| 1 | /* |
| 2 | * html.c: implementation of html.h. |
| 3 | */ |
| 4 | |
| 5 | #include "agedu.h" |
| 6 | #include "html.h" |
| 7 | #include "alloc.h" |
| 8 | #include "trie.h" |
| 9 | #include "index.h" |
| 10 | |
| 11 | #define MAXCOLOUR 511 |
| 12 | |
| 13 | struct html { |
| 14 | char *buf; |
| 15 | size_t buflen, bufsize; |
| 16 | const void *t; |
| 17 | unsigned long long totalsize, oldest, newest; |
| 18 | char *path2; |
| 19 | char *oururi; |
| 20 | size_t hreflen; |
| 21 | const char *uriformat; |
| 22 | unsigned long long thresholds[MAXCOLOUR]; |
| 23 | char *titletexts[MAXCOLOUR+1]; |
| 24 | time_t now; |
| 25 | }; |
| 26 | |
| 27 | static void vhtprintf(struct html *ctx, const char *fmt, va_list ap) |
| 28 | { |
| 29 | va_list ap2; |
| 30 | int size, size2; |
| 31 | char testbuf[2]; |
| 32 | |
| 33 | va_copy(ap2, ap); |
| 34 | /* |
| 35 | * Some C libraries (Solaris, I'm looking at you) don't like |
| 36 | * an output buffer size of zero in vsnprintf, but will return |
| 37 | * sensible values given any non-zero buffer size. Hence, we |
| 38 | * use testbuf to gauge the length of the string. |
| 39 | */ |
| 40 | size = vsnprintf(testbuf, 1, fmt, ap2); |
| 41 | va_end(ap2); |
| 42 | |
| 43 | if (ctx->buflen + size >= ctx->bufsize) { |
| 44 | ctx->bufsize = (ctx->buflen + size) * 3 / 2 + 1024; |
| 45 | ctx->buf = sresize(ctx->buf, ctx->bufsize, char); |
| 46 | } |
| 47 | size2 = vsnprintf(ctx->buf + ctx->buflen, ctx->bufsize - ctx->buflen, |
| 48 | fmt, ap); |
| 49 | assert(size == size2); |
| 50 | ctx->buflen += size; |
| 51 | } |
| 52 | |
| 53 | static void htprintf(struct html *ctx, const char *fmt, ...) |
| 54 | { |
| 55 | va_list ap; |
| 56 | va_start(ap, fmt); |
| 57 | vhtprintf(ctx, fmt, ap); |
| 58 | va_end(ap); |
| 59 | } |
| 60 | |
| 61 | static unsigned long long round_and_format_age(struct html *ctx, |
| 62 | unsigned long long age, |
| 63 | char *buf, int direction) |
| 64 | { |
| 65 | struct tm tm, tm2; |
| 66 | char newbuf[80]; |
| 67 | unsigned long long ret, newret; |
| 68 | int i; |
| 69 | int ym; |
| 70 | static const int minutes[] = { 5, 10, 15, 30, 45 }; |
| 71 | |
| 72 | tm = *localtime(&ctx->now); |
| 73 | ym = tm.tm_year * 12 + tm.tm_mon; |
| 74 | |
| 75 | ret = ctx->now; |
| 76 | strcpy(buf, "Now"); |
| 77 | |
| 78 | for (i = 0; i < lenof(minutes); i++) { |
| 79 | newret = ctx->now - minutes[i] * 60; |
| 80 | sprintf(newbuf, "%d minutes", minutes[i]); |
| 81 | if (newret < age) |
| 82 | goto finish; |
| 83 | strcpy(buf, newbuf); |
| 84 | ret = newret; |
| 85 | } |
| 86 | |
| 87 | for (i = 1; i < 24; i++) { |
| 88 | newret = ctx->now - i * (60*60); |
| 89 | sprintf(newbuf, "%d hour%s", i, i==1 ? "" : "s"); |
| 90 | if (newret < age) |
| 91 | goto finish; |
| 92 | strcpy(buf, newbuf); |
| 93 | ret = newret; |
| 94 | } |
| 95 | |
| 96 | for (i = 1; i < 7; i++) { |
| 97 | newret = ctx->now - i * (24*60*60); |
| 98 | sprintf(newbuf, "%d day%s", i, i==1 ? "" : "s"); |
| 99 | if (newret < age) |
| 100 | goto finish; |
| 101 | strcpy(buf, newbuf); |
| 102 | ret = newret; |
| 103 | } |
| 104 | |
| 105 | for (i = 1; i < 4; i++) { |
| 106 | newret = ctx->now - i * (7*24*60*60); |
| 107 | sprintf(newbuf, "%d week%s", i, i==1 ? "" : "s"); |
| 108 | if (newret < age) |
| 109 | goto finish; |
| 110 | strcpy(buf, newbuf); |
| 111 | ret = newret; |
| 112 | } |
| 113 | |
| 114 | for (i = 1; i < 11; i++) { |
| 115 | tm2 = tm; /* structure copy */ |
| 116 | tm2.tm_year = (ym - i) / 12; |
| 117 | tm2.tm_mon = (ym - i) % 12; |
| 118 | newret = mktime(&tm2); |
| 119 | sprintf(newbuf, "%d month%s", i, i==1 ? "" : "s"); |
| 120 | if (newret < age) |
| 121 | goto finish; |
| 122 | strcpy(buf, newbuf); |
| 123 | ret = newret; |
| 124 | } |
| 125 | |
| 126 | for (i = 1;; i++) { |
| 127 | tm2 = tm; /* structure copy */ |
| 128 | tm2.tm_year = (ym - i*12) / 12; |
| 129 | tm2.tm_mon = (ym - i*12) % 12; |
| 130 | newret = mktime(&tm2); |
| 131 | sprintf(newbuf, "%d year%s", i, i==1 ? "" : "s"); |
| 132 | if (newret < age) |
| 133 | goto finish; |
| 134 | strcpy(buf, newbuf); |
| 135 | ret = newret; |
| 136 | } |
| 137 | |
| 138 | finish: |
| 139 | if (direction > 0) { |
| 140 | /* |
| 141 | * Round toward newest, i.e. use the existing (buf,ret). |
| 142 | */ |
| 143 | } else if (direction < 0) { |
| 144 | /* |
| 145 | * Round toward oldest, i.e. use (newbuf,newret); |
| 146 | */ |
| 147 | strcpy(buf, newbuf); |
| 148 | ret = newret; |
| 149 | } else { |
| 150 | /* |
| 151 | * Round to nearest. |
| 152 | */ |
| 153 | if (ret - age > age - newret) { |
