle) segment. */
 229                 if (trailing_whitespace - written > 0) {
 230                         fputs(set, stream);
 231                         fwrite(line + written,
 232                             trailing_whitespace - written, 1, stream);
 233                         fputs(reset, stream);
 234                 }
 235
 236                 /* Highlight errors in trailing whitespace. */
 237                 if (trailing_whitespace != len) {
 238                         fputs(ws, stream);
 239                         fwrite(line + trailing_whitespace,
 240                             len - trailing_whitespace, 1, stream);
 241                         fputs(reset, stream);
 242                 }
 243                 if (trailing_carriage_return)
 244                         fputc('\r', stream);
 245                 if (trailing_newline)
 246                         fputc('\n', stream);
 247         }
 248         return result;
 249 }
 250
 251 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 252                    FILE *stream, const char *set,
 253                    const char *reset, const char *ws)
 254 {
 255         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 256 }
 257
 258 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 259 {
 260         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 261 }
 262
 263 int ws_blank_line(const char *line, int len)
 264 {
 265         /*
 266          * We _might_ want to treat CR differently from other
 267          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 268          * for now we just use this stupid definition.
 269          */
 270         while (len-- > 0) {
 271                 if (!isspace(*line))
 272                         return 0;
 273                 line++;
 274         }
 275         return 1;
 276 }
 277
 278 /* Copy the line onto the end of the strbuf while fixing whitespaces */
 279 void ws_fix_copy(struct strbuf *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 280 {
 281         /*
 282          * len is number of bytes to be copied from src, starting
 283          * at src.  Typically src[len-1] is '\n', unless this is
 284          * the incomplete last line.
 285          */
 286         int i;
 287         int add_nl_to_tail = 0;
 288         int add_cr_to_tail = 0;
 289         int fixed = 0;
 290         int last_tab_in_indent = -1;
 291         int last_space_in_indent = -1;
 292         int need_fix_leading_space = 0;
 293
 294         /*
 295          * Strip trailing whitespace
 296          */
 297         if (ws_rule & WS_BLANK_AT_EOL) {
 298                 if (0 < len && src[len - 1] == '\n') {
 299                         add_nl_to_tail = 1;
 300                         len--;
 301                         if (0 < len && src[len - 1] == '\r') {
 302                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 303                                 len--;
 304                         }
 305                 }
 306                 if (0 < len && isspace(src[len - 1])) {
 307                         while (0 < len && isspace(src[len-1]))
 308                                 len--;
 309                         fixed = 1;
 310                 }
 311         }
 312
 313         /*
 314          * Check leading whitespaces (indent)
 315          */
 316         for (i = 0; i < len; i++) {
 317                 char ch = src[i];
 318                 if (ch == '\t') {
 319                         last_tab_in_indent = i;
 320                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 321                             0 <= last_space_in_indent)
 322                             need_fix_leading_space = 1;
 323                 } else if (ch == ' ') {
 324                         last_space_in_indent = i;
 325                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 326                             ws_tab_width(ws_rule) <= i - last_tab_in_indent)
 327                                 need_fix_leading_space = 1;
 328                 } else
 329                         break;
 330         }
 331
 332         if (need_fix_leading_space) {
 333                 /* Process indent ourselves */
 334                 int consecutive_spaces = 0;
 335                 int last = last_tab_in_indent + 1;
 336
 337                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 338                         /* have "last" point at one past the indent */
 339                         if (last_tab_in_indent < last_space_in_indent)
 340                                 last = last_space_in_indent + 1;
 341                         else
 342                                 last = last_tab_in_indent + 1;
 343                 }
 344
 345                 /*
 346                  * between src[0..last-1], strip the funny spaces,
 347                  * updating them to tab as needed.
 348                  */
 349                 for (i = 0; i < last; i++) {
 350                         char ch = src[i];
 351                         if (ch != ' ') {
 352                                 consecutive_spaces = 0;
 353                                 strbuf_addch(dst, ch);
 354                         } else {
 355                                 consecutive_spaces++;
 356                                 if (consecutive_spaces == ws_tab_width(ws_rule)) {
 357                                         strbuf_addch(dst, '\t');
 358                                         consecutive_spaces = 0;
 359                                 }
 360                         }
 361                 }
 362                 while (0 < consecutive_spaces--)
 363                         strbuf_addch(dst, ' ');
 364                 len -= last;
 365                 src += last;
 366                 fixed = 1;
 367         } else if ((ws_rule & WS_TAB_IN_INDENT) && last_tab_in_indent >= 0) {
 368                 /* Expand tabs into spaces */
 369                 int start = dst->len;
 370                 int last = last_tab_in_indent + 1;
 371                 for (i = 0; i < last; i++) {
 372                         if (src[i] == '\t')
 373                                 do {
 374                                         strbuf_addch(dst, ' ');
 375                                 } while ((dst->len - start) % ws_tab_width(ws_rule));
 376                         else
 377                                 strbuf_addch(dst, src[i]);
 378                 }
 379                 len -= last;
 380                 src += last;
 381                 fixed = 1;
 382         }
 383
 384         strbuf_add(dst, src, len);
 385         if (add_cr_to_tail)
 386                 strbuf_addch(dst, '\r');
 387         if (add_nl_to_tail)
 388                 strbuf_addch(dst, '\n');
 389         if (fixed && error_count)
 390                 (*error_count)++;
 391 }