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.h"
37 #include "output/chart-item.h"
38 #include "output/driver-provider.h"
39 #include "output/message-item.h"
40 #include "output/options.h"
41 #include "output/output-item-provider.h"
42 #include "output/table-provider.h"
43 #include "output/table-item.h"
44 #include "output/text-item.h"
45
46 #include "gl/minmax.h"
47 #include "gl/xalloc.h"
48
49 #include "gettext.h"
50 #define _(msgid) gettext (msgid)
51
52 struct html_driver
53 {
54 struct output_driver driver;
55 #ifdef HAVE_CAIRO
56 struct cell_color fg;
57 struct cell_color bg;
58 #endif
59 struct file_handle *handle;
60 char *chart_file_name;
61
62 FILE *file;
63 size_t chart_cnt;
64
65 bool css;
66 bool borders;
67 };
68
69 static const struct output_driver_class html_driver_class;
70
71 static void html_output_table (struct html_driver *, const struct table_item *);
72 static void escape_string (FILE *file, const char *text,
73 const char *space, const char *newline);
74 static void print_title_tag (FILE *file, const char *name,
75 const char *content);
76
77 static struct html_driver *
html_driver_cast(struct output_driver * driver)78 html_driver_cast (struct output_driver *driver)
79 {
80 assert (driver->class == &html_driver_class);
81 return UP_CAST (driver, struct html_driver, driver);
82 }
83
84 static struct driver_option *
opt(struct output_driver * d,struct string_map * options,const char * key,const char * default_value)85 opt (struct output_driver *d, struct string_map *options, const char *key,
86 const char *default_value)
87 {
88 return driver_option_get (d, options, key, default_value);
89 }
90
91 static struct output_driver *
html_create(struct file_handle * fh,enum settings_output_devices device_type,struct string_map * o)92 html_create (struct file_handle *fh, enum settings_output_devices device_type,
93 struct string_map *o)
94 {
95 struct output_driver *d;
96 struct html_driver *html;
97
98 html = xzalloc (sizeof *html);
99 d = &html->driver;
100 output_driver_init (&html->driver, &html_driver_class, fh_get_file_name (fh),
101 device_type);
102 html->css = parse_boolean (opt (d, o, "css", "true"));
103 html->borders = parse_boolean (opt (d, o, "borders", "true"));
104
105 html->handle = fh;
106 html->chart_file_name = parse_chart_file_name (opt (d, o, "charts",
107 fh_get_file_name (fh)));
108 html->file = NULL;
109 html->chart_cnt = 1;
110 #ifdef HAVE_CAIRO
111 parse_color (d, o, "background-color", "#FFFFFFFFFFFF", &html->bg);
112 parse_color (d, o, "foreground-color", "#000000000000", &html->fg);
113 #endif
114 html->file = fn_open (html->handle, "w");
115 if (html->file == NULL)
116 {
117 msg_error (errno, _("error opening output file `%s'"), fh_get_file_name (html->handle));
118 goto error;
119 }
120
121 fputs ("<!doctype html>\n", html->file);
122 fprintf (html->file, "<html");
123 char *ln = get_language ();
124 if (ln)
125 fprintf (html->file, " lang=\"%s\"", ln);
126 free (ln);
127 fprintf (html->file, ">\n");
128 fputs ("<head>\n", html->file);
129 print_title_tag (html->file, "title", _("PSPP Output"));
130 fprintf (html->file, "<meta name=\"generator\" content=\"%s\">\n", version);
131 fputs ("<meta http-equiv=\"content-type\" "
132 "content=\"text/html; charset=utf-8\">\n", html->file);
133
134 if (html->css)
135 {
136 fputs ("<style>\n"
137 "<!--\n"
138 "body {\n"
139 " background: white;\n"
140 " color: black;\n"
141 " padding: 0em 12em 0em 3em;\n"
