1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2017, 2018 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
19 #include "output/spv/spv.h"
23 #include <libxml/HTMLparser.h>
24 #include <libxml/xmlreader.h>
28 #include "libpspp/assertion.h"
29 #include "libpspp/cast.h"
30 #include "libpspp/hash-functions.h"
31 #include "libpspp/message.h"
32 #include "libpspp/str.h"
33 #include "libpspp/zip-reader.h"
34 #include "output/page-setup-item.h"
35 #include "output/pivot-table.h"
36 #include "output/spv/detail-xml-parser.h"
37 #include "output/spv/light-binary-parser.h"
38 #include "output/spv/spv-css-parser.h"
39 #include "output/spv/spv-legacy-data.h"
40 #include "output/spv/spv-legacy-decoder.h"
41 #include "output/spv/spv-light-decoder.h"
42 #include "output/spv/spv-table-look.h"
43 #include "output/spv/structure-xml-parser.h"
45 #include "gl/c-ctype.h"
46 #include "gl/intprops.h"
47 #include "gl/minmax.h"
48 #include "gl/xalloc.h"
49 #include "gl/xvasprintf.h"
53 #define _(msgid) gettext (msgid)
54 #define N_(msgid) (msgid)
58 struct string zip_errs;
59 struct zip_reader *zip;
60 struct spv_item *root;
61 struct page_setup *page_setup;
64 const struct page_setup *
65 spv_get_page_setup (const struct spv_reader *spv)
67 return spv->page_setup;
71 spv_item_type_to_string (enum spv_item_type type)
75 case SPV_ITEM_HEADING: return "heading";
76 case SPV_ITEM_TEXT: return "text";
77 case SPV_ITEM_TABLE: return "table";
78 case SPV_ITEM_GRAPH: return "graph";
79 case SPV_ITEM_MODEL: return "model";
80 case SPV_ITEM_IMAGE: return "image";
81 default: return "**error**";
86 spv_item_class_to_string (enum spv_item_class class)
90 #define SPV_CLASS(ENUM, NAME) case SPV_CLASS_##ENUM: return NAME;
98 spv_item_class_from_string (const char *name)
100 #define SPV_CLASS(ENUM, NAME) \
101 if (!strcmp (name, NAME)) return SPV_CLASS_##ENUM;
105 return (enum spv_item_class) SPV_N_CLASSES;
109 spv_item_get_type (const struct spv_item *item)
115 spv_item_get_class (const struct spv_item *item)
117 const char *label = spv_item_get_label (item);
123 case SPV_ITEM_HEADING:
124 return SPV_CLASS_HEADINGS;
127 return (!strcmp (label, "Title") ? SPV_CLASS_OUTLINEHEADERS
128 : !strcmp (label, "Log") ? SPV_CLASS_LOGS
129 : !strcmp (label, "Page Title") ? SPV_CLASS_PAGETITLE
133 return (!strcmp (label, "Warnings") ? SPV_CLASS_WARNINGS
134 : !strcmp (label, "Notes") ? SPV_CLASS_NOTES
138 return SPV_CLASS_CHARTS;
141 return SPV_CLASS_MODELS;
144 return SPV_CLASS_OTHER;
147 return SPV_CLASS_TREES;
150 return SPV_CLASS_UNKNOWN;
155 spv_item_get_label (const struct spv_item *item)
161 spv_item_is_heading (const struct spv_item *item)
163 return item->type == SPV_ITEM_HEADING;
167 spv_item_get_n_children (const struct spv_item *item)
169 return item->n_children;
173 spv_item_get_child (const struct spv_item *item, size_t idx)
175 assert (idx < item->n_children);
176 return item->children[idx];
180 spv_item_is_table (const struct spv_item *item)
182 return item->type == SPV_ITEM_TABLE;
186 spv_item_is_text (const struct spv_item *item)
188 return item->type == SPV_ITEM_TEXT;
191 const struct pivot_value *
192 spv_item_get_text (const struct spv_item *item)
194 assert (spv_item_is_text (item));
199 spv_item_is_image (const struct spv_item *item)
