/* PSPP - a program for statistical analysis.
- Copyright (C) 2007, 2009, 2010, 2011, 2012 Free Software Foundation, Inc.
+ Copyright (C) 2007, 2009, 2010, 2011, 2012, 2013 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
int maxcol;
int maxrow;
-
- z_off_t offset;
};
{
struct spreadsheet spreadsheet;
+ /* The libxml reader for this instance */
xmlTextReaderPtr xtr;
- gzFile gz;
+ /* An internal state variable */
enum reader_state state;
int row;
int col;
int min_col;
int node_type;
- int sheet_index;
+ int current_sheet;
int start_col;
int stop_col;
gnumeric_get_sheet_name (struct spreadsheet *s, int n)
{
struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
- assert (n < s->sheets);
+ assert (n < s->n_sheets);
- return gr->sheets[n].name;
+ return gr->sheets[n].name; // Kludge: Assumes the encoding is utf8
}
int ret;
struct gnumeric_reader *gr = (struct gnumeric_reader *) s;
- assert (n < s->sheets);
+ assert (n < s->n_sheets);
while (
(gr->sheets[n].stop_col == -1)
process_node (gr);
}
-
return create_cell_ref (
gr->sheets[n].start_col,
gr->sheets[n].start_row,
static void
gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
{
- int i;
struct gnumeric_reader *r = r_;
if ( r == NULL)
return ;
if ( r->xtr)
xmlFreeTextReader (r->xtr);
+ r->xtr = NULL;
if ( ! r->used_first_case )
case_unref (r->first_case);
caseproto_unref (r->proto);
- for (i = 0; i < r->spreadsheet.sheets; ++i)
+#if 0
+ for (i = 0; i < r->spreadsheet.n_sheets; ++i)
{
xmlFree (r->sheets[i].name);
}
free (r->sheets);
+
free (r);
+#endif
}
static void
if (name == NULL)
name = xmlStrdup (_xml ("--"));
-
r->node_type = xmlTextReaderNodeType (r->xtr);
- switch ( r->state)
+ switch (r->state)
{
case STATE_PRE_INIT:
- r->sheet_index = -1;
+ r->current_sheet = -1;
if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
XML_READER_TYPE_ELEMENT == r->node_type)
{
r->state = STATE_SHEET_COUNT;
- r->spreadsheet.sheets = 0;
}
break;
if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) &&
XML_READER_TYPE_ELEMENT == r->node_type)
{
- struct sheet_detail *sd ;
- r->spreadsheet.sheets++;
- r->sheets = xrealloc (r->sheets, r->spreadsheet.sheets * sizeof *r->sheets);
- sd = &r->sheets[r->spreadsheet.sheets - 1];
- sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -1;
- sd->offset = -1;
+ ++r->current_sheet;
+ if (r->current_sheet + 1 > r->spreadsheet.n_sheets)
+ {
+ struct sheet_detail *sd ;
+ r->sheets = xrealloc (r->sheets, (r->current_sheet + 1) * sizeof *r->sheets);
+ sd = &r->sheets[r->current_sheet];
+ sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -1;
+ r->spreadsheet.n_sheets = r->current_sheet + 1;
+ }
}
else if (0 == xmlStrcasecmp (name, _xml("gnm:SheetNameIndex")) &&
XML_READER_TYPE_END_ELEMENT == r->node_type)
{
r->state = STATE_INIT;
+ r->current_sheet = -1;
}
else if (XML_READER_TYPE_TEXT == r->node_type)
{
- r->sheets [r->spreadsheet.sheets - 1].name = xmlTextReaderValue (r->xtr);
+ r->sheets [r->spreadsheet.n_sheets - 1].name = xmlTextReaderValue (r->xtr);
}
break;
if (0 == xmlStrcasecmp (name, _xml("gnm:Sheet")) &&
XML_READER_TYPE_ELEMENT == r->node_type)
{
- ++r->sheet_index;
+ ++r->current_sheet;
r->state = STATE_SHEET_START;
- r->sheets[r->sheet_index].offset = gztell (r->gz);
}
break;
case STATE_SHEET_START:
r->state = STATE_SHEET_FOUND;
free (value);
}
- else if (r->target_sheet_index == r->sheet_index + 1)
+ else if (r->target_sheet_index == r->current_sheet + 1)
{
r->state = STATE_SHEET_FOUND;
}
else if (r->node_type == XML_READER_TYPE_TEXT)
{
xmlChar *value = xmlTextReaderValue (r->xtr);
- r->sheets[r->sheet_index].maxrow = _xmlchar_to_int (value);
+ r->sheets[r->current_sheet].maxrow = _xmlchar_to_int (value);
xmlFree (value);
}
break;
else if (r->node_type == XML_READER_TYPE_TEXT)
{
xmlChar *value = xmlTextReaderValue (r->xtr);
- r->sheets[r->sheet_index].maxcol = _xmlchar_to_int (value);
+ r->sheets[r->current_sheet].maxcol = _xmlchar_to_int (value);
xmlFree (value);
}
break;
XML_READER_TYPE_ELEMENT == r->node_type)
{
xmlChar *attr = NULL;
- r->state = STATE_CELL;
attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Col"));
r->col = _xmlchar_to_int (attr);
attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row"));
r->row = _xmlchar_to_int (attr);
free (attr);
- if (r->sheets[r->sheet_index].start_row == -1)
+
+ if (r->sheets[r->current_sheet].start_row == -1)
