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) */
105 /* The libxml reader for this instance */
106 xmlTextReaderPtr xtr;
108 /* An internal state variable */
109 enum reader_state state;
120 struct gnumeric_reader
122 struct spreadsheet spreadsheet;
125 struct state_data rsd;
126 struct state_data msd;
133 struct sheet_detail *sheets;
135 const xmlChar *target_sheet;
136 int target_sheet_index;
138 struct caseproto *proto;
139 struct dictionary *dict;
140 struct ccase *first_case;
141 bool used_first_case;
146 gnumeric_destroy (struct spreadsheet *s)
148 struct gnumeric_reader *r = (struct gnumeric_reader *) s;
151 if (0 == --r->ref_cnt)
155 for (i = 0; i < s->n_sheets; ++i)
157 xmlFree (r->sheets[i].name);
169 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
171 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
172 assert (n < s->n_sheets);
174 return gr->sheets[n].name;
178 static void process_node (struct gnumeric_reader *r, struct state_data *sd);
183 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
186 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
188 assert (n < s->n_sheets);
191 (gr->sheets[n].stop_col == -1)
193 (1 == (ret = xmlTextReaderRead (gr->msd.xtr)))
196 process_node (gr, &gr->msd);
199 return create_cell_ref (
200 gr->sheets[n].start_col,
201 gr->sheets[n].start_row,
202 gr->sheets[n].stop_col,
203 gr->sheets[n].stop_row);
208 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
210 struct gnumeric_reader *r = r_;
217 xmlFreeTextReader (r->rsd.xtr);
220 if ( ! r->used_first_case )
221 case_unref (r->first_case);
223 caseproto_unref (r->proto);
225 gnumeric_destroy (&r->spreadsheet);
231 process_node (struct gnumeric_reader *r, struct state_data *sd)
233 xmlChar *name = xmlTextReaderName (sd->xtr);
235 name = xmlStrdup (_xml ("--"));
237 sd->node_type = xmlTextReaderNodeType (sd->xtr);
242 sd->current_sheet = -1;
243 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
244 XML_READER_TYPE_ELEMENT == sd->node_type)
246 sd->state = STATE_SHEET_COUNT;
250 case STATE_SHEET_COUNT:
251 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
252 XML_READER_TYPE_ELEMENT == sd->node_type)
255 if (sd->current_sheet + 1 > r->spreadsheet.n_sheets)
257 struct sheet_detail *detail ;
258 r->sheets = xrealloc (r->sheets, (sd->current_sheet + 1) * sizeof *r->sheets);
259 detail = &r->sheets[sd->current_sheet];
260 detail->start_col = detail->stop_col = detail->start_row = detail->stop_row = -1;
261 r->spreadsheet.n_sheets = sd->current_sheet + 1;
264 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
265 XML_READER_TYPE_END_ELEMENT == sd->node_type)
267 sd->state = STATE_INIT;
268 sd->current_sheet = -1;
270 else if (XML_READER_TYPE_TEXT == sd->node_type)
272 r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
277 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
278 XML_READER_TYPE_ELEMENT == sd->node_type)
281 sd->state = STATE_SHEET_START;
284 case STATE_SHEET_START:
285 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
286 XML_READER_TYPE_ELEMENT == sd->node_type)
288 sd->state = STATE_SHEET_NAME;
291 case STATE_SHEET_NAME:
292 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
293 XML_READER_TYPE_END_ELEMENT == sd->node_type)
295 sd->state = STATE_INIT;
297 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
298 XML_READER_TYPE_END_ELEMENT == sd->node_type)
300 sd->state = STATE_INIT;
302 else if (XML_READER_TYPE_TEXT == sd->node_type)
304 if ( r->target_sheet != NULL)
306 xmlChar *value = xmlTextReaderValue (sd->xtr);
307 if ( 0 == xmlStrcmp (value, r->target_sheet))
308 sd->state = STATE_SHEET_FOUND;
311 else if (r->target_sheet_index == sd->current_sheet + 1)
313 sd->state = STATE_SHEET_FOUND;
315 else if (r->target_sheet_index == -1)
317 sd->state = STATE_SHEET_FOUND;
321 case STATE_SHEET_FOUND:
322 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
323 XML_READER_TYPE_ELEMENT == sd->node_type)
325 sd->min_col = INT_MAX;
326 if (! xmlTextReaderIsEmptyElement (sd->xtr))
327 sd->state = STATE_CELLS_START;
329 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
330 XML_READER_TYPE_ELEMENT == sd->node_type)
332 sd->state = STATE_MAXROW;
334 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
335 XML_READER_TYPE_ELEMENT == sd->node_type)
337 sd->state = STATE_MAXCOL;
339 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
340 XML_READER_TYPE_END_ELEMENT == sd->node_type)
342 sd->state = STATE_INIT;
346 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
347 XML_READER_TYPE_END_ELEMENT == sd->node_type)
349 sd->state = STATE_SHEET_FOUND;
351 else if (sd->node_type == XML_READER_TYPE_TEXT)
353 xmlChar *value = xmlTextReaderValue (sd->xtr);
