Indexes redesigned to return data reference directly for cleaner code

This commit is contained in:
Simon Rozman 2016-03-08 14:09:23 +01:00
parent 4d72a5bc4c
commit e13133d0e4
6 changed files with 106 additions and 86 deletions

View File

@ -76,13 +76,13 @@ bool wxZRColaKeyHandler::ProcessEvent(wxEvent& event)
ks->chr = 0; ks->chr = 0;
ks->seq_len = n; ks->seq_len = n;
memcpy(ks->seq, m_seq.data(), sizeof(ZRCola::keyseq_db::keyseq::key_t)*n); memcpy(ks->seq, m_seq.data(), sizeof(ZRCola::keyseq_db::keyseq::key_t)*n);
found = m_ks_db.idxKey.find((const unsigned __int16&)*ks, start, end); found = m_ks_db.idxKey.find(*ks, start, end);
delete ks; delete ks;
} }
if (found) { if (found) {
// The exact key sequence found. // The exact key sequence found.
const ZRCola::keyseq_db::keyseq &ks = (const ZRCola::keyseq_db::keyseq&)m_ks_db.data[m_ks_db.idxKey[start]]; const ZRCola::keyseq_db::keyseq &ks = m_ks_db.idxKey[start];
m_seq.clear(); m_seq.clear();
wxObject *obj = event.GetEventObject(); wxObject *obj = event.GetEventObject();
@ -95,7 +95,7 @@ bool wxZRColaKeyHandler::ProcessEvent(wxEvent& event)
return true; return true;
} }
} else if (start < m_ks_db.idxKey.size() && } else if (start < m_ks_db.idxKey.size() &&
ZRCola::keyseq_db::keyseq::CompareSequence(m_seq.data(), m_seq.size(), ((const ZRCola::keyseq_db::keyseq&)m_ks_db.data[m_ks_db.idxKey[start]]).seq, std::min<unsigned __int16>(((const ZRCola::keyseq_db::keyseq&)m_ks_db.data[m_ks_db.idxKey[start]]).seq_len, m_seq.size())) == 0) ZRCola::keyseq_db::keyseq::CompareSequence(m_seq.data(), m_seq.size(), m_ks_db.idxKey[start].seq, std::min<unsigned __int16>(m_ks_db.idxKey[start].seq_len, m_seq.size())) == 0)
{ {
// The sequence is a partial match. Continue watching. // The sequence is a partial match. Continue watching.
event.StopPropagation(); event.StopPropagation();

View File

@ -305,8 +305,8 @@ int _tmain(int argc, _TCHAR *argv[])
// Check key sequences. // Check key sequences.
for (std::vector<unsigned __int32>::size_type i = 1, n = db.idxKey.size(); i < n; i++) { for (std::vector<unsigned __int32>::size_type i = 1, n = db.idxKey.size(); i < n; i++) {
const ZRCola::keyseq_db::keyseq const ZRCola::keyseq_db::keyseq
&ks1 = (const ZRCola::keyseq_db::keyseq&)db.data[db.idxKey[i - 1]], &ks1 = db.idxKey[i - 1],
&ks2 = (const ZRCola::keyseq_db::keyseq&)db.data[db.idxKey[i ]]; &ks2 = db.idxKey[i ];
if (ZRCola::keyseq_db::keyseq::CompareSequence(ks1.seq, ks1.seq_len, ks2.seq, ks2.seq_len) == 0) { if (ZRCola::keyseq_db::keyseq::CompareSequence(ks1.seq, ks1.seq_len, ks2.seq, ks2.seq_len) == 0) {
std::wstring seq_str; std::wstring seq_str;

