SF.net SVN: frontierkernel: [1327] Frontier/trunk/Common/source

[<set...@us...>]

Revision: 1327
Author:   sethdill
Date:     2006-04-23 18:03:36 -0700 (Sun, 23 Apr 2006)
ViewCVS:  http://svn.sourceforge.net/frontierkernel/?rev=1327&view=rev

Log Message:
-----------
First half of changes to re-work string.convertCharset(), so that it now takes internet naems (IANA names) instead of platform-specific codes.
 - also, added new verb string.ischarsetavailable() to make life a little easier for the scripter. Takes one argument, the internet-friendly name of the character set whose existence is being tested (like 'utf-8').

Modified Paths:
--------------
    Frontier/trunk/Common/headers/strings.h
    Frontier/trunk/Common/resources/Mac/kernelverbs.r
    Frontier/trunk/Common/source/strings.c
    Frontier/trunk/Common/source/stringverbs.c
Modified: Frontier/trunk/Common/headers/strings.h
===================================================================
--- Frontier/trunk/Common/headers/strings.h	2006-04-22 21:06:49 UTC (rev 1326)
+++ Frontier/trunk/Common/headers/strings.h	2006-04-24 01:03:36 UTC (rev 1327)
@@ -266,9 +266,9 @@
 
 extern boolean utf8tomacroman (Handle, Handle); /* 2006-02-25 creedon */
 
-extern boolean gettextencodingid (const Handle, long *);  /* 2006-04-20 smd */
+extern boolean convertCharset( Handle, Handle, const bigstring, const bigstring );
 
-extern boolean converttextencoding (Handle, Handle, long, long);  /* 2006-04-14 smd */
+extern boolean isTextEncodingAvailable( const bigstring ); /* 2006-04-23 smd */
 
 #endif
 

Modified: Frontier/trunk/Common/resources/Mac/kernelverbs.r
===================================================================
--- Frontier/trunk/Common/resources/Mac/kernelverbs.r	2006-04-22 21:06:49 UTC (rev 1326)
+++ Frontier/trunk/Common/resources/Mac/kernelverbs.r	2006-04-24 01:03:36 UTC (rev 1327)
@@ -601,7 +601,8 @@
 		"multipleReplaceAll",
 		"macromantoutf8", // 2006-02-25 creedon
 		"utf8tomacroman", // 2006-02-25 creedon
-		"convertcharset"  /* 2006-04-13 smd */
+		"convertcharset",		/* 2006-04-13 smd */
+		"ischarsetavailable"	/* 2006-04-23 smd */
 		}
 	}
 };

Modified: Frontier/trunk/Common/source/strings.c
===================================================================
--- Frontier/trunk/Common/source/strings.c	2006-04-22 21:06:49 UTC (rev 1326)
+++ Frontier/trunk/Common/source/strings.c	2006-04-24 01:03:36 UTC (rev 1327)
@@ -56,8 +56,19 @@
 static byte bshexprefix [] = STR_hexprefix;
 
 
+/* declarations */
 
+static boolean converttextencoding( Handle, Handle, const long, const long );
+static boolean getTextEncodingID( const bigstring, long * );
 
+#ifdef WIN95VERSION
+	static HRESULT UnicodeToAnsi(Handle, Handle, unsigned long);
+	static HRESULT AnsiToUnicode(Handle, Handle, unsigned long);
+#endif
+
+
+/* definitions */
+
 boolean equalstrings (const bigstring bs1, const bigstring bs2) {
 
 	/*
@@ -2341,21 +2352,15 @@
 
 /* convert an "internet name" for an encoding into a platform-specific id for the character set */
 
-boolean gettextencodingid (const Handle h, long * encodingId) {
+static boolean getTextEncodingID( const bigstring bsEncodingName, long * encodingId ) {
 
 #ifdef MACVERSION
 
 	OSStatus err;
 	TextEncoding oneEncoding;
-	bigstring ianaName;
-	long ct, ix = 0;
 	
-	ct = gethandlesize (h);
-	ianaName [0] = (byte) ct;
-	loadfromhandle (h, &ix, ct, ianaName + 1);
+	err = TECGetTextEncodingFromInternetName( &oneEncoding, bsEncodingName );
 	
-	err = TECGetTextEncodingFromInternetName( &oneEncoding, ianaName );
-	
 	if ( err != noErr )
 		return (false);
 	
@@ -2366,8 +2371,27 @@
 	return (true);
 }
 
+/*
+	determine if the specified character set is recognized by the OS
+	bsEncodingName is an IANA-friendly name like "utf-8" , "macintosh", or "iso-8859-1"
+*/
+boolean isTextEncodingAvailable( const bigstring bsEncodingName )
+{
+	long encodingId;
+	boolean fl;
+	
+	fl = getTextEncodingID( bsEncodingName, &encodingId );
+	
+	if ( ! fl )
+		return (false);
+	
+	if ( encodingId < 0 )
+		return (false);
+	
+	return (true);
+}
 
-boolean converttextencoding (Handle h, Handle hresult, long inputcharset, long outputcharset) {
+static boolean converttextencoding( Handle h, Handle hresult, const long inputcharset, const long outputcharset) {
 
 #ifdef MACVERSION /*Mac character conversion common routine*/
 
@@ -2378,10 +2402,6 @@
 	long pullBytes = 0;
 	long lentext = gethandlesize (h);
 
