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");
42 gnumeric_make_reader (struct spreadsheet *spreadsheet,
43 const struct spreadsheet_read_options *opts)
49 gnumeric_unref (struct spreadsheet *r)
56 #include "data/gnumeric-reader.h"
61 #include <libxml/xmlreader.h>
64 #include "data/format.h"
65 #include "data/data-in.h"
66 #include "data/case.h"
67 #include "data/casereader-provider.h"
68 #include "data/dictionary.h"
69 #include "data/identifier.h"
70 #include "data/value.h"
71 #include "data/variable.h"
72 #include "libpspp/i18n.h"
73 #include "libpspp/str.h"
75 #include "gl/xalloc.h"
78 /* Shamelessly lifted from the Gnumeric sources:
79 https://git.gnome.org/browse/gnumeric/tree/src/value.h
85 VALUE_INTEGER = 30, /* Note, this was removed from gnumeric in 2006 - old versions may of
86 course still be around. New ones are supposed to use float.*/
96 static void gnm_file_casereader_destroy (struct casereader *, void *);
98 static struct ccase *gnm_file_casereader_read (struct casereader *, void *);
101 static const struct casereader_class gnm_file_casereader_class =
103 gnm_file_casereader_read,
104 gnm_file_casereader_destroy,
111 STATE_PRE_INIT = 0, /* Initial state */
112 STATE_SHEET_COUNT, /* Found the sheet index */
113 STATE_INIT , /* Other Initial state */
114 STATE_SHEET_START, /* Found the start of a sheet */
115 STATE_SHEET_NAME, /* Found the sheet name */
118 STATE_SHEET_FOUND, /* Found the sheet that we actually want */
119 STATE_CELLS_START, /* Found the start of the cell array */
120 STATE_CELL /* Found a cell */
125 /* The name of the sheet (utf8 encoding) */
139 /* The libxml reader for this instance */
140 xmlTextReaderPtr xtr;
142 /* An internal state variable */
143 enum reader_state state;
156 state_data_destroy (struct state_data *sd)
158 xmlFreeTextReader (sd->xtr);
162 struct gnumeric_reader
164 struct spreadsheet spreadsheet;
166 struct state_data rsd;
167 struct state_data msd;
174 struct sheet_detail *sheets;
176 const xmlChar *target_sheet;
177 int target_sheet_index;
179 struct caseproto *proto;
180 struct dictionary *dict;
181 struct ccase *first_case;
182 bool used_first_case;
184 enum gnm_value_type vtype;
189 gnumeric_unref (struct spreadsheet *s)
191 struct gnumeric_reader *r = (struct gnumeric_reader *) s;
193 if (0 == --s->ref_cnt)
197 for (i = 0; i < s->n_sheets; ++i)
199 xmlFree (r->sheets[i].name);
204 state_data_destroy (&r->msd);
206 dict_destroy (r->dict);
216 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
218 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
219 assert (n < s->n_sheets);
221 return gr->sheets[n].name;
225 static void process_node (struct gnumeric_reader *r, struct state_data *sd);
230 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
233 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
235 assert (n < s->n_sheets);
238 (gr->sheets[n].stop_col == -1)
240 (1 == (ret = xmlTextReaderRead (gr->msd.xtr)))
243 process_node (gr, &gr->msd);
246 return create_cell_range (
247 gr->sheets[n].start_col,
248 gr->sheets[n].start_row,
249 gr->sheets[n].stop_col,
250 gr->sheets[n].stop_row);
255 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
257 struct gnumeric_reader *r = r_;
262 state_data_destroy (&r->rsd);
264 if (r->first_case && ! r->used_first_case )
265 case_unref (r->first_case);
268 caseproto_unref (r->proto);
270 gnumeric_unref (&r->spreadsheet);
275 process_node (struct gnumeric_reader *r, struct state_data *sd)
277 xmlChar *name = xmlTextReaderName (sd->xtr);
279 name = xmlStrdup (_xml ("--"));
281 sd->node_type = xmlTextReaderNodeType (sd->xtr);
286 sd->current_sheet = -1;
287 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
288 XML_READER_TYPE_ELEMENT == sd->node_type)
290 sd->state = STATE_SHEET_COUNT;
294 case STATE_SHEET_COUNT:
295 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
296 XML_READER_TYPE_ELEMENT == sd->node_type)
299 if (sd->current_sheet + 1 > r->spreadsheet.n_sheets)
301 struct sheet_detail *detail ;
302 r->sheets = xrealloc (r->sheets, (sd->current_sheet + 1) * sizeof *r->sheets);
303 detail = &r->sheets[sd->current_sheet];
304 detail->start_col = detail->stop_col = detail->start_row = detail->stop_row = -1;
306 r->spreadsheet.n_sheets = sd->current_sheet + 1;
309 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
310 XML_READER_TYPE_END_ELEMENT == sd->node_type)
312 sd->state = STATE_INIT;
313 sd->current_sheet = -1;
315 else if (XML_READER_TYPE_TEXT == sd->node_type)