View File

@ -55,7 +55,7 @@ namespace ZRCola {
/// ///
/// Memory index /// Memory index
/// ///
template <class T, class T_idx = unsigned __int32> template <class T, class T_idx = unsigned __int32, class T_data = T>
class index : public std::vector<T_idx> class index : public std::vector<T_idx>
{ {
protected: protected:
@ -70,6 +70,58 @@ namespace ZRCola {
index(_In_ std::vector<T> &h) : host(h) {} index(_In_ std::vector<T> &h) : host(h) {}
///
/// Returns data at given position according to the index
///
/// \param[in] pos Position
///
/// \returns Data reference
///
inline const T_data& at(size_type pos) const
{
return (const T_data&)host.at(std::vector<T_idx>::at(pos));
}
///
/// Returns data at given position according to the index
///
/// \param[in] pos Position
///
/// \returns Data reference
///
inline T_data& at(size_type pos)
{
return (T_data&)host.at(std::vector<T_idx>::at(pos));
}
///
/// Returns data at given position according to the index
///
/// \param[in] pos Position
///
/// \returns Data reference
///
inline const T_data& operator[](size_type pos) const
{
return (const T_data&)host[std::vector<T_idx>::at(pos)];
}
///
/// Returns data at given position according to the index
///
/// \param[in] pos Position
///
/// \returns Data reference
///
inline T_data& operator[](size_type pos)
{
return (T_data&)host[std::vector<T_idx>::at(pos)];
}
/// ///
/// Sorts index /// Sorts index
/// ///
@ -90,7 +142,7 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare(_In_ const T &a, _In_ const T &b) const = 0; virtual int compare(_In_ const T_data &a, _In_ const T_data &b) const = 0;
/// ///
@ -104,7 +156,7 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare_sort(_In_ const T &a, _In_ const T &b) const = 0; virtual int compare_sort(_In_ const T_data &a, _In_ const T_data &b) const = 0;
/// ///
@ -119,26 +171,26 @@ namespace ZRCola {
/// - true if found /// - true if found
/// - false otherwise /// - false otherwise
/// ///
bool find(_In_ const T &el, _Out_ size_type &start, _Out_ size_type &end) const bool find(_In_ const T_data &el, _Out_ size_type &start, _Out_ size_type &end) const
{ {
// Start with the full search area. // Start with the full search area.
for (start = 0, end = size(); start < end; ) { for (start = 0, end = size(); start < end; ) {
size_type m = (start + end) / 2; size_type m = (start + end) / 2;
int r = compare(el, host[at(m)]); int r = compare(el, at(m));
if (r < 0) end = m; if (r < 0) end = m;
else if (r > 0) start = m + 1; else if (r > 0) start = m + 1;
else { else {
// Narrow the search area on the left to start at the first element in the run. // Narrow the search area on the left to start at the first element in the run.
for (size_type end2 = m; start < end2;) { for (size_type end2 = m; start < end2;) {
size_type m = (start + end2) / 2; size_type m = (start + end2) / 2;
int r = compare(el, host[at(m)]); int r = compare(el, at(m));
if (r <= 0) end2 = m; else start = m + 1; if (r <= 0) end2 = m; else start = m + 1;
} }
// Narrow the search area on the right to end at the first element not in the run. // Narrow the search area on the right to end at the first element not in the run.
for (size_type start2 = m + 1; start2 < end;) { for (size_type start2 = m + 1; start2 < end;) {
size_type m = (start2 + end) / 2; size_type m = (start2 + end) / 2;
int r = compare(el, host[at(m)]); int r = compare(el, at(m));
if (0 <= r) start2 = m + 1; else end = m; if (0 <= r) start2 = m + 1; else end = m;
} }
@ -152,8 +204,8 @@ namespace ZRCola {
private: private:
static int __cdecl compare_s(void *p, const void *a, const void *b) static int __cdecl compare_s(void *p, const void *a, const void *b)
{ {
const index<T, T_idx> *t = (const index<T, T_idx>*)p; const index<T, T_idx, T_data> *t = (const index<T, T_idx, T_data>*)p;
return t->compare_sort(t->host[*(const T_idx*)a], t->host[*(const T_idx*)b]); return t->compare_sort((const T_data&)t->host[*(const T_idx*)a], (const T_data&)t->host[*(const T_idx*)b]);
} }
}; };

