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