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"
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 */
104 struct gnumeric_reader
106 struct spreadsheet spreadsheet;
108 /* The libxml reader for this instance */
109 xmlTextReaderPtr xtr;
111 /* An internal state variable */
112 enum reader_state state;
125 struct sheet_detail *sheets;
127 const xmlChar *target_sheet;
128 int target_sheet_index;
130 struct caseproto *proto;
131 struct dictionary *dict;
132 struct ccase *first_case;
133 bool used_first_case;
138 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
140 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
141 assert (n < s->n_sheets);
143 return gr->sheets[n].name;
147 static void process_node (struct gnumeric_reader *r);
151 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
154 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
156 assert (n < s->n_sheets);
159 (gr->sheets[n].stop_col == -1)
161 (1 == (ret = xmlTextReaderRead (gr->xtr)))
167 return create_cell_ref (
168 gr->sheets[n].start_col,
169 gr->sheets[n].start_row,
170 gr->sheets[n].stop_col,
171 gr->sheets[n].stop_row);
176 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
179 struct gnumeric_reader *r = r_;
184 xmlFreeTextReader (r->xtr);
187 if ( ! r->used_first_case )
188 case_unref (r->first_case);
190 caseproto_unref (r->proto);
193 for (i = 0; i < r->spreadsheet.n_sheets; ++i)
195 xmlFree (r->sheets[i].name);
206 process_node (struct gnumeric_reader *r)
208 xmlChar *name = xmlTextReaderName (r->xtr);
210 name = xmlStrdup (_xml ("--"));
213 r->node_type = xmlTextReaderNodeType (r->xtr);
218 r->current_sheet = -1;
219 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
220 XML_READER_TYPE_ELEMENT == r->node_type)
222 r->state = STATE_SHEET_COUNT;
226 case STATE_SHEET_COUNT:
227 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
228 XML_READER_TYPE_ELEMENT == r->node_type)
231 if (r->current_sheet + 1 > r->spreadsheet.n_sheets)
233 struct sheet_detail *sd ;
234 r->sheets = xrealloc (r->sheets, (r->current_sheet + 1) * sizeof *r->sheets);
235 sd = &r->sheets[r->current_sheet];
236 sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -1;
237 r->spreadsheet.n_sheets = r->current_sheet + 1;
240 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
241 XML_READER_TYPE_END_ELEMENT == r->node_type)
243 r->state = STATE_INIT;
244 r->current_sheet = -1;
246 else if (XML_READER_TYPE_TEXT == r->node_type)
248 r->sheets [r->spreadsheet.n_sheets - 1].name = xmlTextReaderValue (r->xtr);
253 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
254 XML_READER_TYPE_ELEMENT == r->node_type)
257 r->state = STATE_SHEET_START;
260 case STATE_SHEET_START:
261 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
262 XML_READER_TYPE_ELEMENT == r->node_type)
264 r->state = STATE_SHEET_NAME;
267 case STATE_SHEET_NAME:
268 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
269 XML_READER_TYPE_END_ELEMENT == r->node_type)
271 r->state = STATE_INIT;
273 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
274 XML_READER_TYPE_END_ELEMENT == r->node_type)
276 r->state = STATE_INIT;
278 else if (XML_READER_TYPE_TEXT == r->node_type)
280 if ( r->target_sheet != NULL)
282 xmlChar *value = xmlTextReaderValue (r->xtr);
283 if ( 0 == xmlStrcmp (value, r->target_sheet))
284 r->state = STATE_SHEET_FOUND;
287 else if (r->target_sheet_index == r->current_sheet + 1)
289 r->state = STATE_SHEET_FOUND;
291 else if (r->target_sheet_index == -1)
293 r->state = STATE_SHEET_FOUND;
