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"
34 gnumeric_open_reader (const struct spreadsheet_read_options *opts, struct dictionary **dict)
36 msg (ME, _("Support for %s files was not compiled into this installation of PSPP"), "Gnumeric");
43 #include "data/gnumeric-reader.h"
48 #include <libxml/xmlreader.h>
51 #include "data/case.h"
52 #include "data/casereader-provider.h"
53 #include "data/dictionary.h"
54 #include "data/identifier.h"
55 #include "data/value.h"
56 #include "data/variable.h"
57 #include "libpspp/i18n.h"
58 #include "libpspp/str.h"
60 #include "gl/xalloc.h"
62 static void gnm_file_casereader_destroy (struct casereader *, void *);
64 static struct ccase *gnm_file_casereader_read (struct casereader *, void *);
67 static const struct casereader_class gnm_file_casereader_class =
69 gnm_file_casereader_read,
70 gnm_file_casereader_destroy,
77 STATE_PRE_INIT = 0, /* Initial state */
78 STATE_SHEET_COUNT, /* Found the sheet index */
79 STATE_INIT , /* Other Initial state */
80 STATE_SHEET_START, /* Found the start of a sheet */
81 STATE_SHEET_NAME, /* Found the sheet name */
84 STATE_SHEET_FOUND, /* Found the sheet that we actually want */
85 STATE_CELLS_START, /* Found the start of the cell array */
86 STATE_CELL /* Found a cell */
91 /* The name of the sheet (utf8 encoding) */
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_destroy (struct spreadsheet *s)
140 struct gnumeric_reader *r = s;
143 for (i = 0; i < s->n_sheets; ++i)
145 xmlFree (r->sheets[i].name);
155 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
157 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
158 assert (n < s->n_sheets);
160 return gr->sheets[n].name;
164 static void process_node (struct gnumeric_reader *r);
169 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
172 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
174 assert (n < s->n_sheets);
177 (gr->sheets[n].stop_col == -1)
179 (1 == (ret = xmlTextReaderRead (gr->xtr)))
185 return create_cell_ref (
186 gr->sheets[n].start_col,
187 gr->sheets[n].start_row,
188 gr->sheets[n].stop_col,
189 gr->sheets[n].stop_row);
194 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
196 struct gnumeric_reader *r = r_;
201 xmlFreeTextReader (r->xtr);
204 if ( ! r->used_first_case )
205 case_unref (r->first_case);
207 caseproto_unref (r->proto);
209 gnumeric_destroy (r);
214 process_node (struct gnumeric_reader *r)
216 xmlChar *name = xmlTextReaderName (r->xtr);
218 name = xmlStrdup (_xml ("--"));
220 r->node_type = xmlTextReaderNodeType (r->xtr);
225 r->current_sheet = -1;
226 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
227 XML_READER_TYPE_ELEMENT == r->node_type)
229 r->state = STATE_SHEET_COUNT;
233 case STATE_SHEET_COUNT:
234 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
235 XML_READER_TYPE_ELEMENT == r->node_type)
238 if (r->current_sheet + 1 > r->spreadsheet.n_sheets)
240 struct sheet_detail *sd ;
241 r->sheets = xrealloc (r->sheets, (r->current_sheet + 1) * sizeof *r->sheets);
242 sd = &r->sheets[r->current_sheet];
243 sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -1;
244 r->spreadsheet.n_sheets = r->current_sheet + 1;
247 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
248 XML_READER_TYPE_END_ELEMENT == r->node_type)
250 r->state = STATE_INIT;
251 r->current_sheet = -1;
253 else if (XML_READER_TYPE_TEXT == r->node_type)
255 r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (r->xtr));
260 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
261 XML_READER_TYPE_ELEMENT == r->node_type)
264 r->state = STATE_SHEET_START;
267 case STATE_SHEET_START:
268 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
269 XML_READER_TYPE_ELEMENT == r->node_type)
271 r->state = STATE_SHEET_NAME;
274 case STATE_SHEET_NAME:
275 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
276 XML_READER_TYPE_END_ELEMENT == r->node_type)
278 r->state = STATE_INIT;
280 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
281 XML_READER_TYPE_END_ELEMENT == r->node_type)
283 r->state = STATE_INIT;
285 else if (XML_READER_TYPE_TEXT == r->node_type)
