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 };
  19
  20 unsigned parse_whitespace_rule(const char *string)
  21 {
  22         unsigned rule = WS_DEFAULT_RULE;
  23
  24         while (string) {
  25                 int i;
  26                 size_t len;
  27                 const char *ep;
  28                 int negated = 0;
  29
  30                 string = string + strspn(string, ", \t\n\r");
  31                 ep = strchr(string, ',');
  32                 if (!ep)
  33                         len = strlen(string);
  34                 else
  35                         len = ep - string;
  36
  37                 if (*string == '-') {
  38                         negated = 1;
  39                         string++;
  40                         len--;
  41                 }
  42                 if (!len)
  43                         break;
  44                 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
  45                         if (strncmp(whitespace_rule_names[i].rule_name,
  46                                     string, len))
  47                                 continue;
  48                         if (negated)
  49                                 rule &= ~whitespace_rule_names[i].rule_bits;
  50                         else
  51                                 rule |= whitespace_rule_names[i].rule_bits;
  52                         break;
  53                 }
  54                 string = ep;
  55         }
  56         return rule;
  57 }
  58
  59 static void setup_whitespace_attr_check(struct git_attr_check *check)
  60 {
  61         static struct git_attr *attr_whitespace;
  62
  63         if (!attr_whitespace)
  64                 attr_whitespace = git_attr("whitespace", 10);
  65         check[0].attr = attr_whitespace;
  66 }
  67
  68 unsigned whitespace_rule(const char *pathname)
  69 {
  70         struct git_attr_check attr_whitespace_rule;
  71
  72         setup_whitespace_attr_check(&attr_whitespace_rule);
  73         if (!git_checkattr(pathname, 1, &attr_whitespace_rule)) {
  74                 const char *value;
  75
  76                 value = attr_whitespace_rule.value;
  77                 if (ATTR_TRUE(value)) {
  78                         /* true (whitespace) */
  79                         unsigned all_rule = 0;
  80                         int i;
  81                         for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  82                                 all_rule |= whitespace_rule_names[i].rule_bits;
  83                         return all_rule;
  84                 } else if (ATTR_FALSE(value)) {
  85                         /* false (-whitespace) */
  86                         return 0;
  87                 } else if (ATTR_UNSET(value)) {
  88                         /* reset to default (!whitespace) */
  89                         return whitespace_rule_cfg;
  90                 } else {
  91                         /* string */
  92                         return parse_whitespace_rule(value);
  93                 }
  94         } else {
  95                 return whitespace_rule_cfg;
  96         }
  97 }
  98
  99 /* The returned string should be freed by the caller. */
 100 char *whitespace_error_string(unsigned ws)
 101 {
 102         struct strbuf err = STRBUF_INIT;
 103         if (ws & WS_TRAILING_SPACE)
 104                 strbuf_addstr(&err, "trailing whitespace");
 105         if (ws & WS_SPACE_BEFORE_TAB) {
 106                 if (err.len)
 107                         strbuf_addstr(&err, ", ");
 108                 strbuf_addstr(&err, "space before tab in indent");
 109         }
 110         if (ws & WS_INDENT_WITH_NON_TAB) {
 111                 if (err.len)
 112                         strbuf_addstr(&err, ", ");
 113                 strbuf_addstr(&err, "indent with spaces");
 114         }
 115         return strbuf_detach(&err, NULL);
 116 }
 117
 118 /* If stream is non-NULL, emits the line after checking. */
 119 static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule,
 120                                 FILE *stream, const char *set,
 121                                 const char *reset, const char *ws)
 122 {
 123         unsigned result = 0;
 124         int written = 0;
 125         int trailing_whitespace = -1;
 126         int trailing_newline = 0;
 127         int trailing_carriage_return = 0;
 128         int i;
 129
 130         /* Logic is simpler if we temporarily ignore the trailing newline. */
 131         if (len > 0 && line[len - 1] == '\n') {
 132                 trailing_newline = 1;
 133                 len--;
 134         }
 135         if ((ws_rule & WS_CR_AT_EOL) &&
 136             len > 0 && line[len - 1] == '\r') {
 137                 trailing_carriage_return = 1;
 138                 len--;
 139         }
 140
 141         /* Check for trailing whitespace. */
 142         if (ws_rule & WS_TRAILING_SPACE) {
 143                 for (i = len - 1; i >= 0; i--) {
 144                         if (isspace(line[i])) {
 145                                 trailing_whitespace = i;
 146                                 result |= WS_TRAILING_SPACE;
 147                         }
 148                         else
 149                                 break;
 150                 }
 151         }
 152
 153         /* Check for space before tab in initial indent. */
 154         for (i = 0; i < len; i++) {
 155                 if (line[i] == ' ')
 156                         continue;
 157                 if (line[i] != '\t')
 158                         break;
 159                 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
 160                         result |= WS_SPACE_BEFORE_TAB;
 161                         if (stream) {
 162                                 fputs(ws, stream);
 163                                 fwrite(line + written, i - written, 1, stream);
 164                                 fputs(reset, stream);
 165                         }
 166                 } else if (stream)
 167                         fwrite(line + written, i - written, 1, stream);
 168                 if (stream)
 169                         fwrite(line + i, 1, 1, stream);
 170                 written = i + 1;
 171         }
 172
 173         /* Check for indent using non-tab. */
 174         if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= 8) {
 175                 result |= WS_INDENT_WITH_NON_TAB;
 176                 if (stream) {
 177                         fputs(ws, stream);
 178                         fwrite(line + written, i - written, 1, stream);
 179                         fputs(reset, stream);
 180                 }
 181                 written = i;
 182         }
 183
 184         if (stream) {
 185                 /*
 186                  * Now the rest of the line starts at "written".
