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 */
104 struct gnumeric_reader
106 struct spreadsheet spreadsheet;
108 xmlTextReaderPtr xtr;
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->sheets);
141 return gr->sheets[n].name;
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->sheets);
157 (gr->sheets[n].stop_col == -1)
159 (1 == (ret = xmlTextReaderRead (gr->xtr)))
166 return create_cell_ref (
167 gr->sheets[n].start_col,
168 gr->sheets[n].start_row,
169 gr->sheets[n].stop_col,
170 gr->sheets[n].stop_row);
175 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
178 struct gnumeric_reader *r = r_;
183 xmlFreeTextReader (r->xtr);
185 if ( ! r->used_first_case )
186 case_unref (r->first_case);
188 caseproto_unref (r->proto);
190 for (i = 0; i < r->spreadsheet.sheets; ++i)
192 xmlFree (r->sheets[i].name);
201 process_node (struct gnumeric_reader *r)
203 xmlChar *name = xmlTextReaderName (r->xtr);
205 name = xmlStrdup (_xml ("--"));
208 r->node_type = xmlTextReaderNodeType (r->xtr);
214 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
215 XML_READER_TYPE_ELEMENT == r->node_type)
217 r->state = STATE_SHEET_COUNT;
218 r->spreadsheet.sheets = 0;
222 case STATE_SHEET_COUNT:
223 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
224 XML_READER_TYPE_ELEMENT == r->node_type)
226 struct sheet_detail *sd ;
227 r->spreadsheet.sheets++;
228 r->sheets = xrealloc (r->sheets, r->spreadsheet.sheets * sizeof *r->sheets);
229 sd = &r->sheets[r->spreadsheet.sheets - 1];
230 sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -1;
232 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
233 XML_READER_TYPE_END_ELEMENT == r->node_type)
235 r->state = STATE_INIT;
237 else if (XML_READER_TYPE_TEXT == r->node_type)
239 r->sheets [r->spreadsheet.sheets - 1].name = xmlTextReaderValue (r->xtr);
244 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
245 XML_READER_TYPE_ELEMENT == r->node_type)
248 r->state = STATE_SHEET_START;
251 case STATE_SHEET_START:
252 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
253 XML_READER_TYPE_ELEMENT == r->node_type)
255 r->state = STATE_SHEET_NAME;
258 case STATE_SHEET_NAME:
259 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
260 XML_READER_TYPE_END_ELEMENT == r->node_type)
262 r->state = STATE_INIT;
264 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
265 XML_READER_TYPE_END_ELEMENT == r->node_type)
267 r->state = STATE_INIT;
269 else if (XML_READER_TYPE_TEXT == r->node_type)
271 #if MODEL_SHOULD_WORK
272 if ( r->target_sheet != NULL)
274 xmlChar *value = xmlTextReaderValue (r->xtr);
275 if ( 0 == xmlStrcmp (value, r->target_sheet))
276 r->state = STATE_SHEET_FOUND;
279 else if (r->target_sheet_index == r->sheet_index + 1)
281 r->state = STATE_SHEET_FOUND;
284 r->state = STATE_SHEET_FOUND;
288 case STATE_SHEET_FOUND:
289 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
290 XML_READER_TYPE_ELEMENT == r->node_type)
292 r->min_col = INT_MAX;
293 if (! xmlTextReaderIsEmptyElement (r->xtr))
294 r->state = STATE_CELLS_START;
296 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
297 XML_READER_TYPE_ELEMENT == r->node_type)
299 r->state = STATE_MAXROW;
301 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
302 XML_READER_TYPE_ELEMENT == r->node_type)
304 r->state = STATE_MAXCOL;
306 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
307 XML_READER_TYPE_END_ELEMENT == r->node_type)
309 r->state = STATE_INIT;
313 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
314 XML_READER_TYPE_END_ELEMENT == r->node_type)
316 r->state = STATE_SHEET_FOUND;
318 else if (r->node_type == XML_READER_TYPE_TEXT)
320 xmlChar *value = xmlTextReaderValue (r->xtr);
321 r->sheets[r->sheet_index].maxrow = _xmlchar_to_int (value);
326 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
327 XML_READER_TYPE_END_ELEMENT == r->node_type)
329 r->state = STATE_SHEET_FOUND;
331 else if (r->node_type == XML_READER_TYPE_TEXT)
333 xmlChar *value = xmlTextReaderValue (r->xtr);
334 r->sheets[r->sheet_index].maxcol = _xmlchar_to_int (value);
338 case STATE_CELLS_START:
339 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
340 XML_READER_TYPE_ELEMENT == r->node_type)
342 xmlChar *attr = NULL;
343 r->state = STATE_CELL;
345 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