| 154 | strcpy(buf, newbuf); |
| 155 | ret = newret; |
| 156 | } |
| 157 | } |
| 158 | return ret; |
| 159 | } |
| 160 | |
| 161 | static void get_indices(const void *t, char *path, |
| 162 | unsigned long *xi1, unsigned long *xi2) |
| 163 | { |
| 164 | size_t pathlen = strlen(path); |
| 165 | int c1 = path[pathlen], c2 = (pathlen > 0 ? path[pathlen-1] : 0); |
| 166 | |
| 167 | *xi1 = trie_before(t, path); |
| 168 | make_successor(path); |
| 169 | *xi2 = trie_before(t, path); |
| 170 | path[pathlen] = c1; |
| 171 | if (pathlen > 0) |
| 172 | path[pathlen-1] = c2; |
| 173 | } |
| 174 | |
| 175 | static unsigned long long fetch_size(const void *t, |
| 176 | unsigned long xi1, unsigned long xi2, |
| 177 | unsigned long long atime) |
| 178 | { |
| 179 | if (xi2 - xi1 == 1) { |
| 180 | /* |
| 181 | * We are querying an individual file, so we should not |
| 182 | * depend on the index entries either side of the node, |
| 183 | * since they almost certainly don't both exist. Instead, |
| 184 | * just look up the file's size and atime in the main trie. |
| 185 | */ |
| 186 | const struct trie_file *f = trie_getfile(t, xi1); |
| 187 | if (f->atime < atime) |
| 188 | return f->size; |
| 189 | else |
| 190 | return 0; |
| 191 | } else { |
| 192 | return index_query(t, xi2, atime) - index_query(t, xi1, atime); |
| 193 | } |
| 194 | } |
| 195 | |
| 196 | static void htescape(struct html *ctx, const char *s, int n, int italics) |
| 197 | { |
| 198 | while (n > 0 && *s) { |
| 199 | unsigned char c = (unsigned char)*s++; |
| 200 | |
| 201 | if (c == '&') |
| 202 | htprintf(ctx, "&"); |
| 203 | else if (c == '<') |
| 204 | htprintf(ctx, "<"); |
| 205 | else if (c == '>') |
| 206 | htprintf(ctx, ">"); |
| 207 | else if (c >= ' ' && c < '\177') |
| 208 | htprintf(ctx, "%c", c); |
| 209 | else { |
| 210 | if (italics) htprintf(ctx, "<i>"); |
| 211 | htprintf(ctx, "[%02x]", c); |
| 212 | if (italics) htprintf(ctx, "</i>"); |
| 213 | } |
| 214 | |
| 215 | n--; |
| 216 | } |
| 217 | } |
| 218 | |
| 219 | static void begin_colour_bar(struct html *ctx) |
| 220 | { |
| 221 | htprintf(ctx, "<table cellspacing=0 cellpadding=0" |
| 222 | " style=\"border:0\">\n<tr>\n"); |
| 223 | } |
| 224 | |
| 225 | static void add_to_colour_bar(struct html *ctx, int colour, int pixels) |
| 226 | { |
| 227 | int r, g, b; |
| 228 | |
| 229 | if (colour >= 0 && colour < 256) /* red -> yellow fade */ |
| 230 | r = 255, g = colour, b = 0; |
| 231 | else if (colour >= 256 && colour <= 511) /* yellow -> green fade */ |
| 232 | r = 511 - colour, g = 255, b = 0; |
| 233 | else /* background grey */ |
| 234 | r = g = b = 240; |
| 235 | |
| 236 | if (pixels > 0) { |
| 237 | htprintf(ctx, "<td style=\"width:%dpx; height:1em; " |
| 238 | "background-color:#%02x%02x%02x\"", |
| 239 | pixels, r, g, b); |
| 240 | if (colour >= 0) |
| 241 | htprintf(ctx, " title=\"%s\"", ctx->titletexts[colour]); |
| 242 | htprintf(ctx, "></td>\n"); |
| 243 | } |
| 244 | } |
| 245 | |
| 246 | static void end_colour_bar(struct html *ctx) |
| 247 | { |
| 248 | htprintf(ctx, "</tr>\n</table>\n"); |
| 249 | } |
| 250 | |
| 251 | struct vector { |
| 252 | int want_href, essential; |
| 253 | char *name; |
| 254 | int literal; /* should the name be formatted in fixed-pitch? */ |
| 255 | unsigned long index; |
| 256 | unsigned long long sizes[MAXCOLOUR+1]; |
| 257 | }; |
| 258 | |
| 259 | int vec_compare(const void *av, const void *bv) |
| 260 | { |
| 261 | const struct vector *a = *(const struct vector **)av; |
| 262 | const struct vector *b = *(const struct vector **)bv; |
| 263 | |
| 264 | if (a->sizes[MAXCOLOUR] > b->sizes[MAXCOLOUR]) |
| 265 | return -1; |
| 266 | else if (a->sizes[MAXCOLOUR] < b->sizes[MAXCOLOUR]) |
| 267 | return +1; |
| 268 | else if (a->want_href < b->want_href) |
| 269 | return +1; |
| 270 | else if (a->want_href > b->want_href) |
| 271 | return -1; |
| 272 | else if (a->want_href) |
| 273 | return strcmp(a->name, b->name); |
| 274 | else if (a->index < b->index) |
| 275 | return -1; |
| 276 | else if (a->index > b->index) |
| 277 | return +1; |
| 278 | else if (a->essential < b->essential) |
| 279 | return +1; |
| 280 | else if (a->essential > b->essential) |
| 281 | return -1; |
| 282 | return 0; |
| 283 | } |
| 284 | |
| 285 | static struct vector *make_vector(struct html *ctx, char *path, |
| 286 | int want_href, int essential, |
| 287 | char *name, int literal) |
| 288 | { |
| 289 | unsigned long xi1, xi2; |
| 290 | struct vector *vec = snew(struct vector); |
