kakoune/src/ranked_match.cc

242 lines
7.4 KiB
C++
Raw Normal View History

#include "ranked_match.hh"
2016-11-29 00:53:50 +01:00
#include "flags.hh"
#include "unit_tests.hh"
2016-11-29 00:53:50 +01:00
#include "utf8_iterator.hh"
#include <algorithm>
namespace Kakoune
{
2016-03-25 21:35:57 +01:00
UsedLetters used_letters(StringView str)
{
UsedLetters res = 0;
for (auto c : str)
{
if (c >= 'a' and c <= 'z')
2016-11-22 23:20:30 +01:00
res |= 1uLL << (c - 'a');
2016-03-25 21:35:57 +01:00
else if (c >= 'A' and c <= 'Z')
2016-11-22 23:20:30 +01:00
res |= 1uLL << (c - 'A' + 26);
2016-03-25 21:35:57 +01:00
else if (c == '_')
2016-11-22 23:20:30 +01:00
res |= 1uLL << 53;
2016-03-25 21:35:57 +01:00
else if (c == '-')
2016-11-22 23:20:30 +01:00
res |= 1uLL << 54;
2016-03-25 21:35:57 +01:00
else
2016-11-22 23:20:30 +01:00
res |= 1uLL << 63;
2016-03-25 21:35:57 +01:00
}
return res;
}
bool matches(UsedLetters query, UsedLetters letters)
{
return (query & letters) == query;
}
using Utf8It = utf8::iterator<const char*>;
static int count_word_boundaries_match(StringView candidate, StringView query)
{
int count = 0;
2016-02-18 00:05:08 +01:00
Utf8It query_it{query.begin(), query};
Codepoint prev = 0;
for (Utf8It it{candidate.begin(), candidate}; it != candidate.end(); ++it)
{
const Codepoint c = *it;
const bool is_word_boundary = prev == 0 or
(!iswalnum((wchar_t)prev) and iswalnum((wchar_t)c)) or
(iswlower((wchar_t)prev) and iswupper((wchar_t)c));
prev = c;
if (not is_word_boundary)
continue;
const Codepoint lc = to_lower(c);
2016-02-18 00:05:08 +01:00
for (auto qit = query_it; qit != query.end(); ++qit)
{
const Codepoint qc = *qit;
if (qc == (iswlower((wchar_t)qc) ? lc : c))
{
++count;
2016-02-18 00:05:08 +01:00
query_it = qit+1;
break;
}
}
2016-02-18 00:05:08 +01:00
if (query_it == query.end())
break;
}
return count;
}
static bool smartcase_eq(Codepoint candidate, Codepoint query)
{
return query == (iswlower((wchar_t)query) ? to_lower(candidate) : candidate);
}
struct SubseqRes
{
int max_index;
bool single_word;
};
2017-01-30 12:22:14 +01:00
static Optional<SubseqRes> subsequence_match_smart_case(StringView str, StringView subseq)
{
bool single_word = true;
int max_index = -1;
auto it = str.begin();
int index = 0;
2016-03-28 16:18:15 +02:00
for (auto subseq_it = subseq.begin(); subseq_it != subseq.end();)
{
if (it == str.end())
2017-01-30 12:22:14 +01:00
return {};
2016-03-28 16:18:15 +02:00
const Codepoint c = utf8::read_codepoint(subseq_it, subseq.end());
while (true)
{
auto str_c = utf8::read_codepoint(it, str.end());
if (smartcase_eq(str_c, c))
break;
if (max_index != -1 and single_word and not is_word(str_c))
single_word = false;
++index;
if (it == str.end())
2017-01-30 12:22:14 +01:00
return {};
}
max_index = index++;
}
2017-01-30 12:22:14 +01:00
return SubseqRes{max_index, single_word};
}
2016-03-25 21:35:57 +01:00
template<typename TestFunc>
RankedMatch::RankedMatch(StringView candidate, StringView query, TestFunc func)
2015-10-27 22:25:18 +01:00
{
if (candidate.empty() or query.length() > candidate.length())
return;
if (query.empty())
2015-10-27 22:25:18 +01:00
{
m_candidate = candidate;
return;
}
2016-03-25 21:35:57 +01:00
if (not func())
return;
auto res = subsequence_match_smart_case(candidate, query);
2017-01-30 12:22:14 +01:00
if (not res)
return;
m_candidate = candidate;
2017-01-30 12:22:14 +01:00
m_max_index = res->max_index;
2017-01-30 12:22:14 +01:00
if (res->single_word)
m_flags |= Flags::SingleWord;
if (smartcase_eq(candidate[0], query[0]))
m_flags |= Flags::FirstCharMatch;
auto it = std::search(candidate.begin(), candidate.end(),
query.begin(), query.end(), smartcase_eq);
if (it != candidate.end())
{
m_flags |= Flags::Contiguous;
