/* PSPP - a program for statistical analysis.
- Copyright (C) 2007, 2009, 2010, 2011, 2012, 2013 Free Software Foundation, Inc.
+ Copyright (C) 2007, 2009, 2010, 2011, 2012, 2013, 2016 Free Software Foundation, Inc.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
#include "spreadsheet-reader.h"
-#if !GNM_SUPPORT
+#if !GNM_READ_SUPPORT
struct casereader *
gnumeric_open_reader (const struct spreadsheet_read_options *opts, struct dictionary **dict)
#include <libxml/xmlreader.h>
#include <zlib.h>
+#include "data/format.h"
+#include "data/data-in.h"
#include "data/case.h"
#include "data/casereader-provider.h"
#include "data/dictionary.h"
#include "gl/xalloc.h"
+
+/* Shamelessly lifted from the Gnumeric sources:
+ https://git.gnome.org/browse/gnumeric/tree/src/value.h
+ */
+enum gnm_value_type
+{
+ VALUE_EMPTY = 10,
+ VALUE_BOOLEAN = 20,
+ VALUE_INTEGER = 30, /* Note, this was removed from gnumeric in 2006 - old versions may of
+ course still be around. New ones are supposed to use float.*/
+ VALUE_FLOAT = 40,
+ VALUE_ERROR = 50,
+ VALUE_STRING = 60,
+ VALUE_CELLRANGE = 70,
+ VALUE_ARRAY = 80
+};
+
+
+
static void gnm_file_casereader_destroy (struct casereader *, void *);
static struct ccase *gnm_file_casereader_read (struct casereader *, void *);
int min_col;
};
+
+static void
+state_data_destroy (struct state_data *sd)
+{
+ xmlFreeTextReader (sd->xtr);
+}
+
+
struct gnumeric_reader
{
struct spreadsheet spreadsheet;
- int ref_cnt;
struct state_data rsd;
struct state_data msd;
struct dictionary *dict;
struct ccase *first_case;
bool used_first_case;
+
+ enum gnm_value_type vtype;
};
void
-gnumeric_destroy (struct spreadsheet *s)
+gnumeric_unref (struct spreadsheet *s)
{
struct gnumeric_reader *r = (struct gnumeric_reader *) s;
-#if 0
- if (0 == --r->ref_cnt)
+ if (0 == --s->ref_cnt)
{
int i;
{
xmlFree (r->sheets[i].name);
}
-
+
+
free (r->sheets);
+ state_data_destroy (&r->msd);
+
+ dict_destroy (r->dict);
+
+ free (s->file_name);
free (r);
}
-#endif
}
process_node (gr, &gr->msd);
}
- return create_cell_ref (
+ return create_cell_range (
gr->sheets[n].start_col,
gr->sheets[n].start_row,
gr->sheets[n].stop_col,
if ( r == NULL)
return ;
-#if 0
- if ( r->rsd.xtr)
- xmlFreeTextReader (r->rsd.xtr);
- r->rsd.xtr = NULL;
+ state_data_destroy (&r->rsd);
- if ( ! r->used_first_case )
+ if (r->first_case && ! r->used_first_case )
case_unref (r->first_case);
- caseproto_unref (r->proto);
+ if (r->proto)
+ caseproto_unref (r->proto);
- gnumeric_destroy (&r->spreadsheet);
-#endif
+ gnumeric_unref (&r->spreadsheet);
}
r->sheets = xrealloc (r->sheets, (sd->current_sheet + 1) * sizeof *r->sheets);
detail = &r->sheets[sd->current_sheet];
detail->start_col = detail->stop_col = detail->start_row = detail->stop_row = -1;
+ detail->name = NULL;
r->spreadsheet.n_sheets = sd->current_sheet + 1;
}
}
}
else if (XML_READER_TYPE_TEXT == sd->node_type)
{
- r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
+ if ( r->sheets [r->spreadsheet.n_sheets - 1].name == NULL)
+ r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (sd->xtr));
}
break;
*/
static void
convert_xml_string_to_value (struct ccase *c, const struct variable *var,
- const xmlChar *xv)
+ const xmlChar *xv, enum gnm_value_type type, int col, int row)
{
union value *v = case_data_rw (c, var);
value_set_missing (v, var_get_width (var));
else if ( var_is_alpha (var))
value_copy_str_rpad (v, var_get_width (var), xv, ' ');
- else
+ else if (type == VALUE_FLOAT || type == VALUE_INTEGER)
{
const char *text = CHAR_CAST (const char *, xv);
char *endptr;
if ( errno != 0 || endptr == text)
v->f = SYSMIS;
}
+ else
+ {
+ const char *text = CHAR_CAST (const char *, xv);
+
+ const struct fmt_spec *fmt = var_get_write_format (var);
+
+ char *m = data_in (ss_cstr (text), "UTF-8",
+ fmt->type,
+ v,
+ var_get_width (var),
+ "UTF-8");
+
+ if (m)
+ {
+ char buf [FMT_STRING_LEN_MAX + 1];
+ char *cell = create_cell_ref (col, row);
+
+ msg (MW, _("Cannot convert the value in the spreadsheet cell %s to format (%s): %s"),
+ cell, fmt_to_string (fmt, buf), m);
+ free (cell);
+ }
+ free (m);
+ }
}
struct var_spec
char *name;
int width;
xmlChar *first_value;
+ int first_type;
};
static struct gnumeric_reader *
gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
{
- int ret;
- struct state_data *sd;// = filename ? &r->msd : &r->rsd;
+ int ret = -1;
+ struct state_data *sd;
xmlTextReaderPtr xtr;
gzFile gz;
{
r = xzalloc (sizeof *r);
r->spreadsheet.n_sheets = -1;
- r->spreadsheet.file_name = filename;
+ r->spreadsheet.file_name = strdup (filename);
sd = &r->msd;
}
else
sd->row = sd->col = -1;
sd->state = STATE_PRE_INIT;
sd->xtr = xtr;
- r->ref_cnt++;
+ r->spreadsheet.ref_cnt++;
+
/* Advance to the start of the workbook.
