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 "data/gnumeric-reader.h"
20 #include "spreadsheet-reader.h"
25 #include <libxml/xmlreader.h>
28 #include "data/case.h"
29 #include "data/casereader-provider.h"
30 #include "data/data-in.h"
31 #include "data/dictionary.h"
32 #include "data/format.h"
33 #include "data/identifier.h"
34 #include "data/value.h"
35 #include "data/variable.h"
36 #include "libpspp/i18n.h"
37 #include "libpspp/message.h"
38 #include "libpspp/misc.h"
39 #include "libpspp/str.h"
41 #include "gl/c-strtod.h"
42 #include "gl/minmax.h"
43 #include "gl/xalloc.h"
46 #define _(msgid) gettext (msgid)
47 #define N_(msgid) (msgid)
49 /* Shamelessly lifted from the Gnumeric sources:
50 https://git.gnome.org/browse/gnumeric/tree/src/value.h
56 VALUE_INTEGER = 30, /* Note, this was removed from gnumeric in 2006 - old versions may of
57 course still be around. New ones are supposed to use float.*/
67 static void gnm_file_casereader_destroy (struct casereader *, void *);
69 static struct ccase *gnm_file_casereader_read (struct casereader *, void *);
72 static const struct casereader_class gnm_file_casereader_class =
74 gnm_file_casereader_read,
75 gnm_file_casereader_destroy,
82 STATE_PRE_INIT = 0, /* Initial state */
83 STATE_SHEET_COUNT, /* Found the sheet index */
84 STATE_INIT , /* Other Initial state */
85 STATE_SHEET_START, /* Found the start of a sheet */
86 STATE_SHEET_NAME, /* Found the sheet name */
89 STATE_SHEET_FOUND, /* Found the sheet that we actually want */
90 STATE_CELLS_START, /* Found the start of the cell array */
91 STATE_CELL /* Found a cell */
96 /* The name of the sheet (utf8 encoding) */
110 /* The libxml reader for this instance */
111 xmlTextReaderPtr xtr;
113 /* An internal state variable */
114 enum reader_state state;
127 state_data_destroy (struct state_data *sd)
129 xmlFreeTextReader (sd->xtr);
133 struct gnumeric_reader
135 struct spreadsheet spreadsheet;
137 struct state_data rsd;
138 struct state_data msd;
145 struct sheet_detail *sheets;
147 const xmlChar *target_sheet;
148 int target_sheet_index;
150 struct caseproto *proto;
151 struct dictionary *dict;
152 struct ccase *first_case;
153 bool used_first_case;
155 enum gnm_value_type vtype;
160 gnumeric_unref (struct spreadsheet *s)
162 struct gnumeric_reader *r = (struct gnumeric_reader *) s;
164 if (0 == --s->ref_cnt)
168 for (i = 0; i < s->n_sheets; ++i)
170 xmlFree (r->sheets[i].name);
175 state_data_destroy (&r->msd);
177 dict_unref (r->dict);
187 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
189 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
190 assert (n < s->n_sheets);
192 return gr->sheets[n].name;
196 static void process_node (struct gnumeric_reader *r, struct state_data *sd);
201 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
204 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
206 assert (n < s->n_sheets);
209 (gr->sheets[n].stop_col == -1)
211 (1 == (ret = xmlTextReaderRead (gr->msd.xtr)))
214 process_node (gr, &gr->msd);
217 return create_cell_range (
218 gr->sheets[n].start_col,
219 gr->sheets[n].start_row,
220 gr->sheets[n].stop_col,
221 gr->sheets[n].stop_row);
226 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
228 struct gnumeric_reader *r = r_;
233 state_data_destroy (&r->rsd);
235 if (r->first_case && ! r->used_first_case )
236 case_unref (r->first_case);
239 caseproto_unref (r->proto);
241 gnumeric_unref (&r->spreadsheet);
246 process_node (struct gnumeric_reader *r, struct state_data *sd)
248 xmlChar *name = xmlTextReaderName (sd->xtr);
250 name = xmlStrdup (_xml ("--"));
252 sd->node_type = xmlTextReaderNodeType (sd->xtr);
257 sd->current_sheet = -1;
258 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
259 XML_READER_TYPE_ELEMENT == sd->node_type)
261 sd->state = STATE_SHEET_COUNT;
265 case STATE_SHEET_COUNT:
266 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
267 XML_READER_TYPE_ELEMENT == sd->node_type)
270 if (sd->current_sheet + 1 > r->spreadsheet.n_sheets)
272 struct sheet_detail *detail ;
273 r->sheets = xrealloc (r->sheets, (sd->current_sheet + 1) * sizeof *r->sheets);
274 detail = &r->sheets[sd->current_sheet];
275 detail->start_col = detail->stop_col = detail->start_row = detail->stop_row = -1;
