lib/icu/{Converter,Collate}: return AllocatedString

This commit is contained in:
Max Kellermann
2015-06-25 22:43:55 +02:00
parent 037d1d9ad1
commit 8d11577ff2
10 changed files with 64 additions and 55 deletions

View File

@@ -19,6 +19,7 @@
#include "config.h"
#include "Collate.hxx"
#include "util/AllocatedString.hxx"
#ifdef HAVE_ICU
#include "Util.hxx"
@@ -140,7 +141,7 @@ IcuCollate(const char *a, const char *b)
#endif
}
std::string
AllocatedString<>
IcuCaseFold(const char *src)
{
#ifdef HAVE_ICU
@@ -152,7 +153,7 @@ IcuCaseFold(const char *src)
const auto u = UCharFromUTF8(src);
if (u.IsNull())
return std::string(src);
return AllocatedString<>::Duplicate(src);
size_t folded_capacity = u.size * 2u;
UChar *folded = new UChar[folded_capacity];
@@ -165,20 +166,17 @@ IcuCaseFold(const char *src)
delete[] u.data;
if (folded_length == 0 || error_code != U_ZERO_ERROR) {
delete[] folded;
return std::string(src);
return AllocatedString<>::Duplicate(src);
}
auto result2 = UCharToUTF8({folded, folded_length});
auto result = UCharToUTF8({folded, folded_length});
delete[] folded;
if (result2.IsNull())
return std::string(src);
std::string result(result2.data, result2.size);
delete[] result2.data;
return result;
#elif defined(HAVE_GLIB)
char *tmp = g_utf8_casefold(src, -1);
std::string result(tmp);
auto result = AllocatedString<>::Duplicate(tmp);
g_free(tmp);
return result;
#else
size_t size = strlen(src) + 1;
auto buffer = new char[size];
@@ -194,9 +192,7 @@ IcuCaseFold(const char *src)
assert(nbytes < size);
assert(buffer[nbytes] == 0);
std::string result(buffer, nbytes);
delete[] buffer;
return AllocatedString<>::Donate(buffer);
#endif
return result;
}

View File

@@ -26,6 +26,7 @@
#include <string>
class Error;
template<typename T> class AllocatedString;
bool
IcuCollateInit(Error &error);
@@ -38,7 +39,7 @@ int
IcuCollate(const char *a, const char *b);
gcc_pure gcc_nonnull_all
std::string
AllocatedString<char>
IcuCaseFold(const char *src);
#endif

View File

@@ -22,6 +22,7 @@
#include "Error.hxx"
#include "util/Error.hxx"
#include "util/Macros.hxx"
#include "util/AllocatedString.hxx"
#include "util/WritableBuffer.hxx"
#include "util/ConstBuffer.hxx"
@@ -80,7 +81,7 @@ IcuConverter::Create(const char *charset, Error &error)
#ifdef HAVE_ICU
#elif defined(HAVE_GLIB)
static std::string
static AllocatedString<char>
DoConvert(GIConv conv, const char *src)
{
// TODO: dynamic buffer?
@@ -93,14 +94,14 @@ DoConvert(GIConv conv, const char *src)
size_t n = g_iconv(conv, &in, &in_left, &out, &out_left);
if (n == static_cast<size_t>(-1) || in_left > 0)
return std::string();
return nullptr;
return std::string(buffer, sizeof(buffer) - out_left);
return AllocatedString::Duplicate(buffer, sizeof(buffer) - out_left);
}
#endif
std::string
AllocatedString<char>
IcuConverter::ToUTF8(const char *s) const
{
#ifdef HAVE_ICU
@@ -118,23 +119,16 @@ IcuConverter::ToUTF8(const char *s) const
&source, source + strlen(source),
nullptr, true, &code);
if (code != U_ZERO_ERROR)
return std::string();
return nullptr;
const size_t target_length = target - buffer;
const auto u = UCharToUTF8({buffer, target_length});
if (u.IsNull())
return std::string();
std::string result(u.data, u.size);
delete[] u.data;
return result;
return UCharToUTF8({buffer, target_length});
#elif defined(HAVE_GLIB)
return DoConvert(to_utf8, s);
#endif
}
std::string
AllocatedString<char>
IcuConverter::FromUTF8(const char *s) const
{
#ifdef HAVE_ICU
@@ -142,7 +136,7 @@ IcuConverter::FromUTF8(const char *s) const
const auto u = UCharFromUTF8(s);
if (u.IsNull())
return std::string();
return nullptr;
ucnv_resetFromUnicode(converter);
@@ -157,9 +151,9 @@ IcuConverter::FromUTF8(const char *s) const
delete[] u.data;
if (code != U_ZERO_ERROR)
return std::string();
return nullptr;
return std::string(buffer, target);
return AllocatedString<>::Duplicate(buffer, target);
#elif defined(HAVE_GLIB)
return DoConvert(from_utf8, s);

View File

@@ -33,14 +33,14 @@
#ifdef HAVE_ICU_CONVERTER
#include <string>
class Error;
#ifdef HAVE_ICU
struct UConverter;
#endif
template<typename T> class AllocatedString;
/**
* This class can convert strings with a certain character set to and
* from UTF-8.
@@ -77,17 +77,19 @@ public:
/**
* Convert the string to UTF-8.
* Returns empty string on error.
*
* Returns AllocatedString::Null() on error.
*/
gcc_pure gcc_nonnull_all
std::string ToUTF8(const char *s) const;
AllocatedString<char> ToUTF8(const char *s) const;
/**
* Convert the string from UTF-8.
* Returns empty string on error.
*
* Returns AllocatedString::Null() on error.
*/
gcc_pure gcc_nonnull_all
std::string FromUTF8(const char *s) const;
AllocatedString<char> FromUTF8(const char *s) const;
};
#endif

View File

@@ -19,6 +19,7 @@
#include "config.h"
#include "Util.hxx"
#include "util/AllocatedString.hxx"
#include "util/WritableBuffer.hxx"
#include "util/ConstBuffer.hxx"
@@ -49,7 +50,7 @@ UCharFromUTF8(const char *src)
return { dest, size_t(dest_length) };
}
WritableBuffer<char>
AllocatedString<>
UCharToUTF8(ConstBuffer<UChar> src)
{
assert(!src.IsNull());
@@ -57,7 +58,7 @@ UCharToUTF8(ConstBuffer<UChar> src)
/* worst-case estimate */
size_t dest_capacity = 4 * src.size;
char *dest = new char[dest_capacity];
char *dest = new char[dest_capacity + 1];
UErrorCode error_code = U_ZERO_ERROR;
int32_t dest_length;
@@ -68,5 +69,6 @@ UCharToUTF8(ConstBuffer<UChar> src)
return nullptr;
}
return { dest, size_t(dest_length) };
dest[dest_length] = 0;
return AllocatedString<>::Donate(dest);
}

View File

@@ -26,6 +26,7 @@
template<typename T> struct WritableBuffer;
template<typename T> struct ConstBuffer;
template<typename T> class AllocatedString;
/**
* Wrapper for u_strFromUTF8(). The returned pointer must be freed
@@ -38,7 +39,7 @@ UCharFromUTF8(const char *src);
* Wrapper for u_strToUTF8(). The returned pointer must be freed with
* delete[].
*/
WritableBuffer<char>
AllocatedString<char>
UCharToUTF8(ConstBuffer<UChar> src);
#endif