| 291 | int i; |
| 292 | |
| 293 | vec->want_href = want_href; |
| 294 | vec->essential = essential; |
| 295 | vec->name = name ? dupstr(name) : NULL; |
| 296 | vec->literal = literal; |
| 297 | |
| 298 | get_indices(ctx->t, path, &xi1, &xi2); |
| 299 | |
| 300 | vec->index = xi1; |
| 301 | |
| 302 | for (i = 0; i <= MAXCOLOUR; i++) { |
| 303 | unsigned long long atime; |
| 304 | if (i == MAXCOLOUR) |
| 305 | atime = ULLONG_MAX; |
| 306 | else |
| 307 | atime = ctx->thresholds[i]; |
| 308 | vec->sizes[i] = fetch_size(ctx->t, xi1, xi2, atime); |
| 309 | } |
| 310 | |
| 311 | return vec; |
| 312 | } |
| 313 | |
| 314 | static void print_heading(struct html *ctx, const char *title) |
| 315 | { |
| 316 | htprintf(ctx, "<tr style=\"padding: 0.2em; background-color:#e0e0e0\">\n" |
| 317 | "<td colspan=4 align=center>%s</td>\n</tr>\n", title); |
| 318 | } |
| 319 | |
| 320 | static void compute_display_size(unsigned long long size, |
| 321 | const char **fmt, double *display_size) |
| 322 | { |
| 323 | static const char *const fmts[] = { |
| 324 | "%g b", "%g Kb", "%#.1f Mb", "%#.1f Gb", "%#.1f Tb", |
| 325 | "%#.1f Pb", "%#.1f Eb", "%#.1f Zb", "%#.1f Yb" |
| 326 | }; |
| 327 | int shift = 0; |
| 328 | unsigned long long tmpsize; |
| 329 | double denominator; |
| 330 | |
| 331 | tmpsize = size; |
| 332 | denominator = 1.0; |
| 333 | while (tmpsize >= 1024 && shift < lenof(fmts)-1) { |
| 334 | tmpsize >>= 10; |
| 335 | denominator *= 1024.0; |
| 336 | shift++; |
| 337 | } |
| 338 | *display_size = size / denominator; |
| 339 | *fmt = fmts[shift]; |
| 340 | } |
| 341 | |
| 342 | struct format_option { |
| 343 | const char *prefix, *suffix; /* may include '%%' */ |
| 344 | int prefixlen, suffixlen; /* does not count '%%' */ |
| 345 | char fmttype; /* 0 for none, or 'n' or 'p' */ |
| 346 | int translate_pathsep; /* pathsep rendered as '/'? */ |
| 347 | int shorten_path; /* omit common prefix? */ |
| 348 | }; |
| 349 | |
| 350 | /* |
| 351 | * Gets the next format option from a format string. Advances '*fmt' |
| 352 | * past it, or sets it to NULL if nothing is left. |
| 353 | */ |
| 354 | struct format_option get_format_option(const char **fmt) |
| 355 | { |
| 356 | struct format_option ret; |
| 357 | |
| 358 | /* |
| 359 | * Scan for prefix of format. |
| 360 | */ |
| 361 | ret.prefix = *fmt; |
| 362 | ret.prefixlen = 0; |
| 363 | while (1) { |
| 364 | if (**fmt == '\0') { |
| 365 | /* |
| 366 | * No formatting directive, and this is the last option. |
| 367 | */ |
| 368 | ret.suffix = *fmt; |
| 369 | ret.suffixlen = 0; |
| 370 | ret.fmttype = '\0'; |
| 371 | *fmt = NULL; |
| 372 | return ret; |
| 373 | } else if (**fmt == '%') { |
| 374 | if ((*fmt)[1] == '%') { |
| 375 | (*fmt) += 2; /* just advance one extra */ |
| 376 | ret.prefixlen++; |
| 377 | } else if ((*fmt)[1] == '|') { |
| 378 | /* |
| 379 | * No formatting directive. |
| 380 | */ |
| 381 | ret.suffix = *fmt; |
| 382 | ret.suffixlen = 0; |
| 383 | ret.fmttype = '\0'; |
| 384 | (*fmt) += 2; /* advance to start of next option */ |
| 385 | return ret; |
| 386 | } else { |
| 387 | break; |
| 388 | } |
| 389 | } else { |
| 390 | (*fmt)++; /* normal character */ |
| 391 | ret.prefixlen++; |
| 392 | } |
| 393 | } |
| 394 | |
| 395 | /* |
| 396 | * Interpret formatting directive with flags. |
| 397 | */ |
| 398 | (*fmt)++; |
| 399 | ret.translate_pathsep = ret.shorten_path = 1; |
| 400 | while (1) { |
| 401 | char c = *(*fmt)++; |
| 402 | assert(c); |
| 403 | if (c == '/') { |
| 404 | ret.translate_pathsep = 0; |
| 405 | } else if (c == '-') { |
| 406 | ret.shorten_path = 0; |
| 407 | } else { |
| 408 | assert(c == 'n' || c == 'p'); |
| 409 | ret.fmttype = c; |
| 410 | break; |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | /* |
| 415 | * Scan for suffix. |
| 416 | */ |
| 417 | ret.suffix = *fmt; |
| 418 | ret.suffixlen = 0; |
| 419 | while (1) { |
| 420 | if (**fmt == '\0') { |
| 421 | /* |
| 422 | * This is the last option. |
| 423 | */ |
| 424 | *fmt = NULL; |
| 425 | return ret; |
| 426 | } else if (**fmt != '%') { |
| 427 | (*fmt)++; /* normal character */ |
| 428 | ret.suffixlen++; |
| 429 | } else { |
| 430 | if ((*fmt)[1] == '%') { |
| 431 | (*fmt) += 2; /* just advance one extra */ |
| 432 | ret.suffixlen++; |
| 433 | } else { |
| 434 | assert((*fmt)[1] == '|'); |
| 435 | (*fmt) += 2; /* advance to start of next option */ |
| 436 | return ret; |
| 437 | } |
| 438 | } |
| 439 | } |
| 440 | } |
| 441 | |
| 442 | char *format_string(const char *fmt, unsigned long index, const void *t) |