{
- r->sheets[r->sheet_index].start_row = r->row;
+ r->sheets[r->current_sheet].start_row = r->row;
}
- if (r->sheets[r->sheet_index].start_col == -1)
+ if (r->sheets[r->current_sheet].start_col == -1)
{
- r->sheets[r->sheet_index].start_col = r->col;
+ r->sheets[r->current_sheet].start_col = r->col;
}
+ if (! xmlTextReaderIsEmptyElement (r->xtr))
+ r->state = STATE_CELL;
}
- else if (0 == xmlStrcasecmp (name, _xml("gnm:Cells")) &&
- XML_READER_TYPE_END_ELEMENT == r->node_type)
+ else if ( (0 == xmlStrcasecmp (name, _xml("gnm:Cells"))) && (XML_READER_TYPE_END_ELEMENT == r->node_type) )
{
- r->sheets[r->sheet_index].stop_col = r->col;
- r->sheets[r->sheet_index].stop_row = r->row;
+ r->sheets[r->current_sheet].stop_col = r->col;
+ r->sheets[r->current_sheet].stop_row = r->row;
r->state = STATE_SHEET_NAME;
}
break;
case STATE_CELL:
- if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) &&
- XML_READER_TYPE_END_ELEMENT == r->node_type)
+ if (0 == xmlStrcasecmp (name, _xml("gnm:Cell")) && XML_READER_TYPE_END_ELEMENT == r->node_type)
{
r->state = STATE_CELLS_START;
}
void
gnumeric_destroy (struct spreadsheet *s)
{
- struct gnumeric_reader *r = (struct gnumeric *) s;
gnm_file_casereader_destroy (NULL, s);
}
-struct spreadsheet *
-gnumeric_probe (const char *filename)
+
+static void
+gnumeric_error_handler (void *ctx, const char *mesg,
+ UNUSED xmlParserSeverities sev, xmlTextReaderLocatorPtr loc)
{
+ struct gnumeric_reader *r = ctx;
+
+ msg (MW, _("There was a problem whilst reading the Gnumeric file `%s' (near line %d): `%s'"),
+ r->spreadsheet.file_name,
+ xmlTextReaderLocatorLineNumber (loc),
+ mesg);
+}
+
+static struct gnumeric_reader *
+gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors)
+{
int ret;
- struct gnumeric_reader *r = NULL;
- xmlTextReaderPtr xtr;
+ xmlTextReaderPtr xtr;
gzFile gz;
- gz = gzopen (filename, "r");
+
+ assert (r == NULL || filename == NULL);
+
+ if (r && r->xtr)
+ xmlFreeTextReader (r->xtr);
+
+ if (filename)
+ gz = gzopen (filename, "r");
+ else
+ gz = gzopen ( r->spreadsheet.file_name, "r");
if (NULL == gz)
return NULL;
+
xtr = xmlReaderForIO ((xmlInputReadCallback) gzread,
- (xmlInputCloseCallback) gzclose, gz,
- NULL, NULL, 0);
+ (xmlInputCloseCallback) gzclose, gz,
+ NULL, NULL,
+ show_errors ? 0 : (XML_PARSE_NOERROR | XML_PARSE_NOWARNING) );
if (xtr == NULL)
{
return NULL;
}
- r = xzalloc (sizeof *r);
+ if (r == NULL)
+ {
+ r = xzalloc (sizeof *r);
+ r->spreadsheet.n_sheets = -1;
+ r->spreadsheet.file_name = filename;
+ }
- r->gz = gz;
- r->xtr = xtr;
- r->spreadsheet.sheets = -1;
- r->state = STATE_PRE_INIT;
+ if (show_errors)
+ xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r);
r->target_sheet = NULL;
r->target_sheet_index = -1;
+ r->row = r->col = -1;
+ r->state = STATE_PRE_INIT;
+ r->xtr = xtr;
/* Advance to the start of the workbook.
This gives us some confidence that we are actually dealing with a gnumeric
process_node (r);
}
- if (ret != 1)
+ if ( ret != 1)
{
- /* Not a gnumeric spreadsheet */
+ /* Does not seem to be a gnumeric file */
xmlFreeTextReader (r->xtr);
free (r);
return NULL;
}
-
+
r->spreadsheet.type = SPREADSHEET_GNUMERIC;
- r->spreadsheet.file_name = filename;
-
+
+ return r;
+}
+
+
+struct spreadsheet *
+gnumeric_probe (const char *filename, bool report_errors)
+{
+ struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors);
+
return &r->spreadsheet;
}
struct casereader *
gnumeric_make_reader (struct spreadsheet *spreadsheet,
const struct spreadsheet_read_info *gri,
- struct spreadsheet_read_options *opts)
+ const struct spreadsheet_read_options *opts)
{
struct gnumeric_reader *r = NULL;
unsigned long int vstart = 0;
r = (struct gnumeric_reader *) (spreadsheet);
+ if (r->row != -1)
+ r = gnumeric_reopen (r, NULL, true);
+
if ( opts->cell_range )
{
if ( ! convert_cell_ref (opts->cell_range,
r->target_sheet = BAD_CAST opts->sheet_name;
r->target_sheet_index = opts->sheet_index;
r->row = r->col = -1;
- r->sheet_index = -1;
+ r->current_sheet = -1;
/* Advance to the start of the cells for the target sheet */
while ( (r->state != STATE_CELL || r->row < r->start_row )
free (var_spec);
-
+
+#if 0
if (opts->cell_range == NULL)
{
opts->cell_range = c_xasprintf ("%c%d:%c%ld",
r->stop_col + 'A' + caseproto_get_n_widths (r->proto),
r->start_row + n_cases);
}
+#endif
return casereader_create_sequential
(NULL,