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 */
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 */
90 struct gnumeric_reader
92 struct spreadsheet spreadsheet;
96 enum reader_state state;
104 xmlChar **sheet_names;
106 const xmlChar *target_sheet;
107 int target_sheet_index;
114 struct caseproto *proto;
115 struct dictionary *dict;
116 struct ccase *first_case;
117 bool used_first_case;
122 gnumeric_get_sheet_name (struct spreadsheet *s, int n)
124 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
125 assert (n < s->sheets);
127 return gr->sheet_names[n];
131 gnumeric_get_sheet_range (struct spreadsheet *s, int n)
133 struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
134 assert (n < s->sheets);
136 return "I havent the fogiest idea";
140 static void process_node (struct gnumeric_reader *r);
144 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
147 struct gnumeric_reader *r = r_;
152 xmlFreeTextReader (r->xtr);
154 if ( ! r->used_first_case )
155 case_unref (r->first_case);
157 caseproto_unref (r->proto);
159 for (i = 0; i < r->spreadsheet.sheets; ++i)
161 xmlFree (r->sheet_names[i]);
164 free (r->sheet_names);
170 process_node (struct gnumeric_reader *r)
172 xmlChar *name = xmlTextReaderName (r->xtr);
174 name = xmlStrdup (_xml ("--"));
177 r->node_type = xmlTextReaderNodeType (r->xtr);
182 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
183 XML_READER_TYPE_ELEMENT == r->node_type)
185 r->state = STATE_SHEET_COUNT;
186 r->spreadsheet.sheets = 0;
190 case STATE_SHEET_COUNT:
191 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
192 XML_READER_TYPE_ELEMENT == r->node_type)
194 r->spreadsheet.sheets++;
195 r->sheet_names = xrealloc (r->sheet_names, r->spreadsheet.sheets * sizeof *r->sheet_names);
197 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
198 XML_READER_TYPE_END_ELEMENT == r->node_type)
200 r->state = STATE_INIT;
202 else if (XML_READER_TYPE_TEXT == r->node_type)
204 r->sheet_names [r->spreadsheet.sheets - 1] = xmlTextReaderValue (r->xtr);
209 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
210 XML_READER_TYPE_ELEMENT == r->node_type)
213 r->state = STATE_SHEET_START;
216 case STATE_SHEET_START:
217 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
218 XML_READER_TYPE_ELEMENT == r->node_type)
220 r->state = STATE_SHEET_NAME;
223 case STATE_SHEET_NAME:
224 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
225 XML_READER_TYPE_END_ELEMENT == r->node_type)
227 r->state = STATE_INIT;
229 else if (XML_READER_TYPE_TEXT == r->node_type)
231 if ( r->target_sheet != NULL)
233 xmlChar *value = xmlTextReaderValue (r->xtr);
234 if ( 0 == xmlStrcmp (value, r->target_sheet))
235 r->state = STATE_SHEET_FOUND;
238 else if (r->target_sheet_index == r->sheet_index)
240 r->state = STATE_SHEET_FOUND;
244 case STATE_SHEET_FOUND:
245 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
246 XML_READER_TYPE_ELEMENT == r->node_type)
248 r->min_col = INT_MAX;
249 if (! xmlTextReaderIsEmptyElement (r->xtr))
250 r->state = STATE_CELLS_START;
252 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
253 XML_READER_TYPE_ELEMENT == r->node_type)
255 r->state = STATE_MAXROW;
257 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
258 XML_READER_TYPE_END_ELEMENT == r->node_type)
260 r->state = STATE_INIT;
264 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
265 XML_READER_TYPE_END_ELEMENT == r->node_type)
267 r->state = STATE_SHEET_FOUND;
269 case STATE_CELLS_START:
270 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
271 XML_READER_TYPE_ELEMENT == r->node_type)
273 xmlChar *attr = NULL;
274 r->state = STATE_CELL;
276 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
277 r->col = _xmlchar_to_int (attr);
280 if (r->col < r->min_col)
283 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
284 r->row = _xmlchar_to_int (attr);
287 else if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
288 XML_READER_TYPE_END_ELEMENT == r->node_type)
289 r->state = STATE_SHEET_NAME;
293 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) &&
294 XML_READER_TYPE_END_ELEMENT == r->node_type)
296 r->state = STATE_CELLS_START;
308 Sets the VAR of case C, to the value corresponding to the xml string XV
311 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
314 union value *v = case_data_rw (c, var);
317 value_set_missing (v, var_get_width (var));
318 else if ( var_is_alpha (var))
319 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
322 const char *text = CHAR_CAST (const char *, xv);
326 v->f = c_strtod (text, &endptr);
327 if ( errno != 0 || endptr == text)
336 xmlChar *first_value;
341 gnumeric_destroy (struct spreadsheet *s)
343 struct gnumeric_reader *r = s;
344 gnm_file_casereader_destroy (NULL, s);
348 gnumeric_probe (const char *filename)
351 struct gnumeric_reader *r = NULL;
352 xmlTextReaderPtr xtr;
354 gzFile gz = gzopen (filename, "r");
359 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
360 (xmlInputCloseCallback) gzclose, gz,
366 r = xzalloc (sizeof *r);
369 r->spreadsheet.sheets = -1;
370 r->state = STATE_PRE_INIT;
373 /* Advance to the start of the workbook.