297 case STATE_SHEET_FOUND:
298 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
299 XML_READER_TYPE_ELEMENT == r->node_type)
301 r->min_col = INT_MAX;
302 if (! xmlTextReaderIsEmptyElement (r->xtr))
303 r->state = STATE_CELLS_START;
305 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
306 XML_READER_TYPE_ELEMENT == r->node_type)
308 r->state = STATE_MAXROW;
310 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
311 XML_READER_TYPE_ELEMENT == r->node_type)
313 r->state = STATE_MAXCOL;
315 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
316 XML_READER_TYPE_END_ELEMENT == r->node_type)
318 r->state = STATE_INIT;
322 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
323 XML_READER_TYPE_END_ELEMENT == r->node_type)
325 r->state = STATE_SHEET_FOUND;
327 else if (r->node_type == XML_READER_TYPE_TEXT)
329 xmlChar *value = xmlTextReaderValue (r->xtr);
330 r->sheets[r->current_sheet].maxrow = _xmlchar_to_int (value);
335 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
336 XML_READER_TYPE_END_ELEMENT == r->node_type)
338 r->state = STATE_SHEET_FOUND;
340 else if (r->node_type == XML_READER_TYPE_TEXT)
342 xmlChar *value = xmlTextReaderValue (r->xtr);
343 r->sheets[r->current_sheet].maxcol = _xmlchar_to_int (value);
347 case STATE_CELLS_START:
348 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
349 XML_READER_TYPE_ELEMENT == r->node_type)
351 xmlChar *attr = NULL;
353 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
354 r->col = _xmlchar_to_int (attr);
357 if (r->col < r->min_col)
360 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
361 r->row = _xmlchar_to_int (attr);
364 if (r->sheets[r->current_sheet].start_row == -1)
366 r->sheets[r->current_sheet].start_row = r->row;
369 if (r->sheets[r->current_sheet].start_col == -1)
371 r->sheets[r->current_sheet].start_col = r->col;
373 if (! xmlTextReaderIsEmptyElement (r->xtr))
374 r->state = STATE_CELL;
376 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == r->node_type) )
378 r->sheets[r->current_sheet].stop_col = r->col;
379 r->sheets[r->current_sheet].stop_row = r->row;
380 r->state = STATE_SHEET_NAME;
384 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == r->node_type)
386 r->state = STATE_CELLS_START;
398 Sets the VAR of case C, to the value corresponding to the xml string XV
401 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
404 union value *v = case_data_rw (c, var);
407 value_set_missing (v, var_get_width (var));
408 else if ( var_is_alpha (var))
409 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
412 const char *text = CHAR_CAST (const char *, xv);
416 v->f = c_strtod (text, &endptr);
417 if ( errno != 0 || endptr == text)
426 xmlChar *first_value;
431 gnumeric_destroy (struct spreadsheet *s)
433 struct gnumeric_reader *r = (struct gnumeric *) s;
434 gnm_file_casereader_destroy (NULL, s);
438 static struct gnumeric_reader *
439 gnumeric_reopen (struct gnumeric_reader *r, const char *filename)
443 xmlTextReaderPtr xtr;
446 assert (r == NULL || filename == NULL);
449 xmlFreeTextReader (r->xtr);
452 gz = gzopen (filename, "r");
454 gz = gzopen ( r->spreadsheet.file_name, "r");
459 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
460 (xmlInputCloseCallback) gzclose, gz,
471 r = xzalloc (sizeof *r);
472 r->spreadsheet.n_sheets = -1;
473 r->spreadsheet.file_name = filename;
476 r->target_sheet = NULL;
477 r->target_sheet_index = -1;
479 r->row = r->col = -1;
480 r->state = STATE_PRE_INIT;
483 /* Advance to the start of the workbook.
