Handle more non-ascii characters

nigelhorne · Oct 1, 2024 · b23ecec · b23ecec
1 parent 192aa34
commit b23ecec
Showing 1 changed file with 18 additions and 12 deletions.
diff --git a/ged2site b/ged2site
@@ -2205,7 +2205,7 @@ sub print_person
 		print $index_html '</html>';
 		close $index_html;
 		print $csv '1!';
-		chmod 0444, $index_html;
+		chmod 0444, $filename;
 	} else {
 		if(!-e $filename) {
 			open(my $index_html, '>', $filename);
@@ -2222,7 +2222,7 @@ sub print_person
 				"<body><h1>Family History</h1><a href=\"history.html\">Family History</a></body>";
 			print $index_html '</html>';
 			close $index_html;
-			chmod 0444, $index_html;
+			chmod 0444, $filename;
 		}
 		print $csv '0!';
 	}
@@ -5228,7 +5228,7 @@ sub print_person
 				}
 
 				if(($count == 0) || !places_are_the_same({ person => $person, first => $residence, second => $residencelist[$count - 1] })) {
-					if($birth && $dateofbirth && $rdate && (datecmp($rdate, $dateofbirth) == 0) &&
+					if($birth && $dateofbirth && $rdate && ($rdate !~ /^\d{3,4}$/) && (datecmp($rdate, $dateofbirth) == 0) &&
 					   places_are_the_same({ person => $person, first => $birth, second => $residence})) {
 						# This residence record is for the place of birth, which is
 						# printed elsewhere
@@ -8365,16 +8365,16 @@ sub print_person
 				print $html "<li>$f</li>";
 				print $csv person_line_csv({ person => $father, year => $year });
 				print $xml "\n\t\t<father>\n\t\t\t<xref>", $father->xref(),
-					"</xref>\n\t\t\t<first_name>", given_names($father), "</first_name>\n\t\t\t<surname>",
-					normalize_name($father->surname()), "</surname>\n\t\t</father>";
+					"</xref>\n\t\t\t<first_name>", wide_to_xml(given_names($father)), "</first_name>\n\t\t\t<surname>",
+					wide_to_xml(normalize_name($father->surname())), "</surname>\n\t\t</father>";
 			}
 			print $csv '!';
 			if($mother && (my $m = person_line_html({ person => $mother, year => $year }))) {
 				print $html "<li>$m</li>";
 				print $csv person_line_csv({ person => $mother, year => $year });
 				print $xml "\n\t\t<mother>\n\t\t\t<xref>", $mother->xref(),
-					"</xref>\n\t\t\t<first_name>", given_names($mother), "</first_name>\n\t\t\t<surname>",
-					normalize_name($mother->surname()), "</surname>\n\t\t</mother>";
+					"</xref>\n\t\t\t<first_name>", wide_to_xml(given_names($mother)), "</first_name>\n\t\t\t<surname>",
+					wide_to_xml(normalize_name($mother->surname())), "</surname>\n\t\t</mother>";
 			}
 			print $html '</ol>';
 		} else {
@@ -14979,7 +14979,7 @@ sub wide_to_html
 	$string =~ s/\xc3\x81/&Aacute;/g;	# Á
 	$string =~ s/\xc3\x83/&Icirc;/g;	# Î
 	$string =~ s/\xc3\x9e/&THORN;/g;	# Þ
-	$string =~ s/\xc3\xa1/&aacute;/g;
+	$string =~ s/\xc3\xa1/&aacute;/g;	# á
 	$string =~ s/\xc3\xa2/&acirc;/g;
 	$string =~ s/\xc3\xa4/&auml;/g;
 	$string =~ s/\xc3\xa9/&eacute;/g;
@@ -15030,7 +15030,7 @@ sub wide_to_html
 	$string =~ s/\N{U+017E}/&zcaron;/g;
 	$string =~ s/\N{U+00D6}/&Ouml;/g;	# Ö
 	$string =~ s/\N{U+00DF}/&szlig;/g;	# ß
-	$string =~ s/\N{U+00E1}/&aacute;;/g;
+	$string =~ s/\N{U+00E1}/&aacute;/g;
 	$string =~ s/\N{U+00E2}/&acirc;/g;
 	$string =~ s/\N{U+00E4}/&auml;/g;
 	$string =~ s/\N{U+00E5}/&aring;/g;	# å
@@ -15041,6 +15041,7 @@ sub wide_to_html
 	$string =~ s/\N{U+00EE}/&icirc;/g;
 	$string =~ s/\N{U+00EF}/&iuml;/g;	# ï
 	$string =~ s/\N{U+00F0}/&eth;/g;	# ð
+	$string =~ s/\N{U+00F1}/&ntilde;/g;	# ñ
 	$string =~ s/\N{U+00F4}/&ocirc;/g;	# ô
 	$string =~ s/\N{U+00F6}/&ouml;/g;
 	$string =~ s/\N{U+00F8}/&oslash;/g;	# ø
@@ -15280,11 +15281,11 @@ sub wide_to_xml
 	$string =~ s/\xc2\xae/&#x0AE;/g;
 	$string =~ s/\xc3\x81/&#x0C1;/g;	# Á
 	$string =~ s/\xc3\x8e/&#x0CE;/g;	# Î
-	$string =~ s/\xc3\xa1/&#x0E1;/g;
+	$string =~ s/\xc3\xa1/&#x0E1;/g;	# á
 	$string =~ s/\xc3\xa5/&#x0E5;/g;	# å
 	$string =~ s/\xc3\xa9/&#x0E9;/g;
 	$string =~ s/\xc3\xaf/&#x0EF;/g;	# ï
-	$string =~ s/\xc3\xb1/&#x0F1;/g;	# ntilde
+	$string =~ s/\xc3\xb1/&#x0F1;/g;	# ntilde ñ
 	$string =~ s/\xc5\xa1/&#x161;/g;
 	$string =~ s/\xc4\x8d/&#x10D;/g;
 	$string =~ s/\xc5\xbe/&#x17E;/g;	# ž
@@ -15317,6 +15318,7 @@ sub wide_to_xml
 	$string =~ s/\xe2\x80\x9d/&quot;/g;
 	$string =~ s/\xe2\x80\xa6/.../g;
 	$string =~ s/\xe2\x97\x8f/&#x25CF;/g;	# ●
+	$string =~ s/\xe3\xb1/&#x0F1;/g;	# ntilde ñ - what's this one?
 	# $string =~ s/\xe4\x8d/&#x10D;/g;	# ? ACOM strangeness
 	# $string =~ s/\N{U+0161}/&#x161;/g;
 	# $string =~ s/\N{U+010D}/&#x10D;/g;
@@ -15447,6 +15449,10 @@ sub wide_to_xml
 
 	# print STDERR __LINE__, ": ($string)\n";
 	if($string =~ /[^[:ascii:]]/) {
+		print STDERR (unpack 'H*', $string);
+		print STDERR __LINE__, ': ';
+		print STDERR (sprintf '%v02X', $string);
+		print STDERR "\n";
 		my $i = 0;
 		while((my @call_details = caller($i++))) {
 			print STDERR "\t", colored($call_details[2] . ' of ' . $call_details[1], 'red'), "\n";
@@ -15682,7 +15688,7 @@ sub dbpedia
 											}
 										}
 									}
-									warn ">>>>>>>>>> dbpedia allowig possibility $uri" if($opts{'w'});
+									warn ">>>>>>>>>> dbpedia allowing possibility $uri" if($opts{'w'});
 								}
 							}
 							$Data::Dumper::Maxdepth = 0;