Small algorithm improvements to avoid allocations
TL depends on cppcoro now
This commit is contained in:
@@ -105,49 +105,7 @@ inline void to_upper_ascii(string& str) noexcept
|
||||
}
|
||||
|
||||
template <class String>
|
||||
tl::vector<String> split_on_any(const String& str, const char* separators, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
{
|
||||
return split_on_any(str, String(separators), token_policy);
|
||||
}
|
||||
|
||||
template <class String, class Delim>
|
||||
tl::vector<String> split_on_any(const String& str, const Delim& separators, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
{
|
||||
tl::vector<String> dst;
|
||||
dst.reserve(32);
|
||||
split_on_any(str,
|
||||
separators,
|
||||
[&dst](const String& token)
|
||||
{
|
||||
dst.push_back(token);
|
||||
},
|
||||
token_policy);
|
||||
return dst;
|
||||
}
|
||||
|
||||
template <class String, class Delim>
|
||||
tl::vector<String> split_on_all(const String& str, const Delim& separator, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
{
|
||||
tl::vector<String> dst;
|
||||
dst.reserve(32);
|
||||
split_on_all(str,
|
||||
separator,
|
||||
[&dst](const String& token)
|
||||
{
|
||||
dst.push_back(token);
|
||||
},
|
||||
token_policy);
|
||||
return dst;
|
||||
}
|
||||
|
||||
template <class String, class Func>
|
||||
void split_on_any(const String& str, const char* delimiters, const Func& functor, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
{
|
||||
return split_on_any(str, String(delimiters), functor, token_policy);
|
||||
}
|
||||
|
||||
template <class String, class Func>
|
||||
void split_on_any(const String& str, char delimiter, const Func& functor, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
cppcoro::generator<const String&> split_on_any(const String& str, char sep, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
{
|
||||
String token;
|
||||
const char* cstr = str.data();
|
||||
@@ -155,35 +113,39 @@ void split_on_any(const String& str, char delimiter, const Func& functor, empty_
|
||||
const size_t endOffset = str.size();
|
||||
while (offset <= endOffset)
|
||||
{
|
||||
const char* nextCstr = static_cast<const char*>(::memchr(cstr + offset, delimiter, endOffset - offset));
|
||||
const char* nextCstr = static_cast<const char*>(::memchr(cstr + offset, sep, endOffset - offset));
|
||||
const size_t nextOffset = nextCstr ? nextCstr - cstr : endOffset;
|
||||
|
||||
if (offset < nextOffset || token_policy == empty_token_policy::keep)
|
||||
{
|
||||
token.clear();
|
||||
token.append(cstr + offset, cstr + nextOffset);
|
||||
functor(token);
|
||||
co_yield token;
|
||||
}
|
||||
offset = nextOffset + 1;
|
||||
}
|
||||
}
|
||||
|
||||
template <class String, class Delim, class Func>
|
||||
void split_on_any(const String& str, const Delim& delimiters, const Func& functor, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
template <class String>
|
||||
cppcoro::generator<const String&> split_on_any(const String& str, tl::string_view sep, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
{
|
||||
const size_t k_delims_size = delimiters.size();
|
||||
if (k_delims_size == 1)
|
||||
return split_on_any(str, delimiters[0], functor, token_policy);
|
||||
|
||||
if (k_delims_size == 0)
|
||||
const size_t sep_size = sep.size();
|
||||
if (sep_size == 1)
|
||||
{
|
||||
TL_PLAIN_FAIL("'delimiters' cannot be empty");
|
||||
return;
|
||||
for (auto const& s: split_on_any(str, sep[0], token_policy))
|
||||
co_yield s;
|
||||
co_return;
|
||||
}
|
||||
|
||||
if (sep_size == 0)
|
||||
{
|
||||
TL_PLAIN_FAIL("'sep' cannot be empty");
|
||||
co_return;
|
||||
}
|
||||
|
||||
String token;
|
||||
const char* cstr = str.data();
|
||||
const char* delimsStr = delimiters.data();
|
||||
const char* sepStr = sep.data();
|
||||
size_t offset = 0;
|
||||
const size_t endOffset = str.size();
|
||||
while (offset <= endOffset)
|
||||
@@ -191,7 +153,7 @@ void split_on_any(const String& str, const Delim& delimiters, const Func& functo
|
||||
size_t nextOffset = offset;
|
||||
for (; nextOffset < endOffset; ++nextOffset)
|
||||
{
|
||||
if (::memchr(delimsStr, cstr[nextOffset], k_delims_size) != nullptr)
|
||||
if (::memchr(sepStr, cstr[nextOffset], sep_size) != nullptr)
|
||||
break;
|
||||
}
|
||||
|
||||
@@ -199,47 +161,47 @@ void split_on_any(const String& str, const Delim& delimiters, const Func& functo
|
||||
{
|
||||
token.clear();
|
||||
token.append(cstr + offset, cstr + nextOffset);
|
||||
functor(token);
|
||||
co_yield token;
|
||||
}
|
||||
offset = nextOffset + 1;
|
||||
}
|
||||
}
|
||||
|
||||
template <class String, class Delim, class Func>
|
||||
void split_on_all(const String& str, const Delim& sep, const Func& functor, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
template <class String>
|
||||
cppcoro::generator<const String&> split_on_all(const String& str, tl::string_view sep, empty_token_policy token_policy = empty_token_policy::discard) noexcept
|
||||
{
|
||||
String separator(sep);
|
||||
if (separator.empty())
|
||||
if (sep.empty())
|
||||
{
|
||||
TL_PLAIN_FAIL("'delimiters' cannot be empty");
|
||||
return;
|
||||
TL_PLAIN_FAIL("'sep' cannot be empty");
|
