src/output/html.c

   1 /* PSPP - a program for statistical analysis.
   2    Copyright (C) 1997-9, 2000, 2009, 2010, 2011, 2012, 2013, 2014, 2017,
   3    2020 Free Software Foundation, Inc.
   4
   5    This program is free software: you can redistribute it and/or modify
   6    it under the terms of the GNU General Public License as published by
   7    the Free Software Foundation, either version 3 of the License, or
   8    (at your option) any later version.
   9
  10    This program is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13    GNU General Public License for more details.
  14
  15    You should have received a copy of the GNU General Public License
  16    along with this program.  If not, see <http://www.gnu.org/licenses/>. */
  17
  18 #include <config.h>
  19
  20 #include <errno.h>
  21 #include <stdint.h>
  22 #include <stdlib.h>
  23 #include <ctype.h>
  24 #include <time.h>
  25 #include <unistd.h>
  26 #include <locale.h>
  27
  28 #include "data/file-name.h"
  29 #include "data/file-handle-def.h"
  30 #include "libpspp/assertion.h"
  31 #include "libpspp/cast.h"
  32 #include "libpspp/compiler.h"
  33 #include "libpspp/i18n.h"
  34 #include "libpspp/message.h"
  35 #include "libpspp/version.h"
  36 #include "output/cairo-chart.h"
  37 #include "output/chart.h"
  38 #include "output/driver-provider.h"
  39 #include "output/options.h"
  40 #include "output/output-item.h"
  41 #include "output/pivot-output.h"
  42 #include "output/pivot-table.h"
  43 #include "output/table-provider.h"
  44
  45 #include "gl/minmax.h"
  46 #include "gl/xalloc.h"
  47
  48 #include "gettext.h"
  49 #define _(msgid) gettext (msgid)
  50
  51 /* This file uses TABLE_HORZ and TABLE_VERT enough to warrant abbreviating. */
  52 #define H TABLE_HORZ
  53 #define V TABLE_VERT
  54
  55 struct html_driver
  56   {
  57     struct output_driver driver;
  58     struct cell_color fg;
  59     struct cell_color bg;
  60     struct file_handle *handle;
  61     char *chart_file_name;
  62
  63     FILE *file;
  64     size_t n_charts;
  65
  66     bool bare;
  67     bool css;
  68     bool borders;
  69   };
  70
  71 static const struct output_driver_class html_driver_class;
  72
  73 static void html_output_table (struct html_driver *,
  74                                const struct output_item *);
  75 static void escape_string (FILE *file, const char *text,
  76                            const char *space, const char *newline);
  77 static void print_title_tag (FILE *file, const char *name,
  78                              const char *content);
  79
  80 static struct html_driver *
  81 html_driver_cast (struct output_driver *driver)
  82 {
  83   assert (driver->class == &html_driver_class);
  84   return UP_CAST (driver, struct html_driver, driver);
  85 }
  86
  87 static struct driver_option *
  88 opt (struct output_driver *d, struct string_map *options, const char *key,
  89      const char *default_value)
  90 {
  91   return driver_option_get (d, options, key, default_value);
  92 }
  93
  94 static void
  95 put_header (struct html_driver *html)
  96 {
  97   fputs ("<!doctype html>\n", html->file);
  98   fprintf (html->file, "<html");
  99   char *ln = get_language ();
 100   if (ln)
 101     fprintf (html->file, " lang=\"%s\"", ln);
 102   free (ln);
 103   fprintf (html->file, ">\n");
 104   fputs ("<head>\n", html->file);
 105   print_title_tag (html->file, "title", _("PSPP Output"));
 106   fprintf (html->file, "<meta name=\"generator\" content=\"%s\">\n", version);
 107   fputs ("<meta http-equiv=\"content-type\" "
 108          "content=\"text/html; charset=utf-8\">\n", html->file);
 109
 110   if (html->css)
 111     {
 112       fputs ("<style>\n"
 113              "<!--\n"
 114              "body {\n"
 115              "  background: white;\n"
 116              "  color: black;\n"
 117              "  padding: 0em 12em 0em 3em;\n"
 118              "  margin: 0\n"
 119              "}\n"