346 r->col = _xmlchar_to_int (attr);
349 if (r->col < r->min_col)
352 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
353 r->row = _xmlchar_to_int (attr);
355 if (r->sheets[r->sheet_index].start_row == -1)
357 r->sheets[r->sheet_index].start_row = r->row;
360 if (r->sheets[r->sheet_index].start_col == -1)
362 r->sheets[r->sheet_index].start_col = r->col;
365 else if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
366 XML_READER_TYPE_END_ELEMENT == r->node_type)
368 r->sheets[r->sheet_index].stop_col = r->col;
369 r->sheets[r->sheet_index].stop_row = r->row;
370 r->state = STATE_SHEET_NAME;
374 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) &&
375 XML_READER_TYPE_END_ELEMENT == r->node_type)
377 r->state = STATE_CELLS_START;
389 Sets the VAR of case C, to the value corresponding to the xml string XV
392 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
395 union value *v = case_data_rw (c, var);
398 value_set_missing (v, var_get_width (var));
399 else if ( var_is_alpha (var))
400 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
403 const char *text = CHAR_CAST (const char *, xv);
407 v->f = c_strtod (text, &endptr);
408 if ( errno != 0 || endptr == text)
417 xmlChar *first_value;
422 gnumeric_destroy (struct spreadsheet *s)
424 struct gnumeric_reader *r = (struct gnumeric *) s;
425 gnm_file_casereader_destroy (NULL, s);
429 gnumeric_probe (const char *filename)
432 struct gnumeric_reader *r = NULL;
433 xmlTextReaderPtr xtr;
435 gzFile gz = gzopen (filename, "r");
440 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
441 (xmlInputCloseCallback) gzclose, gz,
447 r = xzalloc (sizeof *r);
450 r->spreadsheet.sheets = -1;
451 r->state = STATE_PRE_INIT;
454 /* Advance to the start of the workbook.
455 This gives us some confidence that we are actually dealing with a gnumeric
458 while ( (r->state != STATE_INIT )
459 && 1 == (ret = xmlTextReaderRead (r->xtr)))
466 /* Not a gnumeric spreadsheet */
472 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
473 r->spreadsheet.file_name = filename;
475 return &r->spreadsheet;
480 gnumeric_make_reader (struct spreadsheet *spreadsheet,
481 const struct spreadsheet_read_info *gri,
482 struct spreadsheet_read_options *opts)
484 struct gnumeric_reader *r = NULL;
485 unsigned long int vstart = 0;
487 casenumber n_cases = CASENUMBER_MAX;
489 struct var_spec *var_spec = NULL;
492 r = (struct gnumeric_reader *) (spreadsheet);
494 if ( opts->cell_range )
496 if ( ! convert_cell_ref (opts->cell_range,
497 &r->start_col, &r->start_row,
498 &r->stop_col, &r->stop_row))
500 msg (SE, _("Invalid cell range `%s'"),
513 r->target_sheet = BAD_CAST opts->sheet_name;
514 r->target_sheet_index = opts->sheet_index;
515 r->row = r->col = -1;
518 /* Advance to the start of the cells for the target sheet */
519 while ( (r->state != STATE_CELL || r->row < r->start_row )
520 && 1 == (ret = xmlTextReaderRead (r->xtr)))
524 value = xmlTextReaderValue (r->xtr);
526 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
528 n_cases = 1 + _xmlchar_to_int (value) ;
533 /* If a range has been given, then use that to calculate the number
535 if ( opts->cell_range)
537 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
540 if ( gri->read_names )
546 /* Read in the first row of cells,
547 including the headers if read_names was set */
549 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
550 && (ret = xmlTextReaderRead (r->xtr))
556 if ( r->row > r->start_row ) break;
558 if ( r->col < r->start_col ||
559 (r->stop_col != -1 && r->col > r->stop_col))
562 idx = r->col - r->start_col;
564 if ( idx >= n_var_specs )
567 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
568 for (i = n_var_specs; i <= idx; ++i)
570 var_spec [i].name = NULL;
571 var_spec [i].width = -1;
572 var_spec [i].first_value = NULL;
574 n_var_specs = idx + 1 ;
577 if ( r->node_type == XML_READER_TYPE_TEXT )
579 xmlChar *value = xmlTextReaderValue (r->xtr);
580 const char *text = CHAR_CAST (const char *, value);
582 if ( r->row < r->start_row)
584 if ( gri->read_names )
586 var_spec [idx].name = xstrdup (text);
591 var_spec [idx].first_value = xmlStrdup (value);
593 if (-1 == var_spec [idx].width )
594 var_spec [idx].width = (gri->asw == -1) ?
