Support special ${U+1}...${U+FFFFFFFF} Unicode character defines

git-svn-id: https://svn.code.sf.net/p/nsis/code/NSIS/trunk@6324 212acab6-be3b-0410-9dea-997c60f758d6

Docs/src/compiler.but

@@ -213,6 +213,13 @@ NSIS version as a 32 bit number.
 \c   !error "NSIS 2.47 or higher is required to build this installer!"
 \c !endif
 
+\S1{preunicodecodepoint} $\{U+1\}...$\{U+FFFFFFFF\}
+
+A Unicode (UCS-4) character.
+
+\c !define U+ABC "SIS" # Define will override
+\c DetailPrint "${U+2115}${U+ABC}" # DOUBLE-STRUCK CAPITAL N + "SIS"
+
 \S1{scopepredefines} Scope Predefines
 
 Standard predefines that contain information of the current code scope.

Docs/src/history.but

@@ -24,6 +24,8 @@ Released on ?, 2013
 
 \b The index of the changed section is stored in $0 during .onSelChange callbacks (\W{http://sourceforge.net/support/tracker.php?aid=1634936}{RFE #1634936})
 
+\b $\{U+1\}...$\{U+FFFFFFFF\} are treated as a Unicode codepoint unless there is already a define with that name
+
 \S2{} Minor Changes
 
 \b %temp%\\Low will be used if the installer cannot write to %temp% nor %windir%\\Temp (\W{http://sourceforge.net/support/tracker.php?aid=2909242}{bug #2909242}, \W{http://sourceforge.net/support/tracker.php?aid=2912824}{patch #2912824})

Source/script.cpp

@@ -684,6 +684,23 @@ void CEXEBuild::ps_addtoline(const TCHAR *str, GrowBuf &linedata, StringList &hi
           ps_addtoline(s,defname,hist);
           defname.add(_T(""),sizeof(_T("")));
           t=definedlist.find((TCHAR*)defname.get());
+          TCHAR unichar[4+1];
+          if (!t && _T('U')==s[0] && _T('+')==s[1])
+          {
+            TCHAR *n=s+2;
+            unsigned long utf32=_tcstoul(n,&t,16);
+            // We only want to accept "${U+HEXDIGITS}" and not "${U+ -HEXDIGITS }"
+            if (*t || _T('-')==*n || _T('+')==*n) t=0;
+            if (_T(' ')==*n || _T('\t')==*n) t=0; // TODO: _istspace()?
+            if (!utf32) t=0; // Don't allow "${U+0}"
+            if (t)
+            {
+              UINT32 codpts[]={utf32,UNICODE_REPLACEMENT_CHARACTER,'?'};
+              for(UINT i=0; i < COUNTOF(codpts); ++i)
+                if (WCFromCodePoint(unichar,COUNTOF(unichar),codpts[i])) break;
+              t=unichar;
+            }
+          }
           if (t && hist.find((TCHAR*)defname.get(),0)<0)
           {
             in+=_tcslen(s)+2;

Source/utf.cpp

@@ -56,6 +56,33 @@ inline UINT UTF8ToWC_Prepare(LPCSTR StrU8,UINT cbU8)
   return UTF8ToWC_Convert(StrU8,cbU8,0,0);
 }
 
+UINT WCFromCodePoint(wchar_t*Dest,UINT cchDest,UINT32 CodPt)
+{
+  // Don't allow half surrogate pairs
+  if (CodPt >= 0xd800 && CodPt <= 0xdfff) CodPt = UNICODE_REPLACEMENT_CHARACTER;
+#ifdef _WIN32
+  if (CodPt <= 0xffff && cchDest)
+  {
+    *Dest = (wchar_t) CodPt;
+    return 1;
+  }
+  else if (cchDest >= 2)
+  {
+    const UINT32 lead_offset = 0xd800 - (0x10000 >> 10);
+    UINT16 lead = lead_offset + (CodPt >> 10), trail = 0xdc00 + (CodPt & 0x3ff);
+    Dest[0] = lead, Dest[1] = trail;
+    return 2;
+  }
+  return 0;
+#else
+  iconvdescriptor iconvd;
+  if (!iconvd.Open("wchar_t",iconvd::GetHostEndianUCS4Code())) return 0;
+  size_t inleft = 4;
+  UINT cchW = iconvd.Convert(&codpt,&inleft,Dest,cchDest*sizeof(wchar_t)) / sizeof(wchar_t);
+  return !inleft ? cchW : 0;
+#endif
+}
+
 wchar_t* DupWCFromBytes(void*Buffer,UINT cbBuffer,WORD SrcCP)
 {
   /*\

Source/utf.h

@@ -23,6 +23,8 @@
 #include <stdio.h>
 #include "util.h" // For my_fopen
 
+const UINT16 UNICODE_REPLACEMENT_CHARACTER = 0xfffd;
+
 #define TSTR_INPUTCHARSET _T("ACP|OEM|CP#|UTF8|UTF16LE")
 
 
@@ -54,6 +56,7 @@ inline UINT32 CodePointFromUTF16SurrogatePair(unsigned short lea,unsigned short
 UINT StrLenUTF16LE(const void*str);
 bool StrSetUTF16LE(tstring&dest, const void*src);
 
+UINT WCFromCodePoint(wchar_t*Dest,UINT cchDest,UINT32 CodPt);
 wchar_t* DupWCFromBytes(void*Buffer,UINT cbBuffer,WORD SrcCP);
 UINT DetectUTFBOM(FILE*strm);
 WORD GetEncodingFromString(const TCHAR*s);