Added insertbom and called from ansi2unicode.sh

2 files changed, 82 insertions, 0 deletions

diff --git a/ansi2unicode.sh b/ansi2unicode.sh
index 246530e..bb8e0fb 100755
--- a/ansi2unicode.sh
+++ b/ansi2unicode.sh
@@ -17,4 +17,5 @@ for i in $@ ; do
     TMP="$i.$$"
     iconv -f $FROM -t $TO "$i" > "$TMP"
     mv "$TMP" "$i"
+    insertbom "$i"
 done
diff --git a/insertbom.c b/insertbom.c
new file mode 100644
index 0000000..ff5833e
--- /dev/null
+++ b/insertbom.c
@@ -0,0 +1,81 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <ctype.h>
+#include <string.h>
+#include <sys/stat.h>
+
+static off_t Size(const char *name)
+{
+    struct stat st = {0};
+
+    stat(name, &st);
+
+    return st.st_size;
+}
+
+static void Handle(const char *name)
+{
+    FILE *fp = NULL;
+    off_t len = Size(name);
+    unsigned char *buff = NULL;
+    int write = 0;
+
+    if (len < 3)
+    {
+    	printf("%s: too short\n", name);
+	return;
+    }
+
+    buff = malloc(len + 2);
+    fp = fopen(name, "rb");
+    fread(buff, 1, len, fp);
+    fclose(fp);
+
+    if ((buff[0] == 0xffu && buff[1] == 0xfeu) ||
+    	(buff[0] == 0xfeu && buff[1] == 0xffu))
+    {
+    	printf("%s: already has BOM\n", name);
+	free(buff);
+	return;
+    }
+
+    if (buff[0] == 0 && isprint(buff[1]))
+    {
+    	printf("%s: Guessing BE UTF-16\n", name);
+	write = 1;
+	memmove(buff, buff + 2, len);
+	buff[0] = 0xfe;
+	buff[1] = 0xff;
+    }
+    else if (buff[1] == 0 && isprint(buff[0]))
+    {
+    	printf("%s: Guessing LE UTF-16\n", name);
+	write = 1;
+	memmove(buff, buff + 2, len);
+	buff[0] = 0xff;
+	buff[1] = 0xfe;
+    }
+    else
+    {
+    	printf("%s: Leaving alone\n", name);
+    }
+
+    if (write)
+    {
+    	fp = fopen(name, "wb");
+	fwrite(buff, 1, len + 2, fp);
+	fclose(fp);
+    }
+
+    free(buff);
+}
+
+int main(int argc, char *argv[])
+{
+    int f;
+
+    for(f = 1; f < argc; f++)
+    {
+    	Handle(argv[f]);
+    }
+}