354 r->sheets[sd->current_sheet].maxrow = _xmlchar_to_int (value);
359 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxCol")) &&
360 XML_READER_TYPE_END_ELEMENT == sd->node_type)
362 sd->state = STATE_SHEET_FOUND;
364 else if (sd->node_type == XML_READER_TYPE_TEXT)
366 xmlChar *value = xmlTextReaderValue (sd->xtr);
367 r->sheets[sd->current_sheet].maxcol = _xmlchar_to_int (value);
371 case STATE_CELLS_START:
372 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
373 XML_READER_TYPE_ELEMENT == sd->node_type)
375 xmlChar *attr = NULL;
377 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Col"));
378 sd->col = _xmlchar_to_int (attr);
381 if (sd->col < sd->min_col)
382 sd->min_col = sd->col;
384 attr = xmlTextReaderGetAttribute (sd->xtr, _xml ("Row"));
385 sd->row = _xmlchar_to_int (attr);
388 if (r->sheets[sd->current_sheet].start_row == -1)
390 r->sheets[sd->current_sheet].start_row = sd->row;
393 if (r->sheets[sd->current_sheet].start_col == -1)
395 r->sheets[sd->current_sheet].start_col = sd->col;
397 if (! xmlTextReaderIsEmptyElement (sd->xtr))
398 sd->state = STATE_CELL;
400 else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == sd->node_type) )
402 r->sheets[sd->current_sheet].stop_col = sd->col;
403 r->sheets[sd->current_sheet].stop_row = sd->row;
404 sd->state = STATE_SHEET_NAME;
408 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == sd->node_type)
410 sd->state = STATE_CELLS_START;
422 Sets the VAR of case C, to the value corresponding to the xml string XV
425 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
428 union value *v = case_data_rw (c, var);
431 value_set_missing (v, var_get_width (var));
432 else if ( var_is_alpha (var))
433 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
436 const char *text = CHAR_CAST (const char *, xv);
440 v->f = c_strtod (text, &endptr);
441 if ( errno != 0 || endptr == text)
450 xmlChar *first_value;
455 gnumeric_error_handler (void *ctx, const char *mesg,
456 UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
458 struct gnumeric_reader *r = ctx;
460 msg (MW, _("There was a problem whilst reading the %s file `%s' (near line %d): `%s'"),
462 r->spreadsheet.file_name,
463 xmlTextReaderLocatorLineNumber (loc),
467 static struct gnumeric_reader *
468 gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
471 struct state_data *sd;// = filename ? &r->msd : &r->rsd;
473 xmlTextReaderPtr xtr;
476 assert (r == NULL || filename == NULL);
480 gz = gzopen (filename, "r");
484 gz = gzopen (r->spreadsheet.file_name, "r");
491 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
492 (xmlInputCloseCallback) gzclose, gz,
494 show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
504 r = xzalloc (sizeof *r);
505 r->spreadsheet.n_sheets = -1;
506 r->spreadsheet.file_name = filename;
515 xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
517 r->target_sheet = NULL;
518 r->target_sheet_index = -1;
520 sd->row = sd->col = -1;
521 sd->state = STATE_PRE_INIT;
525 /* Advance to the start of the workbook.
526 This gives us some confidence that we are actually dealing with a gnumeric
529 while ( (sd->state != STATE_INIT )
530 && 1 == (ret = xmlTextReaderRead (sd->xtr)))
532 process_node (r, sd);
538 /* Does not seem to be a gnumeric file */
539 xmlFreeTextReader (sd->xtr);
544 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
548 const xmlChar *enc = xmlTextReaderConstEncoding (sd->xtr);
549 xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc));
551 if ( XML_CHAR_ENCODING_UTF8 != xce)
553 /* I have been told that ALL gnumeric files are UTF8 encoded. If that is correct, this
555 msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. "
556 "Any non-ascii characters will be incorrectly imported."),
557 r->spreadsheet.file_name,
567 gnumeric_probe (const char *filename, bool report_errors)
569 struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
571 return &r->spreadsheet;
576 gnumeric_make_reader (struct spreadsheet *spreadsheet,
577 const struct spreadsheet_read_options *opts)
580 struct gnumeric_reader *r = NULL;
581 unsigned long int vstart = 0;
583 casenumber n_cases = CASENUMBER_MAX;
585 struct var_spec *var_spec = NULL;
588 r = (struct gnumeric_reader *) (spreadsheet);
590 if (r->rsd.row != -1)
591 r = gnumeric_reopen (r, NULL, true);
595 if ( opts->cell_range )
597 if ( ! convert_cell_ref (opts->cell_range,
598 &r->start_col, &r->start_row,
599 &r->stop_col, &r->stop_row))
601 msg (SE, _("Invalid cell range `%s'"),
614 r->target_sheet = BAD_CAST opts->sheet_name;
615 r->target_sheet_index = opts->sheet_index;
616 r->rsd.row = r->rsd.col = -1;
617 r->rsd.current_sheet = -1;
619 /* Advance to the start of the cells for the target sheet */
