diff --git a/lib/modules/Locale.pmod/Charset.pmod/module.pmod b/lib/modules/Locale.pmod/Charset.pmod/module.pmod
index c597d33667f7b3da8b874c9857654a31f093a067..aeea12752dde9806b8ee76d101f2b404f90dcd8e 100644
--- a/lib/modules/Locale.pmod/Charset.pmod/module.pmod
+++ b/lib/modules/Locale.pmod/Charset.pmod/module.pmod
@@ -106,6 +106,15 @@ protected private inherit _Charset;
//! }
class Decoder
{
+ //! @decl string charset;
+ //!
+ //! Name of the charset - giving this name to @[decoder] returns an
+ //! instance of the same class as this object.
+ //!
+ //! @note
+ //! This is not necessarily the same name that was actually given to
+ //! @[decoder] to produce this object.
+
//! Feeds a string to the decoder.
//!
//! @param s
@@ -116,6 +125,13 @@ class Decoder
//! of calls.
this_program feed(string s);
+ // FIXME: There ought to be a finish(string s) function. Now it's
+ // possible that certain kinds of coding errors are simply ignored
+ // if they occur last in a string. E.g.
+ // Locale.Charset.decoder("utf8")->feed("\345")->drain() returns ""
+ // instead of throwing an error about the incomplete UTF8 sequence,
+ // and there's no good way to get this error.
+
//! Get the decoded data, and reset buffers.
//!
//! @returns
@@ -136,6 +152,15 @@ class Encoder
//! An encoder only differs from a decoder in that it has an extra function.
inherit Decoder;
+ //! @decl string charset;
+ //!
+ //! Name of the charset - giving this name to @[encoder] returns
+ //! an instance of the same class as this one.
+ //!
+ //! @note
+ //! This is not necessarily the same name that was actually given to
+ //! @[encoder] to produce this object.
+
//! Change the replacement callback function.
//!
//! @param rc
@@ -145,6 +170,7 @@ class Encoder
}
private class ASCIIDec {
+ constant charset = "iso88591";
protected private string s = "";
this_program feed(string ss)
{
@@ -166,6 +192,7 @@ private class ASCIIDec {
private class UTF16dec {
inherit ASCIIDec;
+ constant charset = "utf16";
protected int check_bom=1, le=0;
string drain() {
string s = ::drain();
@@ -190,6 +217,7 @@ private class UTF16dec {
private class UTF16LEdec {
inherit UTF16dec;
+ constant charset = "utf16le";
protected void create() { le=1; }
}
@@ -316,12 +344,12 @@ Decoder decoder(string name)
])[name[3..]];
if(sub)
- return EUCDec(sub);
+ return EUCDec(sub, name);
}
if( (< "extendedunixcodepackedformatforjapanese",
"eucpkdfmtjapanese" >)[ name ] )
- return EUCDec("x0208");
+ return EUCDec("x0208", "eucpkdfmtjapanese");
if( (< "gb18030", "gbk", "936", "949" >)[ name ] )
@@ -342,11 +370,12 @@ Decoder decoder(string name)
private class ASCIIEnc
{
+ constant charset = "iso88591";
protected string s = "";
protected string|void replacement;
protected function(string:string)|void repcb;
protected string low_convert(string s, string|void r,
- function(string:string)|void rc)
+ function(string:string)|void rc)
{
int i = sizeof(s);
string rr;
@@ -357,7 +386,7 @@ private class ASCIIEnc
else if(r)
s=s[..i-1]+low_convert(r)+s[i+1..];
else
- error("Character unsupported by encoding.\n");
+ encode_error (s, i, charset, "Character unsupported by encoding.\n");
return s;
}
this_program feed(string ss)
@@ -389,8 +418,9 @@ private class ASCIIEnc
private class UTF16enc {
inherit ASCIIEnc;
+ constant charset = "utf16";
protected private string low_convert(string s, string|void r,
- function(string:string)|void rc)
+ function(string:string)|void rc)
{
int i = sizeof(s);
string rr;
@@ -401,7 +431,7 @@ private class UTF16enc {
else if(r)
s=s[..i-1]+low_convert(r)+s[i+1..];
else
- error("Character unsupported by encoding.\n");
+ encode_error (s, i, charset, "Character unsupported by encoding.\n");
return s;
}
this_program feed(string ss) {
@@ -421,6 +451,7 @@ private class UTF16enc {
private class UTF16LEenc {
inherit UTF16enc;
+ constant charset = "utf16le";
string drain() {
return map(::drain()/2, reverse)*"";
}
@@ -501,6 +532,8 @@ Encoder encoder(string name, string|void replacement,
"latin1", "l1", "ansix341968", "iso646irv1991", "iso646us",
"isoir6", "us", "usascii", "ascii", "367", "819",
"isolatin1">)[name])
+ // FIXME: This doesn't accurately check the range of valid
+ // characters according to the chosen charset.
