From 30a1bd63b44a9d59f02231ed7b013164e957da52 Mon Sep 17 00:00:00 2001
From: Mikko Rasa <tdb@tdb.fi>
Date: Wed, 8 Jun 2011 16:07:31 +0300
Subject: [PATCH] Improve exceptions for Regex and RegMatch

---
 source/strings/regex.cpp    | 61 ++++++++++++++++++++++++++-----------
 source/strings/regex.h      | 14 ++++++++-
 source/strings/regmatch.cpp |  4 +--
 3 files changed, 58 insertions(+), 21 deletions(-)

diff --git a/source/strings/regex.cpp b/source/strings/regex.cpp
index bfd25c7..5b04764 100644
--- a/source/strings/regex.cpp
+++ b/source/strings/regex.cpp
@@ -1,6 +1,6 @@
-#include <stack>
 #include <limits>
-#include <msp/core/except.h>
+#include <list>
+#include <stack>
 #include "format.h"
 #include "regex.h"
 
@@ -31,6 +31,28 @@ T read_int(Msp::Regex::Code::const_iterator &c)
 
 namespace Msp {
 
+bad_regex::bad_regex(const string &w, const string &e, const string::const_iterator &i):
+	logic_error(w+"\n"+make_where(e, i))
+{ }
+
+string bad_regex::make_where(const string &e, const string::const_iterator &i)
+{
+	string result;
+	string::size_type offset = i-e.begin();
+	if(offset>40)
+	{
+		result = e.substr(offset-40, 60);
+		offset = 40;
+	}
+	else
+		result = e.substr(0, 60);
+	result += '\n';
+	result.append(offset, ' ');
+	result += '^';
+	return result;
+}
+
+
 Regex::Regex(const string &expr)
 {
 	n_groups = 0;
@@ -42,7 +64,7 @@ Regex::Regex(const string &expr)
 Regex::Code Regex::compile(const string &expr, string::const_iterator &iter, unsigned &group, bool branch)
 {
 	bool has_branches = false;
-	unsigned level = 0;
+	stack<string::const_iterator> parens;
 	bool escape = false;
 	unsigned bracket = 0;
 	string::const_iterator end;
@@ -62,19 +84,19 @@ Regex::Code Regex::compile(const string &expr, string::const_iterator &iter, uns
 		else if(*end=='\\')
 			escape = true;
 		else if(*end=='(')
-			++level;
+			parens.push(end);
 		else if(*end==')')
 		{
-			if(level==0)
+			if(parens.empty())
 			{
 				if(group==0)
-					throw InvalidParameterValue("Unexpected )");
+					throw bad_regex("unmatched ')'", expr, end);
 				else
 					break;
 			}
-			--level;
+			parens.pop();
 		}
-		else if(*end=='|' && level==0)
+		else if(*end=='|' && parens.empty())
 		{
 			if(branch)
 				break;
@@ -85,8 +107,8 @@ Regex::Code Regex::compile(const string &expr, string::const_iterator &iter, uns
 			bracket = 1;
 	}
 
-	if(level>0)
-		throw InvalidParameterValue("Unmatched (");
+	if(!parens.empty())
+		throw bad_regex("unmatched '('", expr, parens.top());
 
 	Code result;
 
@@ -107,7 +129,7 @@ Regex::Code Regex::compile(const string &expr, string::const_iterator &iter, uns
 
 			Count repeat_min = 1;
 			Count repeat_max = 1;
-			parse_repeat(i, repeat_min, repeat_max);
+			parse_repeat(expr, i, repeat_min, repeat_max);
 
 			for(unsigned j=0; j<repeat_min; ++j)
 				result += atom;
@@ -189,14 +211,14 @@ Regex::Code Regex::parse_atom(const string &expr, string::const_iterator &i, uns
 	if(*i=='\\')
 	{
 		if(++i==expr.end())
-			throw InvalidParameterValue("Stray backslash");
+			throw bad_regex("stray backslash", expr, i-1);
 		flag = true;
 	}
 
 	if(!flag)
 	{
 		if(*i=='*' || *i=='{' || *i=='}' || *i=='+' || *i=='?' || *i=='|' || *i==')')
-			throw InvalidParameterValue("Invalid atom");
+			throw bad_regex("invalid atom", expr, i);
 		else if(*i=='[')
 			return parse_brackets(expr, i);
 		else if(*i=='.')
@@ -225,7 +247,7 @@ Regex::Code Regex::parse_atom(const string &expr, string::const_iterator &i, uns
 	return result;
 }
 
