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/case.h"
52 #include "data/casereader-provider.h"
53 #include "data/dictionary.h"
54 #include "data/identifier.h"
55 #include "data/value.h"
56 #include "data/variable.h"
57 #include "libpspp/i18n.h"
58 #include "libpspp/str.h"
60 #include "gl/xalloc.h"
62 static void gnm_file_casereader_destroy (struct casereader *, void *);
64 static struct ccase *gnm_file_casereader_read (struct casereader *, void *);
67 static const struct casereader_class gnm_file_casereader_class =
69 gnm_file_casereader_read,
70 gnm_file_casereader_destroy,
77 STATE_PRE_INIT = 0, /* Initial state */
78 STATE_SHEET_COUNT, /* Found the sheet index */
79 STATE_INIT , /* Other Initial state */
80 STATE_SHEET_START, /* Found the start of a sheet */
81 STATE_SHEET_NAME, /* Found the sheet name */
84 STATE_SHEET_FOUND, /* Found the sheet that we actually want */
85 STATE_CELLS_START, /* Found the start of the cell array */
86 STATE_CELL /* Found a cell */
91 /* The name of the sheet (utf8 encoding) */
105 /* The libxml reader for this instance */
106 xmlTextReaderPtr xtr;
108 /* An internal state variable */
109 enum reader_state state;
122 state_data_destroy (struct state_data *sd)
124 xmlFreeTextReader (sd->xtr);
128 struct gnumeric_reader
130 struct spreadsheet spreadsheet;
133 struct state_data rsd;
134 struct state_data msd;
141 struct sheet_detail *sheets;
143 const xmlChar *target_sheet;
144 int target_sheet_index;
146 struct caseproto *proto;
147 struct dictionary *dict;
148 struct ccase *first_case;
149 bool used_first_case;
154 gnumeric_destroy (struct spreadsheet *s)
156 struct gnumeric_reader *r = (struct gnumeric_reader *) s;
158 if (0 == --r->ref_cnt)
162 for (i = 0; i < s->n_sheets; ++i)
164 xmlFree (r->sheets[i].name);
168 state_data_destroy (&r->msd);
176 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
178 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
179 assert (n < s->n_sheets);
181 return gr->sheets[n].name;
185 static void process_node (struct gnumeric_reader *r, struct state_data *sd);
190 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
193 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
195 assert (n < s->n_sheets);
198 (gr->sheets[n].stop_col == -1)
200 (1 == (ret = xmlTextReaderRead (gr->msd.xtr)))
203 process_node (gr, &gr->msd);
206 return create_cell_ref (
207 gr->sheets[n].start_col,
208 gr->sheets[n].start_row,
209 gr->sheets[n].stop_col,
210 gr->sheets[n].stop_row);
215 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
217 struct gnumeric_reader *r = r_;
222 state_data_destroy (&r->rsd);
224 if (r->first_case && ! r->used_first_case )
225 case_unref (r->first_case);
228 caseproto_unref (r->proto);
230 gnumeric_destroy (&r->spreadsheet);
235 process_node (struct gnumeric_reader *r, struct state_data *sd)
237 xmlChar *name = xmlTextReaderName (sd->xtr);
239 name = xmlStrdup (_xml ("--"));
241 sd->node_type = xmlTextReaderNodeType (sd->xtr);
246 sd->current_sheet = -1;
247 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
248 XML_READER_TYPE_ELEMENT == sd->node_type)
250 sd->state = STATE_SHEET_COUNT;
254 case STATE_SHEET_COUNT:
255 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
256 XML_READER_TYPE_ELEMENT == sd->node_type)
259 if (sd->current_sheet + 1 > r->spreadsheet.n_sheets)
261 struct sheet_detail *detail ;
262 r->sheets = xrealloc (r->sheets, (sd->current_sheet + 1) * sizeof *r->sheets);
263 detail = &r->sheets[sd->current_sheet];
264 detail->start_col = detail->stop_col = detail->start_row = detail->stop_row = -1;
266 r->spreadsheet.n_sheets = sd->current_sheet + 1;
269 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
270 XML_READER_TYPE_END_ELEMENT == sd->node_type)
272 sd->state = STATE_INIT;
273 sd->current_sheet = -1;
275 else if (XML_READER_TYPE_TEXT == sd->node_type)
277 if ( r->sheets [r->spreadsheet.n_sheets - 1].name == NULL)
278 r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
283 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
