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, 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, 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     return index_query(t, xi2, atime) - index_query(t, xi1, atime);
 180 }
 181
 182 static void htescape(struct html *ctx, const char *s, int n, int italics)
 183 {
 184     while (n > 0 && *s) {
 185         unsigned char c = (unsigned char)*s++;
 186
 187         if (c == '&')
 188             htprintf(ctx, "&amp;");
 189         else if (c == '<')
 190             htprintf(ctx, "&lt;");
 191         else if (c == '>')
 192             htprintf(ctx, "&gt;");
 193         else if (c >= ' ' && c < '\177')
 194             htprintf(ctx, "%c", c);
 195         else {
 196             if (italics) htprintf(ctx, "<i>");
 197             htprintf(ctx, "[%02x]", c);
 198             if (italics) htprintf(ctx, "</i>");
 199         }
 200
 201         n--;
 202     }
 203 }
 204
 205 static void begin_colour_bar(struct html *ctx)
 206 {
 207     htprintf(ctx, "<table cellspacing=0 cellpadding=0"
 208              " style=\"border:0\">\n<tr>\n");
 209 }
 210
 211 static void add_to_colour_bar(struct html *ctx, int colour, int pixels)
 212 {
 213     int r, g, b;
 214
 215     if (colour >= 0 && colour < 256)   /* red -> yellow fade */
 216         r = 255, g = colour, b = 0;
 217     else if (colour >= 256 && colour <= 511)   /* yellow -> green fade */
 218         r = 511 - colour, g = 255, b = 0;
 219     else                               /* background grey */
 220         r = g = b = 240;
 221
 222     if (pixels > 0) {
 223         htprintf(ctx, "<td style=\"width:%dpx; height:1em; "
 224                  "background-color:#%02x%02x%02x\"",
 225                  pixels, r, g, b);
 226         if (colour >= 0)
 227             htprintf(ctx, " title=\"%s\"", ctx->titletexts[colour]);
 228         htprintf(ctx, "></td>\n");
 229     }
 230 }
 231
 232 static void end_colour_bar(struct html *ctx)
 233 {
 234     htprintf(ctx, "</tr>\n</table>\n");
 235 }
 236
 237 struct vector {
 238     int want_href;
 239     char *name;
 240     int literal; /* should the name be formatted in fixed-pitch? */
 241     unsigned long index;
 242     unsigned long long sizes[MAXCOLOUR+1];
 243 };
 244
 245 int vec_compare(const void *av, const void *bv)
 246 {
 247     const struct vector *a = *(const struct vector **)av;
 248     const struct vector *b = *(const struct vector **)bv;
 249
 250     if (a->sizes[MAXCOLOUR] > b->sizes[MAXCOLOUR])
 251         return -1;
 252     else if (a->sizes[MAXCOLOUR] < b->sizes[MAXCOLOUR])
 253         return +1;
 254     else if (a->want_href < b->want_href)
 255         return +1;
 256     else if (a->want_href > b->want_href)
 257         return -1;
 258     else if (a->want_href)
 259         return strcmp(a->name, b->name);
 260     else if (a->index < b->index)
 261         return -1;
 262     else if (a->index > b->index)
 263         return +1;
 264     return 0;
 265 }
 266
 267 static struct vector *make_vector(struct html *ctx, char *path,
 268                                   int want_href, char *name, int literal)
 269 {
 270     unsigned long xi1, xi2;
 271     struct vector *vec = snew(struct vector);
 272     int i;
 273
 274     vec->want_href = want_href;
 275     vec->name = name ? dupstr(name) : NULL;
 276     vec->literal = literal;
 277
 278     get_indices(ctx->t, path, &xi1, &xi2);
 279
 280     vec->index = xi1;
 281
 282     for (i = 0; i <= MAXCOLOUR; i++) {
 283         unsigned long long atime;
 284         if (i == MAXCOLOUR)
 285             atime = ULLONG_MAX;
 286         else
 287             atime = ctx->thresholds[i];
 288         vec->sizes[i] = fetch_size(ctx->t, xi1, xi2, atime);
 289     }
 290
 291     return vec;
 292 }
 293
 294 static void print_heading(struct html *ctx, const char *title)
 295 {
 296     htprintf(ctx, "<tr style=\"padding: 0.2em; background-color:#e0e0e0\">\n"
 297              "<td colspan=4 align=center>%s</td>\n</tr>\n", title);
 298 }
 299
 300 #define PIXEL_SIZE 600                 /* FIXME: configurability? */
 301 static void write_report_line(struct html *ctx, struct vector *vec)
 302 {
 303     unsigned long long size, asize, divisor;
 304     int pix, newpix;
 305     int i;
 306
 307     /*
 308      * A line with literally zero space usage should not be
 309      * printed at all if it's a link to a subdirectory (since it
 310      * probably means the whole thing was excluded by some
 311      * --exclude-path wildcard). If it's [files] or the top-level
 312      * line, though, we must always print _something_, and in that
 313      * case we must fiddle about to prevent divisions by zero in
 314      * the code below.
