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->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->sheets);
159 (gr->sheets[n].stop_col == -1)
161 (1 == (ret = xmlTextReaderRead (gr->xtr)))
168 return create_cell_ref (
169 gr->sheets[n].start_col,
170 gr->sheets[n].start_row,
171 gr->sheets[n].stop_col,
172 gr->sheets[n].stop_row);
177 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
180 struct gnumeric_reader *r = r_;
185 xmlFreeTextReader (r->xtr);
187 if ( ! r->used_first_case )
188 case_unref (r->first_case);
190 caseproto_unref (r->proto);
192 for (i = 0; i < r->spreadsheet.sheets; ++i)
194 xmlFree (r->sheets[i].name);
203 process_node (struct gnumeric_reader *r)
205 xmlChar *name = xmlTextReaderName (r->xtr);
207 name = xmlStrdup (_xml ("--"));
210 r->node_type = xmlTextReaderNodeType (r->xtr);
216 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
217 XML_READER_TYPE_ELEMENT == r->node_type)
219 r->state = STATE_SHEET_COUNT;
220 r->spreadsheet.sheets = 0;
224 case STATE_SHEET_COUNT:
225 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
226 XML_READER_TYPE_ELEMENT == r->node_type)
228 struct sheet_detail *sd ;
229 r->spreadsheet.sheets++;
230 r->sheets = xrealloc (r->sheets, r->spreadsheet.sheets * sizeof *r->sheets);
231 sd = &r->sheets[r->spreadsheet.sheets - 1];
232 sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -1;
234 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
235 XML_READER_TYPE_END_ELEMENT == r->node_type)
237 r->state = STATE_INIT;
239 else if (XML_READER_TYPE_TEXT == r->node_type)
241 r->sheets [r->spreadsheet.sheets - 1].name = xmlTextReaderValue (r->xtr);
246 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
247 XML_READER_TYPE_ELEMENT == r->node_type)
250 r->state = STATE_SHEET_START;
253 case STATE_SHEET_START:
254 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
255 XML_READER_TYPE_ELEMENT == r->node_type)
257 r->state = STATE_SHEET_NAME;
260 case STATE_SHEET_NAME:
261 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
262 XML_READER_TYPE_END_ELEMENT == r->node_type)
264 r->state = STATE_INIT;
266 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
267 XML_READER_TYPE_END_ELEMENT == r->node_type)
269 r->state = STATE_INIT;
271 else if (XML_READER_TYPE_TEXT == r->node_type)
273 if ( r->target_sheet != NULL)
275 xmlChar *value = xmlTextReaderValue (r->xtr);
276 if ( 0 == xmlStrcmp (value, r->target_sheet))
277 r->state = STATE_SHEET_FOUND;
280 else if (r->target_sheet_index == r->sheet_index + 1)
282 r->state = STATE_SHEET_FOUND;
284 else if (r->target_sheet_index == -1)
286 r->state = STATE_SHEET_FOUND;
290 case STATE_SHEET_FOUND:
291 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
292 XML_READER_TYPE_ELEMENT == r->node_type)
294 r->min_col = INT_MAX;
295 if (! xmlTextReaderIsEmptyElement (r->xtr))
296 r->state = STATE_CELLS_START;
298 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
299 XML_READER_TYPE_ELEMENT == r->node_type)
301 r->state = STATE_MAXROW;
303 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
304 XML_READER_TYPE_ELEMENT == r->node_type)
306 r->state = STATE_MAXCOL;
308 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
309 XML_READER_TYPE_END_ELEMENT == r->node_type)
311 r->state = STATE_INIT;
315 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
316 XML_READER_TYPE_END_ELEMENT == r->node_type)
318 r->state = STATE_SHEET_FOUND;
320 else if (r->node_type == XML_READER_TYPE_TEXT)
322 xmlChar *value = xmlTextReaderValue (r->xtr);
323 r->sheets[r->sheet_index].maxrow = _xmlchar_to_int (value);
328 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
329 XML_READER_TYPE_END_ELEMENT == r->node_type)
331 r->state = STATE_SHEET_FOUND;
333 else if (r->node_type == XML_READER_TYPE_TEXT)
335 xmlChar *value = xmlTextReaderValue (r->xtr);
336 r->sheets[r->sheet_index].maxcol = _xmlchar_to_int (value);
340 case STATE_CELLS_START:
