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