mdw@git.distorted.org.uk Git - sgt/agedu/blob - html.c

   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 *href;
  20     size_t hreflen;
  21     const char *format;
  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, "&amp;");
 203         else if (c == '<')
 204             htprintf(ctx, "&lt;");
 205         else if (c == '>')
 206             htprintf(ctx, "&gt;");
 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
 329     while (size >= 1024 && shift < lenof(fmts)-1) {
 330         size >>= 10;
 331         shift++;
 332     }
 333     *display_size = (double)size;
 334     *fmt = fmts[shift];
 335 }
 336
 337 #define PIXEL_SIZE 600                 /* FIXME: configurability? */
 338 static void write_report_line(struct html *ctx, struct vector *vec)
 339 {
 340     unsigned long long size, asize, divisor;
 341     double display_size;
 342     int pix, newpix;
 343     int i;
 344     const char *unitsfmt;
 345
 346     /*
 347      * A line with literally zero space usage should not be
 348      * printed at all if it's a link to a subdirectory (since it
 349      * probably means the whole thing was excluded by some
 350      * --exclude-path wildcard). If it's [files] or the top-level
 351      * line, though, we must always print _something_, and in that
 352      * case we must fiddle about to prevent divisions by zero in
 353      * the code below.
 354      */
 355     if (!vec->sizes[MAXCOLOUR] && !vec->essential)
 356         return;
 357     divisor = ctx->totalsize;
 358     if (!divisor) {
 359         divisor = 1;
 360     }
 361
 362     /*
 363      * Find the total size of this subdirectory.
 364      */
 365     size = vec->sizes[MAXCOLOUR];
 366     compute_display_size(size, &unitsfmt, &display_size);
 367     htprintf(ctx, "<tr>\n"
 368               "<td style=\"padding: 0.2em; text-align: right\">");
 369     htprintf(ctx, unitsfmt, display_size);
 370     htprintf(ctx, "</td>\n");
 371
 372     /*
 373      * Generate a colour bar.
 374      */
 375     htprintf(ctx, "<td style=\"padding: 0.2em\">\n");
 376     begin_colour_bar(ctx);
 377     pix = 0;
 378     for (i = 0; i <= MAXCOLOUR; i++) {
 379         asize = vec->sizes[i];
 380         newpix = asize * PIXEL_SIZE / divisor;
 381         add_to_colour_bar(ctx, i, newpix - pix);
 382         pix = newpix;
 383     }
 384     add_to_colour_bar(ctx, -1, PIXEL_SIZE - pix);
 385     end_colour_bar(ctx);
 386     htprintf(ctx, "</td>\n");
 387
 388     /*
 389      * Output size as a percentage of totalsize.
 390      */
 391     htprintf(ctx, "<td style=\"padding: 0.2em; text-align: right\">"
 392              "%.2f%%</td>\n", (double)size / divisor * 100.0);
 393
 394     /*
 395      * Output a subdirectory marker.
 396      */
 397     htprintf(ctx, "<td style=\"padding: 0.2em\">");
 398     if (vec->name) {
 399         int doing_href = 0;
 400
 401         if (ctx->format && vec->want_href) {
 402             snprintf(ctx->href, ctx->hreflen, ctx->format, vec->index);
 403             htprintf(ctx, "<a href=\"%s\">", ctx->href);
 404             doing_href = 1;
 405         }
 406         if (vec->literal)
 407             htprintf(ctx, "<code>");
 408         htescape(ctx, vec->name, strlen(vec->name), 1);
 409         if (vec->literal)
 410             htprintf(ctx, "</code>");
 411         if (doing_href)
 412             htprintf(ctx, "</a>");
 413     }
 414     htprintf(ctx, "</td>\n</tr>\n");
 415 }
 416
 417 int strcmptrailingpathsep(const char *a, const char *b)
 418 {
 419     while (*a == *b && *a)
 420         a++, b++;
 421
 422     if ((*a == pathsep && !a[1] && !*b) ||
 423         (*b == pathsep && !b[1] && !*a))
 424         return 0;
 425
 426     return (int)(unsigned char)*a - (int)(unsigned char)*b;
 427 }
 428
 429 char *html_query(const void *t, unsigned long index,
 430                  const struct html_config *cfg)
 431 {
 432     struct html actx, *ctx = &actx;
 433     char *path, *path2, *p, *q, *href;
 434     char agebuf1[80], agebuf2[80];
 435     size_t pathlen, subdirpos, hreflen;
 436     unsigned long index2;
 437     int i;
 438     struct vector **vecs;
 439     int nvecs, vecsize;
 440     unsigned long xi1, xi2, xj1, xj2;
 441
 442     if (index >= trie_count(t))
 443         return NULL;
 444
 445     ctx->buf = NULL;
 446     ctx->buflen = ctx->bufsize = 0;
 447     ctx->t = t;
 448     ctx->format = cfg->format;
 449     htprintf(ctx, "<html>\n");
 450
 451     path = snewn(1+trie_maxpathlen(t), char);
 452     ctx->path2 = path2 = snewn(1+trie_maxpathlen(t), char);
 453     if (cfg->format) {
 454         hreflen = strlen(cfg->format) + 100;
 455         href = snewn(hreflen, char);
 456     } else {
 457         hreflen = 0;
 458         href = NULL;
 459     }
 460     ctx->hreflen = hreflen;
 461     ctx->href = href;
 462
 463     /*
 464      * HEAD section.