317 if ( r->sheets [r->spreadsheet.n_sheets - 1].name == NULL)
318 r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
323 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
324 XML_READER_TYPE_ELEMENT == sd->node_type)
327 sd->state = STATE_SHEET_START;
330 case STATE_SHEET_START:
331 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
332 XML_READER_TYPE_ELEMENT == sd->node_type)
334 sd->state = STATE_SHEET_NAME;
337 case STATE_SHEET_NAME:
338 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
339 XML_READER_TYPE_END_ELEMENT == sd->node_type)
341 sd->state = STATE_INIT;
343 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
344 XML_READER_TYPE_END_ELEMENT == sd->node_type)
346 sd->state = STATE_INIT;
348 else if (XML_READER_TYPE_TEXT == sd->node_type)
350 if ( r->target_sheet != NULL)
352 xmlChar *value = xmlTextReaderValue (sd->xtr);
353 if ( 0 == xmlStrcmp (value, r->target_sheet))
354 sd->state = STATE_SHEET_FOUND;
357 else if (r->target_sheet_index == sd->current_sheet + 1)
359 sd->state = STATE_SHEET_FOUND;
361 else if (r->target_sheet_index == -1)
363 sd->state = STATE_SHEET_FOUND;
367 case STATE_SHEET_FOUND:
368 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
369 XML_READER_TYPE_ELEMENT == sd->node_type)
371 sd->min_col = INT_MAX;
372 if (! xmlTextReaderIsEmptyElement (sd->xtr))
373 sd->state = STATE_CELLS_START;
375 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
376 XML_READER_TYPE_ELEMENT == sd->node_type)
378 sd->state = STATE_MAXROW;
380 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
381 XML_READER_TYPE_ELEMENT == sd->node_type)
383 sd->state = STATE_MAXCOL;
385 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
386 XML_READER_TYPE_END_ELEMENT == sd->node_type)
388 sd->state = STATE_INIT;
392 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
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].maxrow = _xmlchar_to_int (value);
405 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
406 XML_READER_TYPE_END_ELEMENT == sd->node_type)
408 sd->state = STATE_SHEET_FOUND;
410 else if (sd->node_type == XML_READER_TYPE_TEXT)
412 xmlChar *value = xmlTextReaderValue (sd->xtr);
413 r->sheets[sd->current_sheet].maxcol = _xmlchar_to_int (value);
417 case STATE_CELLS_START:
418 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
419 XML_READER_TYPE_ELEMENT == sd->node_type)
421 xmlChar *attr = NULL;
423 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Col"));
424 sd->col = _xmlchar_to_int (attr);
427 if (sd->col < sd->min_col)
428 sd->min_col = sd->col;
430 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Row"));
431 sd->row = _xmlchar_to_int (attr);
434 if (r->sheets[sd->current_sheet].start_row == -1)
436 r->sheets[sd->current_sheet].start_row = sd->row;
439 if (r->sheets[sd->current_sheet].start_col == -1)
441 r->sheets[sd->current_sheet].start_col = sd->col;
443 if (! xmlTextReaderIsEmptyElement (sd->xtr))
444 sd->state = STATE_CELL;
446 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == sd->node_type) )
448 r->sheets[sd->current_sheet].stop_col = sd->col;
449 r->sheets[sd->current_sheet].stop_row = sd->row;
450 sd->state = STATE_SHEET_NAME;
454 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == sd->node_type)
456 sd->state = STATE_CELLS_START;
468 Sets the VAR of case C, to the value corresponding to the xml string XV
471 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
472 const xmlChar *xv, enum gnm_value_type type, int col, int row)
474 union value *v = case_data_rw (c, var);
477 value_set_missing (v, var_get_width (var));
478 else if ( var_is_alpha (var))
479 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
480 else if (type == VALUE_FLOAT || type == VALUE_INTEGER)
482 const char *text = CHAR_CAST (const char *, xv);
486 v->f = c_strtod (text, &endptr);
487 if ( errno != 0 || endptr == text)
492 const char *text = CHAR_CAST (const char *, xv);
494 const struct fmt_spec *fmt = var_get_write_format (var);
496 char *m = data_in (ss_cstr (text), "UTF-8",
504 char buf [FMT_STRING_LEN_MAX + 1];
505 char *cell = create_cell_ref (col, row);
507 msg (MW, _("Cannot convert the value in the spreadsheet cell %s to format (%s): %s"),
508 cell, fmt_to_string (fmt, buf), m);
519 xmlChar *first_value;
525 gnumeric_error_handler (void *ctx, const char *mesg,
526 UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
528 struct gnumeric_reader *r = ctx;
530 msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
532 r->spreadsheet.file_name,