 315      */
 316     if (!vec->sizes[MAXCOLOUR] && vec->want_href)
 317         return;
 318     divisor = ctx->totalsize;
 319     if (!divisor) {
 320         divisor = 1;
 321     }
 322
 323     /*
 324      * Find the total size of this subdirectory.
 325      */
 326     size = vec->sizes[MAXCOLOUR];
 327     htprintf(ctx, "<tr>\n"
 328              "<td style=\"padding: 0.2em; text-align: right\">%lluMb</td>\n",
 329              ((size + ((1<<20)-1)) >> 20)); /* convert to Mb, rounding up */
 330
 331     /*
 332      * Generate a colour bar.
 333      */
 334     htprintf(ctx, "<td style=\"padding: 0.2em\">\n");
 335     begin_colour_bar(ctx);
 336     pix = 0;
 337     for (i = 0; i <= MAXCOLOUR; i++) {
 338         asize = vec->sizes[i];
 339         newpix = asize * PIXEL_SIZE / divisor;
 340         add_to_colour_bar(ctx, i, newpix - pix);
 341         pix = newpix;
 342     }
 343     add_to_colour_bar(ctx, -1, PIXEL_SIZE - pix);
 344     end_colour_bar(ctx);
 345     htprintf(ctx, "</td>\n");
 346
 347     /*
 348      * Output size as a percentage of totalsize.
 349      */
 350     htprintf(ctx, "<td style=\"padding: 0.2em; text-align: right\">"
 351              "%.2f%%</td>\n", (double)size / divisor * 100.0);
 352
 353     /*
 354      * Output a subdirectory marker.
 355      */
 356     htprintf(ctx, "<td style=\"padding: 0.2em\">");
 357     if (vec->name) {
 358         int doing_href = 0;
 359
 360         if (ctx->format && vec->want_href) {
 361             snprintf(ctx->href, ctx->hreflen, ctx->format, vec->index);
 362             htprintf(ctx, "<a href=\"%s\">", ctx->href);
 363             doing_href = 1;
 364         }
 365         if (vec->literal)
 366             htprintf(ctx, "<code>");
 367         htescape(ctx, vec->name, strlen(vec->name), 1);
 368         if (vec->literal)
 369             htprintf(ctx, "</code>");
 370         if (doing_href)
 371             htprintf(ctx, "</a>");
 372     }
 373     htprintf(ctx, "</td>\n</tr>\n");
 374 }
 375
 376 int strcmptrailingpathsep(const char *a, const char *b)
 377 {
 378     while (*a == *b && *a)
 379         a++, b++;
 380
 381     if ((*a == pathsep && !a[1] && !*b) ||
 382         (*b == pathsep && !b[1] && !*a))
 383         return 0;
 384
 385     return (int)(unsigned char)*a - (int)(unsigned char)*b;
 386 }
 387
 388 char *html_query(const void *t, unsigned long index,
 389                  const struct html_config *cfg)
 390 {
 391     struct html actx, *ctx = &actx;
 392     char *path, *path2, *p, *q, *href;
 393     char agebuf1[80], agebuf2[80];
 394     size_t pathlen, subdirpos, hreflen;
 395     unsigned long index2;
 396     int i;
 397     struct vector **vecs;
 398     int nvecs, vecsize;
 399     unsigned long xi1, xi2, xj1, xj2;
 400
 401     if (index >= trie_count(t))
 402         return NULL;
 403
 404     ctx->buf = NULL;
 405     ctx->buflen = ctx->bufsize = 0;
 406     ctx->t = t;
 407     ctx->format = cfg->format;
 408     htprintf(ctx, "<html>\n");
 409
 410     path = snewn(1+trie_maxpathlen(t), char);
 411     ctx->path2 = path2 = snewn(1+trie_maxpathlen(t), char);
 412     if (cfg->format) {
 413         hreflen = strlen(cfg->format) + 100;
 414         href = snewn(hreflen, char);
 415     } else {
 416         hreflen = 0;
 417         href = NULL;
 418     }
 419     ctx->hreflen = hreflen;
 420     ctx->href = href;
 421
 422     /*
 423      * HEAD section.