142 " margin: 0\n"
143 "}\n"
144 "body>p {\n"
145 " margin: 0pt 0pt 0pt 0em\n"
146 "}\n"
147 "body>p + p {\n"
148 " text-indent: 1.5em;\n"
149 "}\n"
150 "h1 {\n"
151 " font-size: 150%;\n"
152 " margin-left: -1.33em\n"
153 "}\n"
154 "h2 {\n"
155 " font-size: 125%;\n"
156 " font-weight: bold;\n"
157 " margin-left: -.8em\n"
158 "}\n"
159 "h3 {\n"
160 " font-size: 100%;\n"
161 " font-weight: bold;\n"
162 " margin-left: -.5em }\n"
163 "h4 {\n"
164 " font-size: 100%;\n"
165 " margin-left: 0em\n"
166 "}\n"
167 "h1, h2, h3, h4, h5, h6 {\n"
168 " font-family: sans-serif;\n"
169 " color: blue\n"
170 "}\n"
171 "html {\n"
172 " margin: 0\n"
173 "}\n"
174 "code {\n"
175 " font-family: sans-serif\n"
176 "}\n"
177 "table {\n"
178 " border-collapse: collapse;\n"
179 " margin-bottom: 1em\n"
180 "}\n"
181 "th { background: #dddddd; font-weight: normal; font-style: oblique }\n"
182 "caption {\n"
183 " text-align: left\n"
184 "}\n"
185
186 "a:link {\n"
187 " color: #1f00ff;\n"
188 "}\n"
189 "a:visited {\n"
190 " color: #9900dd;\n"
191 "}\n"
192 "a:active {\n"
193 " color: red;\n"
194 "}\n"
195 "-->\n"
196 "</style>\n",
197 html->file);
198 }
199 fputs ("</head>\n", html->file);
200 fputs ("<body>\n", html->file);
201
202 return d;
203
204 error:
205 output_driver_destroy (d);
206 return NULL;
207 }
208
209 /* Emits <NAME>CONTENT</NAME> to the output, escaping CONTENT as
210 necessary for HTML. */
211 static void
print_title_tag(FILE * file,const char * name,const char * content)212 print_title_tag (FILE *file, const char *name, const char *content)
213 {
214 if (content != NULL)
215 {
216 fprintf (file, "<%s>", name);
217 escape_string (file, content, " ", " - ");
218 fprintf (file, "</%s>\n", name);
219 }
220 }
221
222 static void
html_destroy(struct output_driver * driver)223 html_destroy (struct output_driver *driver)
224 {
225 struct html_driver *html = html_driver_cast (driver);
226
227 if (html->file != NULL)
228 {
229 fprintf (html->file,
230 "</body>\n"
231 "</html>\n"
232 "<!-- end of file -->\n");
233 fn_close (html->handle, html->file);
234 }
235 free (html->chart_file_name);
236 fh_unref (html->handle);
237 free (html);
238 }
239
240 static void
html_submit(struct output_driver * driver,const struct output_item * output_item)241 html_submit (struct output_driver *driver,
242 const struct output_item *output_item)
243 {
244 struct html_driver *html = html_driver_cast (driver);
245
246 if (is_table_item (output_item))
247 {
248 struct table_item *table_item = to_table_item (output_item);
249 html_output_table (html, table_item);
250 }
251 #ifdef HAVE_CAIRO
252 else if (is_chart_item (output_item) && html->chart_file_name != NULL)
253 {
254 struct chart_item *chart_item = to_chart_item (output_item);
255 char *file_name;
256
257 file_name = xr_draw_png_chart (chart_item, html->chart_file_name,
258 html->chart_cnt++,
259 &html->fg,
260 &html->bg
261 );
262 if (file_name != NULL)
263 {
264 const char *title = chart_item_get_title (chart_item);
265 fprintf (html->file, "<img src=\"%s\" alt=\"chart: %s\">",
266 file_name, title ? title : _("No description"));
267 free (file_name);
268 }
269 }
270 #endif /* HAVE_CAIRO */
271 else if (is_text_item (output_item))
272 {
273 struct text_item *text_item = to_text_item (output_item);
274 const char *s = text_item_get_text (text_item);
275
276 switch (text_item_get_type (text_item))
277 {
278 case TEXT_ITEM_PAGE_TITLE:
279 break;
280
