1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2007, 2009, 2010, 2011, 2012 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"
31 #include "c-xvasprintf.h"
36 gnumeric_open_reader (struct spreadsheet_read_info *gri, struct spreadsheet_read_options *opts, struct dictionary **dict)
38 msg (ME, _("Support for %s files was not compiled into this installation of PSPP"), "Gnumeric");
45 #include "data/gnumeric-reader.h"
50 #include <libxml/xmlreader.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"
64 static void gnm_file_casereader_destroy (struct casereader *, void *);
66 static struct ccase *gnm_file_casereader_read (struct casereader *, void *);
68 static const struct casereader_class gnm_file_casereader_class =
70 gnm_file_casereader_read,
71 gnm_file_casereader_destroy,
78 STATE_PRE_INIT = 0, /* Initial state */
79 STATE_SHEET_COUNT, /* Found the sheet index */
80 STATE_INIT , /* Other Initial state */
81 STATE_SHEET_START, /* Found the start of a sheet */
82 STATE_SHEET_NAME, /* Found the sheet name */
85 STATE_SHEET_FOUND, /* Found the sheet that we actually want */
86 STATE_CELLS_START, /* Found the start of the cell array */
87 STATE_CELL /* Found a cell */
106 struct gnumeric_reader
108 struct spreadsheet spreadsheet;
110 xmlTextReaderPtr xtr;
113 enum reader_state state;
126 struct sheet_detail *sheets;
128 const xmlChar *target_sheet;
129 int target_sheet_index;
131 struct caseproto *proto;
132 struct dictionary *dict;
133 struct ccase *first_case;
134 bool used_first_case;
139 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
141 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
142 assert (n < s->sheets);
144 return gr->sheets[n].name;
148 static void process_node (struct gnumeric_reader *r);
152 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
155 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
157 assert (n < s->sheets);
160 (gr->sheets[n].stop_col == -1)
162 (1 == (ret = xmlTextReaderRead (gr->xtr)))
169 return create_cell_ref (
170 gr->sheets[n].start_col,
171 gr->sheets[n].start_row,
172 gr->sheets[n].stop_col,
173 gr->sheets[n].stop_row);
178 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
181 struct gnumeric_reader *r = r_;
186 xmlFreeTextReader (r->xtr);
188 if ( ! r->used_first_case )
189 case_unref (r->first_case);
191 caseproto_unref (r->proto);
193 for (i = 0; i < r->spreadsheet.sheets; ++i)
195 xmlFree (r->sheets[i].name);
204 process_node (struct gnumeric_reader *r)
206 xmlChar *name = xmlTextReaderName (r->xtr);
208 name = xmlStrdup (_xml ("--"));
211 r->node_type = xmlTextReaderNodeType (r->xtr);
217 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
218 XML_READER_TYPE_ELEMENT == r->node_type)
220 r->state = STATE_SHEET_COUNT;
221 r->spreadsheet.sheets = 0;
225 case STATE_SHEET_COUNT:
226 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
227 XML_READER_TYPE_ELEMENT == r->node_type)
229 struct sheet_detail *sd ;
230 r->spreadsheet.sheets++;
231 r->sheets = xrealloc (r->sheets, r->spreadsheet.sheets * sizeof *r->sheets);
232 sd = &r->sheets[r->spreadsheet.sheets - 1];
233 sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -1;
236 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
237 XML_READER_TYPE_END_ELEMENT == r->node_type)
239 r->state = STATE_INIT;
241 else if (XML_READER_TYPE_TEXT == r->node_type)
243 r->sheets [r->spreadsheet.sheets - 1].name = xmlTextReaderValue (r->xtr);
248 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
249 XML_READER_TYPE_ELEMENT == r->node_type)
252 r->state = STATE_SHEET_START;
253 r->sheets[r->sheet_index].offset = gztell (r->gz);
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->sheet_index + 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->sheet_index].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->sheet_index].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;
348 r->state = STATE_CELL;
350 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
351 r->col = _xmlchar_to_int (attr);
354 if (r->col < r->min_col)
357 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
358 r->row = _xmlchar_to_int (attr);
360 if (r->sheets[r->sheet_index].start_row == -1)
362 r->sheets[r->sheet_index].start_row = r->row;
365 if (r->sheets[r->sheet_index].start_col == -1)
367 r->sheets[r->sheet_index].start_col = r->col;
370 else if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
371 XML_READER_TYPE_END_ELEMENT == r->node_type)
373 r->sheets[r->sheet_index].stop_col = r->col;
374 r->sheets[r->sheet_index].stop_row = r->row;
375 r->state = STATE_SHEET_NAME;
379 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) &&
380 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 struct gnumeric_reader *r = (struct gnumeric *) s;
430 gnm_file_casereader_destroy (NULL, s);
434 gnumeric_probe (const char *filename)
437 struct gnumeric_reader *r = NULL;
438 xmlTextReaderPtr xtr;
441 gz = gzopen (filename, "r");
446 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
447 (xmlInputCloseCallback) gzclose, gz,
456 r = xzalloc (sizeof *r);
460 r->spreadsheet.sheets = -1;
461 r->state = STATE_PRE_INIT;
463 r->target_sheet = NULL;
464 r->target_sheet_index = -1;
467 /* Advance to the start of the workbook.