595 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw;
600 else if ( r->node_type == XML_READER_TYPE_ELEMENT
601 && r->state == STATE_CELL)
603 if ( r->row == r->start_row )
606 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
608 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
609 var_spec [idx].width = 0;
617 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
620 /* Create the dictionary and populate it */
621 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
624 for (i = 0 ; i < n_var_specs ; ++i )
628 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
631 /* Probably no data exists for this variable, so allocate a
633 if ( var_spec[i].width == -1 )
634 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
636 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
637 dict_create_var (r->dict, name, var_spec[i].width);
641 /* Create the first case, and cache it */
642 r->used_first_case = false;
644 if ( n_var_specs == 0 )
646 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
647 spreadsheet->file_name);
651 r->proto = caseproto_ref (dict_get_proto (r->dict));
652 r->first_case = case_create (r->proto);
653 case_set_missing (r->first_case);
656 for ( i = 0 ; i < n_var_specs ; ++i )
658 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
661 const struct variable *var = dict_get_var (r->dict, x++);
663 convert_xml_string_to_value (r->first_case, var,
664 var_spec[i].first_value);
667 for ( i = 0 ; i < n_var_specs ; ++i )
669 free (var_spec[i].first_value);
670 free (var_spec[i].name);
676 if (opts->cell_range == NULL)
678 opts->cell_range = c_xasprintf ("%c%d:%c%ld",
681 r->stop_col + 'A' + caseproto_get_n_widths (r->proto),
682 r->start_row + n_cases);
685 return casereader_create_sequential
689 &gnm_file_casereader_class, r);
693 for ( i = 0 ; i < n_var_specs ; ++i )
695 free (var_spec[i].first_value);
696 free (var_spec[i].name);
700 dict_destroy (spreadsheet->dict);
701 spreadsheet->dict = NULL;
703 gnm_file_casereader_destroy (NULL, r);
709 /* Reads and returns one case from READER's file. Returns a null
710 pointer on failure. */
711 static struct ccase *
712 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
717 struct gnumeric_reader *r = r_;
718 int current_row = r->row;
720 if ( !r->used_first_case )
722 r->used_first_case = true;
723 return r->first_case;
726 c = case_create (r->proto);
727 case_set_missing (c);
729 if (r->start_col == -1)
730 r->start_col = r->min_col;
732 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
733 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
737 if ( r->col < r->start_col || (r->stop_col != -1 &&
738 r->col > r->stop_col))
741 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
744 if ( r->stop_row != -1 && r->row > r->stop_row)
747 if ( r->node_type == XML_READER_TYPE_TEXT )
749 xmlChar *value = xmlTextReaderValue (r->xtr);
751 const int idx = r->col - r->start_col;
753 const struct variable *var = dict_get_var (r->dict, idx);
755 convert_xml_string_to_value (c, var, value);
772 #endif /* GNM_SUPPORT */