281 case TEXT_ITEM_TITLE:
282 {
283 int level = MIN (5, output_get_group_level ()) + 1;
284 char tag[3] = { 'H', level + '1', '\0' };
285 print_title_tag (html->file, tag, s);
286 }
287 break;
288
289 case TEXT_ITEM_SYNTAX:
290 fprintf (html->file, "<pre class=\"syntax\">");
291 escape_string (html->file, s, " ", "<br>");
292 fprintf (html->file, "</pre>\n");
293 break;
294
295 case TEXT_ITEM_LOG:
296 print_title_tag (html->file, "pre", s); /* should be <P><TT> */
297 break;
298
299 case TEXT_ITEM_EJECT_PAGE:
300 /* Nothing to do. */
301 break;
302 }
303 }
304 else if (is_message_item (output_item))
305 {
306 const struct message_item *message_item = to_message_item (output_item);
307 char *s = msg_to_string (message_item_get_msg (message_item));
308 print_title_tag (html->file, "p", s);
309 free (s);
310 }
311 }
312
313 /* Write TEXT to file F, escaping characters as necessary for HTML. Spaces are
314 replaced by SPACE, which should be " " or " " New-lines are replaced by
315 NEWLINE, which might be "<BR>" or "\n" or something else appropriate. */
316 static void
escape_string(FILE * file,const char * text,const char * space,const char * newline)317 escape_string (FILE *file, const char *text,
318 const char *space, const char *newline)
319 {
320 for (;;)
321 {
322 char c = *text++;
323 switch (c)
324 {
325 case 0:
326 return;
327 case '\n':
328 fputs (newline, file);
329 break;
330 case '&':
331 fputs ("&", file);
332 break;
333 case '<':
334 fputs ("<", file);
335 break;
336 case '>':
337 fputs (">", file);
338 break;
339 case ' ':
340 fputs (space, file);
341 break;
342 case '"':
343 fputs (""", file);
344 break;
345 default:
346 putc (c, file);
347 break;
348 }
349 }
350 }
351
352 static void
escape_tag(FILE * file,const char * tag,const char * text,const char * space,const char * newline)353 escape_tag (FILE *file, const char *tag,
354 const char *text, const char *space, const char *newline)
355 {
356 if (!text || !*text)
357 return;
358
359 fprintf (file, "<%s>", tag);
360 escape_string (file, text, space, newline);
361 fprintf (file, "</%s>", tag);
362 }
363
364 static const char *
border_to_css(int border)365 border_to_css (int border)
366 {
367 switch (border)
368 {
369 case TABLE_STROKE_NONE:
370 return NULL;
371
372 case TABLE_STROKE_SOLID:
373 return "solid";
374
375 case TABLE_STROKE_DASHED:
376 return "dashed";
377
378 case TABLE_STROKE_THICK:
379 return "thick solid";
380
381 case TABLE_STROKE_THIN:
382 return "thin solid";
383
384 case TABLE_STROKE_DOUBLE:
385 return "double";
386
387 default:
388 return NULL;
389 }
390
391 }
392
393 struct css_style
394 {
395 FILE *file;
396 int n_styles;
397 };
398
399 static struct css_style *
style_start(FILE * file)400 style_start (FILE *file)
401 {
402 struct css_style *cs = XMALLOC (struct css_style);
403 cs->file = file;
404 cs->n_styles = 0;
405 fputs (" style=\"", file);
406 return cs;
407 }
408
409 static void
style_end(struct css_style * cs)410 style_end (struct css_style *cs)
411 {
412 fputs ("\"", cs->file);
413 free (cs);
414 }
415
416 static void
put_style(struct css_style * st,const char * name,const char * value)417 put_style (struct css_style *st, const char *name, const char *value)
418 {
419 if (st->n_styles++ > 0)
420 fputs ("; ", st->file);
421 fprintf (st->file, "%s: %s", name, value);
422 }
423
424 static void
put_border(struct css_style * st,int style,const char * border_name)425 put_border (struct css_style *st, int style, const char *border_name)