284 XML_READER_TYPE_ELEMENT == sd->node_type)
287 sd->state = STATE_SHEET_START;
290 case STATE_SHEET_START:
291 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
292 XML_READER_TYPE_ELEMENT == sd->node_type)
294 sd->state = STATE_SHEET_NAME;
297 case STATE_SHEET_NAME:
298 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
299 XML_READER_TYPE_END_ELEMENT == sd->node_type)
301 sd->state = STATE_INIT;
303 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
304 XML_READER_TYPE_END_ELEMENT == sd->node_type)
306 sd->state = STATE_INIT;
308 else if (XML_READER_TYPE_TEXT == sd->node_type)
310 if ( r->target_sheet != NULL)
312 xmlChar *value = xmlTextReaderValue (sd->xtr);
313 if ( 0 == xmlStrcmp (value, r->target_sheet))
314 sd->state = STATE_SHEET_FOUND;
317 else if (r->target_sheet_index == sd->current_sheet + 1)
319 sd->state = STATE_SHEET_FOUND;
321 else if (r->target_sheet_index == -1)
323 sd->state = STATE_SHEET_FOUND;
327 case STATE_SHEET_FOUND:
328 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
329 XML_READER_TYPE_ELEMENT == sd->node_type)
331 sd->min_col = INT_MAX;
332 if (! xmlTextReaderIsEmptyElement (sd->xtr))
333 sd->state = STATE_CELLS_START;
335 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
336 XML_READER_TYPE_ELEMENT == sd->node_type)
338 sd->state = STATE_MAXROW;
340 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
341 XML_READER_TYPE_ELEMENT == sd->node_type)
343 sd->state = STATE_MAXCOL;
345 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
346 XML_READER_TYPE_END_ELEMENT == sd->node_type)
348 sd->state = STATE_INIT;
352 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
353 XML_READER_TYPE_END_ELEMENT == sd->node_type)
355 sd->state = STATE_SHEET_FOUND;
357 else if (sd->node_type == XML_READER_TYPE_TEXT)
359 xmlChar *value = xmlTextReaderValue (sd->xtr);
360 r->sheets[sd->current_sheet].maxrow = _xmlchar_to_int (value);
365 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
366 XML_READER_TYPE_END_ELEMENT == sd->node_type)
368 sd->state = STATE_SHEET_FOUND;
370 else if (sd->node_type == XML_READER_TYPE_TEXT)
372 xmlChar *value = xmlTextReaderValue (sd->xtr);
373 r->sheets[sd->current_sheet].maxcol = _xmlchar_to_int (value);
377 case STATE_CELLS_START:
378 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
379 XML_READER_TYPE_ELEMENT == sd->node_type)
381 xmlChar *attr = NULL;
383 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Col"));
384 sd->col = _xmlchar_to_int (attr);
387 if (sd->col < sd->min_col)
388 sd->min_col = sd->col;
390 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Row"));
391 sd->row = _xmlchar_to_int (attr);
394 if (r->sheets[sd->current_sheet].start_row == -1)
396 r->sheets[sd->current_sheet].start_row = sd->row;
399 if (r->sheets[sd->current_sheet].start_col == -1)
401 r->sheets[sd->current_sheet].start_col = sd->col;
403 if (! xmlTextReaderIsEmptyElement (sd->xtr))
404 sd->state = STATE_CELL;
406 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == sd->node_type) )
408 r->sheets[sd->current_sheet].stop_col = sd->col;
409 r->sheets[sd->current_sheet].stop_row = sd->row;
410 sd->state = STATE_SHEET_NAME;
414 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == sd->node_type)
416 sd->state = STATE_CELLS_START;
428 Sets the VAR of case C, to the value corresponding to the xml string XV
431 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
434 union value *v = case_data_rw (c, var);
437 value_set_missing (v, var_get_width (var));
438 else if ( var_is_alpha (var))
439 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
442 const char *text = CHAR_CAST (const char *, xv);
446 v->f = c_strtod (text, &endptr);
447 if ( errno != 0 || endptr == text)
456 xmlChar *first_value;
461 gnumeric_error_handler (void *ctx, const char *mesg,
462 UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
464 struct gnumeric_reader *r = ctx;
466 msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
468 r->spreadsheet.file_name,
469 xmlTextReaderLocatorLineNumber (loc),
473 static struct gnumeric_reader *