201 return item->type == SPV_ITEM_IMAGE;
204 static cairo_status_t
205 read_from_zip_member (void *zm_, unsigned char *data, unsigned int length)
207 struct zip_member *zm = zm_;
209 return CAIRO_STATUS_READ_ERROR;
213 int n = zip_member_read (zm, data, length);
215 return CAIRO_STATUS_READ_ERROR;
221 return CAIRO_STATUS_SUCCESS;
225 spv_item_get_image (const struct spv_item *item_)
227 struct spv_item *item = CONST_CAST (struct spv_item *, item_);
228 assert (spv_item_is_image (item));
232 struct zip_member *zm = zip_member_open (item->spv->zip,
234 item->image = cairo_image_surface_create_from_png_stream (
235 read_from_zip_member, zm);
237 zip_member_finish (zm);
244 spv_item_next (const struct spv_item *item)
246 if (item->n_children)
247 return item->children[0];
251 size_t idx = item->parent_idx + 1;
253 if (idx < item->n_children)
254 return item->children[idx];
260 const struct spv_item *
261 spv_item_get_parent (const struct spv_item *item)
267 spv_item_get_level (const struct spv_item *item)
270 for (; item->parent; item = item->parent)
276 spv_item_get_command_id (const struct spv_item *item)
278 return item->command_id;
282 spv_item_get_subtype (const struct spv_item *item)
284 return item->subtype;
288 spv_item_is_visible (const struct spv_item *item)
290 return item->visible;
294 spv_item_destroy (struct spv_item *item)
298 free (item->structure_member);
301 free (item->command_id);
303 for (size_t i = 0; i < item->n_children; i++)
304 spv_item_destroy (item->children[i]);
305 free (item->children);
307 pivot_table_unref (item->table);
308 pivot_table_look_unref (item->table_look);
309 free (item->bin_member);
310 free (item->xml_member);
311 free (item->subtype);
313 pivot_value_destroy (item->text);
315 free (item->png_member);
317 cairo_surface_destroy (item->image);
324 spv_heading_add_child (struct spv_item *parent, struct spv_item *child)
326 assert (parent->type == SPV_ITEM_HEADING);
327 assert (!child->parent);
329 child->parent = parent;
330 child->parent_idx = parent->n_children;
332 if (parent->n_children >= parent->allocated_children)
333 parent->children = x2nrealloc (parent->children,
334 &parent->allocated_children,
335 sizeof *parent->children);
336 parent->children[parent->n_children++] = child;
340 find_xml_child_element (xmlNode *parent, const char *child_name)
342 for (xmlNode *node = parent->children; node; node = node->next)
343 if (node->type == XML_ELEMENT_NODE
345 && !strcmp (CHAR_CAST (char *, node->name), child_name))
352 get_xml_attr (const xmlNode *node, const char *name)
354 return CHAR_CAST (char *, xmlGetProp (node, CHAR_CAST (xmlChar *, name)));
358 put_xml_attr (const char *name, const char *value, struct string *dst)
363 ds_put_format (dst, " %s=\"", name);
364 for (const char *p = value; *p; p++)
369 ds_put_cstr (dst, " ");
372 ds_put_cstr (dst, "&");
375 ds_put_cstr (dst, "<");
378 ds_put_cstr (dst, ">");
381 ds_put_cstr (dst, """);
384 ds_put_byte (dst, *p);
388 ds_put_byte (dst, '"');
392 extract_html_text (const xmlNode *node, int base_font_size, struct string *s)
394 if (node->type == XML_ELEMENT_NODE)
396 const char *name = CHAR_CAST (char *, node->name);
397 if (!strcmp (name, "br"))
398 ds_put_byte (s, '\n');
399 else if (strcmp (name, "style"))