| 443 | { |
| 444 | int maxlen; |
| 445 | char *ret = NULL, *p = NULL; |
| 446 | char *path = NULL, *q = NULL; |
| 447 | char pathsep = trie_pathsep(t); |
| 448 | int maxpathlen = trie_maxpathlen(t); |
| 449 | int leading; |
| 450 | |
| 451 | while (fmt) { |
| 452 | struct format_option opt = get_format_option(&fmt); |
| 453 | if (index && !opt.fmttype) |
| 454 | continue; /* option is only good for the root, which this isn't */ |
| 455 | |
| 456 | maxlen = opt.prefixlen + opt.suffixlen + 1; |
| 457 | switch (opt.fmttype) { |
| 458 | case 'n': |
| 459 | maxlen += 40; /* generous length for an integer */ |
| 460 | break; |
| 461 | case 'p': |
| 462 | maxlen += 3*maxpathlen; /* might have to escape everything */ |
| 463 | break; |
| 464 | } |
| 465 | ret = snewn(maxlen, char); |
| 466 | p = ret; |
| 467 | while (opt.prefixlen-- > 0) { |
| 468 | if ((*p++ = *opt.prefix++) == '%') |
| 469 | opt.prefix++; |
| 470 | } |
| 471 | switch (opt.fmttype) { |
| 472 | case 'n': |
| 473 | p += sprintf(p, "%lu", index); |
| 474 | break; |
| 475 | case 'p': |
| 476 | path = snewn(1+trie_maxpathlen(t), char); |
| 477 | if (opt.shorten_path) { |
| 478 | trie_getpath(t, 0, path); |
| 479 | q = path + strlen(path); |
| 480 | trie_getpath(t, index, path); |
| 481 | if (*q == pathsep) |
| 482 | q++; |
| 483 | } else { |
| 484 | trie_getpath(t, index, path); |
| 485 | q = path; |
| 486 | } |
| 487 | leading = 1; |
| 488 | while (*q) { |
| 489 | char c = *q++; |
| 490 | if (c == pathsep && opt.translate_pathsep) { |
| 491 | *p++ = '/'; |
| 492 | leading = 1; |
| 493 | } else if (!isalnum((unsigned char)c) && |
| 494 | ((leading && c=='.') || !strchr("-.@_", c))) { |
| 495 | p += sprintf(p, "=%02X", (unsigned char)c); |
| 496 | leading = 0; |
| 497 | } else { |
| 498 | *p++ = c; |
| 499 | leading = 0; |
| 500 | } |
| 501 | } |
| 502 | sfree(path); |
| 503 | break; |
| 504 | } |
| 505 | while (opt.suffixlen-- > 0) { |
| 506 | if ((*p++ = *opt.suffix++) == '%') |
| 507 | opt.suffix++; |
| 508 | } |
| 509 | *p = '\0'; |
| 510 | assert(p - ret < maxlen); |
| 511 | return ret; |
| 512 | } |
| 513 | assert(!"Getting here implies an incomplete set of formats"); |
| 514 | } |
| 515 | |
| 516 | char *html_format_path(const void *t, const struct html_config *cfg, |
| 517 | unsigned long index) |
| 518 | { |
| 519 | return format_string(cfg->uriformat, index, t); |
| 520 | } |
| 521 | |
| 522 | int html_parse_path(const void *t, const char *path, |
| 523 | const struct html_config *cfg, unsigned long *index) |
| 524 | { |
| 525 | int len = strlen(path); |
| 526 | int midlen; |
| 527 | const char *p, *q; |
| 528 | char *r; |
| 529 | char pathsep = trie_pathsep(t); |
| 530 | const char *fmt = cfg->uriformat; |
| 531 | |
| 532 | while (fmt) { |
| 533 | struct format_option opt = get_format_option(&fmt); |
| 534 | |
| 535 | /* |
| 536 | * Check prefix and suffix. |
| 537 | */ |
| 538 | midlen = len - opt.prefixlen - opt.suffixlen; |
| 539 | if (midlen < 0) |
| 540 | continue; /* prefix and suffix don't even fit */ |
| 541 | |
| 542 | p = path; |
| 543 | while (opt.prefixlen > 0) { |
| 544 | char c = *opt.prefix++; |
| 545 | if (c == '%') |
| 546 | opt.prefix++; |
| 547 | if (*p != c) |
| 548 | break; |
| 549 | p++; |
| 550 | opt.prefixlen--; |
| 551 | } |
| 552 | if (opt.prefixlen > 0) |
| 553 | continue; /* prefix didn't match */ |
| 554 | |
| 555 | q = path + len - opt.suffixlen; |
| 556 | while (opt.suffixlen > 0) { |
| 557 | char c = *opt.suffix++; |
| 558 | if (c == '%') |
| 559 | opt.suffix++; |
| 560 | if (*q != c) |
| 561 | break; |
| 562 | q++; |
| 563 | opt.suffixlen--; |
| 564 | } |
| 565 | if (opt.suffixlen > 0) |
| 566 | continue; /* suffix didn't match */ |
| 567 | |
| 568 | /* |
| 569 | * Check the data in between. p points at it, and it's midlen |
| 570 | * characters long. |
| 571 | */ |
| 572 | if (opt.fmttype == '\0') { |
| 573 | if (midlen == 0) { |
| 574 | /* |
| 575 | * Successful match against a root format. |
| 576 | */ |
| 577 | *index = 0; |
| 578 | return 1; |
| 579 | } |
| 580 | } else if (opt.fmttype == 'n') { |
| 581 | *index = 0; |
| 582 | while (midlen > 0) { |
| 583 | if (*p >= '0' && *p <= '9') |
| 584 | *index = *index * 10 + (*p - '0'); |
| 585 | else |
| 586 | break; |
| 587 | midlen--; |
| 588 | p++; |
| 589 | } |
| 590 | if (midlen == 0) { |
| 591 | /* |
| 592 | * Successful match against a numeric format. |
| 593 | */ |
| 594 | return 1; |
| 595 | } |
| 596 | } else { |
| 597 | assert(opt.fmttype == 'p'); |
| 598 | |
