1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2007, 2009, 2010, 2011, 2012, 2013, 2016 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 "libpspp/message.h"
20 #include "libpspp/misc.h"
22 #include "gl/minmax.h"
23 #include "gl/c-strtod.h"
26 #define _(msgid) gettext (msgid)
27 #define N_(msgid) (msgid)
29 #include "spreadsheet-reader.h"
34 gnumeric_open_reader (const struct spreadsheet_read_options *opts, struct dictionary **dict)
36 msg (ME, _("Support for %s files was not compiled into this installation of PSPP"), "Gnumeric");
43 #include "data/gnumeric-reader.h"
48 #include <libxml/xmlreader.h>
51 #include "data/format.h"
52 #include "data/data-in.h"
53 #include "data/case.h"
54 #include "data/casereader-provider.h"
55 #include "data/dictionary.h"
56 #include "data/identifier.h"
57 #include "data/value.h"
58 #include "data/variable.h"
59 #include "libpspp/i18n.h"
60 #include "libpspp/str.h"
62 #include "gl/xalloc.h"
65 /* Shamelessly lifted from the Gnumeric sources:
66 https://git.gnome.org/browse/gnumeric/tree/src/value.h
72 VALUE_INTEGER = 30, /* Note, this was removed from gnumeric in 2006 - old versions may of
73 course still be around. New ones are supposed to use float.*/
83 static void gnm_file_casereader_destroy (struct casereader *, void *);
85 static struct ccase *gnm_file_casereader_read (struct casereader *, void *);
88 static const struct casereader_class gnm_file_casereader_class =
90 gnm_file_casereader_read,
91 gnm_file_casereader_destroy,
98 STATE_PRE_INIT = 0, /* Initial state */
99 STATE_SHEET_COUNT, /* Found the sheet index */
100 STATE_INIT , /* Other Initial state */
101 STATE_SHEET_START, /* Found the start of a sheet */
102 STATE_SHEET_NAME, /* Found the sheet name */
105 STATE_SHEET_FOUND, /* Found the sheet that we actually want */
106 STATE_CELLS_START, /* Found the start of the cell array */
107 STATE_CELL /* Found a cell */
112 /* The name of the sheet (utf8 encoding) */
126 /* The libxml reader for this instance */
127 xmlTextReaderPtr xtr;
129 /* An internal state variable */
130 enum reader_state state;
143 state_data_destroy (struct state_data *sd)
145 xmlFreeTextReader (sd->xtr);
149 struct gnumeric_reader
151 struct spreadsheet spreadsheet;
153 struct state_data rsd;
154 struct state_data msd;
161 struct sheet_detail *sheets;
163 const xmlChar *target_sheet;
164 int target_sheet_index;
166 struct caseproto *proto;
167 struct dictionary *dict;
168 struct ccase *first_case;
169 bool used_first_case;
171 enum gnm_value_type vtype;
176 gnumeric_unref (struct spreadsheet *s)
178 struct gnumeric_reader *r = (struct gnumeric_reader *) s;
180 if (0 == --s->ref_cnt)
184 for (i = 0; i < s->n_sheets; ++i)
186 xmlFree (r->sheets[i].name);
191 state_data_destroy (&r->msd);
193 dict_destroy (r->dict);
203 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
205 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
206 assert (n < s->n_sheets);
208 return gr->sheets[n].name;
212 static void process_node (struct gnumeric_reader *r, struct state_data *sd);
217 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
220 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
222 assert (n < s->n_sheets);
225 (gr->sheets[n].stop_col == -1)
227 (1 == (ret = xmlTextReaderRead (gr->msd.xtr)))
230 process_node (gr, &gr->msd);
233 return create_cell_range (
234 gr->sheets[n].start_col,
235 gr->sheets[n].start_row,
236 gr->sheets[n].stop_col,
237 gr->sheets[n].stop_row);
242 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
244 struct gnumeric_reader *r = r_;
249 state_data_destroy (&r->rsd);
251 if (r->first_case && ! r->used_first_case )
252 case_unref (r->first_case);
255 caseproto_unref (r->proto);
257 gnumeric_unref (&r->spreadsheet);
262 process_node (struct gnumeric_reader *r, struct state_data *sd)
264 xmlChar *name = xmlTextReaderName (sd->xtr);
266 name = xmlStrdup (_xml ("--"));
268 sd->node_type = xmlTextReaderNodeType (sd->xtr);
273 sd->current_sheet = -1;
274 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
275 XML_READER_TYPE_ELEMENT == sd->node_type)
277 sd->state = STATE_SHEET_COUNT;
281 case STATE_SHEET_COUNT:
282 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
283 XML_READER_TYPE_ELEMENT == sd->node_type)
286 if (sd->current_sheet + 1 > r->spreadsheet.n_sheets)