 120              "body>p {\n"
 121              "  margin: 0pt 0pt 0pt 0em\n"
 122              "}\n"
 123              "body>p + p {\n"
 124              "  text-indent: 1.5em;\n"
 125              "}\n"
 126              "h1 {\n"
 127              "  font-size: 150%;\n"
 128              "  margin-left: -1.33em\n"
 129              "}\n"
 130              "h2 {\n"
 131              "  font-size: 125%;\n"
 132              "  font-weight: bold;\n"
 133              "  margin-left: -.8em\n"
 134              "}\n"
 135              "h3 {\n"
 136              "  font-size: 100%;\n"
 137              "  font-weight: bold;\n"
 138              "  margin-left: -.5em }\n"
 139              "h4 {\n"
 140              "  font-size: 100%;\n"
 141              "  margin-left: 0em\n"
 142              "}\n"
 143              "h1, h2, h3, h4, h5, h6 {\n"
 144              "  font-family: sans-serif;\n"
 145              "  color: blue\n"
 146              "}\n"
 147              "html {\n"
 148              "  margin: 0\n"
 149              "}\n"
 150              "code {\n"
 151              "  font-family: sans-serif\n"
 152              "}\n"
 153              "table {\n"
 154              "  border-collapse: collapse;\n"
 155              "  margin-bottom: 1em\n"
 156              "}\n"
 157              "caption {\n"
 158              "  text-align: left\n"
 159              "}\n"
 160              "th { font-weight: normal }\n"
 161              "a:link {\n"
 162              "  color: #1f00ff;\n"
 163              "}\n"
 164              "a:visited {\n"
 165              "  color: #9900dd;\n"
 166              "}\n"
 167              "a:active {\n"
 168              "  color: red;\n"
 169              "}\n"
 170              "-->\n"
 171              "</style>\n",
 172              html->file);
 173     }
 174   fputs ("</head>\n", html->file);
 175   fputs ("<body>\n", html->file);
 176 }
 177
 178 static struct output_driver *
 179 html_create (struct file_handle *fh, enum settings_output_devices device_type,
 180              struct string_map *o)
 181 {
 182   struct output_driver *d;
 183   struct html_driver *html = XZALLOC (struct html_driver);
 184   d = &html->driver;
 185   output_driver_init (&html->driver, &html_driver_class, fh_get_file_name (fh),
 186                       device_type);
 187   html->bare = parse_boolean (opt (d, o, "bare", "false"));
 188   html->css = parse_boolean (opt (d, o, "css", "true"));
 189   html->borders = parse_boolean (opt (d, o, "borders", "true"));
 190
 191   html->handle = fh;
 192   html->chart_file_name = parse_chart_file_name (opt (d, o, "charts",
 193                                                       fh_get_file_name (fh)));
 194   html->file = NULL;
 195   html->n_charts = 1;
 196   html->bg = parse_color (opt (d, o, "background-color", "#FFFFFFFFFFFF"));
 197   html->fg = parse_color (opt (d, o, "foreground-color", "#000000000000"));
 198   html->file = fn_open (html->handle, "w");
 199   if (html->file == NULL)
 200     {
 201       msg_error (errno, _("error opening output file `%s'"), fh_get_file_name (html->handle));
 202       goto error;
 203     }
 204
 205   if (!html->bare)
 206     put_header (html);
 207
 208   return d;
 209
 210  error:
 211   output_driver_destroy (d);
 212   return NULL;
 213 }
 214
 215 /* Emits <NAME>CONTENT</NAME> to the output, escaping CONTENT as
 216    necessary for HTML. */
 217 static void
 218 print_title_tag (FILE *file, const char *name, const char *content)
 219 {
 220   if (content != NULL)
 221     {
 222        fprintf (file, "<%s>", name);
 223       escape_string (file, content, " ", " - ");
 224       fprintf (file, "</%s>\n", name);
 225     }
 226 }
 227
 228 static void
 229 html_destroy (struct output_driver *driver)
 230 {
 231   struct html_driver *html = html_driver_cast (driver);
 232
 233   if (html->file != NULL)
 234     {
 235       if (!html->bare)
 236         fprintf (html->file,
