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 (struct spreadsheet_read_info *gri, 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 *);
66 static const struct casereader_class gnm_file_casereader_class =
68 gnm_file_casereader_read,
69 gnm_file_casereader_destroy,
76 STATE_PRE_INIT = 0, /* Initial state */
77 STATE_SHEET_COUNT, /* Found the sheet index */
78 STATE_INIT , /* Other Initial state */
79 STATE_SHEET_START, /* Found the start of a sheet */
80 STATE_SHEET_NAME, /* Found the sheet name */
83 STATE_SHEET_FOUND, /* Found the sheet that we actually want */
84 STATE_CELLS_START, /* Found the start of the cell array */
85 STATE_CELL /* Found a cell */
102 struct gnumeric_reader
104 struct spreadsheet spreadsheet;
106 /* The libxml reader for this instance */
107 xmlTextReaderPtr xtr;
109 /* An internal state variable */
110 enum reader_state state;
123 struct sheet_detail *sheets;
125 const xmlChar *target_sheet;
126 int target_sheet_index;
128 struct caseproto *proto;
129 struct dictionary *dict;
130 struct ccase *first_case;
131 bool used_first_case;
136 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
138 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
139 assert (n < s->n_sheets);
141 return gr->sheets[n].name; // Kludge: Assumes the encoding is utf8
145 static void process_node (struct gnumeric_reader *r);
149 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
152 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
154 assert (n < s->n_sheets);
157 (gr->sheets[n].stop_col == -1)
159 (1 == (ret = xmlTextReaderRead (gr->xtr)))
165 return create_cell_ref (
166 gr->sheets[n].start_col,
167 gr->sheets[n].start_row,
168 gr->sheets[n].stop_col,
169 gr->sheets[n].stop_row);
174 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
176 struct gnumeric_reader *r = r_;
181 xmlFreeTextReader (r->xtr);
184 if ( ! r->used_first_case )
185 case_unref (r->first_case);
187 caseproto_unref (r->proto);
190 for (i = 0; i < r->spreadsheet.n_sheets; ++i)
192 xmlFree (r->sheets[i].name);
203 process_node (struct gnumeric_reader *r)
205 xmlChar *name = xmlTextReaderName (r->xtr);
207 name = xmlStrdup (_xml ("--"));
209 r->node_type = xmlTextReaderNodeType (r->xtr);
214 r->current_sheet = -1;
215 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
216 XML_READER_TYPE_ELEMENT == r->node_type)
218 r->state = STATE_SHEET_COUNT;
222 case STATE_SHEET_COUNT:
223 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
224 XML_READER_TYPE_ELEMENT == r->node_type)
227 if (r->current_sheet + 1 > r->spreadsheet.n_sheets)
229 struct sheet_detail *sd ;
230 r->sheets = xrealloc (r->sheets, (r->current_sheet + 1) * sizeof *r->sheets);
231 sd = &r->sheets[r->current_sheet];
232 sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -1;
233 r->spreadsheet.n_sheets = r->current_sheet + 1;
236 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
237 XML_READER_TYPE_END_ELEMENT == r->node_type)
239 r->state = STATE_INIT;
240 r->current_sheet = -1;
242 else if (XML_READER_TYPE_TEXT == r->node_type)
244 r->sheets [r->spreadsheet.n_sheets - 1].name = xmlTextReaderValue (r->xtr);
249 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
250 XML_READER_TYPE_ELEMENT == r->node_type)
253 r->state = STATE_SHEET_START;
256 case STATE_SHEET_START:
257 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
258 XML_READER_TYPE_ELEMENT == r->node_type)
260 r->state = STATE_SHEET_NAME;
263 case STATE_SHEET_NAME:
264 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
265 XML_READER_TYPE_END_ELEMENT == r->node_type)
267 r->state = STATE_INIT;
269 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
270 XML_READER_TYPE_END_ELEMENT == r->node_type)
272 r->state = STATE_INIT;
274 else if (XML_READER_TYPE_TEXT == r->node_type)
276 if ( r->target_sheet != NULL)
278 xmlChar *value = xmlTextReaderValue (r->xtr);
279 if ( 0 == xmlStrcmp (value, r->target_sheet))
280 r->state = STATE_SHEET_FOUND;
283 else if (r->target_sheet_index == r->current_sheet + 1)
