replace ugly utf8 cache with wrapper around c++ unordered_maps.

amoldeshpande · Jan 11, 2021 · 79a14db · 79a14db
1 parent 7a568a1
commit 79a14db
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 37 deletions.
diff --git a/win32/Makefile.win32 b/win32/Makefile.win32
@@ -67,13 +67,13 @@ BLDTYPE=retail
 
 
 NTLDFLAGS=  -entry:silly_entry  -debug
-NTCFLAGS = -Zi -DNTDBG   -Od
+NTCFLAGS = -Zi -DNTDBG -D_DEBUG  -Od
 LIBES= user32.lib advapi32.lib kernel32.lib $(DEBUG_CRTLIBS) strsafe.lib
 
 !else if "$(BLDTYPE)" == "retail"
 
 NTLDFLAGS=  -entry:silly_entry  -debug
-NTCFLAGS = -Zi -O2 
+NTCFLAGS = -Zi -O2  -DNDEBUG
 LIBES= user32.lib advapi32.lib kernel32.lib  $(CRTLIBS) strsafe.lib
 
 !else
@@ -148,14 +148,14 @@ NTSRCS1 = win32\io.c win32\stdio.c win32\dirent.c win32\signal.c\
 		win32\nt.char.c win32\bogus.c win32\console.c \
 		win32\ntfunc.c win32\ntb1.c win32\ntb2.c win32\globals.c \
 		win32\ps.c win32\nt.const.c win32\clip.c\
-		win32\nt.bind.c  win32\nt.screen.c
+		win32\nt.bind.c  win32\nt.screen.c win32\utf8_chars.cpp
 
 NTSRCS = $(NTSRCS1) win32\fork.c 
 
 NTOBJS = io.$(SUF) stdio.$(SUF) dirent.$(SUF) signal.$(SUF) support.$(SUF) \
 	    nt.char.$(SUF) bogus.$(SUF) console.$(SUF) fork.$(SUF) ntfunc.$(SUF) \
         globals.$(SUF)  ps.$(SUF) \
-		clip.$(SUF) nt.const.$(SUF) nt.bind.$(SUF) nt.screen.$(SUF)
+		clip.$(SUF) nt.const.$(SUF) nt.bind.$(SUF) nt.screen.$(SUF) utf8_chars.$(SUF)
 
 VHSRCS=$(PVSRCS) $(AVSRCS)
 
@@ -200,7 +200,6 @@ chlog:
 .c.$(SUF):
 	$(CC) $(CF) $(CFLAGS) $(REST_WARNING_LEVEL) $(DFLAGS) $(EXTRAFLAGS) $<
 
-
 ed.defns.h: ed.defns.c
 	-@del  $@
 	@echo /* Do not edit this file, make creates it. */ > $@
@@ -321,6 +320,9 @@ CFLAGS_WX = $(CFLAGS) $(WINDOWS_WARNING_LEVEL) -DWINDOWS_ONLY $(ANALYZE_FLAGS)
 {win32}.c.$(SUF): 
 	$(CC) $(CF) $(CFLAGS_WX) $(DFLAGS) $(EXTRAFLAGS) $<
 
+{win32}.cpp.$(SUF): 
+	$(CC) $(CF) $(CFLAGS_WX) $(DFLAGS) $(EXTRAFLAGS) $<
+
 fork.$(SUF):win32\fork.c 
 	$(CC) $(CF) $(CFLAGS_WX) $(DFLAGS) $(EXTRAFLAGS) -GS- win32\fork.c
 

diff --git a/win32/nt.screen.c b/win32/nt.screen.c
@@ -69,8 +69,8 @@ terminit(void)
 int T_ActualWindowSize;
 
 static	void	ReBufferDisplay	(void);
-static uint32_t* utf8_chars = NULL;
-static Char currentIndex = 0;
+Char get_or_cache_utf8_mb(uint32_t inChar);
+uint32_t get_cached_utf8_mb(Char);
 
 
 /*ARGSUSED*/
@@ -120,11 +120,6 @@ ReBufferDisplay(void)
 	b[TermV] = NULL;
 	Vdisplay = b;
 
-#ifdef WINNT_NATIVE_UTF8_SUPPORT
-	if(utf8_chars != NULL) {
-		memset(utf8_chars,0,NT_UTF8_MB);
-	}
-#endif// WINNT_NATIVE_UTF8_SUPPORT
 }
 
 	void
@@ -324,48 +319,26 @@ MoveToChar(int where)
 #ifdef WINNT_NATIVE_UTF8_SUPPORT
 Char nt_make_utf8_multibyte(Char* cp, int len) {
 
-	Char retIdex = 0;
 	uint32_t mbchar = 0;
 
 	if(len == 1){
 		return *cp;
 	}
-	if(utf8_chars == NULL){
-		utf8_chars = xmalloc(sizeof(uint32_t)*NT_UTF8_MB);
-	}
 	for(Char i = 0; i < len;i++) {
 		mbchar <<= 8;
 		mbchar |= *cp;
 		cp++;
 	}
-	for(Char i=0; i < NT_UTF8_MB;i++) {
-		if(utf8_chars[i] == mbchar) {
-			return i | NT_UTF8_MB;
-		}
-		if(utf8_chars[i] ==0) {
-			break;
-		}
-	}
-	retIdex = currentIndex++;
-
-	if(currentIndex == LITERAL){
-		currentIndex++;
-	}
-	//TODO deal with wraparound better
-	if(currentIndex == NT_UTF8_MB){
-		currentIndex = 0;
-	}
-
-	utf8_chars[retIdex] = mbchar;
-	return (retIdex | NT_UTF8_MB);
+	Char i = get_or_cache_utf8_mb(mbchar);
+	return i | NT_UTF8_MB;
 }
 #endif // WINNT_NATIVE_UTF8_SUPPORT
 void putraw_utf8(Char c) {
 #if defined(WINNT_NATIVE_UTF8_SUPPORT)
 	if (c & NT_UTF8_MB) {
 		Char index = c & ~NT_UTF8_MB;
 		if (index >= 0 && index < NT_UTF8_MB) {
-			uint32_t mbchar = utf8_chars[index];
+			uint32_t mbchar = get_cached_utf8_mb(index);
 			int start = 0;
 			//
 			// there have to be at least 2 bytes in the utf8 sequence

diff --git a/win32/utf8_chars.cpp b/win32/utf8_chars.cpp
@@ -0,0 +1,34 @@
+#include <unordered_map>
+#include <sh.h>
+
+static std::unordered_map<uint32_t,Char> utf8_chars;
+static std::unordered_map<Char,uint32_t> reverse_map;
+
+static Char currentIndex;
+extern "C" Char get_or_cache_utf8_mb(uint32_t inChar) {
+	auto iter = utf8_chars.find(inChar);
+	if(iter == utf8_chars.end()) {
+		auto ret = utf8_chars.insert({inChar,currentIndex});	
+		currentIndex++;
+		if(currentIndex == LITERAL ) {
+			currentIndex++;
+		}
+		if (ret.second) {
+			reverse_map.insert({ret.first->second,inChar});
+			return ret.first->second;
+		}
+		return 0;
+	}
+	else {
+		return iter->second;
+	}
+}
+extern "C" uint32_t get_cached_utf8_mb(Char index) {
+
+	auto iter = reverse_map.find(index);
+	if(iter == reverse_map.end()) {
+		return 0x00e29aa0;  //"\U+26A0", Warning sign
+	}
+	return iter->second;
+}
+