View File

@ -81,7 +81,7 @@ namespace ZRCola {
/// ///
/// Composition index /// Composition index
/// ///
class indexComp : public index<unsigned __int16, unsigned __int32> class indexComp : public index<unsigned __int16, unsigned __int32, translation>
{ {
public: public:
/// ///
@ -89,7 +89,7 @@ namespace ZRCola {
/// ///
/// \param[in] h Reference to vector holding the data /// \param[in] h Reference to vector holding the data
/// ///
indexComp(_In_ std::vector<unsigned __int16> &h) : index<unsigned __int16, unsigned __int32>(h) {} indexComp(_In_ std::vector<unsigned __int16> &h) : index<unsigned __int16, unsigned __int32, translation>(h) {}
/// ///
/// Compares two transformations by string (for searching) /// Compares two transformations by string (for searching)
@ -102,13 +102,9 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare(_In_ const unsigned __int16 &a, _In_ const unsigned __int16 &b) const virtual int compare(_In_ const translation &a, _In_ const translation &b) const
{ {
const translation int r = translation::CompareString(a.str, a.str_len, b.str, b.str_len);
&trans_a = (const translation&)a,
&trans_b = (const translation&)b;
int r = translation::CompareString(trans_a.str, trans_a.str_len, trans_b.str, trans_b.str_len);
if (r != 0) return r; if (r != 0) return r;
return 0; return 0;
@ -125,17 +121,13 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare_sort(_In_ const unsigned __int16 &a, _In_ const unsigned __int16 &b) const virtual int compare_sort(_In_ const translation &a, _In_ const translation &b) const
{ {
const translation int r = translation::CompareString(a.str, a.str_len, b.str, b.str_len);
&trans_a = (const translation&)a,
&trans_b = (const translation&)b;
int r = translation::CompareString(trans_a.str, trans_a.str_len, trans_b.str, trans_b.str_len);
if (r != 0) return r; if (r != 0) return r;
if (trans_a.chr < trans_b.chr) return -1; if (a.chr < b.chr) return -1;
else if (trans_a.chr > trans_b.chr) return +1; else if (a.chr > b.chr) return +1;
return 0; return 0;
} }
@ -145,7 +137,7 @@ namespace ZRCola {
/// ///
/// Decomposition index /// Decomposition index
/// ///
class indexDecomp : public index<unsigned __int16, unsigned __int32> class indexDecomp : public index<unsigned __int16, unsigned __int32, translation>
{ {
public: public:
/// ///
@ -153,7 +145,7 @@ namespace ZRCola {
/// ///
/// \param[in] h Reference to vector holding the data /// \param[in] h Reference to vector holding the data
/// ///
indexDecomp(_In_ std::vector<unsigned __int16> &h) : index<unsigned __int16, unsigned __int32>(h) {} indexDecomp(_In_ std::vector<unsigned __int16> &h) : index<unsigned __int16, unsigned __int32, translation>(h) {}
/// ///
/// Compares two transformations by character (for searching) /// Compares two transformations by character (for searching)
@ -166,14 +158,10 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare(_In_ const unsigned __int16 &a, _In_ const unsigned __int16 &b) const virtual int compare(_In_ const translation &a, _In_ const translation &b) const
{ {
const translation if (a.chr < b.chr) return -1;
&trans_a = (const translation&)a, else if (a.chr > b.chr) return +1;
&trans_b = (const translation&)b;
if (trans_a.chr < trans_b.chr) return -1;
else if (trans_a.chr > trans_b.chr) return +1;
return 0; return 0;
} }
@ -189,16 +177,12 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare_sort(_In_ const unsigned __int16 &a, _In_ const unsigned __int16 &b) const virtual int compare_sort(_In_ const translation &a, _In_ const translation &b) const
{ {
const translation if (a.chr < b.chr) return -1;
&trans_a = (const translation&)a, else if (a.chr > b.chr) return +1;
&trans_b = (const translation&)b;
if (trans_a.chr < trans_b.chr) return -1; int r = translation::CompareString(a.str, a.str_len, b.str, b.str_len);
else if (trans_a.chr > trans_b.chr) return +1;
int r = translation::CompareString(trans_a.str, trans_a.str_len, trans_b.str, trans_b.str_len);
if (r != 0) return r; if (r != 0) return r;
return 0; return 0;