 424      */
 425     htprintf(ctx, "<head>\n");
 426     trie_getpath(t, index, path);
 427     htprintf(ctx, "<title>%s: ", PNAME);
 428     htescape(ctx, path, strlen(path), 0);
 429     htprintf(ctx, "</title>\n");
 430     htprintf(ctx, "</head>\n");
 431
 432     /*
 433      * Begin BODY section.
 434      */
 435     htprintf(ctx, "<body>\n");
 436     htprintf(ctx, "<h3 align=center>Disk space breakdown by"
 437              " last-access time</h3>\n");
 438
 439     /*
 440      * Show the pathname we're centred on, with hyperlinks to
 441      * parent directories where available.
 442      */
 443     htprintf(ctx, "<p align=center>\n<code>");
 444     q = path;
 445     for (p = strchr(path, pathsep); p && p[1]; p = strchr(p, pathsep)) {
 446         int doing_href = 0;
 447         char c, *zp;
 448
 449         /*
 450          * See if this path prefix exists in the trie. If so,
 451          * generate a hyperlink.
 452          */
 453         zp = p;
 454         if (p == path)                 /* special case for "/" at start */
 455             zp++;
 456
 457         p++;
 458
 459         c = *zp;
 460         *zp = '\0';
 461         index2 = trie_before(t, path);
 462         trie_getpath(t, index2, path2);
 463         if (!strcmptrailingpathsep(path, path2) && cfg->format) {
 464             snprintf(href, hreflen, cfg->format, index2);
 465             if (!*href)                /* special case that we understand */
 466                 strcpy(href, "./");
 467             htprintf(ctx, "<a href=\"%s\">", href);
 468             doing_href = 1;
 469         }
 470         *zp = c;
 471         htescape(ctx, q, zp - q, 1);
 472         if (doing_href)
 473             htprintf(ctx, "</a>");
 474         htescape(ctx, zp, p - zp, 1);
 475         q = p;
 476     }
 477     htescape(ctx, q, strlen(q), 1);
 478     htprintf(ctx, "</code>\n");
 479
 480     /*
 481      * Decide on the age limit of our colour coding, establish the
 482      * colour thresholds, and write out a key.