 237                  "</body>\n"
 238                  "</html>\n"
 239                  "<!-- end of file -->\n");
 240       fn_close (html->handle, html->file);
 241     }
 242   free (html->chart_file_name);
 243   fh_unref (html->handle);
 244   free (html);
 245 }
 246
 247 static void
 248 html_submit__ (struct output_driver *driver, const struct output_item *item,
 249                int level)
 250 {
 251   struct html_driver *html = html_driver_cast (driver);
 252
 253   switch (item->type)
 254     {
 255     case OUTPUT_ITEM_CHART:
 256       if (html->chart_file_name)
 257         {
 258           char *file_name = xr_draw_png_chart (item->chart,
 259                                                html->chart_file_name,
 260                                                html->n_charts++,
 261                                                &html->fg, &html->bg);
 262           if (file_name != NULL)
 263             {
 264               const char *title = chart_get_title (item->chart);
 265               fprintf (html->file, "<img src=\"%s\" alt=\"chart: %s\">",
 266                        file_name, title ? title : _("No description"));
 267               free (file_name);
 268             }
 269         }
 270       break;
 271
 272     case OUTPUT_ITEM_GROUP:
 273       for (size_t i = 0; i < item->group.n_children; i++)
 274         html_submit__ (driver, item->group.children[i], level + 1);
 275       break;
 276
 277     case OUTPUT_ITEM_IMAGE:
 278       if (html->chart_file_name)
 279         {
 280           char *file_name = xr_write_png_image (
 281             item->image, html->chart_file_name, ++html->n_charts);
 282           if (file_name != NULL)
 283             {
 284               fprintf (html->file, "<img src=\"%s\">", file_name);
 285               free (file_name);
 286             }
 287         }
 288       break;
 289
 290     case OUTPUT_ITEM_MESSAGE:
 291       fprintf (html->file, "<p>");
 292
 293       char *s = msg_to_string (item->message);
 294       escape_string (html->file, s, " ", "<br>");
 295       free (s);
 296
 297       fprintf (html->file, "</p>\n");
 298       break;
 299
 300     case OUTPUT_ITEM_PAGE_BREAK:
 301       break;
 302
 303     case OUTPUT_ITEM_TABLE:
 304       html_output_table (html, item);
 305       break;
 306
 307     case OUTPUT_ITEM_TEXT:
 308       {
 309         char *s = text_item_get_plain_text (item);
 310
 311         switch (item->text.subtype)
 312           {
 313           case TEXT_ITEM_PAGE_TITLE:
 314             break;
 315
 316           case TEXT_ITEM_TITLE:
 317             {
 318               char tag[3] = { 'H', MIN (5, level) + '0', '\0' };
 319               print_title_tag (html->file, tag, s);
 320             }
 321             break;
 322
 323           case TEXT_ITEM_SYNTAX:
 324             fprintf (html->file, "<pre class=\"syntax\">");
 325             escape_string (html->file, s, " ", "<br>");
 326             fprintf (html->file, "</pre>\n");
 327             break;
 328
 329           case TEXT_ITEM_LOG:
 330             fprintf (html->file, "<p>");
 331             escape_string (html->file, s, " ", "<br>");
 332             fprintf (html->file, "</p>\n");
 333             break;
 334           }
 335
 336         free (s);
 337       }
 338       break;
 339     }
 340 }
 341
 342 static void
 343 html_submit (struct output_driver *driver, const struct output_item *item)
 344 {
 345   html_submit__ (driver, item, 1);
 346 }
 347
 348 /* Write TEXT to file F, escaping characters as necessary for HTML.  Spaces are
 349    replaced by SPACE, which should be " " or "&nbsp;" New-lines are replaced by
 350    NEWLINE, which might be "<BR>" or "\n" or something else appropriate. */
 351 static void
 352 escape_string (FILE *file, const char *text,
 353                const char *space, const char *newline)
 354 {
 355   for (;;)
 356     {
 357       char c = *text++;
 358       switch (c)