-bool Regex::parse_repeat(string::const_iterator &i, Count &rmin, Count &rmax)
+bool Regex::parse_repeat(const string &expr, string::const_iterator &i, Count &rmin, Count &rmax)
 {
 	if(*i!='*' && *i!='+' && *i!='?' && *i!='{')
 		return false;
@@ -236,6 +258,8 @@ bool Regex::parse_repeat(string::const_iterator &i, Count &rmin, Count &rmax)
 		rmin = 0;
 	if(*i=='{')
 	{
+		string::const_iterator begin = i;
+
 		rmin = 0;
 		for(++i; isdigit(*i); ++i)
 			rmin = rmin*10+(*i-'0');
@@ -249,7 +273,7 @@ bool Regex::parse_repeat(string::const_iterator &i, Count &rmin, Count &rmax)
 				for(; isdigit(*i); ++i)
 					rmax = rmax*10+(*i-'0');
 				if(rmax<rmin)
-					throw InvalidParameterValue("Invalid bound");
+					throw bad_regex("invalid bound", expr, begin);
 			}
 			else
 				rmax = numeric_limits<Count>::max();
@@ -257,7 +281,7 @@ bool Regex::parse_repeat(string::const_iterator &i, Count &rmin, Count &rmax)
 		else
 			rmax = rmin;
 		if(*i!='}')
-			throw InvalidParameterValue("Invalid bound");
+			throw bad_regex("invalid bound", expr, begin);
 	}
 
 	++i;
@@ -267,6 +291,7 @@ bool Regex::parse_repeat(string::const_iterator &i, Count &rmin, Count &rmax)
 
 Regex::Code Regex::parse_brackets(const string &str, string::const_iterator &iter)
 {
+	string::const_iterator begin = iter;
 	Code result;
 
 	++iter;
@@ -280,7 +305,7 @@ Regex::Code Regex::parse_brackets(const string &str, string::const_iterator &ite
 	string::const_iterator end = iter;
 	for(; (end!=str.end() && (end==iter || *end!=']')); ++end) ;
 	if(end==str.end())
-		throw InvalidParameterValue("Unmatched '['");
+		throw bad_regex("unmatched '['", str, begin);
 
 	unsigned char mask[32] = {0};
 	unsigned type = 0;
@@ -451,7 +476,7 @@ bool Regex::run(const string &str, const string::const_iterator &begin, RegMatch
 						input_consumed = true;
 					}
 					else
-						throw Exception("Invalid instruction");
+						throw logic_error("invalid instruction in regex bytecode");
 
 					if(match_result==negate_match)
 						terminate = true;
diff --git a/source/strings/regex.h b/source/strings/regex.h
index 5b809ab..6792502 100644
--- a/source/strings/regex.h
+++ b/source/strings/regex.h
@@ -1,11 +1,23 @@
 #ifndef MSP_STRINGS_REGEX_H_
 #define MSP_STRINGS_REGEX_H_
 
+#include <stdexcept>
 #include <string>
 #include "regmatch.h"
 
 namespace Msp {
 
+class bad_regex: public std::logic_error
+{
+public:
+	bad_regex(const std::string &, const std::string &, const std::string::const_iterator &);
+	virtual ~bad_regex() throw() { }
+
+private:
+	std::string make_where(const std::string &, const std::string::const_iterator &);
+};
+
+
 /**
 This class provides regular expression matching.  It supports a subset of
 POSIX.2 extended regex syntax.  Character classes, equivalence classes and
@@ -108,7 +120,7 @@ private:
 
 	Code parse_atom(const std::string &, std::string::const_iterator &i, unsigned &);
 	Code parse_brackets(const std::string &, std::string::const_iterator &);
-	bool parse_repeat(std::string::const_iterator &, Count &, Count &);
+	bool parse_repeat(const std::string &, std::string::const_iterator &, Count &, Count &);
 
 public:
 	/** Matches the regex against a string.  Refer to RegMatch documentation for
diff --git a/source/strings/regmatch.cpp b/source/strings/regmatch.cpp
index 61bed84..4799cfb 100644
--- a/source/strings/regmatch.cpp
+++ b/source/strings/regmatch.cpp
@@ -1,4 +1,4 @@
-#include <msp/core/except.h>
+#include <stdexcept>
 #include "regmatch.h"
 
 using namespace std;
@@ -19,7 +19,7 @@ RegMatch::RegMatch(const string &str, const GroupArray &g):
 const RegMatch::Group &RegMatch::group(unsigned i) const
 {
 	if(i>=groups.size())
-		throw InvalidParameterValue("Group index out of range");
+		throw out_of_range("RegMatch::group");
 	return groups[i];
 }
 
-- 
2.45.2