xenia/kakoune
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Regex: Add word boundary assertion support

Browse Source
This commit is contained in:
Maxime Coste 2017-09-18 11:56:14 +09:00
parent a9a04e81b0
commit 9c5d539616
1 changed files with 57 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

62
src/regex_impl.cc
View File

@ -2,6 +2,7 @@
#include "vector.hh" #include "vector.hh"
#include "unit_tests.hh" #include "unit_tests.hh"
#include "string.hh" #include "string.hh"
#include "unicode.hh"
#include "exception.hh" #include "exception.hh"
#include "array_view.hh" #include "array_view.hh"
@ -19,6 +20,8 @@ enum Op : char
Split, Split,
LineStart, LineStart,
LineEnd, LineEnd,
WordBoundary,
NotWordBoundary,
}; };
using Offset = size_t; using Offset = size_t;
@ -42,6 +45,8 @@ enum class Op
Alternation, Alternation,
LineStart, LineStart,
LineEnd, LineEnd,
WordBoundary,
NotWordBoundary,
}; };
struct AstNode struct AstNode
@ -107,7 +112,16 @@ private:
{ {
case '^': ++pos; return make_ast_node(Op::LineStart); case '^': ++pos; return make_ast_node(Op::LineStart);
case '$': ++pos; return make_ast_node(Op::LineEnd); case '$': ++pos; return make_ast_node(Op::LineEnd);
/* TODO: \`, \', \b, \B, look ahead, look behind */ case '\\':
if (pos+1 == end)
return nullptr;
switch (*(pos+1))
{
case 'b': pos += 2; return make_ast_node(Op::WordBoundary);
case 'B': pos += 2; return make_ast_node(Op::NotWordBoundary);
}
break;
/* TODO: \`, \', look ahead, look behind */
} }
return nullptr; return nullptr;
} }
@ -216,6 +230,12 @@ RegexProgram::Offset compile_node(Vector<char>& program, const AstNodePtr& node)
case Op::LineEnd: case Op::LineEnd:
program.push_back(RegexProgram::LineEnd); program.push_back(RegexProgram::LineEnd);
break; break;
case Op::WordBoundary:
program.push_back(RegexProgram::WordBoundary);
break;
case Op::NotWordBoundary:
program.push_back(RegexProgram::NotWordBoundary);
break;
} }
for (auto& offset : goto_end_offsets) for (auto& offset : goto_end_offsets)
@ -279,6 +299,12 @@ void dump(ConstArrayView<char> program)
case RegexProgram::LineEnd: case RegexProgram::LineEnd:
printf("line end\n"); printf("line end\n");
break; break;
case RegexProgram::WordBoundary:
printf("word boundary\n");
break;
case RegexProgram::NotWordBoundary:
printf("not word boundary\n");
break;
case RegexProgram::Match: case RegexProgram::Match:
printf("match\n"); printf("match\n");
} }
@ -299,7 +325,7 @@ struct ThreadedExecutor
{ {
while (true) while (true)
{ {
auto c = m_pos == m_subject.end() ? 0 : *m_pos; char c = m_pos == m_subject.end() ? 0 : *m_pos;
const RegexProgram::Op op = (RegexProgram::Op)*inst++; const RegexProgram::Op op = (RegexProgram::Op)*inst++;
switch (op) switch (op)
{ {
@ -329,6 +355,14 @@ struct ThreadedExecutor
if (not is_line_end()) if (not is_line_end())
return { StepResult::Failed }; return { StepResult::Failed };
break; break;
case RegexProgram::WordBoundary:
if (not is_word_boundary())
return { StepResult::Failed };
break;
case RegexProgram::NotWordBoundary:
if (is_word_boundary())
return { StepResult::Failed };
break;
case RegexProgram::Match: case RegexProgram::Match:
return { StepResult::Matched }; return { StepResult::Matched };
} }
@ -382,6 +416,13 @@ struct ThreadedExecutor
return m_pos == m_subject.end() or *m_pos == '\n'; return m_pos == m_subject.end() or *m_pos == '\n';
} }
bool is_word_boundary() const
{
return m_pos == m_subject.begin() or
m_pos == m_subject.end() or
is_word(*(m_pos-1)) != is_word(*m_pos);
}
ConstArrayView<char> m_program; ConstArrayView<char> m_program;
Vector<const char*> m_threads; Vector<const char*> m_threads;
StringView m_subject; StringView m_subject;
@ -392,7 +433,7 @@ struct ThreadedExecutor
auto test_regex = UnitTest{[]{ auto test_regex = UnitTest{[]{
using Exec = RegexProgram::ThreadedExecutor; using Exec = RegexProgram::ThreadedExecutor;
{ {
StringView re = "a*b"; StringView re = R"(a*b)";
auto program = RegexCompiler::compile(re.begin(), re.end()); auto program = RegexCompiler::compile(re.begin(), re.end());
RegexProgram::dump(program); RegexProgram::dump(program);
Exec exec{program}; Exec exec{program};
@ -404,7 +445,7 @@ auto test_regex = UnitTest{[]{
} }
{ {
StringView re = "^a.*b$"; StringView re = R"(^a.*b$)";
auto program = RegexCompiler::compile(re.begin(), re.end()); auto program = RegexCompiler::compile(re.begin(), re.end());
RegexProgram::dump(program); RegexProgram::dump(program);
Exec exec{program}; Exec exec{program};
@ -415,7 +456,7 @@ auto test_regex = UnitTest{[]{
} }
{ {
StringView re = "^(foo|qux|baz)+(bar)?baz$"; StringView re = R"(^(foo|qux|baz)+(bar)?baz$)";
auto program = RegexCompiler::compile(re.begin(), re.end()); auto program = RegexCompiler::compile(re.begin(), re.end());
RegexProgram::dump(program); RegexProgram::dump(program);
Exec exec{program}; Exec exec{program};
@ -426,6 +467,17 @@ auto test_regex = UnitTest{[]{
kak_assert(exec.match(program, "bazbaz")); kak_assert(exec.match(program, "bazbaz"));
kak_assert(exec.match(program, "quxbaz")); kak_assert(exec.match(program, "quxbaz"));
} }
{
StringView re = R"(.*\b(foo|bar)\b.*)";
auto program = RegexCompiler::compile(re.begin(), re.end());
RegexProgram::dump(program);
Exec exec{program};
kak_assert(exec.match(program, "tchou foo baz"));
kak_assert(not exec.match(program, "tchoufoobaz"));
kak_assert(exec.match(program, "bar"));
kak_assert(not exec.match(program, "foobar"));
}
}}; }};
} }