1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2007, 2009, 2010, 2011, 2012, 2013 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);
190 state_data_destroy (&r->msd);
200 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
202 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
203 assert (n < s->n_sheets);
205 return gr->sheets[n].name;
209 static void process_node (struct gnumeric_reader *r, struct state_data *sd);
214 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
217 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
219 assert (n < s->n_sheets);
222 (gr->sheets[n].stop_col == -1)
224 (1 == (ret = xmlTextReaderRead (gr->msd.xtr)))
227 process_node (gr, &gr->msd);
230 return create_cell_range (
231 gr->sheets[n].start_col,
232 gr->sheets[n].start_row,
233 gr->sheets[n].stop_col,
234 gr->sheets[n].stop_row);
239 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
241 struct gnumeric_reader *r = r_;
246 state_data_destroy (&r->rsd);
248 if (r->first_case && ! r->used_first_case )
249 case_unref (r->first_case);
252 caseproto_unref (r->proto);
254 gnumeric_unref (&r->spreadsheet);
259 process_node (struct gnumeric_reader *r, struct state_data *sd)
261 xmlChar *name = xmlTextReaderName (sd->xtr);
263 name = xmlStrdup (_xml ("--"));
265 sd->node_type = xmlTextReaderNodeType (sd->xtr);
270 sd->current_sheet = -1;
271 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
272 XML_READER_TYPE_ELEMENT == sd->node_type)
274 sd->state = STATE_SHEET_COUNT;
278 case STATE_SHEET_COUNT:
279 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
280 XML_READER_TYPE_ELEMENT == sd->node_type)
283 if (sd->current_sheet + 1 > r->spreadsheet.n_sheets)
285 struct sheet_detail *detail ;
286 r->sheets = xrealloc (r->sheets, (sd->current_sheet + 1) * sizeof *r->sheets);
287 detail = &r->sheets[sd->current_sheet];
288 detail->start_col = detail->stop_col = detail->start_row = detail->stop_row = -1;
290 r->spreadsheet.n_sheets = sd->current_sheet + 1;
293 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
294 XML_READER_TYPE_END_ELEMENT == sd->node_type)
296 sd->state = STATE_INIT;
297 sd->current_sheet = -1;
299 else if (XML_READER_TYPE_TEXT == sd->node_type)
301 if ( r->sheets [r->spreadsheet.n_sheets - 1].name == NULL)
302 r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
307 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
308 XML_READER_TYPE_ELEMENT == sd->node_type)
311 sd->state = STATE_SHEET_START;
314 case STATE_SHEET_START:
315 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
316 XML_READER_TYPE_ELEMENT == sd->node_type)
318 sd->state = STATE_SHEET_NAME;
321 case STATE_SHEET_NAME:
322 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
323 XML_READER_TYPE_END_ELEMENT == sd->node_type)
325 sd->state = STATE_INIT;
327 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
328 XML_READER_TYPE_END_ELEMENT == sd->node_type)
330 sd->state = STATE_INIT;
332 else if (XML_READER_TYPE_TEXT == sd->node_type)
334 if ( r->target_sheet != NULL)
336 xmlChar *value = xmlTextReaderValue (sd->xtr);
337 if ( 0 == xmlStrcmp (value, r->target_sheet))
338 sd->state = STATE_SHEET_FOUND;
341 else if (r->target_sheet_index == sd->current_sheet + 1)
343 sd->state = STATE_SHEET_FOUND;
345 else if (r->target_sheet_index == -1)
347 sd->state = STATE_SHEET_FOUND;
351 case STATE_SHEET_FOUND:
352 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
353 XML_READER_TYPE_ELEMENT == sd->node_type)
355 sd->min_col = INT_MAX;
356 if (! xmlTextReaderIsEmptyElement (sd->xtr))
357 sd->state = STATE_CELLS_START;
359 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
360 XML_READER_TYPE_ELEMENT == sd->node_type)
362 sd->state = STATE_MAXROW;
364 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
365 XML_READER_TYPE_ELEMENT == sd->node_type)
367 sd->state = STATE_MAXCOL;
369 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
370 XML_READER_TYPE_END_ELEMENT == sd->node_type)
372 sd->state = STATE_INIT;
376 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
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].maxrow = _xmlchar_to_int (value);
389 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
390 XML_READER_TYPE_END_ELEMENT == sd->node_type)
392 sd->state = STATE_SHEET_FOUND;
394 else if (sd->node_type == XML_READER_TYPE_TEXT)
396 xmlChar *value = xmlTextReaderValue (sd->xtr);
397 r->sheets[sd->current_sheet].maxcol = _xmlchar_to_int (value);
401 case STATE_CELLS_START:
402 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
403 XML_READER_TYPE_ELEMENT == sd->node_type)
405 xmlChar *attr = NULL;
407 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Col"));
408 sd->col = _xmlchar_to_int (attr);
411 if (sd->col < sd->min_col)
412 sd->min_col = sd->col;
414 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Row"));
