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 } whitespace_rule_names[] = {
  14         { "trailing-space", WS_TRAILING_SPACE },
  15         { "space-before-tab", WS_SPACE_BEFORE_TAB },
  16         { "indent-with-non-tab", WS_INDENT_WITH_NON_TAB },
  17         { "cr-at-eol", WS_CR_AT_EOL },
  18         { "blank-at-eof", WS_BLANK_AT_EOF },
  19 };
  20
  21 unsigned parse_whitespace_rule(const char *string)
  22 {
  23         unsigned rule = WS_DEFAULT_RULE;
  24
  25         while (string) {
  26                 int i;
  27                 size_t len;
  28                 const char *ep;
  29                 int negated = 0;
  30
  31                 string = string + strspn(string, ", \t\n\r");
  32                 ep = strchr(string, ',');
  33                 if (!ep)
  34                         len = strlen(string);
  35                 else
  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                 string = ep;
  56         }
  57         return rule;
  58 }
  59
  60 static void setup_whitespace_attr_check(struct git_attr_check *check)
  61 {
  62         static struct git_attr *attr_whitespace;
  63
  64         if (!attr_whitespace)
  65                 attr_whitespace = git_attr("whitespace", 10);
  66         check[0].attr = attr_whitespace;
  67 }
  68
  69 unsigned whitespace_rule(const char *pathname)
  70 {
  71         struct git_attr_check attr_whitespace_rule;
  72
  73         setup_whitespace_attr_check(&attr_whitespace_rule);
  74         if (!git_checkattr(pathname, 1, &attr_whitespace_rule)) {
  75                 const char *value;
  76
  77                 value = attr_whitespace_rule.value;
  78                 if (ATTR_TRUE(value)) {
  79                         /* true (whitespace) */
  80                         unsigned all_rule = 0;
  81                         int i;
  82                         for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  83                                 all_rule |= whitespace_rule_names[i].rule_bits;
  84                         return all_rule;
  85                 } else if (ATTR_FALSE(value)) {
  86                         /* false (-whitespace) */
  87                         return 0;
  88                 } else if (ATTR_UNSET(value)) {
  89                         /* reset to default (!whitespace) */
  90                         return whitespace_rule_cfg;
  91                 } else {
  92                         /* string */
  93                         return parse_whitespace_rule(value);
  94                 }
  95         } else {
  96                 return whitespace_rule_cfg;
  97         }
  98 }
  99
 100 /* The returned string should be freed by the caller. */
 101 char *whitespace_error_string(unsigned ws)
 102 {
 103         struct strbuf err;
 104         strbuf_init(&err, 0);
 105         if (ws & WS_TRAILING_SPACE)
 106                 strbuf_addstr(&err, "trailing whitespace");
 107         if (ws & WS_SPACE_BEFORE_TAB) {
 108                 if (err.len)
 109                         strbuf_addstr(&err, ", ");
 110                 strbuf_addstr(&err, "space before tab in indent");
 111         }
 112         if (ws & WS_INDENT_WITH_NON_TAB) {
 113                 if (err.len)
 114                         strbuf_addstr(&err, ", ");
 115                 strbuf_addstr(&err, "indent with spaces");
 116         }
 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         return strbuf_detach(&err, NULL);
 123 }
 124
 125 /* If stream is non-NULL, emits the line after checking. */
 126 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 127                                 FILE *stream, const char *set,
 128                                 const char *reset, const char *ws)
 129 {
 130         unsigned result = 0;
 131         int written = 0;
 132         int trailing_whitespace = -1;
 133         int trailing_newline = 0;
 134         int trailing_carriage_return = 0;
 135         int i;
 136
 137         /* Logic is simpler if we temporarily ignore the trailing newline. */
 138         if (len > 0 && line[len - 1] == '\n') {
 139                 trailing_newline = 1;
 140                 len--;
 141         }
 142         if ((ws_rule & WS_CR_AT_EOL) &&
 143             len > 0 && line[len - 1] == '\r') {
 144                 trailing_carriage_return = 1;
 145                 len--;
 146         }
 147
 148         /* Check for trailing whitespace. */
 149         if (ws_rule & WS_TRAILING_SPACE) {
 150                 for (i = len - 1; i >= 0; i--) {
 151                         if (isspace(line[i])) {
 152                                 trailing_whitespace = i;
 153                                 result |= WS_TRAILING_SPACE;
 154                         }
 155                         else
 156                                 break;
 157                 }
 158         }
 159
 160         /* Check for space before tab in initial indent. */
 161         for (i = 0; i < len; i++) {
 162                 if (line[i] == ' ')
 163                         continue;
 164                 if (line[i] != '\t')
 165                         break;
 166                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 167                         result |= WS_SPACE_BEFORE_TAB;
 168                         if (stream) {
 169                                 fputs(ws, stream);
 170                                 fwrite(line + written, i - written, 1, stream);
 171                                 fputs(reset, stream);
 172                         }
 173                 } else if (stream)
 174                         fwrite(line + written, i - written, 1, stream);
 175                 if (stream)
 176                         fwrite(line + i, 1, 1, stream);
 177                 written = i + 1;
 178         }
 179
 180         /* Check for indent using non-tab. */
 181         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= 8) {
 182                 result |= WS_INDENT_WITH_NON_TAB;
 183                 if (stream) {
 184                         fputs(ws, stream);
 185                         fwrite(line + written, i - written, 1, stream);
 186                         fputs(reset, stream);
 187                 }
 188                 written = i;
 189         }
 190
 191         if (stream) {
 192                 /*
 193                  * Now the rest of the line starts at "written".