| 599 | int maxoutlen = trie_maxpathlen(t) + 1; |
| 600 | int maxinlen = midlen + 1; |
| 601 | char triepath[maxinlen+maxoutlen]; |
| 602 | |
| 603 | if (opt.shorten_path) { |
| 604 | trie_getpath(t, 0, triepath); |
| 605 | r = triepath + strlen(triepath); |
| 606 | if (r > triepath && r[-1] != pathsep) |
| 607 | *r++ = pathsep; |
| 608 | } else { |
| 609 | r = triepath; |
| 610 | } |
| 611 | |
| 612 | while (midlen > 0) { |
| 613 | if (*p == '/' && opt.translate_pathsep) { |
| 614 | *r++ = pathsep; |
| 615 | p++; |
| 616 | midlen--; |
| 617 | } else if (*p == '=') { |
| 618 | /* |
| 619 | * We intentionally do not check whether the |
| 620 | * escaped character _should_ have been escaped |
| 621 | * according to the rules in html_format_path. |
| 622 | * |
| 623 | * All clients of this parsing function, after a |
| 624 | * successful parse, call html_format_path to find |
| 625 | * the canonical URI for the same index and return |
| 626 | * an HTTP redirect if the provided URI was not |
| 627 | * exactly equal to that canonical form. This is |
| 628 | * critical when the correction involves adding or |
| 629 | * removing a trailing slash (because then |
| 630 | * relative hrefs on the generated page can be |
| 631 | * computed with respect to the canonical URI |
| 632 | * instead of having to remember what the actual |
| 633 | * URI was), but also has the useful effect that |
| 634 | * if a user attempts to type in (guess) a URI by |
| 635 | * hand they don't have to remember the escaping |
| 636 | * rules - as long as they type _something_ that |
| 637 | * this code can parse into a recognisable |
| 638 | * pathname, it will be automatically 301ed into |
| 639 | * the canonical form. |
| 640 | */ |
| 641 | if (midlen < 3 || |
| 642 | !isxdigit((unsigned char)p[1]) || |
| 643 | !isxdigit((unsigned char)p[2])) |
| 644 | break; /* faulty escape encoding */ |
| 645 | char x[3]; |
| 646 | unsigned cval; |
| 647 | x[0] = p[1]; |
| 648 | x[1] = p[2]; |
| 649 | x[2] = '\0'; |
| 650 | sscanf(x, "%x", &cval); |
| 651 | *r++ = cval; |
| 652 | p += 3; |
| 653 | midlen -= 3; |
| 654 | } else { |
| 655 | *r++ = *p; |
| 656 | p++; |
| 657 | midlen--; |
| 658 | } |
| 659 | } |
| 660 | if (midlen > 0) |
| 661 | continue; /* something went wrong in that loop */ |
| 662 | assert(r - triepath < maxinlen+maxoutlen); |
| 663 | *r = '\0'; |
| 664 | |
| 665 | unsigned long gotidx = trie_before(t, triepath); |
| 666 | if (gotidx >= trie_count(t)) |
| 667 | continue; /* index out of range */ |
| 668 | char retpath[1+maxoutlen]; |
| 669 | trie_getpath(t, gotidx, retpath); |
| 670 | if (strcmp(triepath, retpath)) |
| 671 | continue; /* exact path not found in trie */ |
| 672 | if (!index_has_root(t, gotidx)) |
| 673 | continue; /* path is not a directory */ |
| 674 | |
| 675 | /* |
| 676 | * Successful path-based match. |
| 677 | */ |
| 678 | *index = gotidx; |
| 679 | return 1; |
| 680 | } |
| 681 | } |
| 682 | |
| 683 | return 0; /* no match from any format option */ |
| 684 | } |
| 685 | |
| 686 | char *make_href(const char *source, const char *target) |
| 687 | { |
| 688 | /* |
| 689 | * We insist that both source and target URIs start with a /, or |
| 690 | * else we won't be reliably able to construct relative hrefs |
| 691 | * between them (e.g. because we've got a suffix on the end of |
| 692 | * some CGI pathname that this function doesn't know the final |
| 693 | * component of). |
| 694 | */ |
| 695 | assert(*source == '/'); |
| 696 | assert(*target == '/'); |
| 697 | |
| 698 | /* |
| 699 | * Find the last / in source. Everything up to but not including |
| 700 | * that is the directory to which the output href will be |
| 701 | * relative. We enforce by assertion that there must be a / |
| 702 | * somewhere in source, or else we can't construct a relative href |
| 703 | * at all |
| 704 | */ |
| 705 | const char *sourceend = strrchr(source, '/'); |
| 706 | assert(sourceend != NULL); |
| 707 | |
| 708 | /* |
| 709 | * See how far the target URI agrees with the source one, up to |
| 710 | * and including that /. |
| 711 | */ |
| 712 | const char *s = source, *t = target; |
| 713 | while (s <= sourceend && *s == *t) |
| 714 | s++, t++; |
| 715 | |
| 716 | /* |
| 717 | * We're only interested in agreement of complete path components, |
| 718 | * so back off until we're sitting just after a shared /. |
| 719 | */ |
| 720 | while (s > source && s[-1] != '/') |
| 721 | s--, t--; |
| 722 | assert(s > source); |
| 723 | |
| 724 | /* |
| 725 | * Now we need some number of levels of "../" to get from source |