288 struct sheet_detail *detail ;
289 r->sheets = xrealloc (r->sheets, (sd->current_sheet + 1) * sizeof *r->sheets);
290 detail = &r->sheets[sd->current_sheet];
291 detail->start_col = detail->stop_col = detail->start_row = detail->stop_row = -1;
293 r->spreadsheet.n_sheets = sd->current_sheet + 1;
296 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
297 XML_READER_TYPE_END_ELEMENT == sd->node_type)
299 sd->state = STATE_INIT;
300 sd->current_sheet = -1;
302 else if (XML_READER_TYPE_TEXT == sd->node_type)
304 if ( r->sheets [r->spreadsheet.n_sheets - 1].name == NULL)
305 r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
310 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
311 XML_READER_TYPE_ELEMENT == sd->node_type)
314 sd->state = STATE_SHEET_START;
317 case STATE_SHEET_START:
318 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
319 XML_READER_TYPE_ELEMENT == sd->node_type)
321 sd->state = STATE_SHEET_NAME;
324 case STATE_SHEET_NAME:
325 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
326 XML_READER_TYPE_END_ELEMENT == sd->node_type)
328 sd->state = STATE_INIT;
330 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
331 XML_READER_TYPE_END_ELEMENT == sd->node_type)
333 sd->state = STATE_INIT;
335 else if (XML_READER_TYPE_TEXT == sd->node_type)
337 if ( r->target_sheet != NULL)
339 xmlChar *value = xmlTextReaderValue (sd->xtr);
340 if ( 0 == xmlStrcmp (value, r->target_sheet))
341 sd->state = STATE_SHEET_FOUND;
344 else if (r->target_sheet_index == sd->current_sheet + 1)
346 sd->state = STATE_SHEET_FOUND;
348 else if (r->target_sheet_index == -1)
350 sd->state = STATE_SHEET_FOUND;
354 case STATE_SHEET_FOUND:
355 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
356 XML_READER_TYPE_ELEMENT == sd->node_type)
358 sd->min_col = INT_MAX;
359 if (! xmlTextReaderIsEmptyElement (sd->xtr))
360 sd->state = STATE_CELLS_START;
362 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
363 XML_READER_TYPE_ELEMENT == sd->node_type)
365 sd->state = STATE_MAXROW;
367 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
368 XML_READER_TYPE_ELEMENT == sd->node_type)
370 sd->state = STATE_MAXCOL;
372 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
373 XML_READER_TYPE_END_ELEMENT == sd->node_type)
375 sd->state = STATE_INIT;
379 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
380 XML_READER_TYPE_END_ELEMENT == sd->node_type)
382 sd->state = STATE_SHEET_FOUND;
384 else if (sd->node_type == XML_READER_TYPE_TEXT)
386 xmlChar *value = xmlTextReaderValue (sd->xtr);
387 r->sheets[sd->current_sheet].maxrow = _xmlchar_to_int (value);
392 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
393 XML_READER_TYPE_END_ELEMENT == sd->node_type)
395 sd->state = STATE_SHEET_FOUND;
397 else if (sd->node_type == XML_READER_TYPE_TEXT)
399 xmlChar *value = xmlTextReaderValue (sd->xtr);
400 r->sheets[sd->current_sheet].maxcol = _xmlchar_to_int (value);
404 case STATE_CELLS_START:
405 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
406 XML_READER_TYPE_ELEMENT == sd->node_type)
408 xmlChar *attr = NULL;
410 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Col"));
411 sd->col = _xmlchar_to_int (attr);
414 if (sd->col < sd->min_col)
415 sd->min_col = sd->col;
417 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Row"));
418 sd->row = _xmlchar_to_int (attr);
421 if (r->sheets[sd->current_sheet].start_row == -1)
423 r->sheets[sd->current_sheet].start_row = sd->row;
426 if (r->sheets[sd->current_sheet].start_col == -1)
428 r->sheets[sd->current_sheet].start_col = sd->col;
430 if (! xmlTextReaderIsEmptyElement (sd->xtr))
431 sd->state = STATE_CELL;
433 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == sd->node_type) )
435 r->sheets[sd->current_sheet].stop_col = sd->col;
436 r->sheets[sd->current_sheet].stop_row = sd->row;
437 sd->state = STATE_SHEET_NAME;
441 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == sd->node_type)
443 sd->state = STATE_CELLS_START;
455 Sets the VAR of case C, to the value corresponding to the xml string XV
458 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
459 const xmlChar *xv, enum gnm_value_type type, int col, int row)
461 union value *v = case_data_rw (c, var);
464 value_set_missing (v, var_get_width (var));
465 else if ( var_is_alpha (var))