401 const char *tag = NULL;
402 if (strchr ("biu", name[0]) && name[1] == '\0')
405 ds_put_format (s, "<%s>", tag);
407 else if (!strcmp (name, "font"))
410 ds_put_format (s, "<%s", tag);
412 char *face = get_xml_attr (node, "face");
413 put_xml_attr ("face", face, s);
416 char *color = get_xml_attr (node, "color");
420 put_xml_attr ("color", color, s);
424 if (sscanf (color, "rgb (%"SCNu8", %"SCNu8", %"SCNu8")",
428 snprintf (color2, sizeof color2,
429 "#%02"PRIx8"%02"PRIx8"%02"PRIx8,
431 put_xml_attr ("color", color2, s);
437 char *size_s = get_xml_attr (node, "size");
438 int html_size = size_s ? atoi (size_s) : 0;
440 if (html_size >= 1 && html_size <= 7)
442 static const double scale[7] = {
443 .444, .556, .667, .778, 1.0, 1.33, 2.0
445 double size = base_font_size * scale[html_size - 1];
447 char size2[INT_BUFSIZE_BOUND (int)];
448 snprintf (size2, sizeof size2, "%.0f", size * 1024.);
449 put_xml_attr ("size", size2, s);
452 ds_put_cstr (s, ">");
454 for (const xmlNode *child = node->children; child;
456 extract_html_text (child, base_font_size, s);
458 ds_put_format (s, "</%s>", tag);
461 else if (node->type == XML_TEXT_NODE)
463 /* U+00A0 NONBREAKING SPACE is really, really common in SPV text and it
464 makes it impossible to break syntax across lines. Translate it into a
465 regular space. (Note that U+00A0 is C2 A0 in UTF-8.)
467 Do the same for U+2007 FIGURE SPACE, which also crops out weirdly
469 ds_extend (s, ds_length (s) + xmlStrlen (node->content));
470 for (const uint8_t *p = node->content; *p;)
473 if (p[0] == 0xc2 && p[1] == 0xa0)
478 else if (p[0] == 0xe2 && p[1] == 0x80 && p[2] == 0x87)
488 int last = ds_last (s);
489 if (last != EOF && !c_isspace (last))
493 ds_put_cstr (s, "<");
495 ds_put_cstr (s, ">");
497 ds_put_cstr (s, "&");
505 parse_embedded_html (const xmlNode *node)
507 /* Extract HTML from XML node. */
508 char *html_s = CHAR_CAST (char *, xmlNodeGetContent (node));
512 xmlDoc *html_doc = htmlReadMemory (
513 html_s, strlen (html_s),
514 NULL, "UTF-8", (HTML_PARSE_RECOVER | HTML_PARSE_NOERROR
515 | HTML_PARSE_NOWARNING | HTML_PARSE_NOBLANKS
516 | HTML_PARSE_NONET));
522 /* Given NODE, which should contain HTML content, returns the text within that
523 content as an allocated string. The caller must eventually free the
524 returned string (with xmlFree()). */
526 decode_embedded_html (const xmlNode *node, struct font_style *font_style)
528 struct string markup = DS_EMPTY_INITIALIZER;
529 *font_style = (struct font_style) FONT_STYLE_INITIALIZER;
530 font_style->size = 10;
532 xmlDoc *html_doc = parse_embedded_html (node);
535 xmlNode *root = xmlDocGetRootElement (html_doc);
536 xmlNode *head = root ? find_xml_child_element (root, "head") : NULL;
537 xmlNode *style = head ? find_xml_child_element (head, "style") : NULL;
540 uint8_t *style_s = xmlNodeGetContent (style);
541 spv_parse_css_style (CHAR_CAST (char *, style_s), font_style);
546 extract_html_text (root, font_style->size, &markup);
547 xmlFreeDoc (html_doc);
550 font_style->markup = true;
551 return ds_steal_cstr (&markup);
555 xstrdup_if_nonempty (const char *s)
557 return s && s[0] ? xstrdup (s) : NULL;
561 decode_container_text (const struct spvsx_container_text *ct,
562 struct spv_item *item)
564 item->type = SPV_ITEM_TEXT;
