just converted (indent+comments) from estraier

src/index/csguess.cpp

@@ -0,0 +1,135 @@
+#ifndef lint
+static char	rcsid[] = "@(#$Id: csguess.cpp,v 1.1 2004-12-15 08:21:05 dockes Exp $ (C) 2004 J.F.Dockes";
+#endif
+// This code was converted from estraier / qdbm / myconf.c
+#include <errno.h>
+
+#include <iconv.h>
+#include "csguess.h"
+
+#include <string>
+using std::string;
+
+// The values from estraier were 32768, 256, 0.001
+const int ICONVCHECKSIZ = 4000;
+const int ICONVMISSMAX  = 10;
+const double ICONVALLWRAT = 0.001;
+
+// Try to transcode and count errors (for charset guessing)
+static int transcodeErrCnt(const char *ptr, int size, 
+			   const char *icode, const char *ocode)
+{
+    iconv_t ic;
+    char obuf[ICONVCHECKSIZ], *wp, *rp;
+    size_t isiz, osiz;
+    int miss;
+    isiz = size;
+    if((ic = iconv_open(ocode, icode)) == (iconv_t)-1) return ICONVMISSMAX;
+    miss = 0;
+    rp = (char *)ptr;
+    while(isiz > 0){
+	osiz = ICONVCHECKSIZ;
+	wp = obuf;
+	if(iconv(ic, (const char **)&rp, &isiz, &wp, &osiz) == -1){
+	    if(errno == EILSEQ || errno == EINVAL){
+		rp++;
+		isiz--;
+		miss++;
+		if(miss >= ICONVMISSMAX) 
+		    break;
+	    } else {
+		break;
+	    }
+	}
+    }
+    if(iconv_close(ic) == -1) 
+	return ICONVMISSMAX;
+    return miss;
+}
+
+
+string csguess(const string &in)
+{
+    const char     *hypo;
+    int		i, miss;
+    const char *text = in.c_str();
+    bool cr = false;
+
+    int size = in.length();
+    if (size > ICONVCHECKSIZ)
+	size = ICONVCHECKSIZ;
+
+    // UTF-16 with normal prefix ?
+    if (size >= 2 && (!memcmp(text, "\xfe\xff", 2) || 
+		      !memcmp(text, "\xff\xfe", 2)))
+	return "UTF-16";
+
+    // If we find a zero at an appropriate position, guess it's UTF-16 
+    // anyway. This is a quite expensive test for other texts as we'll 
+    // have to scan the whole thing.
+    for (i = 0; i < size - 1; i += 2) {
+	if (text[i] == 0 && text[i + 1] != 0)
+	    return "UTF-16BE";
+	if (text[i + 1] == 0 && text[i] != 0)
+	    return "UTF-16LE";
+    }
+
+    // Look for iso-2022 specific escape sequences. As iso-2022 begins
+    // in ascii, these succeed fast for a japanese text, but are quite
+    // expensive for any other
+    for (i = 0; i < size - 3; i++) {
+	if (text[i] == 0x1b) {
+	    i++;
+	    if (text[i] == '(' && strchr("BJHI", text[i + 1]))
+		return "ISO-2022-JP";
+	    if (text[i] == '$' && strchr("@B(", text[i + 1]))
+		return "ISO-2022-JP";
+	}
+    }
+
+    // Try conversions from ascii and utf-8. These are unlikely to succeed
+    // by mistake.
+    if (transcodeErrCnt(text, size, "US-ASCII", "UTF-16BE") < 1)
+	return "US-ASCII";
+
+    if (transcodeErrCnt(text, size, "UTF-8", "UTF-16BE") < 1)
+	return "UTF-8";
+
+    hypo = 0;
+    for (i = 0; i < size; i++) {
+	if (text[i] == 0xd) {
+	    cr = true;
+	    break;
+	}
+    }
+
+    if (cr) {
+	if ((miss = transcodeErrCnt(text, size, "Shift_JIS", "EUC-JP")) < 1)
+	    return "Shift_JIS";
+	if (!hypo && miss / (double)size <= ICONVALLWRAT)
+	    hypo = "Shift_JIS";
+	if ((miss = transcodeErrCnt(text, size, "EUC-JP", "UTF-16BE")) < 1)
+	    return "EUC-JP";
+	if (!hypo && miss / (double)size <= ICONVALLWRAT)
+	    hypo = "EUC-JP";
+    } else {
+	if ((miss = transcodeErrCnt(text, size, "EUC-JP", "UTF-16BE")) < 1)
+	    return "EUC-JP";
+	if (!hypo && miss / (double)size <= ICONVALLWRAT)
+	    hypo = "EUC-JP";
+	if ((miss = transcodeErrCnt(text, size, "Shift_JIS", "EUC-JP")) < 1)
+	    return "Shift_JIS";
+	if (!hypo && miss / (double)size <= ICONVALLWRAT)
+	    hypo = "Shift_JIS";
+    }
+    if ((miss = transcodeErrCnt(text, size, "UTF-8", "UTF-16BE")) < 1)
+	return "UTF-8";
+    if (!hypo && miss / (double)size <= ICONVALLWRAT)
+	hypo = "UTF-8";
+    if ((miss = transcodeErrCnt(text, size, "CP932", "UTF-16BE")) < 1)
+	return "CP932";
+    if (!hypo && miss / (double)size <= ICONVALLWRAT)
+	hypo = "CP932";
+
+    return hypo ? hypo : "ISO-8859-1";
+}

src/index/csguess.h

@@ -0,0 +1,12 @@
+#ifndef _CSGUESS_H_INCLUDED_
+#define _CSGUESS_H_INCLUDED_
+/* @(#$Id: csguess.h,v 1.1 2004-12-15 08:21:05 dockes Exp $  (C) 2004 J.F.Dockes */
+
+#include <string>
+
+// Try to guess the character set. This might guess unicode encodings, and
+// some asian charsets, but has no chance, for example, of discriminating
+// betweeen the different iso8859-xx charsets.
+extern std::string csguess(const std::string &in);
+
+#endif /* _CSGUESS_H_INCLUDED_ */