| 726 | * to here, and then we just replicate the rest of 'target'. |
| 727 | */ |
| 728 | int levels = 0; |
| 729 | while (s <= sourceend) { |
| 730 | if (*s == '/') |
| 731 | levels++; |
| 732 | s++; |
| 733 | } |
| 734 | int len = 3*levels + strlen(t); |
| 735 | if (len == 0) { |
| 736 | /* One last special case: if target has no tail _and_ we |
| 737 | * haven't written out any "../". */ |
| 738 | return dupstr("./"); |
| 739 | } else { |
| 740 | char *ret = snewn(len+1, char); |
| 741 | char *p = ret; |
| 742 | while (levels-- > 0) { |
| 743 | *p++ = '.'; |
| 744 | *p++ = '.'; |
| 745 | *p++ = '/'; |
| 746 | } |
| 747 | strcpy(p, t); |
| 748 | return ret; |
| 749 | } |
| 750 | } |
| 751 | |
| 752 | #define PIXEL_SIZE 600 /* FIXME: configurability? */ |
| 753 | static void write_report_line(struct html *ctx, struct vector *vec) |
| 754 | { |
| 755 | unsigned long long size, asize, divisor; |
| 756 | double display_size; |
| 757 | int pix, newpix; |
| 758 | int i; |
| 759 | const char *unitsfmt; |
| 760 | |
| 761 | /* |
| 762 | * A line with literally zero space usage should not be |
| 763 | * printed at all if it's a link to a subdirectory (since it |
| 764 | * probably means the whole thing was excluded by some |
| 765 | * --exclude-path wildcard). If it's [files] or the top-level |
| 766 | * line, though, we must always print _something_, and in that |
| 767 | * case we must fiddle about to prevent divisions by zero in |
| 768 | * the code below. |
| 769 | */ |
| 770 | if (!vec->sizes[MAXCOLOUR] && !vec->essential) |
| 771 | return; |
| 772 | divisor = ctx->totalsize; |
| 773 | if (!divisor) { |
| 774 | divisor = 1; |
| 775 | } |
| 776 | |
| 777 | /* |
| 778 | * Find the total size of this subdirectory. |
| 779 | */ |
| 780 | size = vec->sizes[MAXCOLOUR]; |
| 781 | compute_display_size(size, &unitsfmt, &display_size); |
| 782 | htprintf(ctx, "<tr>\n" |
| 783 | "<td style=\"padding: 0.2em; text-align: right\">"); |
| 784 | htprintf(ctx, unitsfmt, display_size); |
| 785 | htprintf(ctx, "</td>\n"); |
| 786 | |
| 787 | /* |
| 788 | * Generate a colour bar. |
| 789 | */ |
| 790 | htprintf(ctx, "<td style=\"padding: 0.2em\">\n"); |
| 791 | begin_colour_bar(ctx); |
| 792 | pix = 0; |
| 793 | for (i = 0; i <= MAXCOLOUR; i++) { |
| 794 | asize = vec->sizes[i]; |
| 795 | newpix = asize * PIXEL_SIZE / divisor; |
| 796 | add_to_colour_bar(ctx, i, newpix - pix); |
| 797 | pix = newpix; |
| 798 | } |
| 799 | add_to_colour_bar(ctx, -1, PIXEL_SIZE - pix); |
| 800 | end_colour_bar(ctx); |
| 801 | htprintf(ctx, "</td>\n"); |
| 802 | |
| 803 | /* |
| 804 | * Output size as a percentage of totalsize. |
| 805 | */ |
| 806 | htprintf(ctx, "<td style=\"padding: 0.2em; text-align: right\">" |
| 807 | "%.2f%%</td>\n", (double)size / divisor * 100.0); |
| 808 | |
| 809 | /* |
| 810 | * Output a subdirectory marker. |
| 811 | */ |
| 812 | htprintf(ctx, "<td style=\"padding: 0.2em\">"); |
| 813 | if (vec->name) { |
| 814 | int doing_href = 0; |
| 815 | |
| 816 | if (ctx->uriformat && vec->want_href) { |
| 817 | char *targeturi = format_string(ctx->uriformat, vec->index, |
| 818 | ctx->t); |
| 819 | char *link = make_href(ctx->oururi, targeturi); |
| 820 | htprintf(ctx, "<a href=\"%s\">", link); |
| 821 | sfree(link); |
| 822 | sfree(targeturi); |
| 823 | doing_href = 1; |
| 824 | } |
| 825 | if (vec->literal) |
| 826 | htprintf(ctx, "<code>"); |
| 827 | htescape(ctx, vec->name, strlen(vec->name), 1); |
| 828 | if (vec->literal) |
| 829 | htprintf(ctx, "</code>"); |
| 830 | if (doing_href) |
| 831 | htprintf(ctx, "</a>"); |
| 832 | } |
| 833 | htprintf(ctx, "</td>\n</tr>\n"); |
| 834 | } |
| 835 | |
| 836 | int strcmptrailingpathsep(const char *a, const char *b) |
| 837 | { |
| 838 | while (*a == *b && *a) |
| 839 | a++, b++; |
| 840 | |
| 841 | if ((*a == pathsep && !a[1] && !*b) || |
| 842 | (*b == pathsep && !b[1] && !*a)) |
| 843 | return 0; |
| 844 | |
| 845 | return (int)(unsigned char)*a - (int)(unsigned char)*b; |
| 846 | } |
| 847 | |
| 848 | char *html_query(const void *t, unsigned long index, |
| 849 | const struct html_config *cfg, int downlink) |
| 850 | { |
| 851 | struct html actx, *ctx = &actx; |
| 852 | char *path, *path2, *p, *q; |
| 853 | char agebuf1[80], agebuf2[80]; |
| 854 | size_t pathlen, subdirpos; |
| 855 | unsigned long index2; |
| 856 | int i; |
| 857 | struct vector **vecs; |
| 858 | int nvecs, vecsize; |
| 859 | unsigned long xi1, xi2, xj1, xj2; |
| 860 | |
| 861 | if (index >= trie_count(t)) |