565 item->command_id = xstrdup_if_nonempty (ct->command_name);
567 item->text = xzalloc (sizeof *item->text);
568 item->text->type = PIVOT_VALUE_TEXT;
569 item->text->font_style = xmalloc (sizeof *item->text->font_style);
570 item->text->text.local = decode_embedded_html (ct->html->node_.raw,
571 item->text->font_style);
575 decode_page_p (const xmlNode *in, struct page_paragraph *out)
577 char *style = get_xml_attr (in, "style");
578 out->halign = (style && strstr (style, "center") ? TABLE_HALIGN_CENTER
579 : style && strstr (style, "right") ? TABLE_HALIGN_RIGHT
580 : TABLE_HALIGN_LEFT);
583 struct font_style font_style;
584 out->markup = decode_embedded_html (in, &font_style);
585 font_style_uninit (&font_style);
589 decode_page_paragraph (const struct spvsx_page_paragraph *page_paragraph,
590 struct page_heading *ph)
592 memset (ph, 0, sizeof *ph);
594 const struct spvsx_page_paragraph_text *page_paragraph_text
595 = page_paragraph->page_paragraph_text;
596 if (!page_paragraph_text)
599 xmlDoc *html_doc = parse_embedded_html (page_paragraph_text->node_.raw);
603 xmlNode *root = xmlDocGetRootElement (html_doc);
604 xmlNode *body = find_xml_child_element (root, "body");
606 for (const xmlNode *node = body->children; node; node = node->next)
607 if (node->type == XML_ELEMENT_NODE
608 && !strcmp (CHAR_CAST (const char *, node->name), "p"))
610 ph->paragraphs = xrealloc (ph->paragraphs,
611 (ph->n + 1) * sizeof *ph->paragraphs);
612 decode_page_p (node, &ph->paragraphs[ph->n++]);
614 xmlFreeDoc (html_doc);
618 spv_item_load (const struct spv_item *item)
620 if (spv_item_is_table (item))
621 spv_item_get_table (item);
622 else if (spv_item_is_image (item))
623 spv_item_get_image (item);
627 spv_item_is_light_table (const struct spv_item *item)
629 return item->type == SPV_ITEM_TABLE && !item->xml_member;
632 char * WARN_UNUSED_RESULT
633 spv_item_get_raw_light_table (const struct spv_item *item,
634 void **data, size_t *size)
636 return zip_member_read_all (item->spv->zip, item->bin_member, data, size);
639 char * WARN_UNUSED_RESULT
640 spv_item_get_light_table (const struct spv_item *item,
641 struct spvlb_table **tablep)
645 if (!spv_item_is_light_table (item))
646 return xstrdup ("not a light binary table object");
650 char *error = spv_item_get_raw_light_table (item, &data, &size);
654 struct spvbin_input input;
655 spvbin_input_init (&input, data, size);
657 struct spvlb_table *table = NULL;
659 ? xasprintf ("light table member is empty")
660 : !spvlb_parse_table (&input, &table)
661 ? spvbin_input_to_error (&input, NULL)
662 : input.ofs != input.size
663 ? xasprintf ("expected end of file at offset %#zx", input.ofs)
667 struct string s = DS_EMPTY_INITIALIZER;
668 spv_item_format_path (item, &s);
669 ds_put_format (&s, " (%s): %s", item->bin_member, error);
672 error = ds_steal_cstr (&s);
681 pivot_table_open_light (struct spv_item *item)
683 assert (spv_item_is_light_table (item));
685 struct spvlb_table *raw_table;
686 char *error = spv_item_get_light_table (item, &raw_table);
688 error = decode_spvlb_table (raw_table, &item->table);
689 spvlb_free_table (raw_table);
695 spv_item_is_legacy_table (const struct spv_item *item)
697 return item->type == SPV_ITEM_TABLE && item->xml_member;
700 char * WARN_UNUSED_RESULT