341 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
342 XML_READER_TYPE_ELEMENT == r->node_type)
344 xmlChar *attr = NULL;
345 r->state = STATE_CELL;
347 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
348 r->col = _xmlchar_to_int (attr);
351 if (r->col < r->min_col)
354 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
355 r->row = _xmlchar_to_int (attr);
357 if (r->sheets[r->sheet_index].start_row == -1)
359 r->sheets[r->sheet_index].start_row = r->row;
362 if (r->sheets[r->sheet_index].start_col == -1)
364 r->sheets[r->sheet_index].start_col = r->col;
367 else if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
368 XML_READER_TYPE_END_ELEMENT == r->node_type)
370 r->sheets[r->sheet_index].stop_col = r->col;
371 r->sheets[r->sheet_index].stop_row = r->row;
372 r->state = STATE_SHEET_NAME;
376 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) &&
377 XML_READER_TYPE_END_ELEMENT == r->node_type)
379 r->state = STATE_CELLS_START;
391 Sets the VAR of case C, to the value corresponding to the xml string XV
394 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
397 union value *v = case_data_rw (c, var);
400 value_set_missing (v, var_get_width (var));
401 else if ( var_is_alpha (var))
402 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
405 const char *text = CHAR_CAST (const char *, xv);
409 v->f = c_strtod (text, &endptr);
410 if ( errno != 0 || endptr == text)
419 xmlChar *first_value;
424 gnumeric_destroy (struct spreadsheet *s)
426 struct gnumeric_reader *r = (struct gnumeric *) s;
427 gnm_file_casereader_destroy (NULL, s);
431 static struct gnumeric_reader *
432 gnumeric_reopen (struct gnumeric_reader *r, const char *filename)
436 xmlTextReaderPtr xtr;
439 assert (r == NULL || filename == NULL);
442 xmlFreeTextReader (r->xtr);
445 gz = gzopen (filename, "r");
447 gz = gzopen ( r->spreadsheet.file_name, "r");
452 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
453 (xmlInputCloseCallback) gzclose, gz,
464 r = xzalloc (sizeof *r);
465 r->spreadsheet.sheets = -1;
466 r->spreadsheet.file_name = filename;
469 r->target_sheet = NULL;
470 r->target_sheet_index = -1;
472 r->row = r->col = -1;
473 r->state = STATE_PRE_INIT;
476 /* Advance to the start of the workbook.
477 This gives us some confidence that we are actually dealing with a gnumeric
480 while ( (r->state != STATE_INIT )
481 && 1 == (ret = xmlTextReaderRead (r->xtr)))
486 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
493 gnumeric_probe (const char *filename)
495 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename);
497 return &r->spreadsheet;
502 gnumeric_make_reader (struct spreadsheet *spreadsheet,
503 const struct spreadsheet_read_info *gri,
504 struct spreadsheet_read_options *opts)
506 struct gnumeric_reader *r = NULL;
507 unsigned long int vstart = 0;
509 casenumber n_cases = CASENUMBER_MAX;
511 struct var_spec *var_spec = NULL;
514 r = (struct gnumeric_reader *) (spreadsheet);
517 r = gnumeric_reopen (r, NULL);
519 if ( opts->cell_range )
521 if ( ! convert_cell_ref (opts->cell_range,
522 &r->start_col, &r->start_row,
523 &r->stop_col, &r->stop_row))
525 msg (SE, _("Invalid cell range `%s'"),
538 r->target_sheet = BAD_CAST opts->sheet_name;
539 r->target_sheet_index = opts->sheet_index;
540 r->row = r->col = -1;
543 /* Advance to the start of the cells for the target sheet */
544 while ( (r->state != STATE_CELL || r->row < r->start_row )
545 && 1 == (ret = xmlTextReaderRead (r->xtr)))
549 value = xmlTextReaderValue (r->xtr);
551 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
553 n_cases = 1 + _xmlchar_to_int (value) ;
558 /* If a range has been given, then use that to calculate the number
560 if ( opts->cell_range)
562 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
565 if ( gri->read_names )
571 /* Read in the first row of cells,
572 including the headers if read_names was set */
574 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
