X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fgnumeric-reader.c;h=514616ace74b59749cef5930344533a7bda51683;hb=9c2001d5628f95d8f13f83ec8ae4f654e33f40bf;hp=2427acf4ba8208cae59231a4231636f4acf714ad;hpb=28042035f7b1ef3d4a7e33e2ba0a90f4993a337a;p=pspp diff --git a/src/data/gnumeric-reader.c b/src/data/gnumeric-reader.c index 2427acf4ba..514616ace7 100644 --- a/src/data/gnumeric-reader.c +++ b/src/data/gnumeric-reader.c @@ -28,12 +28,10 @@ #include "spreadsheet-reader.h" -#include "c-xvasprintf.h" - #if !GNM_SUPPORT struct casereader * -gnumeric_open_reader (struct spreadsheet_read_info *gri, struct spreadsheet_read_options *opts, struct dictionary **dict) +gnumeric_open_reader (const struct spreadsheet_read_options *opts, struct dictionary **dict) { msg (ME, _("Support for %s files was not compiled into this installation of PSPP"), "Gnumeric"); @@ -89,7 +87,8 @@ enum reader_state struct sheet_detail { - xmlChar *name; + /* The name of the sheet (utf8 encoding) */ + char *name; int start_col; int stop_col; @@ -140,13 +139,14 @@ gnumeric_get_sheet_name (struct spreadsheet *s, int n) struct gnumeric_reader *gr = (struct gnumeric_reader *) s; assert (n < s->n_sheets); - return gr->sheets[n].name; + return gr->sheets[n].name; } static void process_node (struct gnumeric_reader *r); + char * gnumeric_get_sheet_range (struct spreadsheet *s, int n) { @@ -164,7 +164,6 @@ gnumeric_get_sheet_range (struct spreadsheet *s, int n) process_node (gr); } - return create_cell_ref ( gr->sheets[n].start_col, gr->sheets[n].start_row, @@ -176,19 +175,20 @@ gnumeric_get_sheet_range (struct spreadsheet *s, int n) 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); +#if 0 for (i = 0; i < r->spreadsheet.n_sheets; ++i) { xmlFree (r->sheets[i].name); @@ -196,7 +196,9 @@ gnm_file_casereader_destroy (struct casereader *reader UNUSED, void *r_) free (r->sheets); + free (r); +#endif } static void @@ -206,10 +208,9 @@ process_node (struct gnumeric_reader *r) if (name == NULL) name = xmlStrdup (_xml ("--")); - r->node_type = xmlTextReaderNodeType (r->xtr); - switch ( r->state) + switch (r->state) { case STATE_PRE_INIT: r->current_sheet = -1; @@ -217,7 +218,6 @@ process_node (struct gnumeric_reader *r) XML_READER_TYPE_ELEMENT == r->node_type) { r->state = STATE_SHEET_COUNT; - r->spreadsheet.n_sheets = 0; } break; @@ -225,20 +225,25 @@ process_node (struct gnumeric_reader *r) if (0 == xmlStrcasecmp (name, _xml("gnm:SheetName")) && XML_READER_TYPE_ELEMENT == r->node_type) { - struct sheet_detail *sd ; - r->spreadsheet.n_sheets++; - r->sheets = xrealloc (r->sheets, r->spreadsheet.n_sheets * sizeof *r->sheets); - sd = &r->sheets[r->spreadsheet.n_sheets - 1]; - sd->start_col = sd->stop_col = sd->start_row = sd->stop_row = -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.n_sheets - 1].name = xmlTextReaderValue (r->xtr); + r->sheets [r->spreadsheet.n_sheets - 1].name = CHAR_CAST (char *, xmlTextReaderValue (r->xtr)); } break; @@ -342,7 +347,6 @@ process_node (struct gnumeric_reader *r) 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); @@ -354,6 +358,7 @@ process_node (struct gnumeric_reader *r) attr = xmlTextReaderGetAttribute (r->xtr, _xml ("Row")); r->row = _xmlchar_to_int (attr); free (attr); + if (r->sheets[r->current_sheet].start_row == -1) { r->sheets[r->current_sheet].start_row = r->row; @@ -363,9 +368,10 @@ process_node (struct gnumeric_reader *r) { 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->current_sheet].stop_col = r->col; r->sheets[r->current_sheet].stop_row = r->row; @@ -373,8 +379,7 @@ process_node (struct gnumeric_reader *r) } 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; } @@ -423,13 +428,25 @@ struct var_spec void gnumeric_destroy (struct spreadsheet *s) { - struct gnumeric_reader *r = (struct gnumeric *) s; gnm_file_casereader_destroy (NULL, s); } +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 %s file `%s' (near line %d): `%s'"), + "Gnumeric", + r->spreadsheet.file_name, + xmlTextReaderLocatorLineNumber (loc), + mesg); +} + static struct gnumeric_reader * -gnumeric_reopen (struct gnumeric_reader *r, const char *filename) +gnumeric_reopen (struct gnumeric_reader *r, const char *filename, bool show_errors) { int ret; @@ -449,9 +466,11 @@ gnumeric_reopen (struct gnumeric_reader *r, const char *filename) 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) { @@ -465,6 +484,9 @@ gnumeric_reopen (struct gnumeric_reader *r, const char *filename) r->spreadsheet.n_sheets = -1; r->spreadsheet.file_name = filename; } + + if (show_errors) + xmlTextReaderSetErrorHandler (xtr, gnumeric_error_handler, r); r->target_sheet = NULL; r->target_sheet_index = -1; @@ -483,16 +505,41 @@ gnumeric_reopen (struct gnumeric_reader *r, const char *filename) process_node (r); } + + if ( ret != 1) + { + /* Does not seem to be a gnumeric file */ + xmlFreeTextReader (r->xtr); + free (r); + return NULL; + } + r->spreadsheet.type = SPREADSHEET_GNUMERIC; + if (show_errors) + { + const xmlChar *enc = xmlTextReaderConstEncoding (r->xtr); + xmlCharEncoding xce = xmlParseCharEncoding (CHAR_CAST (const char *, enc)); + + if ( XML_CHAR_ENCODING_UTF8 != xce) + { + /* I have been told that ALL gnumeric files are UTF8 encoded. If that is correct, this + can never happen. */ + msg (MW, _("The gnumeric file `%s' is encoded as %s instead of the usual UTF-8 encoding. " + "Any non-ascii characters will be incorrectly imported."), + r->spreadsheet.file_name, + enc); + } + } + return r; } struct spreadsheet * -gnumeric_probe (const char *filename) +gnumeric_probe (const char *filename, bool report_errors) { - struct gnumeric_reader *r = gnumeric_reopen (NULL, filename); + struct gnumeric_reader *r = gnumeric_reopen (NULL, filename, report_errors); return &r->spreadsheet; } @@ -500,9 +547,9 @@ gnumeric_probe (const char *filename) 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) { + int x = 0; struct gnumeric_reader *r = NULL; unsigned long int vstart = 0; int ret; @@ -514,7 +561,7 @@ gnumeric_make_reader (struct spreadsheet *spreadsheet, r = (struct gnumeric_reader *) (spreadsheet); if (r->row != -1) - r = gnumeric_reopen (r, NULL); + r = gnumeric_reopen (r, NULL, true); if ( opts->cell_range ) { @@ -562,7 +609,7 @@ gnumeric_make_reader (struct spreadsheet *spreadsheet, n_cases = MIN (n_cases, r->stop_row - r->start_row + 1); } - if ( gri->read_names ) + if ( opts->read_names ) { r->start_row++; n_cases --; @@ -606,7 +653,7 @@ gnumeric_make_reader (struct spreadsheet *spreadsheet, if ( r->row < r->start_row) { - if ( gri->read_names ) + if ( opts->read_names ) { var_spec [idx].name = xstrdup (text); } @@ -616,8 +663,8 @@ gnumeric_make_reader (struct spreadsheet *spreadsheet, var_spec [idx].first_value = xmlStrdup (value); if (-1 == var_spec [idx].width ) - var_spec [idx].width = (gri->asw == -1) ? - ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : gri->asw; + var_spec [idx].width = (opts->asw == -1) ? + ROUND_UP (strlen(text), SPREADSHEET_DEFAULT_WIDTH) : opts->asw; } free (value); @@ -677,13 +724,15 @@ gnumeric_make_reader (struct spreadsheet *spreadsheet, r->first_case = case_create (r->proto); case_set_missing (r->first_case); - int x = 0; + for ( i = 0 ; i < n_var_specs ; ++i ) { + const struct variable *var; + if ( (var_spec[i].name == NULL) && (var_spec[i].first_value == NULL)) continue; - const struct variable *var = dict_get_var (r->dict, x++); + var = dict_get_var (r->dict, x++); convert_xml_string_to_value (r->first_case, var, var_spec[i].first_value); @@ -697,16 +746,7 @@ gnumeric_make_reader (struct spreadsheet *spreadsheet, free (var_spec); - - if (opts->cell_range == NULL) - { - opts->cell_range = c_xasprintf ("%c%d:%c%ld", - r->start_col + 'A', - r->start_row, - r->stop_col + 'A' + caseproto_get_n_widths (r->proto), - r->start_row + n_cases); - } - + return casereader_create_sequential (NULL, r->proto,