701 spv_item_get_raw_legacy_data (const struct spv_item *item,
702 void **data, size_t *size)
704 if (!spv_item_is_legacy_table (item))
705 return xstrdup ("not a legacy table object");
707 return zip_member_read_all (item->spv->zip, item->bin_member, data, size);
710 char * WARN_UNUSED_RESULT
711 spv_item_get_legacy_data (const struct spv_item *item, struct spv_data *data)
715 char *error = spv_item_get_raw_legacy_data (item, &raw, &size);
718 error = spv_legacy_data_decode (raw, size, data);
725 static char * WARN_UNUSED_RESULT
726 spv_read_xml_member (struct spv_reader *spv, const char *member_name,
727 bool keep_blanks, const char *root_element_name,
732 struct zip_member *zm = zip_member_open (spv->zip, member_name);
734 return ds_steal_cstr (&spv->zip_errs);
736 xmlParserCtxt *parser;
737 xmlKeepBlanksDefault (keep_blanks);
738 parser = xmlCreatePushParserCtxt(NULL, NULL, NULL, 0, NULL);
741 zip_member_finish (zm);
742 return xasprintf (_("%s: Failed to create XML parser"), member_name);
747 while ((retval = zip_member_read (zm, buf, sizeof buf)) > 0)
748 xmlParseChunk (parser, buf, retval, false);
749 xmlParseChunk (parser, NULL, 0, true);
751 xmlDoc *doc = parser->myDoc;
752 bool well_formed = parser->wellFormed;
753 xmlFreeParserCtxt (parser);
757 char *error = ds_steal_cstr (&spv->zip_errs);
758 zip_member_finish (zm);
762 zip_member_finish (zm);
767 return xasprintf(_("%s: document is not well-formed"), member_name);
770 const xmlNode *root_node = xmlDocGetRootElement (doc);
771 assert (root_node->type == XML_ELEMENT_NODE);
772 if (strcmp (CHAR_CAST (char *, root_node->name), root_element_name))
775 return xasprintf(_("%s: root node is \"%s\" but \"%s\" was expected"),
777 CHAR_CAST (char *, root_node->name), root_element_name);
784 char * WARN_UNUSED_RESULT
785 spv_item_get_legacy_table (const struct spv_item *item, xmlDoc **docp)
787 assert (spv_item_is_legacy_table (item));
789 return spv_read_xml_member (item->spv, item->xml_member, false,
790 "visualization", docp);
793 char * WARN_UNUSED_RESULT
794 spv_item_get_structure (const struct spv_item *item, struct _xmlDoc **docp)
796 return spv_read_xml_member (item->spv, item->structure_member, false,
801 identify_item (const struct spv_item *item)
803 return (item->label ? item->label
804 : item->command_id ? item->command_id
805 : spv_item_type_to_string (item->type));
809 spv_item_format_path (const struct spv_item *item, struct string *s)
811 enum { MAX_STACK = 32 };
812 const struct spv_item *stack[MAX_STACK];
815 while (item != NULL && item->parent && n < MAX_STACK)
824 ds_put_byte (s, '/');
826 const char *name = identify_item (item);
827 ds_put_cstr (s, name);
833 for (size_t i = 0; i < item->parent->n_children; i++)
835 const struct spv_item *sibling = item->parent->children[i];
838 else if (!strcmp (name, identify_item (sibling)))
842 ds_put_format (s, "[%zu]", index);
847 static char * WARN_UNUSED_RESULT
848 pivot_table_open_legacy (struct spv_item *item)
850 assert (spv_item_is_legacy_table (item));
852 struct spv_data data;
853 char *error = spv_item_get_legacy_data (item, &data);
856 struct string s = DS_EMPTY_INITIALIZER;
857 spv_item_format_path (item, &s);
858 ds_put_format (&s, " (%s): %s", item->bin_member, error);
861 return ds_steal_cstr (&s);