 187                  * The non-highlighted part ends at "trailing_whitespace".
 188                  */
 189                 if (trailing_whitespace == -1)
 190                         trailing_whitespace = len;
 191
 192                 /* Emit non-highlighted (middle) segment. */
 193                 if (trailing_whitespace - written > 0) {
 194                         fputs(set, stream);
 195                         fwrite(line + written,
 196                             trailing_whitespace - written, 1, stream);
 197                         fputs(reset, stream);
 198                 }
 199
 200                 /* Highlight errors in trailing whitespace. */
 201                 if (trailing_whitespace != len) {
 202                         fputs(ws, stream);
 203                         fwrite(line + trailing_whitespace,
 204                             len - trailing_whitespace, 1, stream);
 205                         fputs(reset, stream);
 206                 }
 207                 if (trailing_carriage_return)
 208                         fputc('\r', stream);
 209                 if (trailing_newline)
 210                         fputc('\n', stream);
 211         }
 212         return result;
 213 }
 214
 215 void ws_check_emit(const char *line, int len, unsigned ws_rule,
 216                    FILE *stream, const char *set,
 217                    const char *reset, const char *ws)
 218 {
 219         (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
 220 }
 221
 222 unsigned ws_check(const char *line, int len, unsigned ws_rule)
 223 {
 224         return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
 225 }
 226
 227 int ws_blank_line(const char *line, int len, unsigned ws_rule)
 228 {
 229         /*
 230          * We _might_ want to treat CR differently from other
 231          * whitespace characters when ws_rule has WS_CR_AT_EOL, but
 232          * for now we just use this stupid definition.
 233          */
 234         while (len-- > 0) {
 235                 if (!isspace(*line))
 236                         return 0;
 237                 line++;
 238         }
 239         return 1;
 240 }
 241
 242 /* Copy the line to the buffer while fixing whitespaces */
 243 int ws_fix_copy(char *dst, const char *src, int len, unsigned ws_rule, int *error_count)
 244 {
 245         /*
 246          * len is number of bytes to be copied from src, starting
 247          * at src.  Typically src[len-1] is '\n', unless this is
 248          * the incomplete last line.
 249          */
 250         int i;
 251         int add_nl_to_tail = 0;
 252         int add_cr_to_tail = 0;
 253         int fixed = 0;
 254         int last_tab_in_indent = -1;
 255         int last_space_in_indent = -1;
 256         int need_fix_leading_space = 0;
 257         char *buf;
 258
 259         /*
 260          * Strip trailing whitespace
 261          */
 262         if ((ws_rule & WS_TRAILING_SPACE) &&
 263             (2 <= len && isspace(src[len-2]))) {
 264                 if (src[len - 1] == '\n') {
 265                         add_nl_to_tail = 1;
 266                         len--;
 267                         if (1 < len && src[len - 1] == '\r') {
 268                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 269                                 len--;
 270                         }
 271                 }
 272                 if (0 < len && isspace(src[len - 1])) {
 273                         while (0 < len && isspace(src[len-1]))
 274                                 len--;
 275                         fixed = 1;
 276                 }
 277         }
 278
 279         /*
 280          * Check leading whitespaces (indent)
 281          */
 282         for (i = 0; i < len; i++) {
 283                 char ch = src[i];
 284                 if (ch == '\t') {
 285                         last_tab_in_indent = i;
 286                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 287                             0 <= last_space_in_indent)
 288                             need_fix_leading_space = 1;
 289                 } else if (ch == ' ') {
 290                         last_space_in_indent = i;
 291                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 292                             8 <= i - last_tab_in_indent)
 293                                 need_fix_leading_space = 1;
 294                 } else
 295                         break;
 296         }
 297
 298         buf = dst;
 299         if (need_fix_leading_space) {
 300                 /* Process indent ourselves */
 301                 int consecutive_spaces = 0;
 302                 int last = last_tab_in_indent + 1;
 303
 304                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 305                         /* have "last" point at one past the indent */
 306                         if (last_tab_in_indent < last_space_in_indent)
 307                                 last = last_space_in_indent + 1;
 308                         else
 309                                 last = last_tab_in_indent + 1;
 310                 }
 311
 312                 /*
 313                  * between src[0..last-1], strip the funny spaces,
 314                  * updating them to tab as needed.
 315                  */
 316                 for (i = 0; i < last; i++) {
 317                         char ch = src[i];
 318                         if (ch != ' ') {
 319                                 consecutive_spaces = 0;
 320                                 *dst++ = ch;
 321                         } else {
 322                                 consecutive_spaces++;
 323                                 if (consecutive_spaces == 8) {
 324                                         *dst++ = '\t';
 325                                         consecutive_spaces = 0;
 326                                 }
 327                         }
 328                 }
 329                 while (0 < consecutive_spaces--)
 330                         *dst++ = ' ';
 331                 len -= last;
 332                 src += last;
 333                 fixed = 1;
 334         }
 335
 336         memcpy(dst, src, len);
 337         if (add_cr_to_tail)
 338                 dst[len++] = '\r';
 339         if (add_nl_to_tail)
 340                 dst[len++] = '\n';
 341         if (fixed && error_count)
 342                 (*error_count)++;
 343         return dst + len - buf;
 344 }