-	// sethandlesize (h, lentext + 1);
-
-	// (*h) [lentext] = '\0';
-
 	status = TECCreateConverter (&converter, inputcharset, outputcharset);
 	if ( status != noErr ) {
 		TECDisposeConverter (converter);
@@ -2419,15 +2439,6 @@
 		}
 	}
 
-	//old code
-	// if (((*h) [0] == '\xFF') || ((*h) [0] == '\xEF')) /*pop byte order mark*/
-	//	pullfromhandle (h, 0, 3, NULL);
-
-	/*
-	if (pullBytes > 0)
-		pullfromhandle (h, 0, pullBytes, NULL);
-	*/
-	
 	lentext = gethandlesize (h) - pullBytes;
 
 	status = TECConvertText (converter, (ConstTextPtr)(*h + pullBytes), lentext, &ctorigbytes, (TextPtr)(*hresult), sizeoutputbuffer, &ctoutputbytes);
@@ -2445,10 +2456,6 @@
 	
 	return (true);
 
-	// kwchange 2005-05-23 --- commented out to see if this is the missing trailing char bug
-	// kwchange 2005-05-27 --- ran this the last week and have no complains
-	//sethandlesize (hresult, gethandlesize (hresult) - 1); /*pop trailing terminator*/
-
 #endif /*end Mac character conversion routine(s)*/
 
 #ifdef WIN95VERSION
@@ -2517,6 +2524,34 @@
 	} /*converttextencoding*/
 
 
+boolean convertCharset( Handle hString, Handle hresult, const bigstring charsetIn, const bigstring charsetOut )
+{
+#ifdef MACVERSION
+	
+	TextEncoding teInputSet, teOutputSet;
+	
+	if ( ! getTextEncodingID( charsetIn, (long *) &teInputSet ) )
+	{
+		// FIXME: generate error, encoding not available
+		return (false);
+	}
+	
+	if ( ! getTextEncodingID( charsetOut, (long *) &teOutputSet ) )
+	{
+		// FIXME: generate error, encoding not available
+		return (false);
+	}
+	
+	return converttextencoding( hString, hresult, (long) teInputSet, (long) teOutputSet );
+	
+#endif
+
+#ifdef WIN95VERSION
+
+#endif
+}
+
+
 boolean utf16toansi (Handle h, Handle hresult) {
 
 	/*

Modified: Frontier/trunk/Common/source/stringverbs.c
===================================================================
--- Frontier/trunk/Common/source/stringverbs.c	2006-04-22 21:06:49 UTC (rev 1326)
+++ Frontier/trunk/Common/source/stringverbs.c	2006-04-24 01:03:36 UTC (rev 1327)
@@ -312,6 +312,8 @@
 	utf8tomacromanfunc, // 2006-02-25 creedon
 	
 	convertcharsetfunc, /* 2006-04-14 smd */
+	
+	ischarsetavailablefunc, /* 2006-04-23 smd */
 
 	ctstringverbs
 	} tystringtoken;
@@ -1319,6 +1321,21 @@
 	} /*latintomac*/
 
 
+static boolean isCharsetAvailableVerb( hdltreenode hp1, tyvaluerecord *v )
+{
+	bigstring bscharset;
+	
+	flnextparamislast = true;
+	
+	if ( !getstringvalue( hp1, 1, bscharset ) ) /* IANA names can't be longer than 40 ASCII characters, so this is safe */
+		return (false);
+	
+	(*v).data.flvalue = isTextEncodingAvailable( bscharset );
+	
+	return (true);
+}
+
+
 static boolean stringfunctionvalue (short token, hdltreenode hparam1, tyvaluerecord *vreturned, bigstring bserror) {
 	
 	/*
@@ -2302,32 +2319,39 @@
 			return (setheapvalue (hresult, stringvaluetype, v));
 			}
 		
-		case convertcharsetfunc: {  /* 2006-04-13 smd: convert any character set to any other character set, if the os supports it */
-			
+		case convertcharsetfunc:  /* 2006-04-13 smd: convert any character set to any other character set, if the os supports it */
+		{
 			Handle h, hresult;
-			long charsetIn, charsetOut;
+			bigstring charsetIn, charsetOut;
 			
-			if (!getlongvalue (hp1, 1, &charsetIn))
+			/*
+				the "internet names" of character sets are 40 ASCII chars or less,
+				so this is a safe way to get the params
+			*/
+			if ( !getstringvalue( hp1, 1, charsetIn ) )
 				goto error;
 			
-			if (!getlongvalue (hp1, 2, &charsetOut))
+			if ( !getstringvalue( hp1, 2, charsetOut ) )
 				goto error;
 			
 			flnextparamislast = true;
 			
-			if (!getreadonlytextvalue (hp1, 3, &h))
+			if ( !getreadonlytextvalue( hp1, 3, &h ) )
 				goto error;
 			
-			newemptyhandle (&hresult);
+			newemptyhandle( &hresult );
 			
-			if (! converttextencoding (h, hresult, charsetIn, charsetOut))
+			if ( !convertCharset( h, hresult, charsetIn, charsetOut ) )
 				goto error;
 			
-			// disposehandle(h);
-			
-			return (setheapvalue (hresult, stringvaluetype, v));
-			}
+			return ( setheapvalue( hresult, stringvaluetype, v ) );
+		}
 		
+		case ischarsetavailablefunc:  /* 2006-04-23 smd: determine if a character set is available, based on an internet-friendly (IANA) name */
+		{
+			return isCharsetAvailableVerb( hparam1, v );
+		}
+		
 		default:
 			errornum = notimplementederror;
 			


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.