285 r->state = STATE_SHEET_FOUND;
287 else if (r->target_sheet_index == -1)
289 r->state = STATE_SHEET_FOUND;
293 case STATE_SHEET_FOUND:
294 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
295 XML_READER_TYPE_ELEMENT == r->node_type)
297 r->min_col = INT_MAX;
298 if (! xmlTextReaderIsEmptyElement (r->xtr))
299 r->state = STATE_CELLS_START;
301 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
302 XML_READER_TYPE_ELEMENT == r->node_type)
304 r->state = STATE_MAXROW;
306 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
307 XML_READER_TYPE_ELEMENT == r->node_type)
309 r->state = STATE_MAXCOL;
311 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
312 XML_READER_TYPE_END_ELEMENT == r->node_type)
314 r->state = STATE_INIT;
318 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
319 XML_READER_TYPE_END_ELEMENT == r->node_type)
321 r->state = STATE_SHEET_FOUND;
323 else if (r->node_type == XML_READER_TYPE_TEXT)
325 xmlChar *value = xmlTextReaderValue (r->xtr);
326 r->sheets[r->current_sheet].maxrow = _xmlchar_to_int (value);
331 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
332 XML_READER_TYPE_END_ELEMENT == r->node_type)
334 r->state = STATE_SHEET_FOUND;
336 else if (r->node_type == XML_READER_TYPE_TEXT)
338 xmlChar *value = xmlTextReaderValue (r->xtr);
339 r->sheets[r->current_sheet].maxcol = _xmlchar_to_int (value);
343 case STATE_CELLS_START:
344 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
345 XML_READER_TYPE_ELEMENT == r->node_type)
347 xmlChar *attr = NULL;
349 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
350 r->col = _xmlchar_to_int (attr);
353 if (r->col < r->min_col)
356 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
357 r->row = _xmlchar_to_int (attr);
360 if (r->sheets[r->current_sheet].start_row == -1)
362 r->sheets[r->current_sheet].start_row = r->row;
365 if (r->sheets[r->current_sheet].start_col == -1)
367 r->sheets[r->current_sheet].start_col = r->col;
369 if (! xmlTextReaderIsEmptyElement (r->xtr))
370 r->state = STATE_CELL;
372 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == r->node_type) )
374 r->sheets[r->current_sheet].stop_col = r->col;
375 r->sheets[r->current_sheet].stop_row = r->row;
376 r->state = STATE_SHEET_NAME;
380 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == r->node_type)
382 r->state = STATE_CELLS_START;
394 Sets the VAR of case C, to the value corresponding to the xml string XV
397 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
400 union value *v = case_data_rw (c, var);
403 value_set_missing (v, var_get_width (var));
404 else if ( var_is_alpha (var))
405 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
408 const char *text = CHAR_CAST (const char *, xv);
412 v->f = c_strtod (text, &endptr);
413 if ( errno != 0 || endptr == text)
422 xmlChar *first_value;
427 gnumeric_destroy (struct spreadsheet *s)
429 gnm_file_casereader_destroy (NULL, s);
434 gnumeric_error_handler (void *ctx, const char *mesg,
435 UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
437 struct gnumeric_reader *r = ctx;
439 msg (MW, _("There was a problem whilst reading the Gnumeric file `%s' (near line %d): `%s'"),
440 r->spreadsheet.file_name,
441 xmlTextReaderLocatorLineNumber (loc),
445 static struct gnumeric_reader *
446 gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
450 xmlTextReaderPtr xtr;
453 assert (r == NULL || filename == NULL);
456 xmlFreeTextReader (r->xtr);
459 gz = gzopen (filename, "r");
461 gz = gzopen ( r->spreadsheet.file_name, "r");
467 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
468 (xmlInputCloseCallback) gzclose, gz,
470 show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
480 r = xzalloc (sizeof *r);
481 r->spreadsheet.n_sheets = -1;
482 r->spreadsheet.file_name = filename;
486 xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
488 r->target_sheet = NULL;
489 r->target_sheet_index = -1;
491 r->row = r->col = -1;
492 r->state = STATE_PRE_INIT;
495 /* Advance to the start of the workbook.
