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;
  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 };
  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                                 if (!whitespace_rule_names[i].loosens_error)
  84                                         all_rule |= whitespace_rule_names[i].rule_bits;
  85                         return all_rule;
  86                 } else if (ATTR_FALSE(value)) {
  87                         /* false (-whitespace) */
  88                         return 0;
  89                 } else if (ATTR_UNSET(value)) {
  90                         /* reset to default (!whitespace) */
  91                         return whitespace_rule_cfg;
  92                 } else {
  93                         /* string */
  94                         return parse_whitespace_rule(value);
  95                 }
  96         } else {
  97                 return whitespace_rule_cfg;
  98         }
  99 }
 100
 101 /* The returned string should be freed by the caller. */
 102 char *whitespace_error_string(unsigned ws)
 103 {
 104         struct strbuf err = STRBUF_INIT;
 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         return strbuf_detach(&err, NULL);
 118 }
 119
 120 /* If stream is non-NULL, emits the line after checking. */
 121 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 122                                 FILE *stream, const char *set,
 123                                 const char *reset, const char *ws)
 124 {
 125         unsigned result = 0;
 126         int written = 0;
 127         int trailing_whitespace = -1;
 128         int trailing_newline = 0;
 129         int trailing_carriage_return = 0;
 130         int i;
 131
 132         /* Logic is simpler if we temporarily ignore the trailing newline. */
 133         if (len > 0 && line[len - 1] == '\n') {
 134                 trailing_newline = 1;
 135                 len--;
 136         }
 137         if ((ws_rule & WS_CR_AT_EOL) &&
 138             len > 0 && line[len - 1] == '\r') {
 139                 trailing_carriage_return = 1;
 140                 len--;
 141         }
 142
 143         /* Check for trailing whitespace. */
 144         if (ws_rule & WS_TRAILING_SPACE) {
 145                 for (i = len - 1; i >= 0; i--) {
 146                         if (isspace(line[i])) {
 147                                 trailing_whitespace = i;
 148                                 result |= WS_TRAILING_SPACE;
 149                         }
 150                         else
 151                                 break;
 152                 }
 153         }
 154
 155         /* Check for space before tab in initial indent. */
 156         for (i = 0; i < len; i++) {
 157                 if (line[i] == ' ')
 158                         continue;
 159                 if (line[i] != '\t')
 160                         break;
 161                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 162                         result |= WS_SPACE_BEFORE_TAB;
 163                         if (stream) {
 164                                 fputs(ws, stream);
 165                                 fwrite(line + written, i - written, 1, stream);
 166                                 fputs(reset, stream);
 167                         }
 168                 } else if (stream)
 169                         fwrite(line + written, i - written, 1, stream);
 170                 if (stream)
 171                         fwrite(line + i, 1, 1, stream);
 172                 written = i + 1;
 173         }
 174
 175         /* Check for indent using non-tab. */
 176         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= 8) {
 177                 result |= WS_INDENT_WITH_NON_TAB;
 178                 if (stream) {
 179                         fputs(ws, stream);
 180                         fwrite(line + written, i - written, 1, stream);
 181                         fputs(reset, stream);
 182                 }
 183                 written = i;
 184         }
 185
 186         if (stream) {
 187                 /*
 188                  * Now the rest of the line starts at "written".
 189                  * The non-highlighted part ends at "trailing_whitespace".
 190                  */
 191                 if (trailing_whitespace == -1)
 192                         trailing_whitespace = len;
 193
 194                 /* Emit non-highlighted (middle) segment. */
 195                 if (trailing_whitespace - written > 0) {
 196                         fputs(set, stream);
 197                         fwrite(line + written,
 198                             trailing_whitespace - written, 1, stream);
 199                         fputs(reset, stream);
 200                 }
 201
 202                 /* Highlight errors in trailing whitespace. */
 203                 if (trailing_whitespace != len) {
 204                         fputs(ws, stream);
 205                         fwrite(line + trailing_whitespace,
 206                             len - trailing_whitespace, 1, stream);
 207                         fputs(reset, stream);
 208                 }
 209                 if (trailing_carriage_return)
 210                         fputc('\r', stream);
 211                 if (trailing_newline)
 212                         fputc('\n', stream);
 213         }
 214         return result;
 215 }
 216
 217 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 218                    FILE *stream, const char *set,
 219                    const char *reset, const char *ws)
 220 {
 221         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 222 }
 223
 224 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 225 {
 226         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 227 }
 228
 229 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 230 {
 231         /*
 232          * We _might_ want to treat CR differently from other
 233          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 234          * for now we just use this stupid definition.