277 r->spreadsheet.n_sheets = sd->current_sheet + 1;
280 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
281 XML_READER_TYPE_END_ELEMENT == sd->node_type)
283 sd->state = STATE_INIT;
284 sd->current_sheet = -1;
286 else if (XML_READER_TYPE_TEXT == sd->node_type)
288 if ( r->sheets [r->spreadsheet.n_sheets - 1].name == NULL)
289 r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
294 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
295 XML_READER_TYPE_ELEMENT == sd->node_type)
298 sd->state = STATE_SHEET_START;
301 case STATE_SHEET_START:
302 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
303 XML_READER_TYPE_ELEMENT == sd->node_type)
305 sd->state = STATE_SHEET_NAME;
308 case STATE_SHEET_NAME:
309 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
310 XML_READER_TYPE_END_ELEMENT == sd->node_type)
312 sd->state = STATE_INIT;
314 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
315 XML_READER_TYPE_END_ELEMENT == sd->node_type)
317 sd->state = STATE_INIT;
319 else if (XML_READER_TYPE_TEXT == sd->node_type)
321 if ( r->target_sheet != NULL)
323 xmlChar *value = xmlTextReaderValue (sd->xtr);
324 if ( 0 == xmlStrcmp (value, r->target_sheet))
325 sd->state = STATE_SHEET_FOUND;
328 else if (r->target_sheet_index == sd->current_sheet + 1)
330 sd->state = STATE_SHEET_FOUND;
332 else if (r->target_sheet_index == -1)
334 sd->state = STATE_SHEET_FOUND;
338 case STATE_SHEET_FOUND:
339 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
340 XML_READER_TYPE_ELEMENT == sd->node_type)
342 sd->min_col = INT_MAX;
343 if (! xmlTextReaderIsEmptyElement (sd->xtr))
344 sd->state = STATE_CELLS_START;
346 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
347 XML_READER_TYPE_ELEMENT == sd->node_type)
349 sd->state = STATE_MAXROW;
351 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
352 XML_READER_TYPE_ELEMENT == sd->node_type)
354 sd->state = STATE_MAXCOL;
356 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
357 XML_READER_TYPE_END_ELEMENT == sd->node_type)
359 sd->state = STATE_INIT;
363 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
364 XML_READER_TYPE_END_ELEMENT == sd->node_type)
366 sd->state = STATE_SHEET_FOUND;
368 else if (sd->node_type == XML_READER_TYPE_TEXT)
370 xmlChar *value = xmlTextReaderValue (sd->xtr);
371 r->sheets[sd->current_sheet].maxrow = _xmlchar_to_int (value);
376 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
377 XML_READER_TYPE_END_ELEMENT == sd->node_type)
379 sd->state = STATE_SHEET_FOUND;
381 else if (sd->node_type == XML_READER_TYPE_TEXT)
383 xmlChar *value = xmlTextReaderValue (sd->xtr);
384 r->sheets[sd->current_sheet].maxcol = _xmlchar_to_int (value);
388 case STATE_CELLS_START:
389 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
390 XML_READER_TYPE_ELEMENT == sd->node_type)
392 xmlChar *attr = NULL;
394 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Col"));
395 sd->col = _xmlchar_to_int (attr);
398 if (sd->col < sd->min_col)
399 sd->min_col = sd->col;
401 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Row"));
402 sd->row = _xmlchar_to_int (attr);
405 if (r->sheets[sd->current_sheet].start_row == -1)
407 r->sheets[sd->current_sheet].start_row = sd->row;
410 if (r->sheets[sd->current_sheet].start_col == -1)
412 r->sheets[sd->current_sheet].start_col = sd->col;
414 if (! xmlTextReaderIsEmptyElement (sd->xtr))
415 sd->state = STATE_CELL;
417 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == sd->node_type) )
419 r->sheets[sd->current_sheet].stop_col = sd->col;
420 r->sheets[sd->current_sheet].stop_row = sd->row;
421 sd->state = STATE_SHEET_NAME;
425 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == sd->node_type)
427 sd->state = STATE_CELLS_START;
439 Sets the VAR of case C, to the value corresponding to the xml string XV
442 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
443 const xmlChar *xv, enum gnm_value_type type, int col, int row)
445 union value *v = case_data_rw (c, var);
448 value_set_missing (v, var_get_width (var));
449 else if ( var_is_alpha (var))
450 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
451 else if (type == VALUE_FLOAT || type == VALUE_INTEGER)