287 if ( r->target_sheet != NULL)
289 xmlChar *value = xmlTextReaderValue (r->xtr);
290 if ( 0 == xmlStrcmp (value, r->target_sheet))
291 r->state = STATE_SHEET_FOUND;
294 else if (r->target_sheet_index == r->current_sheet + 1)
296 r->state = STATE_SHEET_FOUND;
298 else if (r->target_sheet_index == -1)
300 r->state = STATE_SHEET_FOUND;
304 case STATE_SHEET_FOUND:
305 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
306 XML_READER_TYPE_ELEMENT == r->node_type)
308 r->min_col = INT_MAX;
309 if (! xmlTextReaderIsEmptyElement (r->xtr))
310 r->state = STATE_CELLS_START;
312 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
313 XML_READER_TYPE_ELEMENT == r->node_type)
315 r->state = STATE_MAXROW;
317 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
318 XML_READER_TYPE_ELEMENT == r->node_type)
320 r->state = STATE_MAXCOL;
322 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
323 XML_READER_TYPE_END_ELEMENT == r->node_type)
325 r->state = STATE_INIT;
329 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
330 XML_READER_TYPE_END_ELEMENT == r->node_type)
332 r->state = STATE_SHEET_FOUND;
334 else if (r->node_type == XML_READER_TYPE_TEXT)
336 xmlChar *value = xmlTextReaderValue (r->xtr);
337 r->sheets[r->current_sheet].maxrow = _xmlchar_to_int (value);
342 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
343 XML_READER_TYPE_END_ELEMENT == r->node_type)
345 r->state = STATE_SHEET_FOUND;
347 else if (r->node_type == XML_READER_TYPE_TEXT)
349 xmlChar *value = xmlTextReaderValue (r->xtr);
350 r->sheets[r->current_sheet].maxcol = _xmlchar_to_int (value);
354 case STATE_CELLS_START:
355 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
356 XML_READER_TYPE_ELEMENT == r->node_type)
358 xmlChar *attr = NULL;
360 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
361 r->col = _xmlchar_to_int (attr);
364 if (r->col < r->min_col)
367 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
368 r->row = _xmlchar_to_int (attr);
371 if (r->sheets[r->current_sheet].start_row == -1)
373 r->sheets[r->current_sheet].start_row = r->row;
376 if (r->sheets[r->current_sheet].start_col == -1)
378 r->sheets[r->current_sheet].start_col = r->col;
380 if (! xmlTextReaderIsEmptyElement (r->xtr))
381 r->state = STATE_CELL;
383 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == r->node_type) )
385 r->sheets[r->current_sheet].stop_col = r->col;
386 r->sheets[r->current_sheet].stop_row = r->row;
387 r->state = STATE_SHEET_NAME;
391 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == r->node_type)
393 r->state = STATE_CELLS_START;
405 Sets the VAR of case C, to the value corresponding to the xml string XV
408 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
411 union value *v = case_data_rw (c, var);
414 value_set_missing (v, var_get_width (var));
415 else if ( var_is_alpha (var))
416 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
419 const char *text = CHAR_CAST (const char *, xv);
423 v->f = c_strtod (text, &endptr);
424 if ( errno != 0 || endptr == text)
433 xmlChar *first_value;
438 gnumeric_error_handler (void *ctx, const char *mesg,
439 UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
441 struct gnumeric_reader *r = ctx;
443 msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
445 r->spreadsheet.file_name,
446 xmlTextReaderLocatorLineNumber (loc),
450 static struct gnumeric_reader *
451 gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
455 xmlTextReaderPtr xtr;
458 assert (r == NULL || filename == NULL);
461 xmlFreeTextReader (r->xtr);
464 gz = gzopen (filename, "r");
466 gz = gzopen ( r->spreadsheet.file_name, "r");
472 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
473 (xmlInputCloseCallback) gzclose, gz,
475 show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
485 r = xzalloc (sizeof *r);
486 r->spreadsheet.n_sheets = -1;
487 r->spreadsheet.file_name = filename;
491 xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
493 r->target_sheet = NULL;
494 r->target_sheet_index = -1;
496 r->row = r->col = -1;
497 r->state = STATE_PRE_INIT;
500 /* Advance to the start of the workbook.