||||
co_return;
|
||||
}
|
||||
|
||||
size_t p0 = 0;
|
||||
size_t p1 = 0;
|
||||
const String empty_token;
|
||||
String token;
|
||||
while ((p1 = str.find(separator, p0)) != String::npos)
|
||||
while ((p1 = str.find(sep, p0)) != String::npos)
|
||||
{
|
||||
if (p1 > p0)
|
||||
{
|
||||
token.clear();
|
||||
token.append(str, p0, p1 - p0);
|
||||
functor(token);
|
||||
co_yield token;
|
||||
}
|
||||
else if (token_policy == empty_token_policy::keep)
|
||||
functor(String{});
|
||||
co_yield empty_token;
|
||||
|
||||
p0 = p1 + separator.size();
|
||||
p0 = p1 + sep.size();
|
||||
}
|
||||
|
||||
if (str.size() > p0)
|
||||
{
|
||||
token.clear();
|
||||
token.append(str, p0, String::npos);
|
||||
functor(token);
|
||||
co_yield token;
|
||||
}
|
||||
else if (token_policy == empty_token_policy::keep)
|
||||
functor(String{});
|
||||
co_yield empty_token;
|
||||
}
|
||||
|
||||
template <class String>
|
||||
@@ -541,13 +503,13 @@ size_t replace_first(const string& text, const string& target, const string& rep
|
||||
namespace detail
|
||||
{
|
||||
template <class Container, class String>
|
||||
String join_worker(const Container& container, const String& delimiter) noexcept
|
||||
String join_worker(const Container& container, const String& sep) noexcept
|
||||
{
|
||||
String dst;
|
||||
//reserve some space based on some heuristics:
|
||||
// We'll have size - 1 delimiters
|
||||
// We'll have size - 1 seps
|
||||
// And an average string size of 2 chars. The small choice is made to help with the small string case only where the allocator overhead is significant.
|
||||
dst.reserve((delimiter.size() + 2) * container.size());
|
||||
dst.reserve((sep.size() + 2) * container.size());
|
||||
|
||||
auto end = std::end(container);
|
||||
auto it = std::begin(container);
|
||||
@@ -558,19 +520,19 @@ String join_worker(const Container& container, const String& delimiter) noexcept
|
||||
if (it == end)
|
||||
break;
|
||||
|
||||
dst += delimiter;
|
||||
dst += sep;
|
||||
}
|
||||
return dst;
|
||||
}
|
||||
|
||||
template <class Container>
|
||||
string join_worker(const Container& container, const string& delimiter) noexcept
|
||||
string join_worker(const Container& container, const string& sep) noexcept
|
||||
{
|
||||
eastl::string buffer;
|
||||
//reserve some space based on some heuristics:
|
||||
// We'll have size - 1 delimiters
|
||||
// We'll have size - 1 seps
|
||||
// And an average string size of 2 chars. The small choice is made to help with the small string case only where the allocator overhead is significant.
|
||||
buffer.reserve((delimiter.size() + 2) * container.size());
|
||||
buffer.reserve((sep.size() + 2) * container.size());
|
||||
|
||||
auto end = std::end(container);
|
||||
auto it = std::begin(container);
|
||||
@@ -582,41 +544,41 @@ string join_worker(const Container& container, const string& delimiter) noexcept
|
||||
if (it == end)
|
||||
break;
|
||||
|
||||
buffer.append(delimiter.data(), delimiter.size());
|
||||
buffer.append(sep.data(), sep.size());
|
||||
}
|
||||
return string(buffer);
|
||||
}
|
||||
|
||||
//joiner used when the Delim and String are of the same type
|
||||
template <class Container, class Delim, class String>
|
||||
//joiner used when the Sep and String are of the same type
|
||||
template <class Container, class Sep, class String>
|
||||
std::enable_if_t<
|
||||
std::is_same_v<
|
||||
std::remove_const_t<std::remove_reference_t<String>>,
|
||||
std::remove_const_t<std::remove_reference_t<Delim>>>,
|
||||
String> join(const Container& container, const Delim& delimiter) noexcept
|
||||
std::remove_const_t<std::remove_reference_t<Sep>>>,
|
||||
String> join(const Container& container, const Sep& sep) noexcept
|
||||
{
|
||||
return join_worker(container, delimiter);
|
||||
return join_worker(container, sep);
|
||||
}
|
||||
|
||||
//joiner used when the Delim and String are different types
|
||||
template <class Container, class Delim, class String>
|
||||
//joiner used when the Sep and String are different types
|
||||
template <class Container, class Sep, class String>
|
||||
std::enable_if_t<
|
||||
!std::is_same_v<
|
||||
std::remove_const_t<std::remove_reference_t<String>>,
|
||||
std::remove_const_t<std::remove_reference_t<Delim>>>,
|
||||
String> join(const Container& container, const Delim& delimiter) noexcept
|
||||
std::remove_const_t<std::remove_reference_t<Sep>>>,
|
||||
String> join(const Container& container, const Sep& sep) noexcept
|
||||
{
|
||||
return join_worker(container, String(delimiter));
|
||||
return join_worker(container, String(sep));
|
||||
}
|
||||
} //namespace detail
|
||||
|
||||
template <class Container, class Delim, class String>
|
||||
String join(const Container& container, const Delim& delimiter) noexcept
|
||||
template <class Container, class Sep, class String>
|
||||
String join(const Container& container, const Sep& sep) noexcept
|
||||
{
|
||||
if (container.empty())
|
||||
return String();
|
||||
|
||||
return detail::join<Container, Delim, String>(container, delimiter);
|
||||
return detail::join<Container, Sep, String>(container, sep);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user