453 const char *text = CHAR_CAST (const char *, xv);
457 v->f = c_strtod (text, &endptr);
458 if ( errno != 0 || endptr == text)
463 const char *text = CHAR_CAST (const char *, xv);
465 const struct fmt_spec *fmt = var_get_write_format (var);
467 char *m = data_in (ss_cstr (text), "UTF-8",
475 char buf [FMT_STRING_LEN_MAX + 1];
476 char *cell = create_cell_ref (col, row);
478 msg (MW, _("Cannot convert the value in the spreadsheet cell %s to format (%s): %s"),
479 cell, fmt_to_string (fmt, buf), m);
490 xmlChar *first_value;
496 gnumeric_error_handler (void *ctx, const char *mesg,
497 xmlParserSeverities sev UNUSED,
498 xmlTextReaderLocatorPtr loc)
500 struct gnumeric_reader *r = ctx;
502 msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
504 r->spreadsheet.file_name,
505 xmlTextReaderLocatorLineNumber (loc),
509 static struct gnumeric_reader *
510 gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
513 struct state_data *sd;
515 xmlTextReaderPtr xtr;
518 assert (r == NULL || filename == NULL);
522 gz = gzopen (filename, "r");
526 gz = gzopen (r->spreadsheet.file_name, "r");
533 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
534 (xmlInputCloseCallback) gzclose, gz,
536 show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
546 r = xzalloc (sizeof *r);
547 r->spreadsheet.n_sheets = -1;
548 r->spreadsheet.file_name = strdup (filename);
557 xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
559 r->target_sheet = NULL;
560 r->target_sheet_index = -1;
562 sd->row = sd->col = -1;
563 sd->state = STATE_PRE_INIT;
565 r->spreadsheet.ref_cnt++;
568 /* Advance to the start of the workbook.
569 This gives us some confidence that we are actually dealing with a gnumeric
572 while ( (sd->state != STATE_INIT )
573 && 1 == (ret = xmlTextReaderRead (sd->xtr)))
575 process_node (r, sd);
581 /* Does not seem to be a gnumeric file */
582 gnumeric_unref (&r->spreadsheet);
586 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
590 const xmlChar *enc = xmlTextReaderConstEncoding (sd->xtr);
591 xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc));
593 if ( XML_CHAR_ENCODING_UTF8 != xce)
595 /* I have been told that ALL gnumeric files are UTF8 encoded. If that is correct, this
597 msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. "
598 "Any non-ascii characters will be incorrectly imported."),
599 r->spreadsheet.file_name,
609 gnumeric_probe (const char *filename, bool report_errors)
611 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
613 return &r->spreadsheet;
618 gnumeric_make_reader (struct spreadsheet *spreadsheet,
619 const struct spreadsheet_read_options *opts)
623 struct gnumeric_reader *r = NULL;
624 unsigned long int vstart = 0;
626 casenumber n_cases = CASENUMBER_MAX;
628 struct var_spec *var_spec = NULL;
631 r = (struct gnumeric_reader *) (spreadsheet);
633 r = gnumeric_reopen (r, NULL, true);
635 if ( opts->cell_range )
637 if ( ! convert_cell_ref (opts->cell_range,
638 &r->start_col, &r->start_row,
639 &r->stop_col, &r->stop_row))
641 msg (SE, _("Invalid cell range `%s'"),
654 r->target_sheet = BAD_CAST opts->sheet_name;
655 r->target_sheet_index = opts->sheet_index;
656 r->rsd.row = r->rsd.col = -1;
657 r->rsd.current_sheet = -1;
658 r->first_case = NULL;
661 /* Advance to the start of the cells for the target sheet */
662 while ( (r->rsd.state != STATE_CELL || r->rsd.row < r->start_row )
663 && 1 == (ret = xmlTextReaderRead (r->rsd.xtr)))
666 process_node (r, &r->rsd);
667 value = xmlTextReaderValue (r->rsd.xtr);
669 if ( r->rsd.state == STATE_MAXROW && r->rsd.node_type == XML_READER_TYPE_TEXT)
671 n_cases = 1 + _xmlchar_to_int (value) ;
676 /* If a range has been given, then use that to calculate the number
678 if ( opts->cell_range)
680 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
683 if ( opts->read_names )
690 /* Read in the first row of cells,
691 including the headers if read_names was set */
693 (( r->rsd.state == STATE_CELLS_START && r->rsd.row <= r->start_row) || r->rsd.state == STATE_CELL )
694 && (ret = xmlTextReaderRead (r->rsd.xtr))
699 if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_TEXT)
702 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
704 type = _xmlchar_to_int (attr);