 359         {
 360         case 0:
 361           return;
 362         case '\n':
 363           fputs (newline, file);
 364           break;
 365         case '&':
 366           fputs ("&amp;", file);
 367           break;
 368         case '<':
 369           fputs ("&lt;", file);
 370           break;
 371         case '>':
 372           fputs ("&gt;", file);
 373           break;
 374         case ' ':
 375           fputs (space, file);
 376           break;
 377         case '"':
 378           fputs ("&quot;", file);
 379           break;
 380         default:
 381           putc (c, file);
 382           break;
 383         }
 384     }
 385 }
 386
 387 static const char *
 388 border_to_css (int border)
 389 {
 390   switch (border)
 391     {
 392     case TABLE_STROKE_NONE:
 393       return NULL;
 394
 395     case TABLE_STROKE_SOLID:
 396       return "solid";
 397
 398     case TABLE_STROKE_DASHED:
 399       return "dashed";
 400
 401     case TABLE_STROKE_THICK:
 402       return "thick solid";
 403
 404     case TABLE_STROKE_THIN:
 405       return "thin solid";
 406
 407     case TABLE_STROKE_DOUBLE:
 408       return "double";
 409
 410     default:
 411       return NULL;
 412     }
 413
 414 }
 415
 416 struct css_style
 417 {
 418   FILE *file;
 419   int n_styles;
 420 };
 421
 422 static void
 423 style_start (struct css_style *cs, FILE *file)
 424 {
 425   *cs = (struct css_style) {
 426     .file = file,
 427     .n_styles = 0,
 428   };
 429 }
 430
 431 static void
 432 style_end (struct css_style *cs)
 433 {
 434   if (cs->n_styles > 0)
 435     fputs ("'", cs->file);
 436 }
 437
 438 static void
 439 next_style (struct css_style *st)
 440 {
 441   bool first = !st->n_styles++;
 442   fputs (first ? " style='" : "; ", st->file);
 443 }
 444
 445 static void
 446 put_style (struct css_style *st, const char *name, const char *value)
 447 {
 448   next_style (st);
 449   fprintf (st->file, "%s: %s", name, value);
 450 }
 451
 452 static bool
 453 format_color (const struct cell_color color,
 454               const struct cell_color default_color,
 455               char *buf, size_t bufsize)
 456 {
 457   bool retval = !cell_color_equal (&color, &default_color);
 458   if (retval)
 459     {
 460       if (color.alpha == 255)
 461         snprintf (buf, bufsize, "#%02x%02x%02x", color.r, color.g, color.b);
 462       else
 463         snprintf (buf, bufsize, "rgba(%d, %d, %d, %.3f)",
 464                   color.r, color.g, color.b, color.alpha / 255.0);
 465     }
 466   return retval;
 467 }
 468
 469 static void
 470 put_border (const struct table *table, const struct table_cell *cell,
 471             struct css_style *style,
 472             enum table_axis axis, int h, int v,
 473             const char *border_name)
 474 {
 475   struct cell_color color;
 476   const char *css = border_to_css (
 477     table_get_rule (table, axis, cell->d[H][h], cell->d[V][v], &color));
 478   if (css)
 479     {
 480       next_style (style);
 481       fprintf (style->file, "border-%s: %s", border_name, css);
 482
 483       char buf[32];
 484       if (format_color (color, (struct cell_color) CELL_COLOR_BLACK,
 485                         buf, sizeof buf))
 486         fprintf (style->file, " %s", buf);
 487     }
 488 }
 489
 490 static void
 491 html_put_table_cell (struct html_driver *html, const struct pivot_table *pt,
 492                      const struct table_cell *cell,
 493                      const char *tag, const struct table *t)
 494 {
 495   fprintf (html->file, "<%s", tag);
 496
 497   struct css_style style;
 498   style_start (&style, html->file);
 499
 500   struct string body = DS_EMPTY_INITIALIZER;
 501   bool numeric = pivot_value_format_body (cell->value, pt, &body);
 502
 503   enum table_halign halign = table_halign_interpret (cell->cell_style->halign,
 504                                                      numeric);
 505
 506   switch (halign)
 507     {