865 error = spv_read_xml_member (item->spv, item->xml_member, false,
866 "visualization", &doc);
869 spv_data_uninit (&data);
873 struct spvxml_context ctx = SPVXML_CONTEXT_INIT (ctx);
874 struct spvdx_visualization *v;
875 spvdx_parse_visualization (&ctx, xmlDocGetRootElement (doc), &v);
876 error = spvxml_context_finish (&ctx, &v->node_);
879 error = decode_spvdx_table (v, item->subtype, item->table_look,
880 &data, &item->table);
884 struct string s = DS_EMPTY_INITIALIZER;
885 spv_item_format_path (item, &s);
886 ds_put_format (&s, " (%s): %s", item->xml_member, error);
889 error = ds_steal_cstr (&s);
892 spv_data_uninit (&data);
893 spvdx_free_visualization (v);
900 const struct pivot_table *
901 spv_item_get_table (const struct spv_item *item_)
903 struct spv_item *item = CONST_CAST (struct spv_item *, item_);
905 assert (spv_item_is_table (item));
908 char *error = (item->xml_member
909 ? pivot_table_open_legacy (item)
910 : pivot_table_open_light (item));
914 msg (ME, "%s", error);
915 item->table = pivot_table_create_for_text (
916 pivot_value_new_text (N_("Error")),
917 pivot_value_new_user_text (error, -1));
925 /* Constructs a new spv_item from XML and stores it in *ITEMP. Returns NULL if
926 successful, otherwise an error message for the caller to use and free (with
929 XML should be a 'heading' or 'container' element. */
930 static char * WARN_UNUSED_RESULT
931 spv_decode_container (const struct spvsx_container *c,
932 const char *structure_member,
933 struct spv_item *parent)
935 struct spv_item *item = xzalloc (sizeof *item);
936 item->spv = parent->spv;
937 item->label = xstrdup (c->label->text);
938 item->visible = c->visibility == SPVSX_VISIBILITY_VISIBLE;
939 item->structure_member = xstrdup (structure_member);
941 assert (c->n_seq == 1);
942 struct spvxml_node *content = c->seq[0];
943 if (spvsx_is_container_text (content))
944 decode_container_text (spvsx_cast_container_text (content), item);
945 else if (spvsx_is_table (content))
947 item->type = SPV_ITEM_TABLE;
949 struct spvsx_table *table = spvsx_cast_table (content);
950 const struct spvsx_table_structure *ts = table->table_structure;
951 item->bin_member = xstrdup (ts->data_path->text);
952 item->command_id = xstrdup_if_nonempty (table->command_name);
953 item->subtype = xstrdup_if_nonempty (table->sub_type);
956 item->xml_member = ts->path ? xstrdup (ts->path->text) : NULL;
957 char *error = (table->table_properties
958 ? spv_table_look_decode (table->table_properties,
960 : xstrdup ("Legacy table lacks tableProperties"));
963 spv_item_destroy (item);
968 else if (spvsx_is_graph (content))
970 struct spvsx_graph *graph = spvsx_cast_graph (content);
971 item->type = SPV_ITEM_GRAPH;
972 item->command_id = xstrdup_if_nonempty (graph->command_name);
975 else if (spvsx_is_model (content))
977 struct spvsx_model *model = spvsx_cast_model (content);
978 item->type = SPV_ITEM_MODEL;
979 item->command_id = xstrdup_if_nonempty (model->command_name);
982 else if (spvsx_is_object (content))
984 struct spvsx_object *object = spvsx_cast_object (content);
985 item->type = SPV_ITEM_IMAGE;
986 item->png_member = xstrdup (object->uri);
988 else if (spvsx_is_image (content))
990 struct spvsx_image *image = spvsx_cast_image (content);