| 862 | return NULL; |
| 863 | |
| 864 | ctx->buf = NULL; |
| 865 | ctx->buflen = ctx->bufsize = 0; |
| 866 | ctx->t = t; |
| 867 | ctx->uriformat = cfg->uriformat; |
| 868 | htprintf(ctx, "<html>\n"); |
| 869 | |
| 870 | path = snewn(1+trie_maxpathlen(t), char); |
| 871 | ctx->path2 = path2 = snewn(1+trie_maxpathlen(t), char); |
| 872 | if (cfg->uriformat) |
| 873 | ctx->oururi = format_string(cfg->uriformat, index, t); |
| 874 | else |
| 875 | ctx->oururi = NULL; |
| 876 | |
| 877 | /* |
| 878 | * HEAD section. |
| 879 | */ |
| 880 | htprintf(ctx, "<head>\n"); |
| 881 | trie_getpath(t, index, path); |
| 882 | htprintf(ctx, "<title>"); |
| 883 | htescape(ctx, cfg->html_title, strlen(cfg->html_title), 0); |
| 884 | htprintf(ctx, ": "); |
| 885 | htescape(ctx, path, strlen(path), 0); |
| 886 | htprintf(ctx, "</title>\n"); |
| 887 | htprintf(ctx, "</head>\n"); |
| 888 | |
| 889 | /* |
| 890 | * Begin BODY section. |
| 891 | */ |
| 892 | htprintf(ctx, "<body>\n"); |
| 893 | htprintf(ctx, "<h3 align=center>Disk space breakdown by" |
| 894 | " last-access time</h3>\n"); |
| 895 | |
| 896 | /* |
| 897 | * Show the pathname we're centred on, with hyperlinks to |
| 898 | * parent directories where available. |
| 899 | */ |
| 900 | htprintf(ctx, "<p align=center>\n<code>"); |
| 901 | q = path; |
| 902 | for (p = strchr(path, pathsep); p && p[1]; p = strchr(p, pathsep)) { |
| 903 | int doing_href = 0; |
| 904 | char c, *zp; |
| 905 | |
| 906 | /* |
| 907 | * See if this path prefix exists in the trie. If so, |
| 908 | * generate a hyperlink. |
| 909 | */ |
| 910 | zp = p; |
| 911 | if (p == path) /* special case for "/" at start */ |
| 912 | zp++; |
| 913 | |
| 914 | p++; |
| 915 | |
| 916 | c = *zp; |
| 917 | *zp = '\0'; |
| 918 | index2 = trie_before(t, path); |
| 919 | trie_getpath(t, index2, path2); |
| 920 | if (!strcmptrailingpathsep(path, path2) && cfg->uriformat) { |
| 921 | char *targeturi = format_string(cfg->uriformat, index2, t); |
| 922 | char *link = make_href(ctx->oururi, targeturi); |
| 923 | htprintf(ctx, "<a href=\"%s\">", link); |
| 924 | sfree(link); |
| 925 | sfree(targeturi); |
| 926 | doing_href = 1; |
| 927 | } |
| 928 | *zp = c; |
| 929 | htescape(ctx, q, zp - q, 1); |
| 930 | if (doing_href) |
| 931 | htprintf(ctx, "</a>"); |
| 932 | htescape(ctx, zp, p - zp, 1); |
| 933 | q = p; |
| 934 | } |
| 935 | htescape(ctx, q, strlen(q), 1); |
| 936 | htprintf(ctx, "</code>\n"); |
| 937 | |
| 938 | /* |
| 939 | * Decide on the age limit of our colour coding, establish the |
| 940 | * colour thresholds, and write out a key. |
| 941 | */ |
| 942 | ctx->now = time(NULL); |
| 943 | if (cfg->autoage) { |
| 944 | ctx->oldest = index_order_stat(t, 0.05); |
| 945 | ctx->newest = index_order_stat(t, 1.0); |
| 946 | ctx->oldest = round_and_format_age(ctx, ctx->oldest, agebuf1, -1); |
| 947 | ctx->newest = round_and_format_age(ctx, ctx->newest, agebuf2, +1); |
| 948 | } else { |
| 949 | ctx->oldest = cfg->oldest; |
| 950 | ctx->newest = cfg->newest; |
| 951 | ctx->oldest = round_and_format_age(ctx, ctx->oldest, agebuf1, 0); |
| 952 | ctx->newest = round_and_format_age(ctx, ctx->newest, agebuf2, 0); |
| 953 | } |
| 954 | for (i = 0; i < MAXCOLOUR; i++) { |
| 955 | ctx->thresholds[i] = |
| 956 | ctx->oldest + (ctx->newest - ctx->oldest) * i / (MAXCOLOUR-1); |
| 957 | } |
| 958 | for (i = 0; i <= MAXCOLOUR; i++) { |
| 959 | char buf[80]; |
| 960 | |
| 961 | if (i == 0) { |
| 962 | strcpy(buf, "< "); |
| 963 | round_and_format_age(ctx, ctx->thresholds[0], buf+5, 0); |
| 964 | } else if (i == MAXCOLOUR) { |
| 965 | strcpy(buf, "> "); |
| 966 | round_and_format_age(ctx, ctx->thresholds[MAXCOLOUR-1], buf+5, 0); |
| 967 | } else { |
| 968 | unsigned long long midrange = |
| 969 | (ctx->thresholds[i-1] + ctx->thresholds[i]) / 2; |
| 970 | round_and_format_age(ctx, midrange, buf, 0); |
| 971 | } |
| 972 | |
| 973 | ctx->titletexts[i] = dupstr(buf); |
| 974 | } |
| 975 | htprintf(ctx, "<p align=center>Key to colour coding (mouse over for more detail):\n"); |
| 976 | htprintf(ctx, "<p align=center style=\"padding: 0; margin-top:0.4em; " |
| 977 | "margin-bottom:1em\">"); |
| 978 | begin_colour_bar(ctx); |
| 979 | htprintf(ctx, "<td style=\"padding-right:1em\">%s</td>\n", agebuf1); |
| 980 | for (i = 0; i < MAXCOLOUR; i++) |
| 981 | add_to_colour_bar(ctx, i, 1); |
| 982 | htprintf(ctx, "<td style=\"padding-left:1em\">%s</td>\n", agebuf2); |
| 983 | end_colour_bar(ctx); |
| 984 | |
| 985 | /* |
| 986 | * Begin the main table. |
| 987 | */ |
