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 };
  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 = strchr(string, ',');
  36                 if (!ep)
  37                         len = strlen(string);
  38                 else
  39                         len = ep - string;
  40
  41                 if (*string == '-') {
  42                         negated = 1;
  43                         string++;
  44                         len--;
  45                 }
  46                 if (!len)
  47                         break;
  48                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
  49                         if (strncmp(whitespace_rule_names[i].rule_name,
  50                                     string, len))
  51                                 continue;
  52                         if (negated)
  53                                 rule &= ~whitespace_rule_names[i].rule_bits;
  54                         else
  55                                 rule |= whitespace_rule_names[i].rule_bits;
  56                         break;
  57                 }
  58                 string = ep;
  59         }
  60         return rule;
  61 }
  62
  63 static void setup_whitespace_attr_check(struct git_attr_check *check)
  64 {
  65         static struct git_attr *attr_whitespace;
  66
  67         if (!attr_whitespace)
  68                 attr_whitespace = git_attr("whitespace");
  69         check[0].attr = attr_whitespace;
  70 }
  71
  72 unsigned whitespace_rule(const char *pathname)
  73 {
  74         struct git_attr_check attr_whitespace_rule;
  75
  76         setup_whitespace_attr_check(&attr_whitespace_rule);
  77         if (!git_checkattr(pathname, 1, &attr_whitespace_rule)) {
  78                 const char *value;
  79
  80                 value = attr_whitespace_rule.value;
  81                 if (ATTR_TRUE(value)) {
  82                         /* true (whitespace) */
  83                         unsigned all_rule = 0;
  84                         int i;
  85                         for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  86                                 if (!whitespace_rule_names[i].loosens_error &&
  87                                     !whitespace_rule_names[i].exclude_default)
  88                                         all_rule |= whitespace_rule_names[i].rule_bits;
  89                         return all_rule;
  90                 } else if (ATTR_FALSE(value)) {
  91                         /* false (-whitespace) */
  92                         return 0;
  93                 } else if (ATTR_UNSET(value)) {
  94                         /* reset to default (!whitespace) */
  95                         return whitespace_rule_cfg;
  96                 } else {
  97                         /* string */
  98                         return parse_whitespace_rule(value);
  99                 }
 100         } else {
 101                 return whitespace_rule_cfg;
 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         return strbuf_detach(&err, NULL);
 131 }
 132
 133 /* If stream is non-NULL, emits the line after checking. */
 134 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 135                                 FILE *stream, const char *set,
 136                                 const char *reset, const char *ws)
 137 {
 138         unsigned result = 0;
 139         int written = 0;
 140         int trailing_whitespace = -1;
 141         int trailing_newline = 0;
 142         int trailing_carriage_return = 0;
 143         int i;
 144
 145         /* Logic is simpler if we temporarily ignore the trailing newline. */
 146         if (len > 0 && line[len - 1] == '\n') {
 147                 trailing_newline = 1;
 148                 len--;
 149         }
 150         if ((ws_rule & WS_CR_AT_EOL) &&
 151             len > 0 && line[len - 1] == '\r') {
 152                 trailing_carriage_return = 1;
 153                 len--;
 154         }
 155
 156         /* Check for trailing whitespace. */
 157         if (ws_rule & WS_BLANK_AT_EOL) {
 158                 for (i = len - 1; i >= 0; i--) {
 159                         if (isspace(line[i])) {
 160                                 trailing_whitespace = i;
 161                                 result |= WS_BLANK_AT_EOL;
 162                         }
 163                         else
 164                                 break;
 165                 }
 166         }
 167
 168         /* Check for space before tab in initial indent. */
 169         for (i = 0; i < len; i++) {
 170                 if (line[i] == ' ')
 171                         continue;
 172                 if (line[i] != '\t')
 173                         break;
 174                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 175                         result |= WS_SPACE_BEFORE_TAB;
 176                         if (stream) {
 177                                 fputs(ws, stream);
 178                                 fwrite(line + written, i - written, 1, stream);
 179                                 fputs(reset, stream);
 180                         }
 181                 } else if (stream)
 182                         fwrite(line + written, i - written, 1, stream);
 183                 if (stream)
 184                         fwrite(line + i, 1, 1, stream);
 185                 written = i + 1;
 186         }
 187
 188         /* Check for indent using non-tab. */
 189         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= 8) {
 190                 result |= WS_INDENT_WITH_NON_TAB;
 191                 if (stream) {
 192                         fputs(ws, stream);
 193                         fwrite(line + written, i - written, 1, stream);
 194                         fputs(reset, stream);
 195                 }
 196                 written = i;
 197         }
 198
 199         if (stream) {
 200                 /*
 201                  * Now the rest of the line starts at "written".
 202                  * The non-highlighted part ends at "trailing_whitespace".
