Multithreading

    - Multithreaded access to recoding functions using same context
    - Engine plugin to select between UTF-8 and ISO8859-1 for Western European Languages
    - Fix: 'rccTo' converting FS classes
    - FS class support in 'rccConfigSizedToCharset'

3 files changed, 77 insertions, 2 deletions

diff --git a/engines/Makefile.am b/engines/Makefile.am
index 7226682..404cc32 100644
--- a/engines/Makefile.am
+++ b/engines/Makefile.am
@@ -1,4 +1,4 @@
-lib_LTLIBRARIES =
+lib_LTLIBRARIES = libwestern.la
 
 libdir = $(pkgdatadir)/engines
 
@@ -8,4 +8,7 @@ librussian_la_SOURCES = russian.c
 librussian_la_LDFLAGS = -module -avoid-version -export-symbols-regex "rccGetInfo"
 endif
 
+libwestern_la_SOURCES = western.c
+libwestern_la_LDFLAGS = -module -avoid-version -export-symbols-regex "rccGetInfo"
+
 AM_CPPFLAGS = -I../src @RCD_INCLUDES@
diff --git a/engines/russian.c b/engines/russian.c
index 08b8310..0df145c 100644
--- a/engines/russian.c
+++ b/engines/russian.c
@@ -5,7 +5,7 @@
 #include <librcc.h>
 
 static rcc_autocharset_id AutoengineRussian(rcc_engine_context ctx, const char *buf, int len) {
-    return (rcc_charset_id)rcdGetRussianCharset(buf,len);
+    return (rcc_autocharset_id)rcdGetRussianCharset(buf,len);
 }
 
 static rcc_engine russian_engine = {
diff --git a/engines/western.c b/engines/western.c
new file mode 100644
index 0000000..4c6e1aa
--- /dev/null
+++ b/engines/western.c
@@ -0,0 +1,72 @@
+#include <stdio.h>
+#include <string.h>
+
+#include <librcc.h>
+
+#define bit(i) (1<<i)
+
+/* 
+ * Latin unicode subset:
+ * 0x100 - 0x17E
+ * 0x180 - 0x24F
+ * 0x1E00 - 0x1EFF
+ */
+
+static rcc_autocharset_id AutoengineWestern(rcc_engine_context ctx, const char *sbuf, int len) {
+    const unsigned char *buf = sbuf;
+    long i,j;
+    int bytes=0,rflag=0;
+    int res=0;
+
+    if (!len) len = strlen(buf);
+    for (i=0;i<len;i++) {
+	if (buf[i]<128) continue;
+	
+	if (bytes>0) {
+	    if ((buf[i]&0xC0)==0x80) {
+		if (rflag) {
+		    // Western is 0x100-0x17e
+		    res++;
+		}
+		bytes--;
+	    } else {
+		res--;
+		bytes=1-bytes;
+		rflag=0;
+	    }
+	} else {
+	    for (j=6;j>=0;j--)
+		if ((buf[i]&bit(j))==0) break;
+	    
+	    if ((j==0)||(j==6)) {
+		if ((j==6)&&(bytes<0)) bytes++;
+		else res--;
+		continue;
+	    }
+	    bytes=6-j;
+	    if (bytes==1) {
+		// Western Languages (C2-C3)
+		if (buf[i]==0xC2) rflag=1;
+		else if (buf[i]==0xC3) rflag=2;
+	    }
+	}
+	    
+	if ((buf[i]==0xC0)||(buf[i]==0xC1)) {
+	    if (i+1==len) break;
+	    	    
+	}
+    }
+
+    if (res > 0) return (rcc_autocharset_id)0;
+    return (rcc_autocharset_id)1;
+}
+
+static rcc_engine western_engine = {
+    "Western", NULL, NULL, &AutoengineWestern, {"UTF-8","ISO8859-1", NULL}
+};
+
+rcc_engine *rccGetInfo(const char *lang) {
+    if (!lang) return NULL;
+
+    return &western_engine;
+}