 508     case TABLE_HALIGN_RIGHT:
 509       put_style (&style, "text-align", "right");
 510       break;
 511     case TABLE_HALIGN_CENTER:
 512       put_style (&style, "text-align", "center");
 513       break;
 514     default:
 515       /* Do nothing */
 516       break;
 517     }
 518
 519   if (cell->options & TABLE_CELL_ROTATE)
 520     put_style (&style, "writing-mode", "sideways-lr");
 521
 522   if (cell->cell_style->valign != TABLE_VALIGN_TOP)
 523     {
 524       put_style (&style, "vertical-align",
 525                  (cell->cell_style->valign == TABLE_VALIGN_BOTTOM
 526                   ? "bottom" : "middle"));
 527     }
 528
 529   const struct font_style *fs = cell->font_style;
 530   char bgcolor[32];
 531   if (format_color (fs->bg[cell->d[V][0] % 2],
 532                     (struct cell_color) CELL_COLOR_WHITE,
 533                     bgcolor, sizeof bgcolor))
 534     put_style (&style, "background", bgcolor);
 535
 536   char fgcolor[32];
 537   if (format_color (fs->fg[cell->d[V][0] % 2],
 538                     (struct cell_color) CELL_COLOR_BLACK,
 539                     fgcolor, sizeof fgcolor))
 540     put_style (&style, "color", fgcolor);
 541
 542   if (fs->typeface)
 543     {
 544       put_style (&style, "font-family", "\"");
 545       escape_string (html->file, fs->typeface, " ", "\n");
 546       putc ('"', html->file);
 547     }
 548   if (fs->bold)
 549     put_style (&style, "font-weight", "bold");
 550   if (fs->italic)
 551     put_style (&style, "font-style", "italic");
 552   if (fs->underline)
 553     put_style (&style, "text-decoration", "underline");
 554   if (fs->size)
 555     {
 556       char buf[32];
 557       snprintf (buf, sizeof buf, "%dpt", fs->size);
 558       put_style (&style, "font-size", buf);
 559     }
 560
 561   if (t && html->borders)
 562     {
 563       put_border (t, cell, &style, V, 0, 0, "top");
 564       put_border (t, cell, &style, H, 0, 0, "left");
 565
 566       if (cell->d[V][1] == t->n[V])
 567         put_border (t, cell, &style, V, 0, 1, "bottom");
 568       if (cell->d[H][1] == t->n[H])
 569         put_border (t, cell, &style, H, 1, 0, "right");
 570     }
 571   style_end (&style);
 572
 573   int colspan = table_cell_colspan (cell);
 574   if (colspan > 1)
 575     fprintf (html->file, " colspan=\"%d\"", colspan);
 576
 577   int rowspan = table_cell_rowspan (cell);
 578   if (rowspan > 1)
 579     fprintf (html->file, " rowspan=\"%d\"", rowspan);
 580
 581   putc ('>', html->file);
 582
 583   const char *s = ds_cstr (&body);
 584   s += strspn (s, CC_SPACES);
 585   escape_string (html->file, s, " ", "<br>");
 586   ds_destroy (&body);
 587
 588   const struct pivot_value_ex *ex = pivot_value_ex (cell->value);
 589   if (ex->n_subscripts)
 590     {
 591       fputs ("<sub>", html->file);
 592       for (size_t i = 0; i < ex->n_subscripts; i++)
 593         {
 594           if (i)
 595             putc (',', html->file);
 596           escape_string (html->file, ex->subscripts[i], "&nbsp;", "<br>");
 597         }
 598       fputs ("</sub>", html->file);
 599     }
 600   if (ex->n_footnotes > 0)
 601     {
 602       fputs ("<sup>", html->file);
 603       size_t n_footnotes = 0;
 604       for (size_t i = 0; i < ex->n_footnotes; i++)
 605         {
 606           const struct pivot_footnote *f
 607             = pt->footnotes[ex->footnote_indexes[i]];
 608           if (f->show)
 609             {
 610               if (n_footnotes++ > 0)
 611                 putc (',', html->file);
 612
 613               char *marker = pivot_footnote_marker_string (f, pt);
 614               escape_string (html->file, marker, " ", "<br>");
 615               free (marker);
 616             }
 617         }
 618       fputs ("</sup>", html->file);
 619     }
 620
 621   /* output </th> or </td>. */
 622   fprintf (html->file, "</%s>\n", tag);
 623 }
 624
 625 static void
 626 html_output_table_layer (struct html_driver *html, const struct pivot_table *pt,
