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 ( r->target_sheet != NULL)
273 xmlChar *value = xmlTextReaderValue (r->xtr);
274 if ( 0 == xmlStrcmp (value, r->target_sheet))
275 r->state = STATE_SHEET_FOUND;
278 else if (r->target_sheet_index == r->sheet_index + 1)
280 r->state = STATE_SHEET_FOUND;
284 case STATE_SHEET_FOUND:
285 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
286 XML_READER_TYPE_ELEMENT == r->node_type)
288 r->min_col = INT_MAX;
289 if (! xmlTextReaderIsEmptyElement (r->xtr))
290 r->state = STATE_CELLS_START;
292 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
293 XML_READER_TYPE_ELEMENT == r->node_type)
295 r->state = STATE_MAXROW;
297 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
298 XML_READER_TYPE_ELEMENT == r->node_type)
300 r->state = STATE_MAXCOL;
302 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
303 XML_READER_TYPE_END_ELEMENT == r->node_type)
305 r->state = STATE_INIT;
309 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
310 XML_READER_TYPE_END_ELEMENT == r->node_type)
312 r->state = STATE_SHEET_FOUND;
314 else if (r->node_type == XML_READER_TYPE_TEXT)
316 xmlChar *value = xmlTextReaderValue (r->xtr);
317 r->sheets[r->sheet_index].maxrow = _xmlchar_to_int (value);
322 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
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->sheet_index].maxcol = _xmlchar_to_int (value);
334 case STATE_CELLS_START:
335 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
336 XML_READER_TYPE_ELEMENT == r->node_type)
338 xmlChar *attr = NULL;
339 r->state = STATE_CELL;
341 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
342 r->col = _xmlchar_to_int (attr);
345 if (r->col < r->min_col)
348 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
349 r->row = _xmlchar_to_int (attr);
351 if (r->sheets[r->sheet_index].start_row == -1)
353 r->sheets[r->sheet_index].start_row = r->row;
356 if (r->sheets[r->sheet_index].start_col == -1)
358 r->sheets[r->sheet_index].start_col = r->col;
361 else if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
362 XML_READER_TYPE_END_ELEMENT == r->node_type)
364 r->sheets[r->sheet_index].stop_col = r->col;
365 r->sheets[r->sheet_index].stop_row = r->row;
366 r->state = STATE_SHEET_NAME;
370 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) &&
371 XML_READER_TYPE_END_ELEMENT == r->node_type)
373 r->state = STATE_CELLS_START;
385 Sets the VAR of case C, to the value corresponding to the xml string XV
388 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
391 union value *v = case_data_rw (c, var);
394 value_set_missing (v, var_get_width (var));
395 else if ( var_is_alpha (var))
396 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
399 const char *text = CHAR_CAST (const char *, xv);
403 v->f = c_strtod (text, &endptr);
404 if ( errno != 0 || endptr == text)
413 xmlChar *first_value;
418 gnumeric_destroy (struct spreadsheet *s)
420 struct gnumeric_reader *r = (struct gnumeric *) s;
421 gnm_file_casereader_destroy (NULL, s);
425 gnumeric_probe (const char *filename)
428 struct gnumeric_reader *r = NULL;
429 xmlTextReaderPtr xtr;
431 gzFile gz = gzopen (filename, "r");
436 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
437 (xmlInputCloseCallback) gzclose, gz,
443 r = xzalloc (sizeof *r);
446 r->spreadsheet.sheets = -1;
447 r->state = STATE_PRE_INIT;
450 /* Advance to the start of the workbook.