View File

@ -49,7 +49,7 @@ void ZRCola::translation_db::Compose(_In_z_count_(inputMax) const wchar_t* input
// Get the j-th character of the composition. // Get the j-th character of the composition.
// All compositions that get short on characters are lexically ordered before. // All compositions that get short on characters are lexically ordered before.
// Thus the j-th character is considered 0. // Thus the j-th character is considered 0.
const translation &trans = (const translation&)data[idxComp[m]]; const translation &trans = idxComp[m];
wchar_t s = j < trans.str_len ? trans.str[j] : 0; wchar_t s = j < trans.str_len ? trans.str[j] : 0;
// Do the bisection test. // Do the bisection test.
@ -61,7 +61,7 @@ void ZRCola::translation_db::Compose(_In_z_count_(inputMax) const wchar_t* input
// Narrow the search area on the left to start at the first composition in the run. // Narrow the search area on the left to start at the first composition in the run.
for (size_t rr = m; l < rr;) { for (size_t rr = m; l < rr;) {
size_t m = (l + rr) / 2; size_t m = (l + rr) / 2;
const translation &trans = (const translation&)data[idxComp[m]]; const translation &trans = idxComp[m];
wchar_t s = j < trans.str_len ? trans.str[j] : 0; wchar_t s = j < trans.str_len ? trans.str[j] : 0;
if (c <= s) rr = m; else l = m + 1; if (c <= s) rr = m; else l = m + 1;
} }
@ -69,7 +69,7 @@ void ZRCola::translation_db::Compose(_In_z_count_(inputMax) const wchar_t* input
// Narrow the search area on the right to end at the first composition not in the run. // Narrow the search area on the right to end at the first composition not in the run.
for (size_t ll = m + 1; ll < r;) { for (size_t ll = m + 1; ll < r;) {
size_t m = (ll + r) / 2; size_t m = (ll + r) / 2;
const translation &trans = (const translation&)data[idxComp[m]]; const translation &trans = idxComp[m];
wchar_t s = j < trans.str_len ? trans.str[j] : 0; wchar_t s = j < trans.str_len ? trans.str[j] : 0;
if (s <= c) ll = m + 1; else r = m; if (s <= c) ll = m + 1; else r = m;
} }
@ -80,7 +80,7 @@ void ZRCola::translation_db::Compose(_In_z_count_(inputMax) const wchar_t* input
if (l >= r) { if (l >= r) {
// The search area is empty. // The search area is empty.
const translation &trans = (const translation&)data[idxComp[l_prev]]; const translation &trans = idxComp[l_prev];
if (j && l_prev < compositionsCount && j == trans.str_len) { if (j && l_prev < compositionsCount && j == trans.str_len) {
// The first composition of the previous run was a match. // The first composition of the previous run was a match.
output += trans.chr; output += trans.chr;
@ -99,7 +99,7 @@ void ZRCola::translation_db::Compose(_In_z_count_(inputMax) const wchar_t* input
} else { } else {
// End of input reached. // End of input reached.
const translation &trans = (const translation&)data[idxComp[l]]; const translation &trans = idxComp[l];
if (l < compositionsCount && j == trans.str_len) { if (l < compositionsCount && j == trans.str_len) {
// The first composition of the previous run was a match. // The first composition of the previous run was a match.
output += trans.chr; output += trans.chr;
@ -143,7 +143,7 @@ void ZRCOLA_API ZRCola::translation_db::Decompose(_In_z_count_(inputMax) const w
for (size_t l = 0, r = decompositionsCount;; ) { for (size_t l = 0, r = decompositionsCount;; ) {
if (l < r) { if (l < r) {
size_t m = (l + r) / 2; size_t m = (l + r) / 2;
const translation &trans = (const translation&)data[idxDecomp[m]]; const translation &trans = idxDecomp[m];
wchar_t decompSrc = trans.chr; wchar_t decompSrc = trans.chr;
if (c < decompSrc) r = m; if (c < decompSrc) r = m;
else if (decompSrc < c) l = m + 1; else if (decompSrc < c) l = m + 1;