533 xmlTextReaderLocatorLineNumber (loc),
537 static struct gnumeric_reader *
538 gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
541 struct state_data *sd;
543 xmlTextReaderPtr xtr;
546 assert (r == NULL || filename == NULL);
550 gz = gzopen (filename, "r");
554 gz = gzopen (r->spreadsheet.file_name, "r");
561 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
562 (xmlInputCloseCallback) gzclose, gz,
564 show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
574 r = xzalloc (sizeof *r);
575 r->spreadsheet.n_sheets = -1;
576 r->spreadsheet.file_name = strdup (filename);
585 xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
587 r->target_sheet = NULL;
588 r->target_sheet_index = -1;
590 sd->row = sd->col = -1;
591 sd->state = STATE_PRE_INIT;
593 r->spreadsheet.ref_cnt++;
596 /* Advance to the start of the workbook.
597 This gives us some confidence that we are actually dealing with a gnumeric
600 while ( (sd->state != STATE_INIT )
601 && 1 == (ret = xmlTextReaderRead (sd->xtr)))
603 process_node (r, sd);
609 /* Does not seem to be a gnumeric file */
610 gnumeric_unref (&r->spreadsheet);
614 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
618 const xmlChar *enc = xmlTextReaderConstEncoding (sd->xtr);
619 xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc));
621 if ( XML_CHAR_ENCODING_UTF8 != xce)
623 /* I have been told that ALL gnumeric files are UTF8 encoded. If that is correct, this
625 msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. "
626 "Any non-ascii characters will be incorrectly imported."),
627 r->spreadsheet.file_name,
637 gnumeric_probe (const char *filename, bool report_errors)
639 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
641 return &r->spreadsheet;
646 gnumeric_make_reader (struct spreadsheet *spreadsheet,
647 const struct spreadsheet_read_options *opts)
651 struct gnumeric_reader *r = NULL;
652 unsigned long int vstart = 0;
654 casenumber n_cases = CASENUMBER_MAX;
656 struct var_spec *var_spec = NULL;
659 r = (struct gnumeric_reader *) (spreadsheet);
661 r = gnumeric_reopen (r, NULL, true);
663 if ( opts->cell_range )
665 if ( ! convert_cell_ref (opts->cell_range,
666 &r->start_col, &r->start_row,
667 &r->stop_col, &r->stop_row))
669 msg (SE, _("Invalid cell range `%s'"),
682 r->target_sheet = BAD_CAST opts->sheet_name;
683 r->target_sheet_index = opts->sheet_index;
684 r->rsd.row = r->rsd.col = -1;
685 r->rsd.current_sheet = -1;
686 r->first_case = NULL;
689 /* Advance to the start of the cells for the target sheet */
690 while ( (r->rsd.state != STATE_CELL || r->rsd.row < r->start_row )
691 && 1 == (ret = xmlTextReaderRead (r->rsd.xtr)))
694 process_node (r, &r->rsd);
695 value = xmlTextReaderValue (r->rsd.xtr);
697 if ( r->rsd.state == STATE_MAXROW && r->rsd.node_type == XML_READER_TYPE_TEXT)
699 n_cases = 1 + _xmlchar_to_int (value) ;
704 /* If a range has been given, then use that to calculate the number
706 if ( opts->cell_range)
708 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
711 if ( opts->read_names )
718 /* Read in the first row of cells,
719 including the headers if read_names was set */
721 (( r->rsd.state == STATE_CELLS_START && r->rsd.row <= r->start_row) || r->rsd.state == STATE_CELL )
722 && (ret = xmlTextReaderRead (r->rsd.xtr))
727 if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_TEXT)
730 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
732 type = _xmlchar_to_int (attr);
737 process_node (r, &r->rsd);
739 if ( r->rsd.row > r->start_row )
742 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
744 r->vtype = _xmlchar_to_int (attr);
750 if ( r->rsd.col < r->start_col ||
751 (r->stop_col != -1 && r->rsd.col > r->stop_col))
754 idx = r->rsd.col - r->start_col;
756 if ( idx >= n_var_specs )
759 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
760 for (i = n_var_specs; i <= idx; ++i)
762 var_spec [i].name = NULL;
763 var_spec [i].width = -1;
764 var_spec [i].first_value = NULL;
765 var_spec [i].first_type = -1;
767 n_var_specs = idx + 1 ;
770 var_spec [idx].first_type = type;
772 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
774 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
775 const char *text = CHAR_CAST (const char *, value);
777 if ( r->rsd.row < r->start_row)
779 if ( opts->read_names )
781 var_spec [idx].name = xstrdup (text);
786 var_spec [idx].first_value = xmlStrdup (value);
788 if (-1 == var_spec [idx].width )
789 var_spec [idx].width = (opts->asw == -1) ?
