X-Git-Url: http://git.tdb.fi/?a=blobdiff_plain;f=source%2Fstrings%2Fregex.cpp;h=5307c6990286ba373291281fb4ee230ff81d2d07;hb=HEAD;hp=6122fe40c557c9716cca6e88e7b2926408ad29ef;hpb=b537410dcd5b7e661625b51ee72b15e40045fd18;p=libs%2Fcore.git

diff --git a/source/strings/regex.cpp b/source/strings/regex.cpp
index 6122fe4..5307c69 100644
--- a/source/strings/regex.cpp
+++ b/source/strings/regex.cpp
@@ -1,6 +1,8 @@
 #include <limits>
 #include <list>
 #include <stack>
+#include <vector>
+#include <msp/core/except.h>
 #include "format.h"
 #include "regex.h"
 
@@ -55,8 +57,7 @@ string bad_regex::make_where(const string &e, const string::const_iterator &i)
 
 Regex::Regex(const string &expr)
 {
-	n_groups = 0;
-	string::const_iterator iter = expr.begin();
+	auto iter = expr.begin();
 	code = compile(expr, iter, n_groups, false);
 	++n_groups;
 }
@@ -123,13 +124,14 @@ Regex::Code Regex::compile(const string &expr, string::const_iterator &iter, uns
 
 	if(!has_branches)
 	{
-		for(string::const_iterator i=iter; i!=end;)
+		for(auto i=iter; i!=end;)
 		{
 			Code atom = parse_atom(expr, i, group);
 
 			Count repeat_min = 1;
 			Count repeat_max = 1;
-			parse_repeat(expr, i, repeat_min, repeat_max);
+			if(i!=end)
+				parse_repeat(expr, i, repeat_min, repeat_max);
 
 			for(unsigned j=0; j<repeat_min; ++j)
 				result += atom;
@@ -142,7 +144,7 @@ Regex::Code Regex::compile(const string &expr, string::const_iterator &iter, uns
 					result += atom;
 				}
 				result += ND_JUMP;
-				write_int<Offset>(-(atom.size()+jump_size), result);
+				write_int<Offset>(-static_cast<Offset>(atom.size()+jump_size), result);
 			}
 			else if(repeat_max>repeat_min)
 			{
@@ -157,8 +159,8 @@ Regex::Code Regex::compile(const string &expr, string::const_iterator &iter, uns
 	}
 	else
 	{
-		list<Code> branches;
-		for(string::const_iterator i=iter;;)
+		vector<Code> branches;
+		for(auto i=iter;;)
 		{
 			branches.push_back(compile(expr, i, group, true));
 			if(i==end)
@@ -169,14 +171,14 @@ Regex::Code Regex::compile(const string &expr, string::const_iterator &iter, uns
 		unsigned n_branches = branches.size();
 
 		Offset offset = (n_branches-1)*jump_size+branches.front().size();
-		for(list<Code>::iterator i=++branches.begin(); i!=branches.end(); ++i)
+		for(auto i=++branches.begin(); i!=branches.end(); ++i)
 		{
 			result += ND_JUMP;
 			write_int<Offset>(offset, result);
 			offset += i->size();
 		}
 
-		for(list<Code>::iterator i=branches.begin(); i!=branches.end();)
+		for(auto i=branches.begin(); i!=branches.end();)
 		{
 			result += *i;
 			offset -= i->size()+jump_size;
@@ -258,7 +260,7 @@ bool Regex::parse_repeat(const string &expr, string::const_iterator &i, Count &r
 		rmin = 0;
 	if(*i=='{')
 	{
-		string::const_iterator begin = i;
+		auto begin = i;
 
 		rmin = 0;
 		for(++i; isdigit(*i); ++i)
@@ -291,7 +293,7 @@ bool Regex::parse_repeat(const string &expr, string::const_iterator &i, Count &r
 
 Regex::Code Regex::parse_brackets(const string &str, string::const_iterator &iter)
 {
-	string::const_iterator begin = iter;
+	auto begin = iter;
 	Code result;
 
 	++iter;
@@ -302,7 +304,7 @@ Regex::Code Regex::parse_brackets(const string &str, string::const_iterator &ite
 		++iter;
 	}
 
-	string::const_iterator end = iter;
+	auto end = iter;
 	for(; (end!=str.end() && (end==iter || *end!=']')); ++end) ;
 	if(end==str.end())
 		throw bad_regex("unmatched '['", str, begin);
@@ -311,7 +313,7 @@ Regex::Code Regex::parse_brackets(const string &str, string::const_iterator &ite
 	unsigned type = 0;
 	bool range = false;
 	unsigned char first = 0, last = 0;
-	for(string::const_iterator i=iter; i!=end; ++i)
+	for(auto i=iter; i!=end; ++i)
 	{
 		unsigned char c = *i;
 		if(range)
@@ -364,16 +366,16 @@ Regex::Code Regex::parse_brackets(const string &str, string::const_iterator &ite
 
 RegMatch Regex::match(const string &str) const
 {
-	RegMatch::GroupArray groups(n_groups);
+	vector<RegMatch::Group> groups(n_groups);
 
-	for(string::const_iterator i=str.begin(); i!=str.end(); ++i)
+	for(auto i=str.begin(); i!=str.end(); ++i)
 		if(run(str, i, groups))
 			return RegMatch(str, groups);
 
 	return RegMatch();
 }
 
-bool Regex::run(const string &str, const string::const_iterator &begin, RegMatch::GroupArray &groups) const
+bool Regex::run(const string &str, const string::const_iterator &begin, vector<RegMatch::Group> &groups) const
 {
 	bool result = false;
 	list<RunContext> ctx;
@@ -381,7 +383,7 @@ bool Regex::run(const string &str, const string::const_iterator &begin, RegMatch
 	ctx.front().citer = code.begin();
 	ctx.front().groups.resize(groups.size());
 
-	for(string::const_iterator i=begin;;)
+	for(auto i=begin;;)
 	{
 		int c;
 		if(i!=str.end())
@@ -389,7 +391,7 @@ bool Regex::run(const string &str, const string::const_iterator &begin, RegMatch
 		else
 			c = -1;
 
-		for(list<RunContext>::iterator j=ctx.begin(); j!=ctx.end();)
+		for(auto j=ctx.begin(); j!=ctx.end();)
 		{
 			bool terminate = false;
 			bool negate_match = false;
@@ -476,7 +478,7 @@ bool Regex::run(const string &str, const string::const_iterator &begin, RegMatch
 						input_consumed = true;
 					}
 					else
-						throw logic_error("invalid instruction in regex bytecode");
+						throw internal_error("invalid instruction in regex bytecode");
 
 					if(match_result==negate_match)
 						terminate = true;
@@ -513,7 +515,7 @@ bool Regex::group_compare(const RegMatch::Group &g1, const RegMatch::Group &g2)
 	// Earlier match is better
 	if(g1.begin<g2.begin)
 		return true;
-	if(g2.begin>g2.begin)
+	if(g1.begin>g2.begin)
 		return false;
 
 	// Longer match at same position is better
@@ -524,9 +526,9 @@ string Regex::disassemble() const
 {
 	string result;
 
-	for(Code::const_iterator i=code.begin(); i!=code.end();)
+	for(auto i=code.begin(); i!=code.end();)
 	{
-		Code::const_iterator j = i;
+		auto j = i;
 		Offset offset = i-code.begin();
 		string decompiled = disassemble_instruction(i);
 		string bytes;