466 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
467 else if (type == VALUE_FLOAT || type == VALUE_INTEGER)
469 const char *text = CHAR_CAST (const char *, xv);
473 v->f = c_strtod (text, &endptr);
474 if ( errno != 0 || endptr == text)
479 const char *text = CHAR_CAST (const char *, xv);
481 const struct fmt_spec *fmt = var_get_write_format (var);
483 char *m = data_in (ss_cstr (text), "UTF-8",
491 char buf [FMT_STRING_LEN_MAX + 1];
492 char *cell = create_cell_ref (col, row);
494 msg (MW, _("Cannot convert the value in the spreadsheet cell %s to format (%s): %s"),
495 cell, fmt_to_string (fmt, buf), m);
506 xmlChar *first_value;
512 gnumeric_error_handler (void *ctx, const char *mesg,
513 UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
515 struct gnumeric_reader *r = ctx;
517 msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
519 r->spreadsheet.file_name,
520 xmlTextReaderLocatorLineNumber (loc),
524 static struct gnumeric_reader *
525 gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
528 struct state_data *sd;
530 xmlTextReaderPtr xtr;
533 assert (r == NULL || filename == NULL);
537 gz = gzopen (filename, "r");
541 gz = gzopen (r->spreadsheet.file_name, "r");
548 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
549 (xmlInputCloseCallback) gzclose, gz,
551 show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
561 r = xzalloc (sizeof *r);
562 r->spreadsheet.n_sheets = -1;
563 r->spreadsheet.file_name = strdup (filename);
572 xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
574 r->target_sheet = NULL;
575 r->target_sheet_index = -1;
577 sd->row = sd->col = -1;
578 sd->state = STATE_PRE_INIT;
580 r->spreadsheet.ref_cnt++;
583 /* Advance to the start of the workbook.
584 This gives us some confidence that we are actually dealing with a gnumeric
587 while ( (sd->state != STATE_INIT )
588 && 1 == (ret = xmlTextReaderRead (sd->xtr)))
590 process_node (r, sd);
596 /* Does not seem to be a gnumeric file */
597 gnumeric_unref (&r->spreadsheet);
601 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
605 const xmlChar *enc = xmlTextReaderConstEncoding (sd->xtr);
606 xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc));
608 if ( XML_CHAR_ENCODING_UTF8 != xce)
610 /* I have been told that ALL gnumeric files are UTF8 encoded. If that is correct, this
612 msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. "
613 "Any non-ascii characters will be incorrectly imported."),
614 r->spreadsheet.file_name,
624 gnumeric_probe (const char *filename, bool report_errors)
626 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
628 return &r->spreadsheet;
633 gnumeric_make_reader (struct spreadsheet *spreadsheet,
634 const struct spreadsheet_read_options *opts)
638 struct gnumeric_reader *r = NULL;
639 unsigned long int vstart = 0;
641 casenumber n_cases = CASENUMBER_MAX;
643 struct var_spec *var_spec = NULL;
646 r = (struct gnumeric_reader *) (spreadsheet);
648 r = gnumeric_reopen (r, NULL, true);
650 if ( opts->cell_range )
652 if ( ! convert_cell_ref (opts->cell_range,
653 &r->start_col, &r->start_row,
654 &r->stop_col, &r->stop_row))
656 msg (SE, _("Invalid cell range `%s'"),
669 r->target_sheet = BAD_CAST opts->sheet_name;
670 r->target_sheet_index = opts->sheet_index;
671 r->rsd.row = r->rsd.col = -1;
672 r->rsd.current_sheet = -1;
673 r->first_case = NULL;
676 /* Advance to the start of the cells for the target sheet */
677 while ( (r->rsd.state != STATE_CELL || r->rsd.row < r->start_row )
678 && 1 == (ret = xmlTextReaderRead (r->rsd.xtr)))
681 process_node (r, &r->rsd);
682 value = xmlTextReaderValue (r->rsd.xtr);
684 if ( r->rsd.state == STATE_MAXROW && r->rsd.node_type == XML_READER_TYPE_TEXT)
686 n_cases = 1 + _xmlchar_to_int (value) ;
691 /* If a range has been given, then use that to calculate the number
693 if ( opts->cell_range)
695 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
698 if ( opts->read_names )
705 /* Read in the first row of cells,
706 including the headers if read_names was set */
708 (( r->rsd.state == STATE_CELLS_START && r->rsd.row <= r->start_row) || r->rsd.state == STATE_CELL )
709 && (ret = xmlTextReaderRead (r->rsd.xtr))
714 if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_TEXT)
717 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