if (it == candidate.begin())
{
m_flags |= Flags::Prefix;
if (query.length() == candidate.length())
{
m_flags |= Flags::SmartFullMatch;
if (candidate == query)
m_flags |= Flags::FullMatch;
}
}
2015-10-27 22:25:18 +01:00
}
m_word_boundary_match_count = count_word_boundaries_match(candidate, query);
if (m_word_boundary_match_count == query.length())
m_flags |= Flags::OnlyWordBoundary;
2016-03-25 21:35:57 +01:00
}
2015-10-27 22:25:18 +01:00
2016-03-25 21:35:57 +01:00
RankedMatch::RankedMatch(StringView candidate, UsedLetters candidate_letters,
StringView query, UsedLetters query_letters)
: RankedMatch{candidate, query, [&] {
return matches(to_lower(query_letters), to_lower(candidate_letters)) and
matches(query_letters & upper_mask, candidate_letters & upper_mask);
}} {}
2016-03-25 21:35:57 +01:00
RankedMatch::RankedMatch(StringView candidate, StringView query)
: RankedMatch{candidate, query, [] { return true; }}
{
2015-10-27 22:25:18 +01:00
}
bool RankedMatch::operator<(const RankedMatch& other) const
{
2016-03-24 23:04:56 +01:00
kak_assert((bool)*this and (bool)other);
const auto diff = m_flags ^ other.m_flags;
// flags are different, use their ordering to return the first match
if (diff != Flags::None)
return (int)(m_flags & diff) > (int)(other.m_flags & diff);
2017-01-31 23:30:23 +01:00
if (not (m_flags & Flags::Prefix) and
m_word_boundary_match_count != other.m_word_boundary_match_count)
return m_word_boundary_match_count > other.m_word_boundary_match_count;
if (m_max_index != other.m_max_index)
return m_max_index < other.m_max_index;
// Reorder codepoints to improve matching behaviour
auto order = [](Codepoint cp) { return cp == '/' ? 0 : cp; };
auto it1 = m_candidate.begin(), it2 = other.m_candidate.begin();
const auto end1 = m_candidate.end(), end2 = other.m_candidate.end();
auto last1 = it1, last2 = it2;
while (true)
{
// find next mismatch
while (it1 != end1 and it2 != end2 and *it1 == *it2)
++it1, ++it2;
if (it1 == end1 or it2 == end2)
return it1 == end1 and it2 != end2;
// compare codepoints
it1 = utf8::character_start(it1, last1);
it2 = utf8::character_start(it2, last2);
const auto cp1 = utf8::read_codepoint(it1, end1);
const auto cp2 = utf8::read_codepoint(it2, end2);
if (cp1 != cp2)
{
const bool low1 = iswlower((wchar_t)cp1);
const bool low2 = iswlower((wchar_t)cp2);
return low1 == low2 ? order(cp1) < order(cp2) : low1;
}
last1 = it1; last2 = it2;
}
2015-10-27 22:25:18 +01:00
}
UnitTest test_ranked_match{[] {
kak_assert(count_word_boundaries_match("run_all_tests", "rat") == 3);
2016-02-18 00:05:08 +01:00
kak_assert(count_word_boundaries_match("run_all_tests", "at") == 2);
kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "wm") == 2);
kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "cobm") == 3);
kak_assert(count_word_boundaries_match("countWordBoundariesMatch", "cWBM") == 4);
kak_assert(RankedMatch{"source", "so"} < RankedMatch{"source_data", "so"});
kak_assert(not (RankedMatch{"source_data", "so"} < RankedMatch{"source", "so"}));
kak_assert(not (RankedMatch{"source", "so"} < RankedMatch{"source", "so"}));
kak_assert(RankedMatch{"single/word", "wo"} < RankedMatch{"multiw/ord", "wo"});
kak_assert(RankedMatch{"foo/bar/foobar", "foobar"} < RankedMatch{"foo/bar/baz", "foobar"});
kak_assert(RankedMatch{"delete-buffer", "db"} < RankedMatch{"debug", "db"});
kak_assert(RankedMatch{"create_task", "ct"} < RankedMatch{"constructor", "ct"});
2017-01-31 23:30:23 +01:00
kak_assert(RankedMatch{"class", "cla"} < RankedMatch{"class::attr", "cla"});
kak_assert(RankedMatch{"meta/", "meta"} < RankedMatch{"meta-a/", "meta"});
}};
2016-03-25 21:35:57 +01:00
UnitTest test_used_letters{[]()
{
kak_assert(used_letters("abcd") == to_lower(used_letters("abcdABCD")));
}};
}