496 This gives us some confidence that we are actually dealing with a gnumeric
499 while ( (r->state != STATE_INIT )
500 && 1 == (ret = xmlTextReaderRead (r->xtr)))
507 /* Does not seem to be a gnumeric file */
508 xmlFreeTextReader (r->xtr);
513 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
520 gnumeric_probe (const char *filename, bool report_errors)
522 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
524 return &r->spreadsheet;
529 gnumeric_make_reader (struct spreadsheet *spreadsheet,
530 const struct spreadsheet_read_info *gri,
531 const struct spreadsheet_read_options *opts)
534 struct gnumeric_reader *r = NULL;
535 unsigned long int vstart = 0;
537 casenumber n_cases = CASENUMBER_MAX;
539 struct var_spec *var_spec = NULL;
542 r = (struct gnumeric_reader *) (spreadsheet);
545 r = gnumeric_reopen (r, NULL, true);
547 if ( opts->cell_range )
549 if ( ! convert_cell_ref (opts->cell_range,
550 &r->start_col, &r->start_row,
551 &r->stop_col, &r->stop_row))
553 msg (SE, _("Invalid cell range `%s'"),
566 r->target_sheet = BAD_CAST opts->sheet_name;
567 r->target_sheet_index = opts->sheet_index;
568 r->row = r->col = -1;
569 r->current_sheet = -1;
571 /* Advance to the start of the cells for the target sheet */
572 while ( (r->state != STATE_CELL || r->row < r->start_row )
573 && 1 == (ret = xmlTextReaderRead (r->xtr)))
577 value = xmlTextReaderValue (r->xtr);
579 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
581 n_cases = 1 + _xmlchar_to_int (value) ;
586 /* If a range has been given, then use that to calculate the number
588 if ( opts->cell_range)
590 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
593 if ( gri->read_names )
599 /* Read in the first row of cells,
600 including the headers if read_names was set */
602 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
603 && (ret = xmlTextReaderRead (r->xtr))
609 if ( r->row > r->start_row ) break;
611 if ( r->col < r->start_col ||
612 (r->stop_col != -1 && r->col > r->stop_col))
615 idx = r->col - r->start_col;
617 if ( idx >= n_var_specs )
620 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
621 for (i = n_var_specs; i <= idx; ++i)
623 var_spec [i].name = NULL;
624 var_spec [i].width = -1;
625 var_spec [i].first_value = NULL;
627 n_var_specs = idx + 1 ;
630 if ( r->node_type == XML_READER_TYPE_TEXT )
632 xmlChar *value = xmlTextReaderValue (r->xtr);
633 const char *text = CHAR_CAST (const char *, value);
635 if ( r->row < r->start_row)
637 if ( gri->read_names )
639 var_spec [idx].name = xstrdup (text);
644 var_spec [idx].first_value = xmlStrdup (value);
646 if (-1 == var_spec [idx].width )
647 var_spec [idx].width = (gri->asw == -1) ?
648 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw;
653 else if ( r->node_type == XML_READER_TYPE_ELEMENT
654 && r->state == STATE_CELL)
656 if ( r->row == r->start_row )
659 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
661 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
662 var_spec [idx].width = 0;
670 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
673 /* Create the dictionary and populate it */
674 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
677 for (i = 0 ; i < n_var_specs ; ++i )
681 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
684 /* Probably no data exists for this variable, so allocate a
686 if ( var_spec[i].width == -1 )
687 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
689 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
690 dict_create_var (r->dict, name, var_spec[i].width);
694 /* Create the first case, and cache it */
695 r->used_first_case = false;
697 if ( n_var_specs == 0 )
699 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
700 spreadsheet->file_name);
704 r->proto = caseproto_ref (dict_get_proto (r->dict));
705 r->first_case = case_create (r->proto);
706 case_set_missing (r->first_case);
709 for ( i = 0 ; i < n_var_specs ; ++i )
711 const struct variable *var;
713 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
716 var = dict_get_var (r->dict, x++);
718 convert_xml_string_to_value (r->first_case, var,
719 var_spec[i].first_value);
722 for ( i = 0 ; i < n_var_specs ; ++i )
724 free (var_spec[i].first_value);
725 free (var_spec[i].name);
731 return casereader_create_sequential
735 &gnm_file_casereader_class, r);
739 for ( i = 0 ; i < n_var_specs ; ++i )
741 free (var_spec[i].first_value);
742 free (var_spec[i].name);
746 dict_destroy (spreadsheet->dict);
747 spreadsheet->dict = NULL;
749 gnm_file_casereader_destroy (NULL, r);
755 /* Reads and returns one case from READER's file. Returns a null
756 pointer on failure. */
757 static struct ccase *
758 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
763 struct gnumeric_reader *r = r_;
764 int current_row = r->row;
766 if ( !r->used_first_case )
768 r->used_first_case = true;
769 return r->first_case;
772 c = case_create (r->proto);
773 case_set_missing (c);
775 if (r->start_col == -1)
776 r->start_col = r->min_col;
778 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
779 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
783 if ( r->col < r->start_col || (r->stop_col != -1 &&
784 r->col > r->stop_col))
787 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
790 if ( r->stop_row != -1 && r->row > r->stop_row)
793 if ( r->node_type == XML_READER_TYPE_TEXT )
795 xmlChar *value = xmlTextReaderValue (r->xtr);
797 const int idx = r->col - r->start_col;
799 const struct variable *var = dict_get_var (r->dict, idx);
801 convert_xml_string_to_value (c, var, value);
818 #endif /* GNM_SUPPORT */