return ASCIIEnc(replacement, repcb);
if(has_prefix(name, "iso2022"))
@@ -536,12 +569,12 @@ Encoder encoder(string name, string|void replacement,
])[name[3..]];
if(sub)
- return EUCEnc(sub, replacement, repcb);
+ return EUCEnc(sub, name, replacement, repcb);
}
if( (< "extendedunixcodepackedformatforjapanese",
"eucpkdfmtjapanese" >)[ name ] )
- return EUCEnc("x0208", replacement, repcb);
+ return EUCEnc("x0208", "eucpkdfmtjapanese", replacement, repcb);
Encoder o = rfc1345(name, 1, replacement, repcb);
@@ -859,7 +892,16 @@ class DecodeError
//! The failing position in @[err_str].
string charset;
- //! The decoding charset.
+ //! The decoding charset, typically as known to
+ //! @[Locale.Charset.decoder].
+ //!
+ //! @note
+ //! Other code may produce errors of this type. In that case this
+ //! name is something that @[Locale.Charset.decoder] does not accept
+ //! (unless it implements exactly the same charset), and it should
+ //! be reasonably certain that @[Locale.Charset.decoder] never
+ //! accepts that name in the future (unless it is extended to
+ //! implement exactly the same charset).
protected void create (string err_str, int err_pos, string charset,
void|string reason, void|array bt)
@@ -873,6 +915,16 @@ class DecodeError
}
}
+void decode_error (string err_str, int err_pos, string charset,
+ void|string reason, void|mixed... args)
+//! Throws a @[DecodeError] exception. See @[DecodeError.create] for
+//! details about the arguments. If @[args] is given then the error
+//! reason is formatted using @expr{sprintf(@[reason], @@@[args])@}.
+{
+ if (sizeof (args)) reason = sprintf (reason, @args);
+ throw (DecodeError (err_str, err_pos, charset, reason, backtrace()[..<1]));
+}
+
class EncodeError
//! Error thrown when encode fails (and no replacement char or
//! replacement callback has been registered).
@@ -893,7 +945,16 @@ class EncodeError
//! The failing position in @[err_str].
string charset;
- //! The encoding charset.
+ //! The encoding charset, typically as known to
+ //! @[Locale.Charset.encoder].
+ //!
+ //! @note
+ //! Other code may produce errors of this type. In that case this
+ //! name is something that @[Locale.Charset.encoder] does not accept
+ //! (unless it implements exactly the same charset), and it should
+ //! be reasonably certain that @[Locale.Charset.encoder] never
+ //! accepts that name in the future (unless it is extended to
+ //! implement exactly the same charset).
protected void create (string err_str, int err_pos, string charset,
void|string reason, void|array bt)
@@ -906,3 +967,13 @@ class EncodeError
bt);
}
}
+
+void encode_error (string err_str, int err_pos, string charset,
+ void|string reason, void|mixed... args)
+//! Throws an @[EncodeError] exception. See @[EncodeError.create] for
+//! details about the arguments. If @[args] is given then the error
+//! reason is formatted using @expr{sprintf(@[reason], @@@[args])@}.
+{
+ if (sizeof (args)) reason = sprintf (reason, @args);
+ throw (EncodeError (err_str, err_pos, charset, reason, backtrace()[..<1]));
+}