474 gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
477 struct state_data *sd;
479 xmlTextReaderPtr xtr;
482 assert (r == NULL || filename == NULL);
486 gz = gzopen (filename, "r");
490 gz = gzopen (r->spreadsheet.file_name, "r");
497 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
498 (xmlInputCloseCallback) gzclose, gz,
500 show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
510 r = xzalloc (sizeof *r);
511 r->spreadsheet.n_sheets = -1;
512 r->spreadsheet.file_name = filename;
521 xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
523 r->target_sheet = NULL;
524 r->target_sheet_index = -1;
526 sd->row = sd->col = -1;
527 sd->state = STATE_PRE_INIT;
531 /* Advance to the start of the workbook.
532 This gives us some confidence that we are actually dealing with a gnumeric
535 while ( (sd->state != STATE_INIT )
536 && 1 == (ret = xmlTextReaderRead (sd->xtr)))
538 process_node (r, sd);
544 /* Does not seem to be a gnumeric file */
545 xmlFreeTextReader (sd->xtr);
550 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
554 const xmlChar *enc = xmlTextReaderConstEncoding (sd->xtr);
555 xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc));
557 if ( XML_CHAR_ENCODING_UTF8 != xce)
559 /* I have been told that ALL gnumeric files are UTF8 encoded. If that is correct, this
561 msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. "
562 "Any non-ascii characters will be incorrectly imported."),
563 r->spreadsheet.file_name,
573 gnumeric_probe (const char *filename, bool report_errors)
575 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
577 return &r->spreadsheet;
582 gnumeric_make_reader (struct spreadsheet *spreadsheet,
583 const struct spreadsheet_read_options *opts)
586 struct gnumeric_reader *r = NULL;
587 unsigned long int vstart = 0;
589 casenumber n_cases = CASENUMBER_MAX;
591 struct var_spec *var_spec = NULL;
594 r = (struct gnumeric_reader *) (spreadsheet);
596 r = gnumeric_reopen (r, NULL, true);
598 if ( opts->cell_range )
600 if ( ! convert_cell_ref (opts->cell_range,
601 &r->start_col, &r->start_row,
602 &r->stop_col, &r->stop_row))
604 msg (SE, _("Invalid cell range `%s'"),
617 r->target_sheet = BAD_CAST opts->sheet_name;
618 r->target_sheet_index = opts->sheet_index;
619 r->rsd.row = r->rsd.col = -1;
620 r->rsd.current_sheet = -1;
621 r->first_case = NULL;
624 /* Advance to the start of the cells for the target sheet */
625 while ( (r->rsd.state != STATE_CELL || r->rsd.row < r->start_row )
626 && 1 == (ret = xmlTextReaderRead (r->rsd.xtr)))
629 process_node (r, &r->rsd);
630 value = xmlTextReaderValue (r->rsd.xtr);
632 if ( r->rsd.state == STATE_MAXROW && r->rsd.node_type == XML_READER_TYPE_TEXT)
634 n_cases = 1 + _xmlchar_to_int (value) ;
639 /* If a range has been given, then use that to calculate the number
641 if ( opts->cell_range)
643 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
646 if ( opts->read_names )
652 /* Read in the first row of cells,
653 including the headers if read_names was set */
655 (( r->rsd.state == STATE_CELLS_START && r->rsd.row <= r->start_row) || r->rsd.state == STATE_CELL )
656 && (ret = xmlTextReaderRead (r->rsd.xtr))
660 process_node (r, &r->rsd);
662 if ( r->rsd.row > r->start_row ) break;
664 if ( r->rsd.col < r->start_col ||
665 (r->stop_col != -1 && r->rsd.col > r->stop_col))
668 idx = r->rsd.col - r->start_col;
670 if ( idx >= n_var_specs )
673 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
674 for (i = n_var_specs; i <= idx; ++i)
676 var_spec [i].name = NULL;
677 var_spec [i].width = -1;
678 var_spec [i].first_value = NULL;
680 n_var_specs = idx + 1 ;
683 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
685 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
686 const char *text = CHAR_CAST (const char *, value);
688 if ( r->rsd.row < r->start_row)
690 if ( opts->read_names )
692 var_spec [idx].name = xstrdup (text);
697 var_spec [idx].first_value = xmlStrdup (value);
699 if (-1 == var_spec [idx].width )
700 var_spec [idx].width = (opts->asw == -1) ?
