Fix 3 non-UTF8 markup characters, fix and update UTF8 whitespace

diff --git a/qcsrc/common/command/markup.qc b/qcsrc/common/command/markup.qc
index bd1b79cbfbdb05dd3b3644b4c79f696f12ec2210..2397982da99288ca829b3cad5f1dc5883f035bfb 100644 (file)
--- a/qcsrc/common/command/markup.qc
+++ b/qcsrc/common/command/markup.qc
@@ -12,48 +12,50 @@ void GenericCommand_markup_init()
        if (markup_init)
                return;
        markup_init = true;
+       // "&<-" doesn't have a corresponding symbol anymore
+       // replaced with similar triangle facing left ("&<|")
        int i = 0;
-       markup_from[i] = "&alien"; markup_to[i] = "\x12"; ++i;
-       markup_from[i] = "&:-)"; markup_to[i] = "\x13"; ++i;
-       markup_from[i] = "&:-("; markup_to[i] = "\x14"; ++i;
-       markup_from[i] = "&x-P"; markup_to[i] = "\x15"; ++i;
-       markup_from[i] = "&:-/"; markup_to[i] = "\x16"; ++i;
-       markup_from[i] = "&:-D"; markup_to[i] = "\x17"; ++i;
-       markup_from[i] = "&<<"; markup_to[i] = "\x18"; ++i;
-       markup_from[i] = "&>>"; markup_to[i] = "\x19"; ++i;
-       markup_from[i] = "&dot"; markup_to[i] = "\x1a"; ++i;
-       markup_from[i] = "&^_"; markup_to[i] = "\x1b"; ++i;
+       markup_from[i] = "&alien";  markup_to[i] = "\x12"; ++i;
+       markup_from[i] = "&:-)";    markup_to[i] = "\x13"; ++i;
+       markup_from[i] = "&:-(";    markup_to[i] = "\x14"; ++i;
+       markup_from[i] = "&x-P";    markup_to[i] = "\x15"; ++i;
+       markup_from[i] = "&:-/";    markup_to[i] = "\x16"; ++i;
+       markup_from[i] = "&:-D";    markup_to[i] = "\x17"; ++i;
+       markup_from[i] = "&<<";     markup_to[i] = "\x18"; ++i;
+       markup_from[i] = "&>>";     markup_to[i] = "\x19"; ++i;
+       markup_from[i] = "&dot";    markup_to[i] = "\x1a"; ++i;
+       markup_from[i] = "&^_";     markup_to[i] = "\x1b"; ++i;
        markup_from[i] = "&ysplat"; markup_to[i] = "\x1c"; ++i;
-       markup_from[i] = "&-]"; markup_to[i] = "\x1d"; ++i;
-       markup_from[i] = "&--"; markup_to[i] = "\x1e"; ++i;
-       markup_from[i] = "&[-"; markup_to[i] = "\x1f"; ++i;
-       markup_from[i] = "&s<"; markup_to[i] = "\x2c"; ++i;
-       markup_from[i] = "&s>"; markup_to[i] = "\x2e"; ++i;
-       markup_from[i] = "&<-"; markup_to[i] = "\x7f"; ++i;
-       markup_from[i] = "&[="; markup_to[i] = "\x80"; ++i;
-       markup_from[i] = "&=="; markup_to[i] = "\x81"; ++i;
-       markup_from[i] = "&=]"; markup_to[i] = "\x82"; ++i;
-       markup_from[i] = "&r!"; markup_to[i] = "\x84"; ++i;
-       markup_from[i] = "&|o|"; markup_to[i] = "\x85"; ++i;
-       markup_from[i] = "&|u|"; markup_to[i] = "\x86"; ++i;
-       markup_from[i] = "&|i|"; markup_to[i] = "\x87"; ++i;
-       markup_from[i] = "&|c|"; markup_to[i] = "\x88"; ++i;
-       markup_from[i] = "&[c]"; markup_to[i] = "\x89"; ++i;
-       markup_from[i] = "&[n]"; markup_to[i] = "\x8a"; ++i;
-       markup_from[i] = "&[]"; markup_to[i] = "\x8b"; ++i;
-       markup_from[i] = "&r?"; markup_to[i] = "\x8c"; ++i;
-       markup_from[i] = "&|>"; markup_to[i] = "\x8d"; ++i;
+       markup_from[i] = "&-]";     markup_to[i] = "\x1d"; ++i;
+       markup_from[i] = "&--";     markup_to[i] = "\x1e"; ++i;
+       markup_from[i] = "&[-";     markup_to[i] = "\x1f"; ++i;