415 sd->row = _xmlchar_to_int (attr);
418 if (r->sheets[sd->current_sheet].start_row == -1)
420 r->sheets[sd->current_sheet].start_row = sd->row;
423 if (r->sheets[sd->current_sheet].start_col == -1)
425 r->sheets[sd->current_sheet].start_col = sd->col;
427 if (! xmlTextReaderIsEmptyElement (sd->xtr))
428 sd->state = STATE_CELL;
430 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == sd->node_type) )
432 r->sheets[sd->current_sheet].stop_col = sd->col;
433 r->sheets[sd->current_sheet].stop_row = sd->row;
434 sd->state = STATE_SHEET_NAME;
438 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == sd->node_type)
440 sd->state = STATE_CELLS_START;
452 Sets the VAR of case C, to the value corresponding to the xml string XV
455 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
456 const xmlChar *xv, enum gnm_value_type type, int col, int row)
458 union value *v = case_data_rw (c, var);
461 value_set_missing (v, var_get_width (var));
462 else if ( var_is_alpha (var))
463 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
464 else if (type == VALUE_FLOAT || type == VALUE_INTEGER)
466 const char *text = CHAR_CAST (const char *, xv);
470 v->f = c_strtod (text, &endptr);
471 if ( errno != 0 || endptr == text)
476 const char *text = CHAR_CAST (const char *, xv);
478 const struct fmt_spec *fmt = var_get_write_format (var);
480 char *m = data_in (ss_cstr (text), "UTF-8",
488 char buf [FMT_STRING_LEN_MAX + 1];
489 char *cell = create_cell_ref (col, row);
491 msg (MW, _("Cannot convert the value in the spreadsheet cell %s to format (%s): %s"),
492 cell, fmt_to_string (fmt, buf), m);
503 xmlChar *first_value;
509 gnumeric_error_handler (void *ctx, const char *mesg,
510 UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
512 struct gnumeric_reader *r = ctx;
514 msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
516 r->spreadsheet.file_name,
517 xmlTextReaderLocatorLineNumber (loc),
521 static struct gnumeric_reader *
522 gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
525 struct state_data *sd;
527 xmlTextReaderPtr xtr;
530 assert (r == NULL || filename == NULL);
534 gz = gzopen (filename, "r");
538 gz = gzopen (r->spreadsheet.file_name, "r");
545 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
546 (xmlInputCloseCallback) gzclose, gz,
548 show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
558 r = xzalloc (sizeof *r);
559 r->spreadsheet.n_sheets = -1;
560 r->spreadsheet.file_name = strdup (filename);
569 xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
571 r->target_sheet = NULL;
572 r->target_sheet_index = -1;
574 sd->row = sd->col = -1;
575 sd->state = STATE_PRE_INIT;
577 r->spreadsheet.ref_cnt++;
580 /* Advance to the start of the workbook.
581 This gives us some confidence that we are actually dealing with a gnumeric
584 while ( (sd->state != STATE_INIT )
585 && 1 == (ret = xmlTextReaderRead (sd->xtr)))
587 process_node (r, sd);
593 /* Does not seem to be a gnumeric file */
594 gnumeric_unref (&r->spreadsheet);
598 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
602 const xmlChar *enc = xmlTextReaderConstEncoding (sd->xtr);
603 xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc));
605 if ( XML_CHAR_ENCODING_UTF8 != xce)
607 /* I have been told that ALL gnumeric files are UTF8 encoded. If that is correct, this
609 msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. "
610 "Any non-ascii characters will be incorrectly imported."),
611 r->spreadsheet.file_name,
621 gnumeric_probe (const char *filename, bool report_errors)
623 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
625 return &r->spreadsheet;
630 gnumeric_make_reader (struct spreadsheet *spreadsheet,
631 const struct spreadsheet_read_options *opts)
635 struct gnumeric_reader *r = NULL;
636 unsigned long int vstart = 0;
638 casenumber n_cases = CASENUMBER_MAX;
640 struct var_spec *var_spec = NULL;
643 r = (struct gnumeric_reader *) (spreadsheet);
645 r = gnumeric_reopen (r, NULL, true);
647 if ( opts->cell_range )
649 if ( ! convert_cell_ref (opts->cell_range,
650 &r->start_col, &r->start_row,
651 &r->stop_col, &r->stop_row))
653 msg (SE, _("Invalid cell range `%s'"),
666 r->target_sheet = BAD_CAST opts->sheet_name;
667 r->target_sheet_index = opts->sheet_index;
668 r->rsd.row = r->rsd.col = -1;
669 r->rsd.current_sheet = -1;
670 r->first_case = NULL;
673 /* Advance to the start of the cells for the target sheet */
674 while ( (r->rsd.state != STATE_CELL || r->rsd.row < r->start_row )
675 && 1 == (ret = xmlTextReaderRead (r->rsd.xtr)))
678 process_node (r, &r->rsd);
679 value = xmlTextReaderValue (r->rsd.xtr);
681 if ( r->rsd.state == STATE_MAXROW && r->rsd.node_type == XML_READER_TYPE_TEXT)
683 n_cases = 1 + _xmlchar_to_int (value) ;
688 /* If a range has been given, then use that to calculate the number