991 item->type = SPV_ITEM_IMAGE;
992 item->png_member = xstrdup (image->data_path->text);
994 else if (spvsx_is_tree (content))
995 item->type = SPV_ITEM_TREE;
999 spv_heading_add_child (parent, item);
1003 static char * WARN_UNUSED_RESULT
1004 spv_decode_children (struct spv_reader *spv, const char *structure_member,
1005 struct spvxml_node **seq, size_t n_seq,
1006 struct spv_item *parent)
1008 for (size_t i = 0; i < n_seq; i++)
1010 const struct spvxml_node *node = seq[i];
1013 if (spvsx_is_container (node))
1015 const struct spvsx_container *container
1016 = spvsx_cast_container (node);
1017 error = spv_decode_container (container, structure_member, parent);
1019 else if (spvsx_is_heading (node))
1021 const struct spvsx_heading *subheading = spvsx_cast_heading (node);
1022 struct spv_item *subitem = xzalloc (sizeof *subitem);
1023 subitem->structure_member = xstrdup (structure_member);
1024 subitem->spv = parent->spv;
1025 subitem->type = SPV_ITEM_HEADING;
1026 subitem->label = xstrdup (subheading->label->text);
1027 if (subheading->command_name)
1028 subitem->command_id = xstrdup (subheading->command_name);
1029 subitem->visible = !subheading->heading_visibility_present;
1030 spv_heading_add_child (parent, subitem);
1032 error = spv_decode_children (spv, structure_member,
1033 subheading->seq, subheading->n_seq,
1046 static struct page_setup *
1047 decode_page_setup (const struct spvsx_page_setup *in, const char *file_name)
1049 struct page_setup *out = xmalloc (sizeof *out);
1050 *out = (struct page_setup) PAGE_SETUP_INITIALIZER;
1052 out->initial_page_number = in->initial_page_number;
1054 if (in->paper_width != DBL_MAX)
1055 out->paper[TABLE_HORZ] = in->paper_width;
1056 if (in->paper_height != DBL_MAX)
1057 out->paper[TABLE_VERT] = in->paper_height;
1059 if (in->margin_left != DBL_MAX)
1060 out->margins[TABLE_HORZ][0] = in->margin_left;
1061 if (in->margin_right != DBL_MAX)
1062 out->margins[TABLE_HORZ][1] = in->margin_right;
1063 if (in->margin_top != DBL_MAX)
1064 out->margins[TABLE_VERT][0] = in->margin_top;
1065 if (in->margin_bottom != DBL_MAX)
1066 out->margins[TABLE_VERT][1] = in->margin_bottom;
1068 if (in->space_after != DBL_MAX)
1069 out->object_spacing = in->space_after;
1072 out->chart_size = (in->chart_size == SPVSX_CHART_SIZE_FULL_HEIGHT
1073 ? PAGE_CHART_FULL_HEIGHT
1074 : in->chart_size == SPVSX_CHART_SIZE_HALF_HEIGHT
1075 ? PAGE_CHART_HALF_HEIGHT
1076 : in->chart_size == SPVSX_CHART_SIZE_QUARTER_HEIGHT
1077 ? PAGE_CHART_QUARTER_HEIGHT
1078 : PAGE_CHART_AS_IS);
1080 decode_page_paragraph (in->page_header->page_paragraph, &out->headings[0]);
1081 decode_page_paragraph (in->page_footer->page_paragraph, &out->headings[1]);
1083 out->file_name = xstrdup (file_name);
1088 static char * WARN_UNUSED_RESULT
1089 spv_heading_read (struct spv_reader *spv,
1090 const char *file_name, const char *member_name)
1093 char *error = spv_read_xml_member (spv, member_name, true, "heading", &doc);
1097 struct spvxml_context ctx = SPVXML_CONTEXT_INIT (ctx);
1098 struct spvsx_root_heading *root;
1099 spvsx_parse_root_heading (&ctx, xmlDocGetRootElement (doc), &root);
1100 error = spvxml_context_finish (&ctx, &root->node_);
1102 if (!error && root->page_setup)
1103 spv->page_setup = decode_page_setup (root->page_setup, file_name);