575 && (ret = xmlTextReaderRead (r->xtr))
581 if ( r->row > r->start_row ) break;
583 if ( r->col < r->start_col ||
584 (r->stop_col != -1 && r->col > r->stop_col))
587 idx = r->col - r->start_col;
589 if ( idx >= n_var_specs )
592 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
593 for (i = n_var_specs; i <= idx; ++i)
595 var_spec [i].name = NULL;
596 var_spec [i].width = -1;
597 var_spec [i].first_value = NULL;
599 n_var_specs = idx + 1 ;
602 if ( r->node_type == XML_READER_TYPE_TEXT )
604 xmlChar *value = xmlTextReaderValue (r->xtr);
605 const char *text = CHAR_CAST (const char *, value);
607 if ( r->row < r->start_row)
609 if ( gri->read_names )
611 var_spec [idx].name = xstrdup (text);
616 var_spec [idx].first_value = xmlStrdup (value);
618 if (-1 == var_spec [idx].width )
619 var_spec [idx].width = (gri->asw == -1) ?
620 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw;
625 else if ( r->node_type == XML_READER_TYPE_ELEMENT
626 && r->state == STATE_CELL)
628 if ( r->row == r->start_row )
631 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
633 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
634 var_spec [idx].width = 0;
642 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
645 /* Create the dictionary and populate it */
646 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
649 for (i = 0 ; i < n_var_specs ; ++i )
653 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
656 /* Probably no data exists for this variable, so allocate a
658 if ( var_spec[i].width == -1 )
659 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
661 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
662 dict_create_var (r->dict, name, var_spec[i].width);
666 /* Create the first case, and cache it */
667 r->used_first_case = false;
669 if ( n_var_specs == 0 )
671 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
672 spreadsheet->file_name);
676 r->proto = caseproto_ref (dict_get_proto (r->dict));
677 r->first_case = case_create (r->proto);
678 case_set_missing (r->first_case);
681 for ( i = 0 ; i < n_var_specs ; ++i )
683 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
686 const struct variable *var = dict_get_var (r->dict, x++);
688 convert_xml_string_to_value (r->first_case, var,
689 var_spec[i].first_value);
692 for ( i = 0 ; i < n_var_specs ; ++i )
694 free (var_spec[i].first_value);
695 free (var_spec[i].name);
701 if (opts->cell_range == NULL)
703 opts->cell_range = c_xasprintf ("%c%d:%c%ld",
706 r->stop_col + 'A' + caseproto_get_n_widths (r->proto),
707 r->start_row + n_cases);
710 return casereader_create_sequential
714 &gnm_file_casereader_class, r);
718 for ( i = 0 ; i < n_var_specs ; ++i )
720 free (var_spec[i].first_value);
721 free (var_spec[i].name);
725 dict_destroy (spreadsheet->dict);
726 spreadsheet->dict = NULL;
728 gnm_file_casereader_destroy (NULL, r);
734 /* Reads and returns one case from READER's file. Returns a null
735 pointer on failure. */
736 static struct ccase *
737 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
742 struct gnumeric_reader *r = r_;
743 int current_row = r->row;
745 if ( !r->used_first_case )
747 r->used_first_case = true;
748 return r->first_case;
751 c = case_create (r->proto);
752 case_set_missing (c);
754 if (r->start_col == -1)
755 r->start_col = r->min_col;
757 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
758 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
762 if ( r->col < r->start_col || (r->stop_col != -1 &&
763 r->col > r->stop_col))
766 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
769 if ( r->stop_row != -1 && r->row > r->stop_row)
772 if ( r->node_type == XML_READER_TYPE_TEXT )
774 xmlChar *value = xmlTextReaderValue (r->xtr);
776 const int idx = r->col - r->start_col;
778 const struct variable *var = dict_get_var (r->dict, idx);
780 convert_xml_string_to_value (c, var, value);
797 #endif /* GNM_SUPPORT */