426 {
427 const char *css = border_to_css (style);
428 if (css)
429 {
430 if (st->n_styles++ > 0)
431 fputs ("; ", st->file);
432 fprintf (st->file, "border-%s: %s", border_name, css);
433 }
434 }
435
436 static void
put_tfoot(struct html_driver * html,const struct table * t,bool * tfoot)437 put_tfoot (struct html_driver *html, const struct table *t, bool *tfoot)
438 {
439 if (!*tfoot)
440 {
441 fputs ("<tfoot>\n", html->file);
442 fputs ("<tr>\n", html->file);
443 fprintf (html->file, "<td colspan=%d>\n", table_nc (t));
444 *tfoot = true;
445 }
446 else
447 fputs ("\n<br>", html->file);
448 }
449
450 static void
html_put_footnote_markers(struct html_driver * html,const struct footnote ** footnotes,size_t n_footnotes)451 html_put_footnote_markers (struct html_driver *html,
452 const struct footnote **footnotes,
453 size_t n_footnotes)
454 {
455 if (n_footnotes > 0)
456 {
457 fputs ("<sup>", html->file);
458 for (size_t i = 0; i < n_footnotes; i++)
459 {
460 const struct footnote *f = footnotes[i];
461
462 if (i > 0)
463 putc (',', html->file);
464 escape_string (html->file, f->marker, " ", "<br>");
465 }
466 fputs ("</sup>", html->file);
467 }
468 }
469
470 static void
html_put_table_item_text(struct html_driver * html,const struct table_item_text * text)471 html_put_table_item_text (struct html_driver *html,
472 const struct table_item_text *text)
473 {
474 escape_string (html->file, text->content, " ", "<br>");
475 html_put_footnote_markers (html, text->footnotes, text->n_footnotes);
476 }
477
478 static void
html_put_table_item_layers(struct html_driver * html,const struct table_item_layers * layers)479 html_put_table_item_layers (struct html_driver *html,
480 const struct table_item_layers *layers)
481 {
482 for (size_t i = 0; i < layers->n_layers; i++)
483 {
484 if (i)
485 fputs ("<br>\n", html->file);
486
487 const struct table_item_layer *layer = &layers->layers[i];
488 escape_string (html->file, layer->content, " ", "<br>");
489 html_put_footnote_markers (html, layer->footnotes, layer->n_footnotes);
490 }
491 }
492
493 static void
html_output_table(struct html_driver * html,const struct table_item * item)494 html_output_table (struct html_driver *html, const struct table_item *item)
495 {
496 const struct table *t = table_item_get_table (item);
497 bool tfoot = false;
498 int y;
499
500 fputs ("<table>\n", html->file);
501
502 const struct table_item_text *caption = table_item_get_caption (item);
503 if (caption)
504 {
505 put_tfoot (html, t, &tfoot);
506 html_put_table_item_text (html, caption);
507 }
508 const struct footnote **f;
509 size_t n_footnotes = table_collect_footnotes (item, &f);
510
511 for (size_t i = 0; i < n_footnotes; i++)
512 {
513 put_tfoot (html, t, &tfoot);
514 escape_tag (html->file, "sup", f[i]->marker, " ", "<br>");
515 escape_string (html->file, f[i]->content, " ", "<br>");
516 }
517 free (f);
518 if (tfoot)
519 {
520 fputs ("</td>\n", html->file);
521 fputs ("</tr>\n", html->file);
522 fputs ("</tfoot>\n", html->file);
523 }
524
525 const struct table_item_text *title = table_item_get_title (item);
526 const struct table_item_layers *layers = table_item_get_layers (item);
527 if (title || layers)
528 {
529 fputs ("<caption>", html->file);
530 if (title)
531 html_put_table_item_text (html, title);
532 if (title && layers)
533 fputs ("<br>\n", html->file);
534 if (layers)
535 html_put_table_item_layers (html, layers);