620 while ( (r->rsd.state != STATE_CELL || r->rsd.row < r->start_row )
621 && 1 == (ret = xmlTextReaderRead (r->rsd.xtr)))
624 process_node (r, &r->rsd);
625 value = xmlTextReaderValue (r->rsd.xtr);
627 if ( r->rsd.state == STATE_MAXROW && r->rsd.node_type == XML_READER_TYPE_TEXT)
629 n_cases = 1 + _xmlchar_to_int (value) ;
634 /* If a range has been given, then use that to calculate the number
636 if ( opts->cell_range)
638 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
641 if ( opts->read_names )
647 /* Read in the first row of cells,
648 including the headers if read_names was set */
650 (( r->rsd.state == STATE_CELLS_START && r->rsd.row <= r->start_row) || r->rsd.state == STATE_CELL )
651 && (ret = xmlTextReaderRead (r->rsd.xtr))
655 process_node (r, &r->rsd);
657 if ( r->rsd.row > r->start_row ) break;
659 if ( r->rsd.col < r->start_col ||
660 (r->stop_col != -1 && r->rsd.col > r->stop_col))
663 idx = r->rsd.col - r->start_col;
665 if ( idx >= n_var_specs )
668 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
669 for (i = n_var_specs; i <= idx; ++i)
671 var_spec [i].name = NULL;
672 var_spec [i].width = -1;
673 var_spec [i].first_value = NULL;
675 n_var_specs = idx + 1 ;
678 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
680 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
681 const char *text = CHAR_CAST (const char *, value);
683 if ( r->rsd.row < r->start_row)
685 if ( opts->read_names )
687 var_spec [idx].name = xstrdup (text);
692 var_spec [idx].first_value = xmlStrdup (value);
694 if (-1 == var_spec [idx].width )
695 var_spec [idx].width = (opts->asw == -1) ?
696 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw;
701 else if ( r->rsd.node_type == XML_READER_TYPE_ELEMENT
702 && r->rsd.state == STATE_CELL)
704 if ( r->rsd.row == r->start_row )
707 xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
709 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
710 var_spec [idx].width = 0;
718 const xmlChar *enc = xmlTextReaderConstEncoding (r->rsd.xtr);
721 /* Create the dictionary and populate it */
722 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
725 for (i = 0 ; i < n_var_specs ; ++i )
729 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
732 /* Probably no data exists for this variable, so allocate a
734 if ( var_spec[i].width == -1 )
735 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
737 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
738 dict_create_var (r->dict, name, var_spec[i].width);
742 /* Create the first case, and cache it */
743 r->used_first_case = false;
745 if ( n_var_specs == 0 )
747 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
748 spreadsheet->file_name);
752 r->proto = caseproto_ref (dict_get_proto (r->dict));
753 r->first_case = case_create (r->proto);
754 case_set_missing (r->first_case);
757 for ( i = 0 ; i < n_var_specs ; ++i )
759 const struct variable *var;
761 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
764 var = dict_get_var (r->dict, x++);
766 convert_xml_string_to_value (r->first_case, var,
767 var_spec[i].first_value);
770 for ( i = 0 ; i < n_var_specs ; ++i )
772 free (var_spec[i].first_value);
773 free (var_spec[i].name);
779 return casereader_create_sequential
783 &gnm_file_casereader_class, r);
787 for ( i = 0 ; i < n_var_specs ; ++i )
789 free (var_spec[i].first_value);
790 free (var_spec[i].name);
794 dict_destroy (spreadsheet->dict);
795 spreadsheet->dict = NULL;
797 gnm_file_casereader_destroy (NULL, r);
803 /* Reads and returns one case from READER's file. Returns a null
804 pointer on failure. */
805 static struct ccase *
806 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
811 struct gnumeric_reader *r = r_;
812 int current_row = r->rsd.row;
814 if ( !r->used_first_case )
816 r->used_first_case = true;
817 return r->first_case;
820 c = case_create (r->proto);
821 case_set_missing (c);
823 if (r->start_col == -1)
824 r->start_col = r->rsd.min_col;
826 while ((r->rsd.state == STATE_CELL || r->rsd.state == STATE_CELLS_START )
827 && r->rsd.row == current_row && (ret = xmlTextReaderRead (r->rsd.xtr)))
829 process_node (r, &r->rsd);
831 if ( r->rsd.col < r->start_col || (r->stop_col != -1 &&
832 r->rsd.col > r->stop_col))
835 if ( r->rsd.col - r->start_col >= caseproto_get_n_widths (r->proto))
838 if ( r->stop_row != -1 && r->rsd.row > r->stop_row)
841 if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
843 xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
845 const int idx = r->rsd.col - r->start_col;
847 const struct variable *var = dict_get_var (r->dict, idx);
849 convert_xml_string_to_value (c, var, value);
866 #endif /* GNM_SUPPORT */