+       markup_from[i] = "&s<";     markup_to[i] = "\xac"; ++i;
+       markup_from[i] = "&s>";     markup_to[i] = "\xae"; ++i;
+       markup_from[i] = "&<-";     markup_to[i] = "\xff"; ++i; // same as "&<|"
+       markup_from[i] = "&[=";     markup_to[i] = "\x80"; ++i;
+       markup_from[i] = "&==";     markup_to[i] = "\x81"; ++i;
+       markup_from[i] = "&=]";     markup_to[i] = "\x82"; ++i;
+       markup_from[i] = "&r!";     markup_to[i] = "\x84"; ++i;
+       markup_from[i] = "&|o|";    markup_to[i] = "\x85"; ++i;
+       markup_from[i] = "&|u|";    markup_to[i] = "\x86"; ++i;
+       markup_from[i] = "&|i|";    markup_to[i] = "\x87"; ++i;
+       markup_from[i] = "&|c|";    markup_to[i] = "\x88"; ++i;
+       markup_from[i] = "&[c]";    markup_to[i] = "\x89"; ++i;
+       markup_from[i] = "&[n]";    markup_to[i] = "\x8a"; ++i;
+       markup_from[i] = "&[]";     markup_to[i] = "\x8b"; ++i;
+       markup_from[i] = "&r?";     markup_to[i] = "\x8c"; ++i;
+       markup_from[i] = "&|>";     markup_to[i] = "\x8d"; ++i;
        markup_from[i] = "&splat0"; markup_to[i] = "\x8e"; ++i;
        markup_from[i] = "&splat1"; markup_to[i] = "\x8f"; ++i;
-       markup_from[i] = "&[["; markup_to[i] = "\x90"; ++i;
-       markup_from[i] = "&]]"; markup_to[i] = "\x91"; ++i;
+       markup_from[i] = "&[[";     markup_to[i] = "\x90"; ++i;
+       markup_from[i] = "&]]";     markup_to[i] = "\x91"; ++i;
        markup_from[i] = "&splat2"; markup_to[i] = "\x9a"; ++i;
-       markup_from[i] = "&)("; markup_to[i] = "\x9b"; ++i;
+       markup_from[i] = "&)(";     markup_to[i] = "\x9b"; ++i;
        markup_from[i] = "&splat3"; markup_to[i] = "\x9c"; ++i;
-       markup_from[i] = "&(."; markup_to[i] = "\x9d"; ++i;
-       markup_from[i] = "&.."; markup_to[i] = "\x9e"; ++i;
-       markup_from[i] = "&.)"; markup_to[i] = "\x9f"; ++i;
-       markup_from[i] = "&<|"; markup_to[i] = "\xff"; ++i;
+       markup_from[i] = "&(.";     markup_to[i] = "\x9d"; ++i;
+       markup_from[i] = "&..";     markup_to[i] = "\x9e"; ++i;
+       markup_from[i] = "&.)";     markup_to[i] = "\x9f"; ++i;
+       markup_from[i] = "&<|";     markup_to[i] = "\xff"; ++i;
 }
 
 string GenericCommand_markup(string s2)

diff --git a/qcsrc/lib/string.qh b/qcsrc/lib/string.qh
index a7508b8fb85c236f61619684dc1e692d5fec0859..946438ca5f836ae29c8a2f8e276e46f8cd1330f6 100644 (file)
--- a/qcsrc/lib/string.qh
+++ b/qcsrc/lib/string.qh
@@ -312,7 +312,10 @@ bool isInvisibleString(string s)
                        case 192:          // charmap space
                                if (!utf8) break;
                                return false;
-                       case 0xE000 + 192: // utf8 charmap space
+                       case 0xE000: // invisible char of the utf8 quake charmap
+                       case 0xE00A: // invisible char of the utf8 quake charmap
+                       case 0xE0A0: // invisible char of the utf8 quake charmap
+                       case 0xE020: // invisible char of the utf8 quake charmap
                        case 0x00A0: // NO-BREAK SPACE
                        //case 0x1680: // OGHAM SPACE MARK
                        case 0x180E: // MONGOLIAN VOWEL SEPARATOR