374 This gives us some confidence that we are actually dealing with a gnumeric
377 while ( (r->state != STATE_INIT )
378 && 1 == (ret = xmlTextReaderRead (r->xtr)))
385 /* Not a gnumeric spreadsheet */
391 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
392 r->spreadsheet.file_name = filename;
394 return &r->spreadsheet;
399 gnumeric_make_reader (struct spreadsheet *spreadsheet,
400 const struct spreadsheet_read_info *gri,
401 struct spreadsheet_read_options *opts)
403 struct gnumeric_reader *r = NULL;
404 unsigned long int vstart = 0;
406 casenumber n_cases = CASENUMBER_MAX;
408 struct var_spec *var_spec = NULL;
411 r = (struct gnumeric_reader *) (spreadsheet);
413 if ( opts->cell_range )
415 if ( ! convert_cell_ref (opts->cell_range,
416 &r->start_col, &r->start_row,
417 &r->stop_col, &r->stop_row))
419 msg (SE, _("Invalid cell range `%s'"),
433 r->target_sheet = BAD_CAST opts->sheet_name;
434 r->target_sheet_index = opts->sheet_index;
435 r->row = r->col = -1;
438 /* Advance to the start of the cells for the target sheet */
439 while ( (r->state != STATE_CELL || r->row < r->start_row )
440 && 1 == (ret = xmlTextReaderRead (r->xtr)))
444 value = xmlTextReaderValue (r->xtr);
446 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
448 n_cases = 1 + _xmlchar_to_int (value) ;
454 /* If a range has been given, then use that to calculate the number
456 if ( opts->cell_range)
458 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
461 if ( gri->read_names )
467 /* Read in the first row of cells,
468 including the headers if read_names was set */
470 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
471 && (ret = xmlTextReaderRead (r->xtr))
477 if ( r->row > r->start_row ) break;
479 if ( r->col < r->start_col ||
480 (r->stop_col != -1 && r->col > r->stop_col))
483 idx = r->col - r->start_col;
485 if ( idx >= n_var_specs )
488 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
489 for (i = n_var_specs; i <= idx; ++i)
491 var_spec [i].name = NULL;
492 var_spec [i].width = -1;
493 var_spec [i].first_value = NULL;
495 n_var_specs = idx + 1 ;
498 if ( r->node_type == XML_READER_TYPE_TEXT )
500 xmlChar *value = xmlTextReaderValue (r->xtr);
501 const char *text = CHAR_CAST (const char *, value);
503 if ( r->row < r->start_row)
505 if ( gri->read_names )
507 var_spec [idx].name = xstrdup (text);
512 var_spec [idx].first_value = xmlStrdup (value);
514 if (-1 == var_spec [idx].width )
515 var_spec [idx].width = (gri->asw == -1) ?
516 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw;
521 else if ( r->node_type == XML_READER_TYPE_ELEMENT
522 && r->state == STATE_CELL)
524 if ( r->row == r->start_row )
527 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
529 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
530 var_spec [idx].width = 0;
538 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
541 /* Create the dictionary and populate it */
542 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
545 for (i = 0 ; i < n_var_specs ; ++i )
549 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
552 /* Probably no data exists for this variable, so allocate a
554 if ( var_spec[i].width == -1 )
555 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
557 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
558 dict_create_var (r->dict, name, var_spec[i].width);
562 /* Create the first case, and cache it */
563 r->used_first_case = false;
565 if ( n_var_specs == 0 )
567 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
568 spreadsheet->file_name);
572 r->proto = caseproto_ref (dict_get_proto (r->dict));
573 r->first_case = case_create (r->proto);
574 case_set_missing (r->first_case);
577 for ( i = 0 ; i < n_var_specs ; ++i )
579 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
582 const struct variable *var = dict_get_var (r->dict, x++);
584 convert_xml_string_to_value (r->first_case, var,
585 var_spec[i].first_value);
588 for ( i = 0 ; i < n_var_specs ; ++i )
590 free (var_spec[i].first_value);
591 free (var_spec[i].name);
597 if (opts->cell_range == NULL)
599 opts->cell_range = c_xasprintf ("%c%d:%c%ld",
602 r->stop_col + 'A' + caseproto_get_n_widths (r->proto),
603 r->start_row + n_cases);
606 return casereader_create_sequential
610 &gnm_file_casereader_class, r);
614 for ( i = 0 ; i < n_var_specs ; ++i )
616 free (var_spec[i].first_value);
617 free (var_spec[i].name);
621 dict_destroy (spreadsheet->dict);
622 spreadsheet->dict = NULL;
624 gnm_file_casereader_destroy (NULL, r);
630 /* Reads and returns one case from READER's file. Returns a null
631 pointer on failure. */
632 static struct ccase *
633 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
638 struct gnumeric_reader *r = r_;
639 int current_row = r->row;
641 if ( !r->used_first_case )
643 r->used_first_case = true;
644 return r->first_case;
647 c = case_create (r->proto);
648 case_set_missing (c);
650 if (r->start_col == -1)
651 r->start_col = r->min_col;
653 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
654 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
658 if ( r->col < r->start_col || (r->stop_col != -1 &&
659 r->col > r->stop_col))
662 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
665 if ( r->stop_row != -1 && r->row > r->stop_row)
668 if ( r->node_type == XML_READER_TYPE_TEXT )
670 xmlChar *value = xmlTextReaderValue (r->xtr);
672 const int idx = r->col - r->start_col;
674 const struct variable *var = dict_get_var (r->dict, idx);
676 convert_xml_string_to_value (c, var, value);
693 #endif /* GNM_SUPPORT */