701 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw;
706 else if ( r->rsd.node_type == XML_READER_TYPE_ELEMENT
707 && r->rsd.state == STATE_CELL)
709 if ( r->rsd.row == r->start_row )
712 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
714 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
715 var_spec [idx].width = 0;
723 const xmlChar *enc = xmlTextReaderConstEncoding (r->rsd.xtr);
726 /* Create the dictionary and populate it */
727 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
730 for (i = 0 ; i < n_var_specs ; ++i )
734 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
737 /* Probably no data exists for this variable, so allocate a
739 if ( var_spec[i].width == -1 )
740 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
742 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
743 dict_create_var (r->dict, name, var_spec[i].width);
747 /* Create the first case, and cache it */
748 r->used_first_case = false;
750 if ( n_var_specs == 0 )
752 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
753 spreadsheet->file_name);
757 r->proto = caseproto_ref (dict_get_proto (r->dict));
758 r->first_case = case_create (r->proto);
759 case_set_missing (r->first_case);
762 for ( i = 0 ; i < n_var_specs ; ++i )
764 const struct variable *var;
766 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
769 var = dict_get_var (r->dict, x++);
771 convert_xml_string_to_value (r->first_case, var,
772 var_spec[i].first_value);
775 for ( i = 0 ; i < n_var_specs ; ++i )
777 free (var_spec[i].first_value);
778 free (var_spec[i].name);
784 return casereader_create_sequential
788 &gnm_file_casereader_class, r);
792 for ( i = 0 ; i < n_var_specs ; ++i )
794 free (var_spec[i].first_value);
795 free (var_spec[i].name);
799 dict_destroy (spreadsheet->dict);
800 spreadsheet->dict = NULL;
802 gnm_file_casereader_destroy (NULL, r);
808 /* Reads and returns one case from READER's file. Returns a null
809 pointer on failure. */
810 static struct ccase *
811 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
816 struct gnumeric_reader *r = r_;
817 int current_row = r->rsd.row;
819 if ( !r->used_first_case )
821 r->used_first_case = true;
822 return r->first_case;
825 c = case_create (r->proto);
826 case_set_missing (c);
828 if (r->start_col == -1)
829 r->start_col = r->rsd.min_col;
831 while ((r->rsd.state == STATE_CELL || r->rsd.state == STATE_CELLS_START )
832 && r->rsd.row == current_row && (ret = xmlTextReaderRead (r->rsd.xtr)))
834 process_node (r, &r->rsd);
836 if ( r->rsd.col < r->start_col || (r->stop_col != -1 &&
837 r->rsd.col > r->stop_col))
840 if ( r->rsd.col - r->start_col >= caseproto_get_n_widths (r->proto))
843 if ( r->stop_row != -1 && r->rsd.row > r->stop_row)
846 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
848 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
850 const int idx = r->rsd.col - r->start_col;
852 const struct variable *var = dict_get_var (r->dict, idx);
854 convert_xml_string_to_value (c, var, value);
870 #endif /* GNM_SUPPORT */