 483      */
 484     ctx->now = time(NULL);
 485     if (cfg->autoage) {
 486         ctx->oldest = index_order_stat(t, 0.05);
 487         ctx->newest = index_order_stat(t, 1.0);
 488         ctx->oldest = round_and_format_age(ctx, ctx->oldest, agebuf1, -1);
 489         ctx->newest = round_and_format_age(ctx, ctx->newest, agebuf2, +1);
 490     } else {
 491         ctx->oldest = cfg->oldest;
 492         ctx->newest = cfg->newest;
 493         ctx->oldest = round_and_format_age(ctx, ctx->oldest, agebuf1, 0);
 494         ctx->newest = round_and_format_age(ctx, ctx->newest, agebuf2, 0);
 495     }
 496     for (i = 0; i < MAXCOLOUR; i++) {
 497         ctx->thresholds[i] =
 498             ctx->oldest + (ctx->newest - ctx->oldest) * i / (MAXCOLOUR-1);
 499     }
 500     for (i = 0; i <= MAXCOLOUR; i++) {
 501         char buf[80];
 502
 503         if (i == 0) {
 504             strcpy(buf, "&lt; ");
 505             round_and_format_age(ctx, ctx->thresholds[0], buf+5, 0);
 506         } else if (i == MAXCOLOUR) {
 507             strcpy(buf, "&gt; ");
 508             round_and_format_age(ctx, ctx->thresholds[MAXCOLOUR-1], buf+5, 0);
 509         } else {
 510             unsigned long long midrange =
 511                 (ctx->thresholds[i-1] + ctx->thresholds[i]) / 2;
 512             round_and_format_age(ctx, midrange, buf, 0);
 513         }
 514
 515         ctx->titletexts[i] = dupstr(buf);
 516     }
 517     htprintf(ctx, "<p align=center>Key to colour coding (mouse over for more detail):\n");
 518     htprintf(ctx, "<p align=center style=\"padding: 0; margin-top:0.4em; "
 519              "margin-bottom:1em\"");
 520     begin_colour_bar(ctx);
 521     htprintf(ctx, "<td style=\"padding-right:1em\">%s</td>\n", agebuf1);
 522     for (i = 0; i < MAXCOLOUR; i++)
 523         add_to_colour_bar(ctx, i, 1);
 524     htprintf(ctx, "<td style=\"padding-left:1em\">%s</td>\n", agebuf2);
 525     end_colour_bar(ctx);
 526
 527     /*
 528      * Begin the main table.
 529      */
 530     htprintf(ctx, "<p align=center>\n<table style=\"margin:0; border:0\">\n");
 531
 532     /*
 533      * Find the total size of our entire subdirectory. We'll use
 534      * that as the scale for all the colour bars in this report.
 535      */
 536     get_indices(t, path, &xi1, &xi2);
 537     ctx->totalsize = fetch_size(t, xi1, xi2, ULLONG_MAX);
 538
 539     /*
 540      * Generate a report line for the whole subdirectory.
 541      */
 542     vecsize = 64;
 543     vecs = snewn(vecsize, struct vector *);
 544     nvecs = 1;
 545     vecs[0] = make_vector(ctx, path, 0, NULL, 0);
 546     print_heading(ctx, "Overall");
 547     write_report_line(ctx, vecs[0]);
 548
 549     /*
 550      * Now generate report lines for all its children, and the
 551      * files contained in it.
 552      */
 553     print_heading(ctx, "Subdirectories");
 554
 555     vecs[0]->name = dupstr("[files]");
 556     get_indices(t, path, &xi1, &xi2);
 557     xi1++;
 558     pathlen = strlen(path);
 559     subdirpos = pathlen + 1;
 560     if (pathlen > 0 && path[pathlen-1] == pathsep)
 561         subdirpos--;
 562     while (xi1 < xi2) {
 563         trie_getpath(t, xi1, path2);
 564         get_indices(t, ctx->path2, &xj1, &xj2);
 565         xi1 = xj2;
 566         if (xj2 - xj1 <= 1)
 567             continue;                  /* skip individual files */
 568         if (nvecs >= vecsize) {
 569             vecsize = nvecs * 3 / 2 + 64;
 570             vecs = sresize(vecs, vecsize, struct vector *);
 571         }
 572         assert(strlen(path2) > pathlen);
 573         vecs[nvecs] = make_vector(ctx, path2, 1, path2 + subdirpos, 1);
 574         for (i = 0; i <= MAXCOLOUR; i++)
 575             vecs[0]->sizes[i] -= vecs[nvecs]->sizes[i];
 576         nvecs++;
 577     }
 578
 579     qsort(vecs, nvecs, sizeof(vecs[0]), vec_compare);
 580
 581     for (i = 0; i < nvecs; i++)
 582         write_report_line(ctx, vecs[i]);
 583
 584     /*
 585      * Close the main table.
 586      */
 587     htprintf(ctx, "</table>\n");
 588
 589     /*
 590      * Finish up and tidy up.
 591      */
 592     htprintf(ctx, "</body>\n");
 593     htprintf(ctx, "</html>\n");
 594     sfree(href);
 595     sfree(path2);
 596     sfree(path);
 597     for (i = 0; i < nvecs; i++) {
 598         sfree(vecs[i]->name);
 599         sfree(vecs[i]);
 600     }
 601     sfree(vecs);
 602
 603     return ctx->buf;
 604 }