536 fputs ("</caption>\n", html->file);
537 }
538
539 fputs ("<tbody>\n", html->file);
540
541 for (y = 0; y < table_nr (t); y++)
542 {
543 int x;
544
545 fputs ("<tr>\n", html->file);
546 for (x = 0; x < table_nc (t);)
547 {
548 struct table_cell cell;
549 const char *tag;
550
551 table_get_cell (t, x, y, &cell);
552 if (x != cell.d[TABLE_HORZ][0] || y != cell.d[TABLE_VERT][0])
553 goto next_1;
554
555 /* output <td> or <th> tag. */
556 bool is_header = (y < table_ht (t)
557 || y >= table_nr (t) - table_hb (t)
558 || x < table_hl (t)
559 || x >= table_nc (t) - table_hr (t));
560 tag = is_header ? "th" : "td";
561 fprintf (html->file, "<%s", tag);
562
563 struct css_style *style = style_start (html->file);
564 enum table_halign halign = table_halign_interpret (
565 cell.style->cell_style.halign, cell.options & TAB_NUMERIC);
566
567 switch (halign)
568 {
569 case TABLE_HALIGN_RIGHT:
570 put_style (style, "text-align", "right");
571 break;
572 case TABLE_HALIGN_CENTER:
573 put_style (style, "text-align", "center");
574 break;
575 default:
576 /* Do nothing */
577 break;
578 }
579
580 if (cell.style->cell_style.valign != TABLE_VALIGN_TOP)
581 {
582 put_style (style, "vertical-align",
583 (cell.style->cell_style.valign == TABLE_VALIGN_BOTTOM
584 ? "bottom" : "middle"));
585 }
586
587 int colspan = table_cell_colspan (&cell);
588 int rowspan = table_cell_rowspan (&cell);
589
590 if (html->borders)
591 {
592 /* Cell borders. */
593 struct cell_color color;
594
595 int top = table_get_rule (t, TABLE_VERT, x, y, &color);
596 put_border (style, top, "top");
597
598 if (y + rowspan == table_nr (t))
599 {
600 int bottom = table_get_rule (t, TABLE_VERT, x, y + rowspan,
601 &color);
602 put_border (style, bottom, "bottom");
603 }
604
605 int left = table_get_rule (t, TABLE_HORZ, x, y, &color);
606 put_border (style, left, "left");
607
608 if (x + colspan == table_nc (t))
609 {
610 int right = table_get_rule (t, TABLE_HORZ, x + colspan, y,
611 &color);
612 put_border (style, right, "right");
613 }
614 }
615 style_end (style);
616
617 if (colspan > 1)
618 fprintf (html->file, " colspan=\"%d\"", colspan);
619
620 if (rowspan > 1)
621 fprintf (html->file, " rowspan=\"%d\"", rowspan);
622
623 putc ('>', html->file);
624
625 /* Output cell contents. */
626 const char *s = cell.text;
627 if (cell.options & TAB_FIX)
628 escape_tag (html->file, "tt", s, " ", "<br>");
629 else
630 {
631 s += strspn (s, CC_SPACES);
632 escape_string (html->file, s, " ", "<br>");
633 }
634
635 if (cell.n_subscripts)
636 {
637 fputs ("<sub>", html->file);
638 for (size_t i = 0; i < cell.n_subscripts; i++)
639 {
640 if (i)
641 putc (',', html->file);
642 escape_string (html->file, cell.subscripts[i],
643 " ", "<br>");
644 }
645 fputs ("</sub>", html->file);
646 }
647 if (cell.superscript)
648 escape_tag (html->file, "sup", cell.superscript, " ", "<br>");
649 html_put_footnote_markers (html, cell.footnotes, cell.n_footnotes);
650
651 /* output </th> or </td>. */
652 fprintf (html->file, "</%s>\n", tag);
653
654 next_1:
655 x = cell.d[TABLE_HORZ][1];
656 }
657 fputs ("</tr>\n", html->file);
658 }
659
660 fputs ("</tbody>\n", html->file);
661 fputs ("</table>\n\n", html->file);
662 }
663
664 struct output_driver_factory html_driver_factory =
665 { "html", "pspp.html", html_create };
666
667 static const struct output_driver_class html_driver_class =
668 {
669 "html",
670 html_destroy,
671 html_submit,
672 NULL,
673 };
674