mirror of
https://codeberg.org/comaps/comaps
synced 2025-12-19 04:53:36 +00:00
@@ -21,7 +21,7 @@ size_t AbsDiff(size_t a, size_t b)
|
|||||||
class TransitionTable
|
class TransitionTable
|
||||||
{
|
{
|
||||||
public:
|
public:
|
||||||
TransitionTable(UniString const & s, std::vector<UniString> const & prefixMisprints, size_t prefixSize)
|
TransitionTable(UniString const & s, std::array<UniString, 11> const & prefixMisprints, size_t prefixSize)
|
||||||
: m_s(s)
|
: m_s(s)
|
||||||
, m_size(s.size())
|
, m_size(s.size())
|
||||||
, m_prefixMisprints(prefixMisprints)
|
, m_prefixMisprints(prefixMisprints)
|
||||||
@@ -108,7 +108,7 @@ private:
|
|||||||
|
|
||||||
UniString const & m_s;
|
UniString const & m_s;
|
||||||
size_t const m_size;
|
size_t const m_size;
|
||||||
std::vector<UniString> const m_prefixMisprints;
|
std::array<UniString, 11> const m_prefixMisprints;
|
||||||
size_t const m_prefixSize;
|
size_t const m_prefixSize;
|
||||||
};
|
};
|
||||||
} // namespace
|
} // namespace
|
||||||
@@ -190,7 +190,7 @@ void LevenshteinDFA::State::Normalize()
|
|||||||
|
|
||||||
// LevenshteinDFA ----------------------------------------------------------------------------------
|
// LevenshteinDFA ----------------------------------------------------------------------------------
|
||||||
// static
|
// static
|
||||||
LevenshteinDFA::LevenshteinDFA(UniString const & s, size_t prefixSize, std::vector<UniString> const & prefixMisprints,
|
LevenshteinDFA::LevenshteinDFA(UniString const & s, size_t prefixSize, std::array<UniString, 11> const & prefixMisprints,
|
||||||
size_t maxErrors)
|
size_t maxErrors)
|
||||||
: m_size(s.size())
|
: m_size(s.size())
|
||||||
, m_maxErrors(maxErrors)
|
, m_maxErrors(maxErrors)
|
||||||
|
|||||||
@@ -98,7 +98,7 @@ public:
|
|||||||
LevenshteinDFA(LevenshteinDFA &&) = default;
|
LevenshteinDFA(LevenshteinDFA &&) = default;
|
||||||
LevenshteinDFA & operator=(LevenshteinDFA &&) = default;
|
LevenshteinDFA & operator=(LevenshteinDFA &&) = default;
|
||||||
|
|
||||||
LevenshteinDFA(UniString const & s, size_t prefixSize, std::vector<UniString> const & prefixMisprints,
|
LevenshteinDFA(UniString const & s, size_t prefixSize, std::array<UniString, 11> const & prefixMisprints,
|
||||||
size_t maxErrors);
|
size_t maxErrors);
|
||||||
LevenshteinDFA(std::string const & s, size_t prefixSize, size_t maxErrors);
|
LevenshteinDFA(std::string const & s, size_t prefixSize, size_t maxErrors);
|
||||||
LevenshteinDFA(UniString const & s, size_t maxErrors);
|
LevenshteinDFA(UniString const & s, size_t maxErrors);
|
||||||
|
|||||||
@@ -21,7 +21,7 @@ using namespace strings;
|
|||||||
|
|
||||||
namespace
|
namespace
|
||||||
{
|
{
|
||||||
std::vector<UniString> const kAllowedMisprints = {
|
std::array<UniString, 11> const kAllowedMisprints = {
|
||||||
MakeUniString("ckq"), MakeUniString("eyjiu"), MakeUniString("gh"), MakeUniString("pf"), MakeUniString("vw"),
|
MakeUniString("ckq"), MakeUniString("eyjiu"), MakeUniString("gh"), MakeUniString("pf"), MakeUniString("vw"),
|
||||||
|
|
||||||
// Russian
|
// Russian
|
||||||