790 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw;
795 else if ( r->rsd.node_type == XML_READER_TYPE_ELEMENT
796 && r->rsd.state == STATE_CELL)
798 if ( r->rsd.row == r->start_row )
801 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
803 if ( NULL == attr || VALUE_STRING != _xmlchar_to_int (attr))
804 var_spec [idx].width = 0;
812 const xmlChar *enc = xmlTextReaderConstEncoding (r->rsd.xtr);
815 /* Create the dictionary and populate it */
816 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
819 for (i = 0 ; i < n_var_specs ; ++i )
823 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
826 /* Probably no data exists for this variable, so allocate a
828 if ( var_spec[i].width == -1 )
829 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
831 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
832 dict_create_var (r->dict, name, var_spec[i].width);
836 /* Create the first case, and cache it */
837 r->used_first_case = false;
839 if ( n_var_specs == 0 )
841 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
842 spreadsheet->file_name);
846 r->proto = caseproto_ref (dict_get_proto (r->dict));
847 r->first_case = case_create (r->proto);
848 case_set_missing (r->first_case);
851 for ( i = 0 ; i < n_var_specs ; ++i )
853 const struct variable *var;
855 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
858 var = dict_get_var (r->dict, x++);
860 convert_xml_string_to_value (r->first_case, var,
861 var_spec[i].first_value,
862 var_spec[i].first_type,
867 for ( i = 0 ; i < n_var_specs ; ++i )
869 free (var_spec[i].first_value);
870 free (var_spec[i].name);
876 return casereader_create_sequential
880 &gnm_file_casereader_class, r);
884 for ( i = 0 ; i < n_var_specs ; ++i )
886 free (var_spec[i].first_value);
887 free (var_spec[i].name);
892 gnm_file_casereader_destroy (NULL, r);
898 /* Reads and returns one case from READER's file. Returns a null
899 pointer on failure. */
900 static struct ccase *
901 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
906 struct gnumeric_reader *r = r_;
907 int current_row = r->rsd.row;
909 if ( !r->used_first_case )
911 r->used_first_case = true;
912 return r->first_case;
915 c = case_create (r->proto);
916 case_set_missing (c);
918 if (r->start_col == -1)
919 r->start_col = r->rsd.min_col;
922 while ((r->rsd.state == STATE_CELL || r->rsd.state == STATE_CELLS_START )
923 && r->rsd.row == current_row && (ret = xmlTextReaderRead (r->rsd.xtr)))
925 process_node (r, &r->rsd);
927 if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_ELEMENT)
930 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
932 r->vtype = _xmlchar_to_int (attr);
937 if ( r->rsd.col < r->start_col || (r->stop_col != -1 &&
938 r->rsd.col > r->stop_col))
941 if ( r->rsd.col - r->start_col >= caseproto_get_n_widths (r->proto))
944 if ( r->stop_row != -1 && r->rsd.row > r->stop_row)
948 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
950 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
951 const int idx = r->rsd.col - r->start_col;
952 const struct variable *var = dict_get_var (r->dict, idx);
954 convert_xml_string_to_value (c, var, value, r->vtype,
955 r->rsd.col, r->rsd.row);
971 #endif /* GNM_READ_SUPPORT */