--- trunk/all2xml.pl	2004/04/02 23:31:25	298
+++ trunk/all2xml.pl	2004/12/31 03:35:43	618
@@ -5,7 +5,6 @@
 use Getopt::Std;
 use Data::Dumper;
 use XML::Simple;
-use Text::Unaccent 1.02;	# 1.01 won't compile on my platform,
 use Text::Iconv;
 use Config::IniFiles;
 use Encode;
@@ -17,7 +16,7 @@
 
 my $config_file = $0;
 $config_file =~ s/\.pl$/.conf/;
-$config_file = $ARGV[0] if (-f $ARGV[0]);
+$config_file = $ARGV[0] if ($ARGV[0] && -f $ARGV[0]);
 die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file);
 
 my $config;
@@ -140,12 +139,14 @@
 		} else {
 			print STDERR "WARNING: field '$field' doesn't have 'name' attribute!";
 		}
+
 		if ($field_name) {
+			$field_name = x($field_name);
 			if (! $last_field_name) {
-				$last_field_name = x($field_name);
+				$last_field_name = $field_name;
 				return $last_field_name;
 			} elsif ($field_name ne $last_field_name) {
-				$last_field_name = x($field_name);
+				$last_field_name = $field_name;
 				return $last_field_name;
 			}
 		}
@@ -169,6 +170,8 @@
 			($s,$se,$d,$i) = (0,1,0,0);
 		} elsif (lc($type) =~ /^lookup/) {
 			($s,$se,$d,$i,$il) = (0,1,0,0,1);
+		} elsif ($type) {
+			print STDERR "WARNING: unknown type: $type\n";
 		}
 		return ($s,$se,$d,$i,$il);
 	}
@@ -526,7 +529,7 @@
 					$swish_data =~ s/ +/ /g;
 					$swish_data =~ s/ +$//g;
 
-					$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
+					$xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data));
 				}
 
 				my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page];
@@ -536,7 +539,7 @@
 
 					# add delimiters before and after word.
 					# That is required to produce exact match
-					$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data));
+					$xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data));
 				}
 				
 				my $idel = $cache->{index_delimiter}->{$field};
@@ -569,7 +572,7 @@
 				$swish_data =~ s/ +/ /g;
 				$swish_data =~ s/ +$//g;
 
-				$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
+				$xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data));
 			}
 
 			if ($swish_exact_data) {
@@ -578,7 +581,7 @@
 
 				# add delimiters before and after word.
 				# That is required to produce exact match
-				$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data));
+				$xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data));
 			}
 		}
 	}
@@ -615,9 +618,18 @@
 
 my $show_progress = $cfg_global->val('global', 'show_progress');
 
-my $unac_filter = $cfg_global->val('global', 'unac_filter');
-if ($unac_filter) {
-	require $unac_filter;
+my $my_unac_filter = $cfg_global->val('global', 'my_unac_filter');
+if ($my_unac_filter) {
+	print STDERR "using $my_unac_filter to filter characters for search\n";
+	require $my_unac_filter;
+} else {
+	print STDERR "### fallback to default my_unac_string!\n";
+	eval q{
+	sub main::my_unac_string($$) {
+		my ($charset, $string) = (@_);
+		return $string;
+	}
+	};
 }
 
 foreach my $database ($cfg->Sections) {
@@ -629,6 +641,10 @@
 	my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional
 	if ($lookup_file) {
 		#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644;
+		if (! -e $lookup_file) {
+			open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!";
+			close(LOOKUP);
+		}
 		tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644;
 		print STDERR "creating lookup file '$lookup_file'\n";
 		# delete memory cache for lookup file
@@ -651,15 +667,30 @@
 
 	$config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 );
 
+	# helper for progress bar
+	sub fmt_time {
+		my $t = shift || 0;
+		my $out = "";
+
+		my ($ss,$mm,$hh) = gmtime($t);
+		$out .= "${hh}h" if ($hh);
+		$out .= sprintf("%02d:%02d", $mm,$ss);
+		$out .= "  " if ($hh == 0);
+		return $out;
+	}
+
 	# output current progress indicator
 	my $last_p = 0;
+	my $start_t = time();
 	sub progress {
 		return if (! $show_progress);
 		my $current = shift;
 		my $total = shift || 1;
 		my $p = int($current * 100 / $total);
 		if ($p != $last_p) {
-			printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p );
+			my $rate = ($current / (time() - $start_t || 1));
+			my $eta = ($total-$current) / ($rate || 1);
+			printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta));
 			$last_p = $p;
 		}
 	}
@@ -799,7 +830,11 @@
 			for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) {
 				my $cell = $oWorksheet->{Cells}[$iR][$iC];
 				if ($cell) {
-					$row->{int2col($iC)} = $cell->Value;
+					# this conversion is a cludge.
+					# Files from Excell could have
+					# characters which don't fit into
+					# destination encoding.
+					$row->{int2col($iC)} = $utf2cp->convert($cell->Value) || $cell->Value;
 				}
 			}