 194                  * The non-highlighted part ends at "trailing_whitespace".
 195                  */
 196                 if (trailing_whitespace == -1)
 197                         trailing_whitespace = len;
 198
 199                 /* Emit non-highlighted (middle) segment. */
 200                 if (trailing_whitespace - written > 0) {
 201                         fputs(set, stream);
 202                         fwrite(line + written,
 203                             trailing_whitespace - written, 1, stream);
 204                         fputs(reset, stream);
 205                 }
 206
 207                 /* Highlight errors in trailing whitespace. */
 208                 if (trailing_whitespace != len) {
 209                         fputs(ws, stream);
 210                         fwrite(line + trailing_whitespace,
 211                             len - trailing_whitespace, 1, stream);
 212                         fputs(reset, stream);
 213                 }
 214                 if (trailing_carriage_return)
 215                         fputc('\r', stream);
 216                 if (trailing_newline)
 217                         fputc('\n', stream);
 218         }
 219         return result;
 220 }
 221
 222 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 223                    FILE *stream, const char *set,
 224                    const char *reset, const char *ws)
 225 {
 226         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 227 }
 228
 229 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 230 {
 231         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 232 }
 233
 234 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 235 {
 236         /*
 237          * We _might_ want to treat CR differently from other
 238          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 239          * for now we just use this stupid definition.
 240          */
 241         while (len-- > 0) {
 242                 if (!isspace(*line))
 243                         return 0;
 244                 line++;
 245         }
 246         return 1;
 247 }
 248
 249 /* Copy the line to the buffer while fixing whitespaces */
 250 int ws_fix_copy(char *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 251 {
 252         /*
 253          * len is number of bytes to be copied from src, starting
 254          * at src.  Typically src[len-1] is '\n', unless this is
 255          * the incomplete last line.
 256          */
 257         int i;
 258         int add_nl_to_tail = 0;
 259         int add_cr_to_tail = 0;
 260         int fixed = 0;
 261         int last_tab_in_indent = -1;
 262         int last_space_in_indent = -1;
 263         int need_fix_leading_space = 0;
 264         char *buf;
 265
 266         /*
 267          * Strip trailing whitespace
 268          */
 269         if ((ws_rule & WS_TRAILING_SPACE) &&
 270             (2 <= len && isspace(src[len-2]))) {
 271                 if (src[len - 1] == '\n') {
 272                         add_nl_to_tail = 1;
 273                         len--;
 274                         if (1 < len && src[len - 1] == '\r') {
 275                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 276                                 len--;
 277                         }
 278                 }
 279                 if (0 < len && isspace(src[len - 1])) {
 280                         while (0 < len && isspace(src[len-1]))
 281                                 len--;
 282                         fixed = 1;
 283                 }
 284         }
 285
 286         /*
 287          * Check leading whitespaces (indent)
 288          */
 289         for (i = 0; i < len; i++) {
 290                 char ch = src[i];
 291                 if (ch == '\t') {
 292                         last_tab_in_indent = i;
 293                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 294                             0 <= last_space_in_indent)
 295                             need_fix_leading_space = 1;
 296                 } else if (ch == ' ') {
 297                         last_space_in_indent = i;
 298                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 299                             8 <= i - last_tab_in_indent)
 300                                 need_fix_leading_space = 1;
 301                 } else
 302                         break;
 303         }
 304
 305         buf = dst;
 306         if (need_fix_leading_space) {
 307                 /* Process indent ourselves */
 308                 int consecutive_spaces = 0;
 309                 int last = last_tab_in_indent + 1;
 310
 311                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 312                         /* have "last" point at one past the indent */
 313                         if (last_tab_in_indent < last_space_in_indent)
 314                                 last = last_space_in_indent + 1;
 315                         else
 316                                 last = last_tab_in_indent + 1;
 317                 }
 318
 319                 /*
 320                  * between src[0..last-1], strip the funny spaces,
 321                  * updating them to tab as needed.
 322                  */
 323                 for (i = 0; i < last; i++) {
 324                         char ch = src[i];
 325                         if (ch != ' ') {
 326                                 consecutive_spaces = 0;
 327                                 *dst++ = ch;
 328                         } else {
 329                                 consecutive_spaces++;
 330                                 if (consecutive_spaces == 8) {
 331                                         *dst++ = '\t';
 332                                         consecutive_spaces = 0;
 333                                 }
 334                         }
 335                 }
 336                 while (0 < consecutive_spaces--)
 337                         *dst++ = ' ';
 338                 len -= last;
 339                 src += last;
 340                 fixed = 1;
 341         }
 342
 343         memcpy(dst, src, len);
 344         if (add_cr_to_tail)
 345                 dst[len++] = '\r';
 346         if (add_nl_to_tail)
 347                 dst[len++] = '\n';
 348         if (fixed && error_count)
 349                 (*error_count)++;
 350         return dst + len - buf;
 351 }