 203                  */
 204                 if (trailing_whitespace == -1)
 205                         trailing_whitespace = len;
 206
 207                 /* Emit non-highlighted (middle) segment. */
 208                 if (trailing_whitespace - written > 0) {
 209                         fputs(set, stream);
 210                         fwrite(line + written,
 211                             trailing_whitespace - written, 1, stream);
 212                         fputs(reset, stream);
 213                 }
 214
 215                 /* Highlight errors in trailing whitespace. */
 216                 if (trailing_whitespace != len) {
 217                         fputs(ws, stream);
 218                         fwrite(line + trailing_whitespace,
 219                             len - trailing_whitespace, 1, stream);
 220                         fputs(reset, stream);
 221                 }
 222                 if (trailing_carriage_return)
 223                         fputc('\r', stream);
 224                 if (trailing_newline)
 225                         fputc('\n', stream);
 226         }
 227         return result;
 228 }
 229
 230 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 231                    FILE *stream, const char *set,
 232                    const char *reset, const char *ws)
 233 {
 234         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 235 }
 236
 237 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 238 {
 239         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 240 }
 241
 242 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 243 {
 244         /*
 245          * We _might_ want to treat CR differently from other
 246          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 247          * for now we just use this stupid definition.
 248          */
 249         while (len-- > 0) {
 250                 if (!isspace(*line))
 251                         return 0;
 252                 line++;
 253         }
 254         return 1;
 255 }
 256
 257 /* Copy the line to the buffer while fixing whitespaces */
 258 int ws_fix_copy(char *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 259 {
 260         /*
 261          * len is number of bytes to be copied from src, starting
 262          * at src.  Typically src[len-1] is '\n', unless this is
 263          * the incomplete last line.
 264          */
 265         int i;
 266         int add_nl_to_tail = 0;
 267         int add_cr_to_tail = 0;
 268         int fixed = 0;
 269         int last_tab_in_indent = -1;
 270         int last_space_in_indent = -1;
 271         int need_fix_leading_space = 0;
 272         char *buf;
 273
 274         /*
 275          * Strip trailing whitespace
 276          */
 277         if (ws_rule & WS_BLANK_AT_EOL) {
 278                 if (0 < len && src[len - 1] == '\n') {
 279                         add_nl_to_tail = 1;
 280                         len--;
 281                         if (0 < len && src[len - 1] == '\r') {
 282                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 283                                 len--;
 284                         }
 285                 }
 286                 if (0 < len && isspace(src[len - 1])) {
 287                         while (0 < len && isspace(src[len-1]))
 288                                 len--;
 289                         fixed = 1;
 290                 }
 291         }
 292
 293         /*
 294          * Check leading whitespaces (indent)
 295          */
 296         for (i = 0; i < len; i++) {
 297                 char ch = src[i];
 298                 if (ch == '\t') {
 299                         last_tab_in_indent = i;
 300                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 301                             0 <= last_space_in_indent)
 302                             need_fix_leading_space = 1;
 303                 } else if (ch == ' ') {
 304                         last_space_in_indent = i;
 305                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 306                             8 <= i - last_tab_in_indent)
 307                                 need_fix_leading_space = 1;
 308                 } else
 309                         break;
 310         }
 311
 312         buf = dst;
 313         if (need_fix_leading_space) {
 314                 /* Process indent ourselves */
 315                 int consecutive_spaces = 0;
 316                 int last = last_tab_in_indent + 1;
 317
 318                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 319                         /* have "last" point at one past the indent */
 320                         if (last_tab_in_indent < last_space_in_indent)
 321                                 last = last_space_in_indent + 1;
 322                         else
 323                                 last = last_tab_in_indent + 1;
 324                 }
 325
 326                 /*
 327                  * between src[0..last-1], strip the funny spaces,
 328                  * updating them to tab as needed.
 329                  */
 330                 for (i = 0; i < last; i++) {
 331                         char ch = src[i];
 332                         if (ch != ' ') {
 333                                 consecutive_spaces = 0;
 334                                 *dst++ = ch;
 335                         } else {
 336                                 consecutive_spaces++;
 337                                 if (consecutive_spaces == 8) {
 338                                         *dst++ = '\t';
 339                                         consecutive_spaces = 0;
 340                                 }
 341                         }
 342                 }
 343                 while (0 < consecutive_spaces--)
 344                         *dst++ = ' ';
 345                 len -= last;
 346                 src += last;
 347                 fixed = 1;
 348         }
 349
 350         memcpy(dst, src, len);
 351         if (add_cr_to_tail)
 352                 dst[len++] = '\r';
 353         if (add_nl_to_tail)
 354                 dst[len++] = '\n';
 355         if (fixed && error_count)
 356                 (*error_count)++;
 357         return dst + len - buf;
 358 }