 627                          const size_t *layer_indexes)
 628 {
 629   struct table *title, *layers, *body, *caption, *footnotes;
 630   pivot_output (pt, layer_indexes, true, &title, &layers, &body,
 631                 &caption, &footnotes, NULL, NULL);
 632
 633   fputs ("<table", html->file);
 634   if (pt->notes)
 635     {
 636       fputs (" title=\"", html->file);
 637       escape_string (html->file, pt->notes, " ", "\n");
 638       putc ('"', html->file);
 639     }
 640   fputs (">\n", html->file);
 641
 642   if (title)
 643     {
 644       struct table_cell cell;
 645       table_get_cell (title, 0, 0, &cell);
 646       html_put_table_cell (html, pt, &cell, "caption", NULL);
 647     }
 648
 649   if (layers)
 650     {
 651       fputs ("<thead>\n", html->file);
 652       for (size_t y = 0; y < layers->n[V]; y++)
 653         {
 654           fputs ("<tr>\n", html->file);
 655
 656           struct table_cell cell;
 657           table_get_cell (layers, 0, y, &cell);
 658           cell.d[H][1] = body->n[H];
 659           html_put_table_cell (html, pt, &cell, "td", NULL);
 660
 661           fputs ("</tr>\n", html->file);
 662         }
 663       fputs ("</thead>\n", html->file);
 664     }
 665
 666   fputs ("<tbody>\n", html->file);
 667   for (int y = 0; y < body->n[V]; y++)
 668     {
 669       fputs ("<tr>\n", html->file);
 670       for (int x = 0; x < body->n[H]; )
 671         {
 672           struct table_cell cell;
 673           table_get_cell (body, x, y, &cell);
 674           if (x == cell.d[TABLE_HORZ][0] && y == cell.d[TABLE_VERT][0])
 675             {
 676               bool is_header = (y < body->h[V][0]
 677                                 || y >= body->n[V] - body->h[V][1]
 678                                 || x < body->h[H][0]
 679                                 || x >= body->n[H] - body->h[H][1]);
 680               const char *tag = is_header ? "th" : "td";
 681               html_put_table_cell (html, pt, &cell, tag, body);
 682             }
 683
 684           x = cell.d[TABLE_HORZ][1];
 685         }
 686       fputs ("</tr>\n", html->file);
 687     }
 688   fputs ("</tbody>\n", html->file);
 689
 690   if (caption || footnotes)
 691     {
 692       fprintf (html->file, "<tfoot>\n");
 693
 694       if (caption)
 695         {
 696           fputs ("<tr>\n", html->file);
 697
 698           struct table_cell cell;
 699           table_get_cell (caption, 0, 0, &cell);
 700           cell.d[H][1] = body->n[H];
 701           html_put_table_cell (html, pt, &cell, "td", NULL);
 702
 703           fputs ("</tr>\n", html->file);
 704         }
 705
 706       if (footnotes)
 707         for (size_t y = 0; y < footnotes->n[V]; y++)
 708           {
 709             fputs ("<tr>\n", html->file);
 710
 711             struct table_cell cell;
 712             table_get_cell (footnotes, 0, y, &cell);
 713             cell.d[H][1] = body->n[H];
 714             html_put_table_cell (html, pt, &cell, "td", NULL);
 715
 716             fputs ("</tr>\n", html->file);
 717           }
 718       fputs ("</tfoot>\n", html->file);
 719     }
 720
 721   fputs ("</table>\n\n", html->file);
 722
 723   table_unref (title);
 724   table_unref (layers);
 725   table_unref (body);
 726   table_unref (caption);
 727   table_unref (footnotes);
 728 }
 729
 730 static void
 731 html_output_table (struct html_driver *html, const struct output_item *item)
 732 {
 733   size_t *layer_indexes;
 734   PIVOT_OUTPUT_FOR_EACH_LAYER (layer_indexes, item->table, true)
 735     html_output_table_layer (html, item->table, layer_indexes);
 736 }
 737
 738 struct output_driver_factory html_driver_factory =
 739   { "html", "pspp.html", html_create };
 740
 741 static const struct output_driver_class html_driver_class =
 742   {
 743     .name = "html",
 744     .destroy = html_destroy,
 745     .submit = html_submit,
 746     .handles_groups = true,
 747   };