468 This gives us some confidence that we are actually dealing with a gnumeric
471 while ( (r->state != STATE_INIT )
472 && 1 == (ret = xmlTextReaderRead (r->xtr)))
479 /* Not a gnumeric spreadsheet */
480 xmlFreeTextReader (r->xtr);
485 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
486 r->spreadsheet.file_name = filename;
488 return &r->spreadsheet;
493 gnumeric_make_reader (struct spreadsheet *spreadsheet,
494 const struct spreadsheet_read_info *gri,
495 struct spreadsheet_read_options *opts)
497 struct gnumeric_reader *r = NULL;
498 unsigned long int vstart = 0;
500 casenumber n_cases = CASENUMBER_MAX;
502 struct var_spec *var_spec = NULL;
505 r = (struct gnumeric_reader *) (spreadsheet);
507 if ( opts->cell_range )
509 if ( ! convert_cell_ref (opts->cell_range,
510 &r->start_col, &r->start_row,
511 &r->stop_col, &r->stop_row))
513 msg (SE, _("Invalid cell range `%s'"),
526 r->target_sheet = BAD_CAST opts->sheet_name;
527 r->target_sheet_index = opts->sheet_index;
528 r->row = r->col = -1;
531 /* Advance to the start of the cells for the target sheet */
532 while ( (r->state != STATE_CELL || r->row < r->start_row )
533 && 1 == (ret = xmlTextReaderRead (r->xtr)))
537 value = xmlTextReaderValue (r->xtr);
539 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
541 n_cases = 1 + _xmlchar_to_int (value) ;
546 /* If a range has been given, then use that to calculate the number
548 if ( opts->cell_range)
550 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
553 if ( gri->read_names )
559 /* Read in the first row of cells,
560 including the headers if read_names was set */
562 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
563 && (ret = xmlTextReaderRead (r->xtr))
569 if ( r->row > r->start_row ) break;
571 if ( r->col < r->start_col ||
572 (r->stop_col != -1 && r->col > r->stop_col))
575 idx = r->col - r->start_col;
577 if ( idx >= n_var_specs )
580 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
581 for (i = n_var_specs; i <= idx; ++i)
583 var_spec [i].name = NULL;
584 var_spec [i].width = -1;
585 var_spec [i].first_value = NULL;
587 n_var_specs = idx + 1 ;
590 if ( r->node_type == XML_READER_TYPE_TEXT )
592 xmlChar *value = xmlTextReaderValue (r->xtr);
593 const char *text = CHAR_CAST (const char *, value);
595 if ( r->row < r->start_row)
597 if ( gri->read_names )
599 var_spec [idx].name = xstrdup (text);
604 var_spec [idx].first_value = xmlStrdup (value);
606 if (-1 == var_spec [idx].width )
607 var_spec [idx].width = (gri->asw == -1) ?
608 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw;
613 else if ( r->node_type == XML_READER_TYPE_ELEMENT
614 && r->state == STATE_CELL)
616 if ( r->row == r->start_row )
619 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
621 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
622 var_spec [idx].width = 0;
630 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
633 /* Create the dictionary and populate it */
634 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
637 for (i = 0 ; i < n_var_specs ; ++i )
641 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
644 /* Probably no data exists for this variable, so allocate a
646 if ( var_spec[i].width == -1 )
647 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
649 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
650 dict_create_var (r->dict, name, var_spec[i].width);
654 /* Create the first case, and cache it */
655 r->used_first_case = false;
657 if ( n_var_specs == 0 )
659 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
660 spreadsheet->file_name);
664 r->proto = caseproto_ref (dict_get_proto (r->dict));
665 r->first_case = case_create (r->proto);
666 case_set_missing (r->first_case);
669 for ( i = 0 ; i < n_var_specs ; ++i )
671 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
674 const struct variable *var = dict_get_var (r->dict, x++);
676 convert_xml_string_to_value (r->first_case, var,
677 var_spec[i].first_value);
680 for ( i = 0 ; i < n_var_specs ; ++i )
682 free (var_spec[i].first_value);
683 free (var_spec[i].name);
689 if (opts->cell_range == NULL)
691 opts->cell_range = c_xasprintf ("%c%d:%c%ld",
694 r->stop_col + 'A' + caseproto_get_n_widths (r->proto),
695 r->start_row + n_cases);
698 return casereader_create_sequential
702 &gnm_file_casereader_class, r);
706 for ( i = 0 ; i < n_var_specs ; ++i )
708 free (var_spec[i].first_value);
709 free (var_spec[i].name);
713 dict_destroy (spreadsheet->dict);
714 spreadsheet->dict = NULL;
716 gnm_file_casereader_destroy (NULL, r);
722 /* Reads and returns one case from READER's file. Returns a null
723 pointer on failure. */
724 static struct ccase *
725 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
730 struct gnumeric_reader *r = r_;
731 int current_row = r->row;
733 if ( !r->used_first_case )
735 r->used_first_case = true;
736 return r->first_case;
739 c = case_create (r->proto);
740 case_set_missing (c);
742 if (r->start_col == -1)
743 r->start_col = r->min_col;
745 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
746 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
750 if ( r->col < r->start_col || (r->stop_col != -1 &&
751 r->col > r->stop_col))
754 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
757 if ( r->stop_row != -1 && r->row > r->stop_row)
760 if ( r->node_type == XML_READER_TYPE_TEXT )
762 xmlChar *value = xmlTextReaderValue (r->xtr);
764 const int idx = r->col - r->start_col;
766 const struct variable *var = dict_get_var (r->dict, idx);
768 convert_xml_string_to_value (c, var, value);
785 #endif /* GNM_SUPPORT */