690 if ( opts->cell_range)
692 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
695 if ( opts->read_names )
702 /* Read in the first row of cells,
703 including the headers if read_names was set */
705 (( r->rsd.state == STATE_CELLS_START && r->rsd.row <= r->start_row) || r->rsd.state == STATE_CELL )
706 && (ret = xmlTextReaderRead (r->rsd.xtr))
711 if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_TEXT)
714 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
716 type = _xmlchar_to_int (attr);
721 process_node (r, &r->rsd);
723 if ( r->rsd.row > r->start_row )
726 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
728 r->vtype = _xmlchar_to_int (attr);
734 if ( r->rsd.col < r->start_col ||
735 (r->stop_col != -1 && r->rsd.col > r->stop_col))
738 idx = r->rsd.col - r->start_col;
740 if ( idx >= n_var_specs )
743 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
744 for (i = n_var_specs; i <= idx; ++i)
746 var_spec [i].name = NULL;
747 var_spec [i].width = -1;
748 var_spec [i].first_value = NULL;
749 var_spec [i].first_type = -1;
751 n_var_specs = idx + 1 ;
754 var_spec [idx].first_type = type;
756 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
758 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
759 const char *text = CHAR_CAST (const char *, value);
761 if ( r->rsd.row < r->start_row)
763 if ( opts->read_names )
765 var_spec [idx].name = xstrdup (text);
770 var_spec [idx].first_value = xmlStrdup (value);
772 if (-1 == var_spec [idx].width )
773 var_spec [idx].width = (opts->asw == -1) ?
774 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw;
779 else if ( r->rsd.node_type == XML_READER_TYPE_ELEMENT
780 && r->rsd.state == STATE_CELL)
782 if ( r->rsd.row == r->start_row )
785 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
787 if ( NULL == attr || VALUE_STRING != _xmlchar_to_int (attr))
788 var_spec [idx].width = 0;
796 const xmlChar *enc = xmlTextReaderConstEncoding (r->rsd.xtr);
799 /* Create the dictionary and populate it */
800 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
803 for (i = 0 ; i < n_var_specs ; ++i )
807 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
810 /* Probably no data exists for this variable, so allocate a
812 if ( var_spec[i].width == -1 )
813 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
815 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
816 dict_create_var (r->dict, name, var_spec[i].width);
820 /* Create the first case, and cache it */
821 r->used_first_case = false;
823 if ( n_var_specs == 0 )
825 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
826 spreadsheet->file_name);
830 r->proto = caseproto_ref (dict_get_proto (r->dict));
831 r->first_case = case_create (r->proto);
832 case_set_missing (r->first_case);
835 for ( i = 0 ; i < n_var_specs ; ++i )
837 const struct variable *var;
839 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
842 var = dict_get_var (r->dict, x++);
844 convert_xml_string_to_value (r->first_case, var,
845 var_spec[i].first_value,
846 var_spec[i].first_type,
851 for ( i = 0 ; i < n_var_specs ; ++i )
853 free (var_spec[i].first_value);
854 free (var_spec[i].name);
860 return casereader_create_sequential
864 &gnm_file_casereader_class, r);
868 for ( i = 0 ; i < n_var_specs ; ++i )
870 free (var_spec[i].first_value);
871 free (var_spec[i].name);
875 dict_destroy (spreadsheet->dict);
876 spreadsheet->dict = NULL;
878 gnm_file_casereader_destroy (NULL, r);
884 /* Reads and returns one case from READER's file. Returns a null
885 pointer on failure. */
886 static struct ccase *
887 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
892 struct gnumeric_reader *r = r_;
893 int current_row = r->rsd.row;
895 if ( !r->used_first_case )
897 r->used_first_case = true;
898 return r->first_case;
901 c = case_create (r->proto);
902 case_set_missing (c);
904 if (r->start_col == -1)
905 r->start_col = r->rsd.min_col;
908 while ((r->rsd.state == STATE_CELL || r->rsd.state == STATE_CELLS_START )
909 && r->rsd.row == current_row && (ret = xmlTextReaderRead (r->rsd.xtr)))
911 process_node (r, &r->rsd);
913 if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_ELEMENT)
916 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
918 r->vtype = _xmlchar_to_int (attr);
923 if ( r->rsd.col < r->start_col || (r->stop_col != -1 &&
924 r->rsd.col > r->stop_col))
927 if ( r->rsd.col - r->start_col >= caseproto_get_n_widths (r->proto))
930 if ( r->stop_row != -1 && r->rsd.row > r->stop_row)
934 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
936 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
937 const int idx = r->rsd.col - r->start_col;
938 const struct variable *var = dict_get_var (r->dict, idx);
940 convert_xml_string_to_value (c, var, value, r->vtype,
941 r->rsd.col, r->rsd.row);
957 #endif /* GNM_READ_SUPPORT */