| 988 | htprintf(ctx, "<p align=center>\n<table style=\"margin:0; border:0\">\n"); |
| 989 | |
| 990 | /* |
| 991 | * Find the total size of our entire subdirectory. We'll use |
| 992 | * that as the scale for all the colour bars in this report. |
| 993 | */ |
| 994 | get_indices(t, path, &xi1, &xi2); |
| 995 | ctx->totalsize = fetch_size(t, xi1, xi2, ULLONG_MAX); |
| 996 | |
| 997 | /* |
| 998 | * Generate a report line for the whole subdirectory. |
| 999 | */ |
| 1000 | vecsize = 64; |
| 1001 | vecs = snewn(vecsize, struct vector *); |
| 1002 | nvecs = 1; |
| 1003 | vecs[0] = make_vector(ctx, path, 0, 1, NULL, 0); |
| 1004 | print_heading(ctx, "Overall"); |
| 1005 | write_report_line(ctx, vecs[0]); |
| 1006 | |
| 1007 | /* |
| 1008 | * Now generate report lines for all its children, and the |
| 1009 | * files contained in it. |
| 1010 | */ |
| 1011 | print_heading(ctx, "Subdirectories"); |
| 1012 | |
| 1013 | vecs[0]->name = dupstr("[files]"); |
| 1014 | get_indices(t, path, &xi1, &xi2); |
| 1015 | xi1++; |
| 1016 | pathlen = strlen(path); |
| 1017 | subdirpos = pathlen + 1; |
| 1018 | if (pathlen > 0 && path[pathlen-1] == pathsep) |
| 1019 | subdirpos--; |
| 1020 | while (xi1 < xi2) { |
| 1021 | trie_getpath(t, xi1, path2); |
| 1022 | get_indices(t, ctx->path2, &xj1, &xj2); |
| 1023 | xi1 = xj2; |
| 1024 | if (!cfg->showfiles && xj2 - xj1 <= 1) |
| 1025 | continue; /* skip individual files */ |
| 1026 | if (nvecs >= vecsize) { |
| 1027 | vecsize = nvecs * 3 / 2 + 64; |
| 1028 | vecs = sresize(vecs, vecsize, struct vector *); |
| 1029 | } |
| 1030 | assert(strlen(path2) > pathlen); |
| 1031 | vecs[nvecs] = make_vector(ctx, path2, downlink && (xj2 - xj1 > 1), 0, |
| 1032 | path2 + subdirpos, 1); |
| 1033 | for (i = 0; i <= MAXCOLOUR; i++) |
| 1034 | vecs[0]->sizes[i] -= vecs[nvecs]->sizes[i]; |
| 1035 | nvecs++; |
| 1036 | } |
| 1037 | |
| 1038 | qsort(vecs, nvecs, sizeof(vecs[0]), vec_compare); |
| 1039 | |
| 1040 | for (i = 0; i < nvecs; i++) |
| 1041 | write_report_line(ctx, vecs[i]); |
| 1042 | |
| 1043 | /* |
| 1044 | * Close the main table. |
| 1045 | */ |
| 1046 | htprintf(ctx, "</table>\n"); |
| 1047 | |
| 1048 | /* |
| 1049 | * Finish up and tidy up. |
| 1050 | */ |
| 1051 | htprintf(ctx, "</body>\n"); |
| 1052 | htprintf(ctx, "</html>\n"); |
| 1053 | sfree(ctx->oururi); |
| 1054 | sfree(path2); |
| 1055 | sfree(path); |
| 1056 | for (i = 0; i < nvecs; i++) { |
| 1057 | sfree(vecs[i]->name); |
| 1058 | sfree(vecs[i]); |
| 1059 | } |
| 1060 | sfree(vecs); |
| 1061 | |
| 1062 | return ctx->buf; |
| 1063 | } |
| 1064 | |
| 1065 | int html_dump(const void *t, unsigned long index, unsigned long endindex, |
| 1066 | int maxdepth, const struct html_config *cfg, |
| 1067 | const char *pathprefix) |
| 1068 | { |
| 1069 | /* |
| 1070 | * Determine the filename for this file. |
| 1071 | */ |
| 1072 | assert(cfg->fileformat != NULL); |
| 1073 | char *filename = format_string(cfg->fileformat, index, t); |
| 1074 | char *path = dupfmt("%s%s", pathprefix, filename); |
| 1075 | sfree(filename); |
| 1076 | |
| 1077 | /* |
| 1078 | * Create the HTML itself. Don't write out downlinks from our |
| 1079 | * deepest level. |
| 1080 | */ |
| 1081 | char *html = html_query(t, index, cfg, maxdepth != 0); |
| 1082 | |
| 1083 | /* |
| 1084 | * Write it out. |
| 1085 | */ |
| 1086 | FILE *fp = fopen(path, "w"); |
| 1087 | if (!fp) { |
| 1088 | fprintf(stderr, "%s: %s: open: %s\n", PNAME, path, strerror(errno)); |
| 1089 | return 1; |
| 1090 | } |
| 1091 | if (fputs(html, fp) < 0) { |
| 1092 | fprintf(stderr, "%s: %s: write: %s\n", PNAME, path, strerror(errno)); |
| 1093 | fclose(fp); |
| 1094 | return 1; |
| 1095 | } |
| 1096 | if (fclose(fp) < 0) { |
| 1097 | fprintf(stderr, "%s: %s: fclose: %s\n", PNAME, path, strerror(errno)); |
| 1098 | return 1; |
| 1099 | } |
| 1100 | sfree(path); |
| 1101 | |
| 1102 | /* |
| 1103 | * Recurse. |
| 1104 | */ |
| 1105 | if (maxdepth != 0) { |
| 1106 | unsigned long subindex, subendindex; |
| 1107 | int newdepth = (maxdepth > 0 ? maxdepth - 1 : maxdepth); |
| 1108 | char rpath[1+trie_maxpathlen(t)]; |
| 1109 | |
| 1110 | index++; |
| 1111 | while (index < endindex) { |
| 1112 | trie_getpath(t, index, rpath); |
| 1113 | get_indices(t, rpath, &subindex, &subendindex); |
| 1114 | index = subendindex; |
| 1115 | if (subendindex - subindex > 1) { |
| 1116 | if (html_dump(t, subindex, subendindex, newdepth, |
| 1117 | cfg, pathprefix)) |
| 1118 | return 1; |
| 1119 | } |
| 1120 | } |
| 1121 | } |
| 1122 | return 0; |
| 1123 | } |