King_DuckZ/kamokan
1
0
Fork 0
mirror of https://github.com/KingDuckZ/kamokan.git synced 2025-02-17 09:35:49 +00:00
Code Issues Releases Wiki Activity

Stock indices (1-based) in escape_bytes instead of just a flag.

Browse source 
The corresponding byte in escape_bytes to every to-be-escaped
characters in the source buffer gets set to the index of the
character in Needle +1. expand() is similar to slow_copy(),
except that it doesn't need to re-compare each character in
the source buffer to check if it needs to be replaced. Each
non-zero byte (that's why indices are 1-based) in parWhich
is the index of the sequence that should be used to replace the
original character (needles[curr_byte-1]).
This commit is contained in:
King_DuckZ 2017-06-19 19:19:42 +01:00
parent 6bee1af080
commit e649e9a196
1 changed files with 20 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

27
src/tawashi/escapist.cpp
View file

@ -24,6 +24,7 @@
#if !defined(HTML_ESCAPE_WITH_HOUDINI) #if !defined(HTML_ESCAPE_WITH_HOUDINI)
# include <algorithm> # include <algorithm>
#endif #endif
#include <climits>
namespace tawashi { namespace tawashi {
namespace { namespace {
@ -73,6 +74,20 @@ namespace tawashi {
parDest.push_back(parSource[z]); parDest.push_back(parSource[z]);
} }
} }
template <char... Needle, typename P, std::size_t... Sizes>
void expand (const char* parSource, std::string& parDest, P parWhich, const char (&...parWith)[Sizes]) {
static_assert(sizeof...(Needle) + 1 <= 0xFF, "Too many search chars, their indices won't fit in a byte");
std::array<const char*, sizeof...(Needle)> withs {parWith...};
std::array<uint32_t, sizeof...(Needle)> sizes {(static_cast<uint32_t>(Sizes) - 1)...};
for (unsigned int z = 0; z < sizeof(P) * CHAR_BIT; z += CHAR_BIT) {
const auto curr = 0xFF bitand (parWhich >> z);
if (curr)
parDest.append(withs[curr - 1], sizes[curr - 1]);
else
parDest.push_back(parSource[z / CHAR_BIT]);
}
}
#endif #endif
#if !defined(HTML_ESCAPE_WITH_HOUDINI) #if !defined(HTML_ESCAPE_WITH_HOUDINI)
@ -142,16 +157,14 @@ namespace tawashi {
for (unsigned int z = pre_bytes; z < inp_size - post_bytes; z += sizeof(packs[0])) { for (unsigned int z = pre_bytes; z < inp_size - post_bytes; z += sizeof(packs[0])) {
const uint32_t& val = *reinterpret_cast<const uint32_t*>(parStr.data() + z); const uint32_t& val = *reinterpret_cast<const uint32_t*>(parStr.data() + z);
uint32_t escape_bytes = 0; uint32_t escape_bytes = 0;
uint8_t char_index = 1; //indices are 1-based
for (uint32_t pack : packs) { for (uint32_t pack : packs) {
const uint32_t t = val xor pack; const uint32_t t = val xor pack;
escape_bytes = (t - c1) bitand compl t bitand c2; const uint32_t placeholders = ((t - c1) bitand compl t bitand c2) >> 7;
if (escape_bytes) assert((escape_bytes bitand (placeholders * 0xFF)) == 0);
break; escape_bytes |= placeholders * char_index++;
} }
if (escape_bytes) expand<Needle...>(parStr.data() + z, retval, escape_bytes, parWith...);
slow_copy<Needle...>(parStr.data() + z, retval, sizeof(packs[0]), parWith...);
else
retval.append(parStr.data() + z, sizeof(packs[0]));
} }
slow_copy<Needle...>(parStr.data() + inp_size - post_bytes, retval, post_bytes, parWith...); slow_copy<Needle...>(parStr.data() + inp_size - post_bytes, retval, post_bytes, parWith...);