 465      */
 466     htprintf(ctx, "<head>\n");
 467     trie_getpath(t, index, path);
 468     htprintf(ctx, "<title>%s: ", PNAME);
 469     htescape(ctx, path, strlen(path), 0);
 470     htprintf(ctx, "</title>\n");
 471     htprintf(ctx, "</head>\n");
 472
 473     /*
 474      * Begin BODY section.
 475      */
 476     htprintf(ctx, "<body>\n");
 477     htprintf(ctx, "<h3 align=center>Disk space breakdown by"
 478              " last-access time</h3>\n");
 479
 480     /*
 481      * Show the pathname we're centred on, with hyperlinks to
 482      * parent directories where available.
 483      */
 484     htprintf(ctx, "<p align=center>\n<code>");
 485     q = path;
 486     for (p = strchr(path, pathsep); p && p[1]; p = strchr(p, pathsep)) {
 487         int doing_href = 0;
 488         char c, *zp;
 489
 490         /*
 491          * See if this path prefix exists in the trie. If so,
 492          * generate a hyperlink.
 493          */
 494         zp = p;
 495         if (p == path)                 /* special case for "/" at start */
 496             zp++;
 497
 498         p++;
 499
 500         c = *zp;
 501         *zp = '\0';
 502         index2 = trie_before(t, path);
 503         trie_getpath(t, index2, path2);
 504         if (!strcmptrailingpathsep(path, path2) && cfg->format) {
 505             snprintf(href, hreflen, cfg->format, index2);
 506             if (!*href)                /* special case that we understand */
 507                 strcpy(href, "./");
 508             htprintf(ctx, "<a href=\"%s\">", href);
 509             doing_href = 1;
 510         }
 511         *zp = c;
 512         htescape(ctx, q, zp - q, 1);
 513         if (doing_href)
 514             htprintf(ctx, "</a>");
 515         htescape(ctx, zp, p - zp, 1);
 516         q = p;
 517     }
 518     htescape(ctx, q, strlen(q), 1);
 519     htprintf(ctx, "</code>\n");
 520
 521     /*
 522      * Decide on the age limit of our colour coding, establish the
 523      * colour thresholds, and write out a key.