451 This gives us some confidence that we are actually dealing with a gnumeric
454 while ( (r->state != STATE_INIT )
455 && 1 == (ret = xmlTextReaderRead (r->xtr)))
462 /* Not a gnumeric spreadsheet */
468 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
469 r->spreadsheet.file_name = filename;
471 return &r->spreadsheet;
476 gnumeric_make_reader (struct spreadsheet *spreadsheet,
477 const struct spreadsheet_read_info *gri,
478 struct spreadsheet_read_options *opts)
480 struct gnumeric_reader *r = NULL;
481 unsigned long int vstart = 0;
483 casenumber n_cases = CASENUMBER_MAX;
485 struct var_spec *var_spec = NULL;
488 r = (struct gnumeric_reader *) (spreadsheet);
490 if ( opts->cell_range )
492 if ( ! convert_cell_ref (opts->cell_range,
493 &r->start_col, &r->start_row,
494 &r->stop_col, &r->stop_row))
496 msg (SE, _("Invalid cell range `%s'"),
509 r->target_sheet = BAD_CAST opts->sheet_name;
510 r->target_sheet_index = opts->sheet_index;
511 r->row = r->col = -1;
514 /* Advance to the start of the cells for the target sheet */
515 while ( (r->state != STATE_CELL || r->row < r->start_row )
516 && 1 == (ret = xmlTextReaderRead (r->xtr)))
520 value = xmlTextReaderValue (r->xtr);
522 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
524 n_cases = 1 + _xmlchar_to_int (value) ;
529 /* If a range has been given, then use that to calculate the number
531 if ( opts->cell_range)
533 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
536 if ( gri->read_names )
542 /* Read in the first row of cells,
543 including the headers if read_names was set */
545 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
546 && (ret = xmlTextReaderRead (r->xtr))
552 if ( r->row > r->start_row ) break;
554 if ( r->col < r->start_col ||
555 (r->stop_col != -1 && r->col > r->stop_col))
558 idx = r->col - r->start_col;
560 if ( idx >= n_var_specs )
563 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
564 for (i = n_var_specs; i <= idx; ++i)
566 var_spec [i].name = NULL;
567 var_spec [i].width = -1;
568 var_spec [i].first_value = NULL;
570 n_var_specs = idx + 1 ;
573 if ( r->node_type == XML_READER_TYPE_TEXT )
575 xmlChar *value = xmlTextReaderValue (r->xtr);
576 const char *text = CHAR_CAST (const char *, value);
578 if ( r->row < r->start_row)
580 if ( gri->read_names )
582 var_spec [idx].name = xstrdup (text);
587 var_spec [idx].first_value = xmlStrdup (value);
589 if (-1 == var_spec [idx].width )
590 var_spec [idx].width = (gri->asw == -1) ?
591 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw;
596 else if ( r->node_type == XML_READER_TYPE_ELEMENT
597 && r->state == STATE_CELL)
599 if ( r->row == r->start_row )
602 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
604 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
605 var_spec [idx].width = 0;
613 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
616 /* Create the dictionary and populate it */
617 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
620 for (i = 0 ; i < n_var_specs ; ++i )
624 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
627 /* Probably no data exists for this variable, so allocate a
629 if ( var_spec[i].width == -1 )
630 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
632 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
633 dict_create_var (r->dict, name, var_spec[i].width);
637 /* Create the first case, and cache it */
638 r->used_first_case = false;
640 if ( n_var_specs == 0 )
642 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
643 spreadsheet->file_name);
647 r->proto = caseproto_ref (dict_get_proto (r->dict));
648 r->first_case = case_create (r->proto);
649 case_set_missing (r->first_case);
652 for ( i = 0 ; i < n_var_specs ; ++i )
654 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
657 const struct variable *var = dict_get_var (r->dict, x++);
659 convert_xml_string_to_value (r->first_case, var,
660 var_spec[i].first_value);
663 for ( i = 0 ; i < n_var_specs ; ++i )
665 free (var_spec[i].first_value);
666 free (var_spec[i].name);
672 if (opts->cell_range == NULL)
674 opts->cell_range = c_xasprintf ("%c%d:%c%ld",
677 r->stop_col + 'A' + caseproto_get_n_widths (r->proto),
678 r->start_row + n_cases);
681 return casereader_create_sequential
685 &gnm_file_casereader_class, r);
689 for ( i = 0 ; i < n_var_specs ; ++i )
691 free (var_spec[i].first_value);
692 free (var_spec[i].name);
696 dict_destroy (spreadsheet->dict);
697 spreadsheet->dict = NULL;
699 gnm_file_casereader_destroy (NULL, r);
705 /* Reads and returns one case from READER's file. Returns a null
706 pointer on failure. */
707 static struct ccase *
708 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
713 struct gnumeric_reader *r = r_;
714 int current_row = r->row;
716 if ( !r->used_first_case )
718 r->used_first_case = true;
719 return r->first_case;
722 c = case_create (r->proto);
723 case_set_missing (c);
725 if (r->start_col == -1)
726 r->start_col = r->min_col;
728 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
729 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
733 if ( r->col < r->start_col || (r->stop_col != -1 &&
734 r->col > r->stop_col))
737 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
740 if ( r->stop_row != -1 && r->row > r->stop_row)
743 if ( r->node_type == XML_READER_TYPE_TEXT )
745 xmlChar *value = xmlTextReaderValue (r->xtr);
747 const int idx = r->col - r->start_col;
749 const struct variable *var = dict_get_var (r->dict, idx);
751 convert_xml_string_to_value (c, var, value);
768 #endif /* GNM_SUPPORT */