View File

@ -89,7 +89,7 @@ namespace ZRCola {
/// ///
/// Character index /// Character index
/// ///
class indexChr : public index<unsigned __int16, unsigned __int32> class indexChr : public index<unsigned __int16, unsigned __int32, keyseq>
{ {
public: public:
/// ///
@ -97,7 +97,7 @@ namespace ZRCola {
/// ///
/// \param[in] h Reference to vector holding the data /// \param[in] h Reference to vector holding the data
/// ///
indexChr(_In_ std::vector<unsigned __int16> &h) : index<unsigned __int16, unsigned __int32>(h) {} indexChr(_In_ std::vector<unsigned __int16> &h) : index<unsigned __int16, unsigned __int32, keyseq>(h) {}
/// ///
/// Compares two key sequences by character (for searching) /// Compares two key sequences by character (for searching)
@ -110,14 +110,10 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare(_In_ const unsigned __int16 &a, _In_ const unsigned __int16 &b) const virtual int compare(_In_ const keyseq &a, _In_ const keyseq &b) const
{ {
const keyseq if (a.chr < b.chr) return -1;
&ks_a = (const keyseq&)a, else if (a.chr > b.chr) return +1;
&ks_b = (const keyseq&)b;
if (ks_a.chr < ks_b.chr) return -1;
else if (ks_a.chr > ks_b.chr) return +1;
return 0; return 0;
} }
@ -133,16 +129,12 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare_sort(_In_ const unsigned __int16 &a, _In_ const unsigned __int16 &b) const virtual int compare_sort(_In_ const keyseq &a, _In_ const keyseq &b) const
{ {
const keyseq if (a.chr < b.chr) return -1;
&ks_a = (const keyseq&)a, else if (a.chr > b.chr) return +1;
&ks_b = (const keyseq&)b;
if (ks_a.chr < ks_b.chr) return -1; int r = keyseq::CompareSequence(a.seq, a.seq_len, b.seq, b.seq_len);
else if (ks_a.chr > ks_b.chr) return +1;
int r = keyseq::CompareSequence(ks_a.seq, ks_a.seq_len, ks_b.seq, ks_b.seq_len);
if (r != 0) return r; if (r != 0) return r;
return 0; return 0;
@ -153,7 +145,7 @@ namespace ZRCola {
/// ///
/// Key index /// Key index
/// ///
class indexKey : public index<unsigned __int16, unsigned __int32> class indexKey : public index<unsigned __int16, unsigned __int32, keyseq>
{ {
public: public:
/// ///
@ -161,7 +153,7 @@ namespace ZRCola {
/// ///
/// \param[in] h Reference to vector holding the data /// \param[in] h Reference to vector holding the data
/// ///
indexKey(_In_ std::vector<unsigned __int16> &h) : index<unsigned __int16, unsigned __int32>(h) {} indexKey(_In_ std::vector<unsigned __int16> &h) : index<unsigned __int16, unsigned __int32, keyseq>(h) {}
/// ///
/// Compares two key sequences by key (for searching) /// Compares two key sequences by key (for searching)
@ -174,13 +166,9 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare(_In_ const unsigned __int16 &a, _In_ const unsigned __int16 &b) const virtual int compare(_In_ const keyseq &a, _In_ const keyseq &b) const
{ {
const keyseq int r = keyseq::CompareSequence(a.seq, a.seq_len, b.seq, b.seq_len);
&ks_a = (const keyseq&)a,
&ks_b = (const keyseq&)b;
int r = keyseq::CompareSequence(ks_a.seq, ks_a.seq_len, ks_b.seq, ks_b.seq_len);
if (r != 0) return r; if (r != 0) return r;
return 0; return 0;
@ -197,17 +185,13 @@ namespace ZRCola {
/// - =0 when a == b /// - =0 when a == b
/// - >0 when a > b /// - >0 when a > b
/// ///
virtual int compare_sort(_In_ const unsigned __int16 &a, _In_ const unsigned __int16 &b) const virtual int compare_sort(_In_ const keyseq &a, _In_ const keyseq &b) const
{ {
const keyseq int r = keyseq::CompareSequence(a.seq, a.seq_len, b.seq, b.seq_len);
&ks_a = (const keyseq&)a,
&ks_b = (const keyseq&)b;
int r = keyseq::CompareSequence(ks_a.seq, ks_a.seq_len, ks_b.seq, ks_b.seq_len);
if (r != 0) return r; if (r != 0) return r;
if (ks_a.chr < ks_b.chr) return -1; if (a.chr < b.chr) return -1;
else if (ks_a.chr > ks_b.chr) return +1; else if (a.chr > b.chr) return +1;
return 0; return 0;
} }