 524      */
 525     ctx->now = time(NULL);
 526     if (cfg->autoage) {
 527         ctx->oldest = index_order_stat(t, 0.05);
 528         ctx->newest = index_order_stat(t, 1.0);
 529         ctx->oldest = round_and_format_age(ctx, ctx->oldest, agebuf1, -1);
 530         ctx->newest = round_and_format_age(ctx, ctx->newest, agebuf2, +1);
 531     } else {
 532         ctx->oldest = cfg->oldest;
 533         ctx->newest = cfg->newest;
 534         ctx->oldest = round_and_format_age(ctx, ctx->oldest, agebuf1, 0);
 535         ctx->newest = round_and_format_age(ctx, ctx->newest, agebuf2, 0);
 536     }
 537     for (i = 0; i < MAXCOLOUR; i++) {
 538         ctx->thresholds[i] =
 539             ctx->oldest + (ctx->newest - ctx->oldest) * i / (MAXCOLOUR-1);
 540     }
 541     for (i = 0; i <= MAXCOLOUR; i++) {
 542         char buf[80];
 543
 544         if (i == 0) {
 545             strcpy(buf, "&lt; ");
 546             round_and_format_age(ctx, ctx->thresholds[0], buf+5, 0);
 547         } else if (i == MAXCOLOUR) {
 548             strcpy(buf, "&gt; ");
 549             round_and_format_age(ctx, ctx->thresholds[MAXCOLOUR-1], buf+5, 0);
 550         } else {
 551             unsigned long long midrange =
 552                 (ctx->thresholds[i-1] + ctx->thresholds[i]) / 2;
 553             round_and_format_age(ctx, midrange, buf, 0);
 554         }
 555
 556         ctx->titletexts[i] = dupstr(buf);
 557     }
 558     htprintf(ctx, "<p align=center>Key to colour coding (mouse over for more detail):\n");
 559     htprintf(ctx, "<p align=center style=\"padding: 0; margin-top:0.4em; "
 560              "margin-bottom:1em\"");
 561     begin_colour_bar(ctx);
 562     htprintf(ctx, "<td style=\"padding-right:1em\">%s</td>\n", agebuf1);
 563     for (i = 0; i < MAXCOLOUR; i++)
 564         add_to_colour_bar(ctx, i, 1);
 565     htprintf(ctx, "<td style=\"padding-left:1em\">%s</td>\n", agebuf2);
 566     end_colour_bar(ctx);
 567
 568     /*
 569      * Begin the main table.
 570      */
 571     htprintf(ctx, "<p align=center>\n<table style=\"margin:0; border:0\">\n");
 572
 573     /*
 574      * Find the total size of our entire subdirectory. We'll use
 575      * that as the scale for all the colour bars in this report.
 576      */
 577     get_indices(t, path, &xi1, &xi2);
 578     ctx->totalsize = fetch_size(t, xi1, xi2, ULLONG_MAX);
 579
 580     /*
 581      * Generate a report line for the whole subdirectory.
 582      */
 583     vecsize = 64;
 584     vecs = snewn(vecsize, struct vector *);
 585     nvecs = 1;
 586     vecs[0] = make_vector(ctx, path, 0, 1, NULL, 0);
 587     print_heading(ctx, "Overall");
 588     write_report_line(ctx, vecs[0]);
 589
 590     /*
 591      * Now generate report lines for all its children, and the
 592      * files contained in it.
 593      */
 594     print_heading(ctx, "Subdirectories");
 595
 596     vecs[0]->name = dupstr("[files]");
 597     get_indices(t, path, &xi1, &xi2);
 598     xi1++;
 599     pathlen = strlen(path);
 600     subdirpos = pathlen + 1;
 601     if (pathlen > 0 && path[pathlen-1] == pathsep)
 602         subdirpos--;
 603     while (xi1 < xi2) {
 604         trie_getpath(t, xi1, path2);
 605         get_indices(t, ctx->path2, &xj1, &xj2);
 606         xi1 = xj2;
 607         if (!cfg->showfiles && xj2 - xj1 <= 1)
 608             continue;                  /* skip individual files */
 609         if (nvecs >= vecsize) {
 610             vecsize = nvecs * 3 / 2 + 64;
 611             vecs = sresize(vecs, vecsize, struct vector *);
 612         }
 613         assert(strlen(path2) > pathlen);
 614         vecs[nvecs] = make_vector(ctx, path2, (xj2 - xj1 > 1), 0,
 615                                   path2 + subdirpos, 1);
 616         for (i = 0; i <= MAXCOLOUR; i++)
 617             vecs[0]->sizes[i] -= vecs[nvecs]->sizes[i];
 618         nvecs++;
 619     }
 620
 621     qsort(vecs, nvecs, sizeof(vecs[0]), vec_compare);
 622
 623     for (i = 0; i < nvecs; i++)
 624         write_report_line(ctx, vecs[i]);
 625
 626     /*
 627      * Close the main table.
 628      */
 629     htprintf(ctx, "</table>\n");
 630
 631     /*
 632      * Finish up and tidy up.
 633      */
 634     htprintf(ctx, "</body>\n");
 635     htprintf(ctx, "</html>\n");
 636     sfree(href);
 637     sfree(path2);
 638     sfree(path);
 639     for (i = 0; i < nvecs; i++) {
 640         sfree(vecs[i]->name);
 641         sfree(vecs[i]);
 642     }
 643     sfree(vecs);
 644
 645     return ctx->buf;
 646 }