1105 for (size_t i = 0; !error && i < root->n_seq; i++)
1106 error = spv_decode_children (spv, member_name, root->seq, root->n_seq,
1111 char *s = xasprintf ("%s: %s", member_name, error);
1116 spvsx_free_root_heading (root);
1123 spv_get_root (const struct spv_reader *spv)
1129 spv_detect__ (struct zip_reader *zip, char **errorp)
1133 const char *member = "META-INF/MANIFEST.MF";
1134 if (!zip_reader_contains_member (zip, member))
1139 *errorp = zip_member_read_all (zip, "META-INF/MANIFEST.MF",
1144 const char *magic = "allowPivoting=true";
1145 bool is_spv = size == strlen (magic) && !memcmp (magic, data, size);
1151 /* Returns NULL if FILENAME is an SPV file, otherwise an error string that the
1152 caller must eventually free(). */
1153 char * WARN_UNUSED_RESULT
1154 spv_detect (const char *filename)
1156 struct string zip_error;
1157 struct zip_reader *zip = zip_reader_create (filename, &zip_error);
1159 return ds_steal_cstr (&zip_error);
1162 if (spv_detect__ (zip, &error) <= 0 && !error)
1163 error = xasprintf("%s: not an SPV file", filename);
1164 zip_reader_destroy (zip);
1165 ds_destroy (&zip_error);
1169 char * WARN_UNUSED_RESULT
1170 spv_open (const char *filename, struct spv_reader **spvp)
1174 struct spv_reader *spv = xzalloc (sizeof *spv);
1175 ds_init_empty (&spv->zip_errs);
1176 spv->zip = zip_reader_create (filename, &spv->zip_errs);
1179 char *error = ds_steal_cstr (&spv->zip_errs);
1185 int detect = spv_detect__ (spv->zip, &error);
1189 return error ? error : xasprintf("%s: not an SPV file", filename);
1192 spv->root = xzalloc (sizeof *spv->root);
1193 spv->root->spv = spv;
1194 spv->root->type = SPV_ITEM_HEADING;
1195 for (size_t i = 0; ; i++)
1197 const char *member_name = zip_reader_get_member_name (spv->zip, i);
1201 struct substring member_name_ss = ss_cstr (member_name);
1202 if (ss_starts_with (member_name_ss, ss_cstr ("outputViewer"))
1203 && ss_ends_with (member_name_ss, ss_cstr (".xml")))
1205 char *error = spv_heading_read (spv, filename, member_name);
1219 spv_close (struct spv_reader *spv)
1223 ds_destroy (&spv->zip_errs);
1224 zip_reader_destroy (spv->zip);
1225 spv_item_destroy (spv->root);
1226 page_setup_destroy (spv->page_setup);
1232 spv_item_set_table_look (struct spv_item *item,
1233 const struct pivot_table_look *look)
1235 /* If this is a table, install the table look in it.
1237 (We can't just set item->table_look because light tables ignore it and
1238 legacy tables sometimes override it.) */
1239 if (spv_item_is_table (item))
1241 spv_item_load (item);
1242 pivot_table_set_look (item->table, look);
1245 for (size_t i = 0; i < item->n_children; i++)
1246 spv_item_set_table_look (item->children[i], look);
1249 char * WARN_UNUSED_RESULT
1250 spv_decode_fmt_spec (uint32_t u32, struct fmt_spec *out)
1253 || (u32 == 0x10000 || u32 == 1 /* both used as string formats */))
1255 *out = fmt_for_output (FMT_F, 40, 2);
1259 uint8_t raw_type = u32 >> 16;
1260 uint8_t w = u32 >> 8;
1264 *out = (struct fmt_spec) { .type = FMT_F, .w = w, .d = d };
1265 bool ok = raw_type >= 40 || fmt_from_io (raw_type, &out->type);
1268 fmt_fix_output (out);
1269 ok = fmt_check_width_compat (out, 0);
1275 *out = fmt_for_output (FMT_F, 40, 2);
1276 return xasprintf ("bad format %#"PRIx32, u32);