@@ -118,11 +118,12 @@ static std::pair<UniString, UniString> const kPreprocessReplacements[] = {
|
|||||||
{MakeUniString("ι.ν"), MakeUniString("ιερός ναός")},
|
{MakeUniString("ι.ν"), MakeUniString("ιερός ναός")},
|
||||||
{MakeUniString("κων/νου"), MakeUniString("κωνσταντίνου")},
|
{MakeUniString("κων/νου"), MakeUniString("κωνσταντίνου")},
|
||||||
{MakeUniString("д-р"), MakeUniString("доктор")},
|
{MakeUniString("д-р"), MakeUniString("доктор")},
|
||||||
{MakeUniString("ж.к"), MakeUniString("жилищен комплекс")},
|
{MakeUniString("ж.к."), MakeUniString("жилищен комплекс")},
|
||||||
{MakeUniString("м-н"), MakeUniString("микрорайон")},
|
{MakeUniString("м-н"), MakeUniString("микрорайон")},
|
||||||
{MakeUniString("наб-я"), MakeUniString("набережная")},
|
{MakeUniString("наб-я"), MakeUniString("набережная")},
|
||||||
{MakeUniString("пр-д"), MakeUniString("проезд")},
|
{MakeUniString("пр-д"), MakeUniString("проезд")},
|
||||||
{MakeUniString("пр-т"), MakeUniString("проспект")},
|
{MakeUniString("пр-т"), MakeUniString("проспект")},
|
||||||
|
{MakeUniString("আ/এ"), MakeUniString("আবাসিক এলাকা")},
|
||||||
};
|
};
|
||||||
|
|
||||||
void TransliterateHiraganaToKatakana(UniString & s)
|
void TransliterateHiraganaToKatakana(UniString & s)
|
||||||
|
|||||||
@@ -55,7 +55,7 @@ unordered_map<string, vector<string>> const kSynonyms = {
|
|||||||
{"accs", {"access"}},
|
{"accs", {"access"}},
|
||||||
{"aceq", {"acequia"}},
|
{"aceq", {"acequia"}},
|
||||||
{"ach", {"ancien chemin"}},
|
{"ach", {"ancien chemin"}},
|
||||||
{"a.d", {"auf der", "an der"}},
|
//{"a.d", {"auf der", "an der"}}, // unreachable
|
||||||
{"adi", {"ambodi"}},
|
{"adi", {"ambodi"}},
|
||||||
{"adm", {"admirała", "admirał"}},
|
{"adm", {"admirała", "admirał"}},
|
||||||
{"afb", {"air force base"}},
|
{"afb", {"air force base"}},
|
||||||
@@ -901,7 +901,7 @@ unordered_map<string, vector<string>> const kSynonyms = {
|
|||||||
{"prk", {"purok"}},
|
{"prk", {"purok"}},
|
||||||
{"prm", {"promenade"}},
|
{"prm", {"promenade"}},
|
||||||
{"profª", {"professora"}},
|
{"profª", {"professora"}},
|
||||||
{"prof", {"professora", "professor", "profesora", "profesorky", "profesora (profesor)"}},
|
{"prof", {"professora", "professor", "profesora", "profesorky", "profesor"}},
|
||||||
{"proj", {"projectada", "projetada"}},
|
{"proj", {"projectada", "projetada"}},
|
||||||
{"prolng", {"prolongamento"}},
|
{"prolng", {"prolongamento"}},
|
||||||
{"prol", {"prolongación"}},
|
{"prol", {"prolongación"}},
|
||||||
@@ -1359,7 +1359,6 @@ unordered_map<string, vector<string>> const kSynonyms = {
|
|||||||
{"туп", {"тупик", "тупік"}},
|
{"туп", {"тупик", "тупік"}},
|
||||||
{"ул", {"улица"}},
|
{"ул", {"улица"}},
|
||||||
{"ш", {"шаша", "шосе", "шоссе"}},
|
{"ш", {"шаша", "шосе", "шоссе"}},
|
||||||
{"আ/এ", {"আবাসিক এলাকা"}},
|
|
||||||
{"ইউপি", {"ইউনিয়ন পরিষদ"}},
|
{"ইউপি", {"ইউনিয়ন পরিষদ"}},
|
||||||
{"একনেক", {"জাতীয় অর্থনৈতিক পরিষদের নির্বাহী কমিটি"}},
|
{"একনেক", {"জাতীয় অর্থনৈতিক পরিষদের নির্বাহী কমিটি"}},
|
||||||
{"টেশিস", {"টেলিফোন শিল্প সংস্থা"}},
|
{"টেশিস", {"টেলিফোন শিল্প সংস্থা"}},
|
||||||
|
|||||||
Reference in New Issue
Block a user