709 process_node (r, &r->rsd);
711 if ( r->rsd.row > r->start_row )
714 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
716 r->vtype = _xmlchar_to_int (attr);
722 if ( r->rsd.col < r->start_col ||
723 (r->stop_col != -1 && r->rsd.col > r->stop_col))
726 idx = r->rsd.col - r->start_col;
728 if ( idx >= n_var_specs )
731 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
732 for (i = n_var_specs; i <= idx; ++i)
734 var_spec [i].name = NULL;
735 var_spec [i].width = -1;
736 var_spec [i].first_value = NULL;
737 var_spec [i].first_type = -1;
739 n_var_specs = idx + 1 ;
742 var_spec [idx].first_type = type;
744 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
746 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
747 const char *text = CHAR_CAST (const char *, value);
749 if ( r->rsd.row < r->start_row)
751 if ( opts->read_names )
753 var_spec [idx].name = xstrdup (text);
758 var_spec [idx].first_value = xmlStrdup (value);
760 if (-1 == var_spec [idx].width )
761 var_spec [idx].width = (opts->asw == -1) ?
762 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw;
767 else if ( r->rsd.node_type == XML_READER_TYPE_ELEMENT
768 && r->rsd.state == STATE_CELL)
770 if ( r->rsd.row == r->start_row )
773 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
775 if ( NULL == attr || VALUE_STRING != _xmlchar_to_int (attr))
776 var_spec [idx].width = 0;
784 const xmlChar *enc = xmlTextReaderConstEncoding (r->rsd.xtr);
787 /* Create the dictionary and populate it */
788 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
791 for (i = 0 ; i < n_var_specs ; ++i )
795 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
798 /* Probably no data exists for this variable, so allocate a
800 if ( var_spec[i].width == -1 )
801 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
803 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
804 dict_create_var (r->dict, name, var_spec[i].width);
808 /* Create the first case, and cache it */
809 r->used_first_case = false;
811 if ( n_var_specs == 0 )
813 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
814 spreadsheet->file_name);
818 r->proto = caseproto_ref (dict_get_proto (r->dict));
819 r->first_case = case_create (r->proto);
820 case_set_missing (r->first_case);
823 for ( i = 0 ; i < n_var_specs ; ++i )
825 const struct variable *var;
827 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
830 var = dict_get_var (r->dict, x++);
832 convert_xml_string_to_value (r->first_case, var,
833 var_spec[i].first_value,
834 var_spec[i].first_type,
839 for ( i = 0 ; i < n_var_specs ; ++i )
841 free (var_spec[i].first_value);
842 free (var_spec[i].name);
848 return casereader_create_sequential
852 &gnm_file_casereader_class, r);
856 for ( i = 0 ; i < n_var_specs ; ++i )
858 free (var_spec[i].first_value);
859 free (var_spec[i].name);
864 gnm_file_casereader_destroy (NULL, r);
870 /* Reads and returns one case from READER's file. Returns a null
871 pointer on failure. */
872 static struct ccase *
873 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
878 struct gnumeric_reader *r = r_;
879 int current_row = r->rsd.row;
881 if ( !r->used_first_case )
883 r->used_first_case = true;
884 return r->first_case;
887 c = case_create (r->proto);
888 case_set_missing (c);
890 if (r->start_col == -1)
891 r->start_col = r->rsd.min_col;
894 while ((r->rsd.state == STATE_CELL || r->rsd.state == STATE_CELLS_START )
895 && r->rsd.row == current_row && (ret = xmlTextReaderRead (r->rsd.xtr)))
897 process_node (r, &r->rsd);
899 if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_ELEMENT)
902 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
904 r->vtype = _xmlchar_to_int (attr);
909 if ( r->rsd.col < r->start_col || (r->stop_col != -1 &&
910 r->rsd.col > r->stop_col))
913 if ( r->rsd.col - r->start_col >= caseproto_get_n_widths (r->proto))
916 if ( r->stop_row != -1 && r->rsd.row > r->stop_row)
920 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
922 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
923 const int idx = r->rsd.col - r->start_col;
924 const struct variable *var = dict_get_var (r->dict, idx);
926 convert_xml_string_to_value (c, var, value, r->vtype,
927 r->rsd.col, r->rsd.row);