thomasave
/
Lexesis
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix re

This commit is contained in:
Robin Jadoul 2016-04-26 12:35:38 +02:00
parent e476591495
commit 5338d91fbf
2 changed files with 20 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
include/Lexesis/re.h
View File

@ -78,7 +78,7 @@ namespace lxs {
std::shared_ptr<RE> e, f; std::shared_ptr<RE> e, f;
}; };
std::shared_ptr<RE> parseRE(std::string& input); std::shared_ptr<RE> parseRE(const std::string& input);
class SyntaxError : public std::runtime_error class SyntaxError : public std::runtime_error
{ {

29
src/re.cpp
View File

@ -14,7 +14,7 @@ namespace lxs {
State EmptyRE::toENFA(ENFA& enfa, State attach) State EmptyRE::toENFA(ENFA& enfa, State attach)
{ {
enfa.numStates = std::max(attach + 1, enfa.numStates); enfa.numStates = attach + 1;
enfa.accepting.clear(); enfa.accepting.clear();
enfa.accepting.insert(attach + 1); enfa.accepting.insert(attach + 1);
return attach + 1; return attach + 1;
@ -134,6 +134,7 @@ namespace lxs {
case '.': case '.':
case '\'': case '\'':
case '"': case '"':
case '-':
break; break;
case 'n': case 'n':
c = '\n'; break; c = '\n'; break;
@ -159,23 +160,24 @@ namespace lxs {
void sumREs(std::vector<std::shared_ptr<RE> >& res) void sumREs(std::vector<std::shared_ptr<RE> >& res)
{ {
for (std::size_t step = 2; step < res.size(); step <<= 1) for (std::size_t step = 1; step < res.size(); step <<= 1)
{ {
for (std::size_t i = 0; i < res.size(); i += step) for (std::size_t i = 0; i < res.size(); i += step * 2)
{ {
if (i + step / 2 < res.size()) if (i + step < res.size())
res[i] = std::make_shared<PlusRE>(res[i], res[i + step / 2]); res[i] = std::make_shared<PlusRE>(res[i], res[i + step]);
} }
} }
} }
std::shared_ptr<RE> parseCharacterClass(string& input, size_t& idx) { std::shared_ptr<RE> parseCharacterClass(const string& input, size_t& idx) {
if (idx >= input.size()) if (idx >= input.size())
throw SyntaxError("Unclosed character class"); throw SyntaxError("Unclosed character class");
std::vector<std::shared_ptr<RE> > chars; std::vector<std::shared_ptr<RE> > chars;
std::set<char> used_chars; std::set<char> used_chars;
bool invert = false; bool invert = false;
int last_char = -1;
if (input[idx] == '^') if (input[idx] == '^')
{ {
@ -191,6 +193,7 @@ namespace lxs {
{ {
used_chars.insert(']'); used_chars.insert(']');
idx++; idx++;
last_char = ']';
} }
if (idx >= input.size()) if (idx >= input.size())
@ -200,6 +203,7 @@ namespace lxs {
{ {
used_chars.insert('-'); used_chars.insert('-');
idx++; idx++;
last_char = '-';
} }
if (idx >= input.size()) if (idx >= input.size())
@ -220,10 +224,13 @@ namespace lxs {
} }
else else
{ {
for (int i = ((SingleRE*)(chars[chars.size() - 1].get()))->c + 1; i <= input[idx]; i++) if (last_char == -1)
throw SyntaxError("Nothing to apply range to");
for (int i = last_char + 1; i <= input[idx]; i++)
{ {
used_chars.insert((char) i); used_chars.insert((char) i);
} }
last_char = -1;
} }
} }
else if (input[idx] == '\\') else if (input[idx] == '\\')
@ -231,11 +238,13 @@ namespace lxs {
idx++; idx++;
if (idx >= input.size()) if (idx >= input.size())
throw SyntaxError("Unclosed character classe"); throw SyntaxError("Unclosed character classe");
used_chars.insert(parseEscapeChar(input[idx])); last_char = parseEscapeChar(input[idx]);
used_chars.insert(last_char);
} }
else else
{ {
used_chars.insert(input[idx]); used_chars.insert(input[idx]);
last_char = input[idx];
} }
} }
@ -263,7 +272,7 @@ namespace lxs {
return any[0]; return any[0];
} }
std::shared_ptr<RE> parseRE(string& input, size_t& idx) std::shared_ptr<RE> parseRE(const string& input, size_t& idx)
{ {
stack<std::shared_ptr<RE> > stk; stack<std::shared_ptr<RE> > stk;
for (; idx < input.length(); idx++) for (; idx < input.length(); idx++)
@ -358,7 +367,7 @@ namespace lxs {
} }
std::shared_ptr<RE> parseRE(string& input) std::shared_ptr<RE> parseRE(const string& input)
{ {
size_t i = 0; size_t i = 0;
std::shared_ptr<RE> res = parseRE(input, i); std::shared_ptr<RE> res = parseRE(input, i);