484 This gives us some confidence that we are actually dealing with a gnumeric
487 while ( (r->state != STATE_INIT )
488 && 1 == (ret = xmlTextReaderRead (r->xtr)))
493 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
500 gnumeric_probe (const char *filename)
502 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename);
504 return &r->spreadsheet;
509 gnumeric_make_reader (struct spreadsheet *spreadsheet,
510 const struct spreadsheet_read_info *gri,
511 struct spreadsheet_read_options *opts)
513 struct gnumeric_reader *r = NULL;
514 unsigned long int vstart = 0;
516 casenumber n_cases = CASENUMBER_MAX;
518 struct var_spec *var_spec = NULL;
521 r = (struct gnumeric_reader *) (spreadsheet);
524 r = gnumeric_reopen (r, NULL);
526 if ( opts->cell_range )
528 if ( ! convert_cell_ref (opts->cell_range,
529 &r->start_col, &r->start_row,
530 &r->stop_col, &r->stop_row))
532 msg (SE, _("Invalid cell range `%s'"),
545 r->target_sheet = BAD_CAST opts->sheet_name;
546 r->target_sheet_index = opts->sheet_index;
547 r->row = r->col = -1;
548 r->current_sheet = -1;
550 /* Advance to the start of the cells for the target sheet */
551 while ( (r->state != STATE_CELL || r->row < r->start_row )
552 && 1 == (ret = xmlTextReaderRead (r->xtr)))
556 value = xmlTextReaderValue (r->xtr);
558 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
560 n_cases = 1 + _xmlchar_to_int (value) ;
565 /* If a range has been given, then use that to calculate the number
567 if ( opts->cell_range)
569 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
572 if ( gri->read_names )
578 /* Read in the first row of cells,
579 including the headers if read_names was set */
581 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
582 && (ret = xmlTextReaderRead (r->xtr))
588 if ( r->row > r->start_row ) break;
590 if ( r->col < r->start_col ||
591 (r->stop_col != -1 && r->col > r->stop_col))
594 idx = r->col - r->start_col;
596 if ( idx >= n_var_specs )
599 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
600 for (i = n_var_specs; i <= idx; ++i)
602 var_spec [i].name = NULL;
603 var_spec [i].width = -1;
604 var_spec [i].first_value = NULL;
606 n_var_specs = idx + 1 ;
609 if ( r->node_type == XML_READER_TYPE_TEXT )
611 xmlChar *value = xmlTextReaderValue (r->xtr);
612 const char *text = CHAR_CAST (const char *, value);
614 if ( r->row < r->start_row)
616 if ( gri->read_names )
618 var_spec [idx].name = xstrdup (text);
623 var_spec [idx].first_value = xmlStrdup (value);
625 if (-1 == var_spec [idx].width )
626 var_spec [idx].width = (gri->asw == -1) ?
627 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw;
632 else if ( r->node_type == XML_READER_TYPE_ELEMENT
633 && r->state == STATE_CELL)
635 if ( r->row == r->start_row )
638 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
640 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
641 var_spec [idx].width = 0;
649 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
652 /* Create the dictionary and populate it */
653 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
656 for (i = 0 ; i < n_var_specs ; ++i )
660 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
663 /* Probably no data exists for this variable, so allocate a
665 if ( var_spec[i].width == -1 )
666 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
668 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
669 dict_create_var (r->dict, name, var_spec[i].width);
673 /* Create the first case, and cache it */
674 r->used_first_case = false;
676 if ( n_var_specs == 0 )
678 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
679 spreadsheet->file_name);
683 r->proto = caseproto_ref (dict_get_proto (r->dict));
684 r->first_case = case_create (r->proto);
685 case_set_missing (r->first_case);
688 for ( i = 0 ; i < n_var_specs ; ++i )
690 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
693 const struct variable *var = dict_get_var (r->dict, x++);
695 convert_xml_string_to_value (r->first_case, var,
696 var_spec[i].first_value);
699 for ( i = 0 ; i < n_var_specs ; ++i )
701 free (var_spec[i].first_value);
702 free (var_spec[i].name);
708 if (opts->cell_range == NULL)
710 opts->cell_range = c_xasprintf ("%c%d:%c%ld",
713 r->stop_col + 'A' + caseproto_get_n_widths (r->proto),
714 r->start_row + n_cases);
717 return casereader_create_sequential
721 &gnm_file_casereader_class, r);
725 for ( i = 0 ; i < n_var_specs ; ++i )
727 free (var_spec[i].first_value);
728 free (var_spec[i].name);
732 dict_destroy (spreadsheet->dict);
733 spreadsheet->dict = NULL;
735 gnm_file_casereader_destroy (NULL, r);
741 /* Reads and returns one case from READER's file. Returns a null
742 pointer on failure. */
743 static struct ccase *
744 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
749 struct gnumeric_reader *r = r_;
750 int current_row = r->row;
752 if ( !r->used_first_case )
754 r->used_first_case = true;
755 return r->first_case;
758 c = case_create (r->proto);
759 case_set_missing (c);
761 if (r->start_col == -1)
762 r->start_col = r->min_col;
764 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
765 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
769 if ( r->col < r->start_col || (r->stop_col != -1 &&
770 r->col > r->stop_col))
773 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
776 if ( r->stop_row != -1 && r->row > r->stop_row)
779 if ( r->node_type == XML_READER_TYPE_TEXT )
781 xmlChar *value = xmlTextReaderValue (r->xtr);
783 const int idx = r->col - r->start_col;
785 const struct variable *var = dict_get_var (r->dict, idx);
787 convert_xml_string_to_value (c, var, value);
804 #endif /* GNM_SUPPORT */