 235          */
 236         while (len-- > 0) {
 237                 if (!isspace(*line))
 238                         return 0;
 239                 line++;
 240         }
 241         return 1;
 242 }
 243
 244 /* Copy the line to the buffer while fixing whitespaces */
 245 int ws_fix_copy(char *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 246 {
 247         /*
 248          * len is number of bytes to be copied from src, starting
 249          * at src.  Typically src[len-1] is '\n', unless this is
 250          * the incomplete last line.
 251          */
 252         int i;
 253         int add_nl_to_tail = 0;
 254         int add_cr_to_tail = 0;
 255         int fixed = 0;
 256         int last_tab_in_indent = -1;
 257         int last_space_in_indent = -1;
 258         int need_fix_leading_space = 0;
 259         char *buf;
 260
 261         /*
 262          * Strip trailing whitespace
 263          */
 264         if ((ws_rule & WS_TRAILING_SPACE) &&
 265             (2 <= len && isspace(src[len-2]))) {
 266                 if (src[len - 1] == '\n') {
 267                         add_nl_to_tail = 1;
 268                         len--;
 269                         if (1 < len && src[len - 1] == '\r') {
 270                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 271                                 len--;
 272                         }
 273                 }
 274                 if (0 < len && isspace(src[len - 1])) {
 275                         while (0 < len && isspace(src[len-1]))
 276                                 len--;
 277                         fixed = 1;
 278                 }
 279         }
 280
 281         /*
 282          * Check leading whitespaces (indent)
 283          */
 284         for (i = 0; i < len; i++) {
 285                 char ch = src[i];
 286                 if (ch == '\t') {
 287                         last_tab_in_indent = i;
 288                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 289                             0 <= last_space_in_indent)
 290                             need_fix_leading_space = 1;
 291                 } else if (ch == ' ') {
 292                         last_space_in_indent = i;
 293                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 294                             8 <= i - last_tab_in_indent)
 295                                 need_fix_leading_space = 1;
 296                 } else
 297                         break;
 298         }
 299
 300         buf = dst;
 301         if (need_fix_leading_space) {
 302                 /* Process indent ourselves */
 303                 int consecutive_spaces = 0;
 304                 int last = last_tab_in_indent + 1;
 305
 306                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 307                         /* have "last" point at one past the indent */
 308                         if (last_tab_in_indent < last_space_in_indent)
 309                                 last = last_space_in_indent + 1;
 310                         else
 311                                 last = last_tab_in_indent + 1;
 312                 }
 313
 314                 /*
 315                  * between src[0..last-1], strip the funny spaces,
 316                  * updating them to tab as needed.
 317                  */
 318                 for (i = 0; i < last; i++) {
 319                         char ch = src[i];
 320                         if (ch != ' ') {
 321                                 consecutive_spaces = 0;
 322                                 *dst++ = ch;
 323                         } else {
 324                                 consecutive_spaces++;
 325                                 if (consecutive_spaces == 8) {
 326                                         *dst++ = '\t';
 327                                         consecutive_spaces = 0;
 328                                 }
 329                         }
 330                 }
 331                 while (0 < consecutive_spaces--)
 332                         *dst++ = ' ';
 333                 len -= last;
 334                 src += last;
 335                 fixed = 1;
 336         }
 337
 338         memcpy(dst, src, len);
 339         if (add_cr_to_tail)
 340                 dst[len++] = '\r';
 341         if (add_nl_to_tail)
 342                 dst[len++] = '\n';
 343         if (fixed && error_count)
 344                 (*error_count)++;
 345         return dst + len - buf;
 346 }