This gives us some confidence that we are actually dealing with a gnumeric
if ( ret != 1)
{
/* Does not seem to be a gnumeric file */
- xmlFreeTextReader (sd->xtr);
- free (r);
+ gnumeric_unref (&r->spreadsheet);
return NULL;
}
gnumeric_make_reader (struct spreadsheet *spreadsheet,
const struct spreadsheet_read_options *opts)
{
+ int type = 0;
int x = 0;
struct gnumeric_reader *r = NULL;
unsigned long int vstart = 0;
r = (struct gnumeric_reader *) (spreadsheet);
- if (r->rsd.row != -1)
- r = gnumeric_reopen (r, NULL, true);
-
-
+ r = gnumeric_reopen (r, NULL, true);
if ( opts->cell_range )
{
r->target_sheet_index = opts->sheet_index;
r->rsd.row = r->rsd.col = -1;
r->rsd.current_sheet = -1;
+ r->first_case = NULL;
+ r->proto = NULL;
/* Advance to the start of the cells for the target sheet */
while ( (r->rsd.state != STATE_CELL || r->rsd.row < r->start_row )
n_cases --;
}
+
/* Read in the first row of cells,
including the headers if read_names was set */
while (
)
{
int idx;
+
+ if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_TEXT)
+ {
+ xmlChar *attr =
+ xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
+
+ type = _xmlchar_to_int (attr);
+
+ xmlFree (attr);
+ }
+
process_node (r, &r->rsd);
- if ( r->rsd.row > r->start_row ) break;
+ if ( r->rsd.row > r->start_row )
+ {
+ xmlChar *attr =
+ xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
+
+ r->vtype = _xmlchar_to_int (attr);
+
+ xmlFree (attr);
+ break;
+ }
if ( r->rsd.col < r->start_col ||
(r->stop_col != -1 && r->rsd.col > r->stop_col))
var_spec [i].name = NULL;
var_spec [i].width = -1;
var_spec [i].first_value = NULL;
+ var_spec [i].first_type = -1;
}
n_var_specs = idx + 1 ;
}
+ var_spec [idx].first_type = type;
+
if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
{
xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
xmlChar *attr =
xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
- if ( NULL == attr || 60 != _xmlchar_to_int (attr))
+ if ( NULL == attr || VALUE_STRING != _xmlchar_to_int (attr))
var_spec [idx].width = 0;
free (attr);
continue;
var = dict_get_var (r->dict, x++);
-
+
convert_xml_string_to_value (r->first_case, var,
- var_spec[i].first_value);
+ var_spec[i].first_value,
+ var_spec[i].first_type,
+ r->rsd.col + i - 1,
+ r->rsd.row - 1);
}
for ( i = 0 ; i < n_var_specs ; ++i )
}
free (var_spec);
- dict_destroy (spreadsheet->dict);
- spreadsheet->dict = NULL;
gnm_file_casereader_destroy (NULL, r);
if (r->start_col == -1)
r->start_col = r->rsd.min_col;
+
while ((r->rsd.state == STATE_CELL || r->rsd.state == STATE_CELLS_START )
&& r->rsd.row == current_row && (ret = xmlTextReaderRead (r->rsd.xtr)))
{
process_node (r, &r->rsd);
+ if (r->rsd.state == STATE_CELL && r->rsd.node_type == XML_READER_TYPE_ELEMENT)
+ {
+ xmlChar *attr =
+ xmlTextReaderGetAttribute (r->rsd.xtr, _xml ("ValueType"));
+
+ r->vtype = _xmlchar_to_int (attr);
+
+ xmlFree (attr);
+ }
+
if ( r->rsd.col < r->start_col || (r->stop_col != -1 &&
r->rsd.col > r->stop_col))
continue;
if ( r->stop_row != -1 && r->rsd.row > r->stop_row)
break;
+
if ( r->rsd.node_type == XML_READER_TYPE_TEXT )
{
xmlChar *value = xmlTextReaderValue (r->rsd.xtr);
-
const int idx = r->rsd.col - r->start_col;
-
const struct variable *var = dict_get_var (r->dict, idx);
- convert_xml_string_to_value (c, var, value);
+ convert_xml_string_to_value (c, var, value, r->vtype,
+ r->rsd.col, r->rsd.row);
- free (value);
+ xmlFree (value);
}
-
}
if (ret == 1)
}
-#endif /* GNM_SUPPORT */
+#endif /* GNM_READ_SUPPORT */
projects / pspp / blobdiff