From: Andrei Zmievski <andrei@php.net>
Date: Mon, 27 Mar 2006 03:19:30 +0000 (+0000)
Subject: Rewrite unicode_encode() and unicode_decode() functions. Apply the new
X-Git-Tag: RELEASE_1_3~246
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=b36d2dfef6d2b35fcc9fe22b8be7144c9d2d7b03;p=php

Rewrite unicode_encode() and unicode_decode() functions. Apply the new
conversion error semantics.
---

diff --git a/ext/unicode/unicode.c b/ext/unicode/unicode.c
index 23c1a489b2..e468feabd8 100644
--- a/ext/unicode/unicode.c
+++ b/ext/unicode/unicode.c
@@ -23,92 +23,112 @@
 
 void php_register_unicode_iterators(TSRMLS_D);
 
-/* {{{ proto unicode unicode_decode(string input, string encoding) U
-   Takes a string in the source encoding and converts it to a UTF-16 unicode string, returning the result */
+/* {{{ proto unicode unicode_decode(binary input, string encoding [, int flags]) U
+   Takes a binary string converts it to a Unicode string using the specifed encoding */
 static PHP_FUNCTION(unicode_decode)
 {
-	union {
-		void *vptr;
-		char *bin;
-	} input;
-	zend_uchar type;
-	int len;
-	char *encoding;
-	int enclen;
+	char *str, *enc;
+	int str_len, enc_len;
+	long flags;
+	UChar *dest;
+	int dest_len;
 	UErrorCode status;
 	UConverter *conv = NULL;
-	UChar *target;
-	int targetlen;
+	int num_conv;
 
-	if (FAILURE == zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ts", &input.vptr, &len, &type, &encoding, &enclen)) {
+	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "Ss|l", &str, &str_len, &enc, &enc_len, &flags)) {
 		return;
 	}
 
-	if (type == IS_UNICODE) {
-		php_error_docref(NULL TSRMLS_CC, E_WARNING, "input string is already unicode");
-		RETURN_FALSE;
+	if (ZEND_NUM_ARGS() > 2) {
+		if ((flags & 0xff) > ZEND_CONV_ERROR_LAST_ENUM) {
+			php_error_docref(NULL TSRMLS_CC, E_WARNING, "illegal value for conversion error mode");
+			RETURN_FALSE;
+		}
+	} else {
+		flags = UG(to_error_mode);
 	}
 
 	status = U_ZERO_ERROR;
-	conv = ucnv_open(encoding, &status);
-	if (!conv) {
-		php_error_docref(NULL TSRMLS_CC, E_WARNING, "could not locate converter for %s", encoding);
+	conv = ucnv_open(enc, &status);
+	if (U_FAILURE(status)) {
+		php_error_docref(NULL TSRMLS_CC, E_WARNING, "could not create converter for '%s' encoding", enc);
 		RETURN_FALSE;
 	}
 
+	zend_set_converter_error_mode(conv, ZEND_TO_UNICODE, flags);
+
 	status = U_ZERO_ERROR;
-	zend_convert_to_unicode(conv, &target, &targetlen, input.bin, len, &status);
+	num_conv = zend_convert_to_unicode(conv, &dest, &dest_len, str, str_len, &status);
 	if (U_FAILURE(status)) {
-		/* TODO: error handling semantics ? */
-		php_error_docref(NULL TSRMLS_CC, E_WARNING, "conversion was not entirely successful: %d", status);
+		zend_raise_conversion_error_ex("could not decode binary string", conv, ZEND_TO_UNICODE, num_conv, (flags & ZEND_CONV_ERROR_EXCEPTION) TSRMLS_CC);
+		efree(dest);
+		ucnv_close(conv);	
+		RETURN_FALSE;
 	}
-	RETVAL_UNICODEL(target, targetlen, 0);
-
 	ucnv_close(conv);	
-}
-/* }}} */
 
-/* {{{ proto bool unicode_semantics() U
-   Check whether unicode semantics are enabled */
-static PHP_FUNCTION(unicode_semantics)
-{
-	RETURN_BOOL(UG(unicode));
+	RETVAL_UNICODEL(dest, dest_len, 0);
 }
 /* }}} */
 
-/* {{{ proto string unicode_encode(unicode input, string encoding) U
-   Takes a unicode string and converts it to a string in the specified encoding */
+/* {{{ proto binary unicode_encode(unicode input, string encoding [, int flags]) U
+   Takes a Unicode string and converts it to a binary string using the specified encoding */
 static PHP_FUNCTION(unicode_encode)
 {
 	UChar *uni;
-	int len;
-	char *encoding;
-	int enclen;
+	char *enc;
+	int uni_len, enc_len;
+	long flags;
+	char *dest;
+	int dest_len;
 	UErrorCode status;
 	UConverter *conv = NULL;
-	char *target;
-	int targetlen;
+	int num_conv;
 
-	if (FAILURE == zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "us", &uni, &len, &encoding, &enclen)) {
+	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "Us|l", &uni, &uni_len, &enc, &enc_len, &flags) == FAILURE) {
 		return;
 	}
 
+	if (ZEND_NUM_ARGS() > 2) {
+		if ((flags & 0xff) > ZEND_CONV_ERROR_LAST_ENUM) {
+			php_error_docref(NULL TSRMLS_CC, E_WARNING, "illegal value for conversion error mode");
+			RETURN_FALSE;
+		}
+	} else {
+		flags = UG(from_error_mode);
+	}
+
 	status = U_ZERO_ERROR;
-	conv = ucnv_open(encoding, &status);
-	if (!conv) {
-		php_error_docref(NULL TSRMLS_CC, E_WARNING, "could not locate converter for %s", encoding);
+	conv = ucnv_open(enc, &status);
+	if (U_FAILURE(status)) {
+		php_error_docref(NULL TSRMLS_CC, E_WARNING, "could not create converter for '%s' encoding", enc);
 		RETURN_FALSE;
 	}
 
+	zend_set_converter_error_mode(conv, ZEND_FROM_UNICODE, flags);
+	zend_set_converter_subst_char(conv, UG(from_subst_char));
+
 	status = U_ZERO_ERROR;
-	zend_convert_from_unicode(conv, &target, &targetlen, uni, len, &status);
+	num_conv = zend_convert_from_unicode(conv, &dest, &dest_len, uni, uni_len, &status);
 	if (U_FAILURE(status)) {
-		/* TODO: error handling semantics ? */
-		php_error_docref(NULL TSRMLS_CC, E_WARNING, "conversion was not entirely successful: %d", status);
+		int32_t offset = u_countChar32(uni, num_conv);
+		zend_raise_conversion_error_ex("could not encode Unicode string", conv, ZEND_FROM_UNICODE, offset, (flags & ZEND_CONV_ERROR_EXCEPTION) TSRMLS_CC);
+		efree(dest);
+		ucnv_close(conv);	
+		RETURN_FALSE;
 	}
-	RETVAL_STRINGL(target, targetlen, 0);
-
 	ucnv_close(conv);	
+
+	RETVAL_STRINGL(dest, dest_len, 0);
+}
+/* }}} */
+
+/* {{{ proto bool unicode_semantics() U
+   Check whether unicode semantics are enabled */
+static PHP_FUNCTION(unicode_semantics)
+{
+	RETURN_BOOL(UG(unicode));
 }
 /* }}} */