719 type = _xmlchar_to_int (attr);
724 process_node (r, &r->rsd);
726 if ( r->rsd.row > r->start_row )
729 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
731 r->vtype = _xmlchar_to_int (attr);
737 if ( r->rsd.col < r->start_col ||
738 (r->stop_col != -1 && r->rsd.col > r->stop_col))
741 idx = r->rsd.col - r->start_col;
743 if ( idx >= n_var_specs )
746 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
747 for (i = n_var_specs; i <= idx; ++i)
749 var_spec [i].name = NULL;
750 var_spec [i].width = -1;
751 var_spec [i].first_value = NULL;
752 var_spec [i].first_type = -1;
754 n_var_specs = idx + 1 ;
757 var_spec [idx].first_type = type;
759 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
761 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
762 const char *text = CHAR_CAST (const char *, value);
764 if ( r->rsd.row < r->start_row)
766 if ( opts->read_names )
768 var_spec [idx].name = xstrdup (text);
773 var_spec [idx].first_value = xmlStrdup (value);
775 if (-1 == var_spec [idx].width )
776 var_spec [idx].width = (opts->asw == -1) ?
777 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw;
782 else if ( r->rsd.node_type == XML_READER_TYPE_ELEMENT
783 && r->rsd.state == STATE_CELL)
785 if ( r->rsd.row == r->start_row )
788 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
790 if ( NULL == attr || VALUE_STRING != _xmlchar_to_int (attr))
791 var_spec [idx].width = 0;
799 const xmlChar *enc = xmlTextReaderConstEncoding (r->rsd.xtr);
802 /* Create the dictionary and populate it */
803 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
806 for (i = 0 ; i < n_var_specs ; ++i )
810 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
813 /* Probably no data exists for this variable, so allocate a
815 if ( var_spec[i].width == -1 )
816 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
818 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
819 dict_create_var (r->dict, name, var_spec[i].width);
823 /* Create the first case, and cache it */
824 r->used_first_case = false;
826 if ( n_var_specs == 0 )
828 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
829 spreadsheet->file_name);
833 r->proto = caseproto_ref (dict_get_proto (r->dict));
834 r->first_case = case_create (r->proto);
835 case_set_missing (r->first_case);
838 for ( i = 0 ; i < n_var_specs ; ++i )
840 const struct variable *var;
842 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
845 var = dict_get_var (r->dict, x++);
847 convert_xml_string_to_value (r->first_case, var,
848 var_spec[i].first_value,
849 var_spec[i].first_type,
854 for ( i = 0 ; i < n_var_specs ; ++i )
856 free (var_spec[i].first_value);
857 free (var_spec[i].name);
863 return casereader_create_sequential
867 &gnm_file_casereader_class, r);
871 for ( i = 0 ; i < n_var_specs ; ++i )
873 free (var_spec[i].first_value);
874 free (var_spec[i].name);
879 gnm_file_casereader_destroy (NULL, r);
885 /* Reads and returns one case from READER's file. Returns a null
886 pointer on failure. */
887 static struct ccase *
888 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
893 struct gnumeric_reader *r = r_;
894 int current_row = r->rsd.row;
896 if ( !r->used_first_case )
898 r->used_first_case = true;
899 return r->first_case;
902 c = case_create (r->proto);
903 case_set_missing (c);
905 if (r->start_col == -1)
906 r->start_col = r->rsd.min_col;
909 while ((r->rsd.state == STATE_CELL || r->rsd.state == STATE_CELLS_START )
910 && r->rsd.row == current_row && (ret = xmlTextReaderRead (r->rsd.xtr)))
912 process_node (r, &r->rsd);
914 if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_ELEMENT)
917 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
919 r->vtype = _xmlchar_to_int (attr);
924 if ( r->rsd.col < r->start_col || (r->stop_col != -1 &&
925 r->rsd.col > r->stop_col))
928 if ( r->rsd.col - r->start_col >= caseproto_get_n_widths (r->proto))
931 if ( r->stop_row != -1 && r->rsd.row > r->stop_row)
935 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
937 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
938 const int idx = r->rsd.col - r->start_col;
939 const struct variable *var = dict_get_var (r->dict, idx);
941 convert_xml_string_to_value (c, var, value, r->vtype,
942 r->rsd.col, r->rsd.row);
958 #endif /* GNM_READ_SUPPORT */