501 This gives us some confidence that we are actually dealing with a gnumeric
504 while ( (r->state != STATE_INIT )
505 && 1 == (ret = xmlTextReaderRead (r->xtr)))
513 /* Does not seem to be a gnumeric file */
514 xmlFreeTextReader (r->xtr);
519 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
523 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
524 xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc));
526 if ( XML_CHAR_ENCODING_UTF8 != xce)
528 /* I have been told that ALL gnumeric files are UTF8 encoded. If that is correct, this
530 msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. "
531 "Any non-ascii characters will be incorrectly imported."),
532 r->spreadsheet.file_name,
542 gnumeric_probe (const char *filename, bool report_errors)
544 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
546 return &r->spreadsheet;
551 gnumeric_make_reader (struct spreadsheet *spreadsheet,
552 const struct spreadsheet_read_options *opts)
555 struct gnumeric_reader *r = NULL;
556 unsigned long int vstart = 0;
558 casenumber n_cases = CASENUMBER_MAX;
560 struct var_spec *var_spec = NULL;
563 r = (struct gnumeric_reader *) (spreadsheet);
566 r = gnumeric_reopen (r, NULL, true);
568 if ( opts->cell_range )
570 if ( ! convert_cell_ref (opts->cell_range,
571 &r->start_col, &r->start_row,
572 &r->stop_col, &r->stop_row))
574 msg (SE, _("Invalid cell range `%s'"),
587 r->target_sheet = BAD_CAST opts->sheet_name;
588 r->target_sheet_index = opts->sheet_index;
589 r->row = r->col = -1;
590 r->current_sheet = -1;
592 /* Advance to the start of the cells for the target sheet */
593 while ( (r->state != STATE_CELL || r->row < r->start_row )
594 && 1 == (ret = xmlTextReaderRead (r->xtr)))
598 value = xmlTextReaderValue (r->xtr);
600 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
602 n_cases = 1 + _xmlchar_to_int (value) ;
607 /* If a range has been given, then use that to calculate the number
609 if ( opts->cell_range)
611 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
614 if ( opts->read_names )
620 /* Read in the first row of cells,
621 including the headers if read_names was set */
623 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
624 && (ret = xmlTextReaderRead (r->xtr))
630 if ( r->row > r->start_row ) break;
632 if ( r->col < r->start_col ||
633 (r->stop_col != -1 && r->col > r->stop_col))
636 idx = r->col - r->start_col;
638 if ( idx >= n_var_specs )
641 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
642 for (i = n_var_specs; i <= idx; ++i)
644 var_spec [i].name = NULL;
645 var_spec [i].width = -1;
646 var_spec [i].first_value = NULL;
648 n_var_specs = idx + 1 ;
651 if ( r->node_type == XML_READER_TYPE_TEXT )
653 xmlChar *value = xmlTextReaderValue (r->xtr);
654 const char *text = CHAR_CAST (const char *, value);
656 if ( r->row < r->start_row)
658 if ( opts->read_names )
660 var_spec [idx].name = xstrdup (text);
665 var_spec [idx].first_value = xmlStrdup (value);
667 if (-1 == var_spec [idx].width )
668 var_spec [idx].width = (opts->asw == -1) ?
669 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw;
674 else if ( r->node_type == XML_READER_TYPE_ELEMENT
675 && r->state == STATE_CELL)
677 if ( r->row == r->start_row )
680 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
682 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
683 var_spec [idx].width = 0;
691 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
694 /* Create the dictionary and populate it */
695 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
698 for (i = 0 ; i < n_var_specs ; ++i )
702 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
705 /* Probably no data exists for this variable, so allocate a
707 if ( var_spec[i].width == -1 )
708 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
710 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
711 dict_create_var (r->dict, name, var_spec[i].width);
715 /* Create the first case, and cache it */
716 r->used_first_case = false;
718 if ( n_var_specs == 0 )
720 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
721 spreadsheet->file_name);
725 r->proto = caseproto_ref (dict_get_proto (r->dict));
726 r->first_case = case_create (r->proto);
727 case_set_missing (r->first_case);
730 for ( i = 0 ; i < n_var_specs ; ++i )
732 const struct variable *var;
734 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
737 var = dict_get_var (r->dict, x++);
739 convert_xml_string_to_value (r->first_case, var,
740 var_spec[i].first_value);
743 for ( i = 0 ; i < n_var_specs ; ++i )
745 free (var_spec[i].first_value);
746 free (var_spec[i].name);
752 return casereader_create_sequential
756 &gnm_file_casereader_class, r);
760 for ( i = 0 ; i < n_var_specs ; ++i )
762 free (var_spec[i].first_value);
763 free (var_spec[i].name);
767 dict_destroy (spreadsheet->dict);
768 spreadsheet->dict = NULL;
770 gnm_file_casereader_destroy (NULL, r);
776 /* Reads and returns one case from READER's file. Returns a null
777 pointer on failure. */
778 static struct ccase *
779 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
784 struct gnumeric_reader *r = r_;
785 int current_row = r->row;
787 if ( !r->used_first_case )
789 r->used_first_case = true;
790 return r->first_case;
793 c = case_create (r->proto);
794 case_set_missing (c);
796 if (r->start_col == -1)
797 r->start_col = r->min_col;
799 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
800 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
804 if ( r->col < r->start_col || (r->stop_col != -1 &&
805 r->col > r->stop_col))
808 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
811 if ( r->stop_row != -1 && r->row > r->stop_row)
814 if ( r->node_type == XML_READER_TYPE_TEXT )
816 xmlChar *value = xmlTextReaderValue (r->xtr);
818 const int idx = r->col - r->start_col;
820 const struct variable *var = dict_get_var (r->dict, idx);
822 convert_xml_string_to_value (c, var, value);
839 #endif /* GNM_SUPPORT */