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;
105 const xmlChar *target_sheet;
106 int target_sheet_index;
113 struct caseproto *proto;
114 struct dictionary *dict;
115 struct ccase *first_case;
116 bool used_first_case;
119 static void process_node (struct gnumeric_reader *r);
123 gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
125 struct gnumeric_reader *r = r_;
130 xmlFreeTextReader (r->xtr);
132 if ( ! r->used_first_case )
133 case_unref (r->first_case);
135 caseproto_unref (r->proto);
141 process_node (struct gnumeric_reader *r)
143 xmlChar *name = xmlTextReaderName (r->xtr);
145 name = xmlStrdup (_xml ("--"));
148 r->node_type = xmlTextReaderNodeType (r->xtr);
153 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
154 XML_READER_TYPE_ELEMENT == r->node_type)
156 r->state = STATE_SHEET_COUNT;
157 r->spreadsheet.sheets = 0;
161 case STATE_SHEET_COUNT:
162 if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
163 XML_READER_TYPE_ELEMENT == r->node_type)
165 r->spreadsheet.sheets++;
167 else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
168 XML_READER_TYPE_END_ELEMENT == r->node_type)
170 r->state = STATE_INIT;
175 if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
176 XML_READER_TYPE_ELEMENT == r->node_type)
179 r->state = STATE_SHEET_START;
182 case STATE_SHEET_START:
183 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
184 XML_READER_TYPE_ELEMENT == r->node_type)
186 r->state = STATE_SHEET_NAME;
189 case STATE_SHEET_NAME:
190 if (0 == xmlStrcasecmp (name, _xml("gnm:Name")) &&
191 XML_READER_TYPE_END_ELEMENT == r->node_type)
193 r->state = STATE_INIT;
195 else if (XML_READER_TYPE_TEXT == r->node_type)
197 if ( r->target_sheet != NULL)
199 xmlChar *value = xmlTextReaderValue (r->xtr);
200 if ( 0 == xmlStrcmp (value, r->target_sheet))
201 r->state = STATE_SHEET_FOUND;
204 else if (r->target_sheet_index == r->sheet_index)
206 r->state = STATE_SHEET_FOUND;
210 case STATE_SHEET_FOUND:
211 if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
212 XML_READER_TYPE_ELEMENT == r->node_type)
214 r->min_col = INT_MAX;
215 if (! xmlTextReaderIsEmptyElement (r->xtr))
216 r->state = STATE_CELLS_START;
218 else if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
219 XML_READER_TYPE_ELEMENT == r->node_type)
221 r->state = STATE_MAXROW;
223 else if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
224 XML_READER_TYPE_END_ELEMENT == r->node_type)
226 r->state = STATE_INIT;
230 if (0 == xmlStrcasecmp (name, _xml("gnm:MaxRow")) &&
231 XML_READER_TYPE_END_ELEMENT == r->node_type)
233 r->state = STATE_SHEET_FOUND;
235 case STATE_CELLS_START:
236 if (0 == xmlStrcasecmp (name, _xml ("gnm:Cell")) &&
237 XML_READER_TYPE_ELEMENT == r->node_type)
239 xmlChar *attr = NULL;
240 r->state = STATE_CELL;
242 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
243 r->col = _xmlchar_to_int (attr);
246 if (r->col < r->min_col)
249 attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
250 r->row = _xmlchar_to_int (attr);
253 else if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
254 XML_READER_TYPE_END_ELEMENT == r->node_type)
255 r->state = STATE_SHEET_NAME;
259 if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) &&
260 XML_READER_TYPE_END_ELEMENT == r->node_type)
262 r->state = STATE_CELLS_START;
274 Sets the VAR of case C, to the value corresponding to the xml string XV
277 convert_xml_string_to_value (struct ccase *c, const struct variable *var,
280 union value *v = case_data_rw (c, var);
283 value_set_missing (v, var_get_width (var));
284 else if ( var_is_alpha (var))
285 value_copy_str_rpad (v, var_get_width (var), xv, ' ');
288 const char *text = CHAR_CAST (const char *, xv);
292 v->f = c_strtod (text, &endptr);
293 if ( errno != 0 || endptr == text)
302 xmlChar *first_value;
306 gnumeric_probe (const char *filename)
309 struct gnumeric_reader *r = NULL;
310 xmlTextReaderPtr xtr;
312 gzFile gz = gzopen (filename, "r");
317 xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
318 (xmlInputCloseCallback) gzclose, gz,
324 r = xzalloc (sizeof *r);
327 r->spreadsheet.sheets = -1;
328 r->state = STATE_PRE_INIT;
331 /* Advance to the start of the workbook.
332 This gives us some confidence that we are actually dealing with a gnumeric
335 while ( (r->state != STATE_INIT )
336 && 1 == (ret = xmlTextReaderRead (r->xtr)))
343 /* Not a gnumeric spreadsheet */
349 r->spreadsheet.type = SPREADSHEET_GNUMERIC;
350 r->spreadsheet.file_name = filename;
352 return &r->spreadsheet;
357 gnumeric_make_reader (struct spreadsheet *spreadsheet,
358 const struct spreadsheet_read_info *gri,
359 struct spreadsheet_read_options *opts)
361 struct gnumeric_reader *r = NULL;
362 unsigned long int vstart = 0;
364 casenumber n_cases = CASENUMBER_MAX;
366 struct var_spec *var_spec = NULL;
369 r = (struct gnumeric_reader *) (spreadsheet);
371 if ( opts->cell_range )
373 if ( ! convert_cell_ref (opts->cell_range,
374 &r->start_col, &r->start_row,
375 &r->stop_col, &r->stop_row))
377 msg (SE, _("Invalid cell range `%s'"),
391 r->target_sheet = BAD_CAST opts->sheet_name;
392 r->target_sheet_index = opts->sheet_index;
393 r->row = r->col = -1;
396 /* Advance to the start of the cells for the target sheet */
397 while ( (r->state != STATE_CELL || r->row < r->start_row )
398 && 1 == (ret = xmlTextReaderRead (r->xtr)))
402 value = xmlTextReaderValue (r->xtr);
404 if ( r->state == STATE_MAXROW && r->node_type == XML_READER_TYPE_TEXT)
406 n_cases = 1 + _xmlchar_to_int (value) ;
412 /* If a range has been given, then use that to calculate the number
414 if ( opts->cell_range)
416 n_cases = MIN (n_cases, r->stop_row - r->start_row + 1);
419 if ( gri->read_names )
425 /* Read in the first row of cells,
426 including the headers if read_names was set */
428 (( r->state == STATE_CELLS_START && r->row <= r->start_row) || r->state == STATE_CELL )
429 && (ret = xmlTextReaderRead (r->xtr))
435 if ( r->row > r->start_row ) break;
437 if ( r->col < r->start_col ||
438 (r->stop_col != -1 && r->col > r->stop_col))
441 idx = r->col - r->start_col;
443 if ( idx >= n_var_specs )
446 var_spec = xrealloc (var_spec, sizeof (*var_spec) * (idx + 1));
447 for (i = n_var_specs; i <= idx; ++i)
449 var_spec [i].name = NULL;
450 var_spec [i].width = -1;
451 var_spec [i].first_value = NULL;
453 n_var_specs = idx + 1 ;
456 if ( r->node_type == XML_READER_TYPE_TEXT )
458 xmlChar *value = xmlTextReaderValue (r->xtr);
459 const char *text = CHAR_CAST (const char *, value);
461 if ( r->row < r->start_row)
463 if ( gri->read_names )
465 var_spec [idx].name = xstrdup (text);
470 var_spec [idx].first_value = xmlStrdup (value);
472 if (-1 == var_spec [idx].width )
473 var_spec [idx].width = (gri->asw == -1) ?
474 ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw;
479 else if ( r->node_type == XML_READER_TYPE_ELEMENT
480 && r->state == STATE_CELL)
482 if ( r->row == r->start_row )
485 xmlTextReaderGetAttribute (r->xtr, _xml ("ValueType"));
487 if ( NULL == attr || 60 != _xmlchar_to_int (attr))
488 var_spec [idx].width = 0;
496 const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr);
499 /* Create the dictionary and populate it */
500 spreadsheet->dict = r->dict = dict_create (CHAR_CAST (const char *, enc));
503 for (i = 0 ; i < n_var_specs ; ++i )
507 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
510 /* Probably no data exists for this variable, so allocate a
512 if ( var_spec[i].width == -1 )
513 var_spec[i].width = SPREADSHEET_DEFAULT_WIDTH;
515 name = dict_make_unique_var_name (r->dict, var_spec[i].name, &vstart);
516 dict_create_var (r->dict, name, var_spec[i].width);
520 /* Create the first case, and cache it */
521 r->used_first_case = false;
523 if ( n_var_specs == 0 )
525 msg (MW, _("Selected sheet or range of spreadsheet `%s' is empty."),
526 spreadsheet->file_name);
530 r->proto = caseproto_ref (dict_get_proto (r->dict));
531 r->first_case = case_create (r->proto);
532 case_set_missing (r->first_case);
535 for ( i = 0 ; i < n_var_specs ; ++i )
537 if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL))
540 const struct variable *var = dict_get_var (r->dict, x++);
542 convert_xml_string_to_value (r->first_case, var,
543 var_spec[i].first_value);
546 for ( i = 0 ; i < n_var_specs ; ++i )
548 free (var_spec[i].first_value);
549 free (var_spec[i].name);
555 if (opts->cell_range == NULL)
557 opts->cell_range = c_xasprintf ("%c%d:%c%ld",
560 r->stop_col + 'A' + caseproto_get_n_widths (r->proto),
561 r->start_row + n_cases);
564 return casereader_create_sequential
568 &gnm_file_casereader_class, r);
572 for ( i = 0 ; i < n_var_specs ; ++i )
574 free (var_spec[i].first_value);
575 free (var_spec[i].name);
579 dict_destroy (spreadsheet->dict);
580 spreadsheet->dict = NULL;
582 gnm_file_casereader_destroy (NULL, r);
588 /* Reads and returns one case from READER's file. Returns a null
589 pointer on failure. */
590 static struct ccase *
591 gnm_file_casereader_read (struct casereader *reader UNUSED, void *r_)
596 struct gnumeric_reader *r = r_;
597 int current_row = r->row;
599 if ( !r->used_first_case )
601 r->used_first_case = true;
602 return r->first_case;
605 c = case_create (r->proto);
606 case_set_missing (c);
608 if (r->start_col == -1)
609 r->start_col = r->min_col;
611 while ((r->state == STATE_CELL || r->state == STATE_CELLS_START )
612 && r->row == current_row && (ret = xmlTextReaderRead (r->xtr)))
616 if ( r->col < r->start_col || (r->stop_col != -1 &&
617 r->col > r->stop_col))
620 if ( r->col - r->start_col >= caseproto_get_n_widths (r->proto))
623 if ( r->stop_row != -1 && r->row > r->stop_row)
626 if ( r->node_type == XML_READER_TYPE_TEXT )
628 xmlChar *value = xmlTextReaderValue (r->xtr);
630 const int idx = r->col - r->start_col;
632 const struct variable *var = dict_get_var (r->dict, idx);
634 convert_xml_string_to_value (c, var, value);
651 #endif /* GNM_SUPPORT */