Restructure code for alternative identifier suggestions

6 files changed, 86 insertions, 50 deletions

diff --git a/libdevcore/StringUtils.cpp b/libdevcore/StringUtils.cpp
index 7b2b4f5e..1dbe151f 100644
--- a/libdevcore/StringUtils.cpp
+++ b/libdevcore/StringUtils.cpp
@@ -29,16 +29,24 @@
 using namespace std;
 using namespace dev;
 
-namespace dev
+bool dev::stringWithinDistance(string const& _str1, string const& _str2, size_t _maxDistance)
 {
-
-bool stringWithinDistance(string const& _name1, string const& _name2, size_t _maxDistance)
-{
-	if (_name1 == _name2)
+	if (_str1 == _str2)
 		return true;
 
-	size_t n1 = _name1.size(), n2 = _name2.size();
-	vector<vector<size_t>> dp(n1 + 1, vector<size_t>(n2 + 1));
+	size_t n1 = _str1.size(), n2 = _str2.size();
+	size_t distance = stringDistance(_str1, _str2);
+
+	// if distance is not greater than _maxDistance, and distance is strictly less than length of both names, they can be considered similar
+	// this is to avoid irrelevant suggestions
+	return distance <=  _maxDistance && distance < n1 && distance < n2;
+}
+
+size_t dev::stringDistance(string const& _str1, string const& _str2)
+{
+	size_t n1 = _str1.size(), n2 = _str2.size();
+	// Optimize by storing only last 2 rows and current row. So first index is considered modulo 3
+	vector<vector<size_t>> dp(3, vector<size_t>(n2 + 1));
 
 	// In this dp formulation of Damerau–Levenshtein distance we are assuming that the strings are 1-based to make base case storage easier.
 	// So index accesser to _name1 and _name2 have to be adjusted accordingly
@@ -46,32 +54,37 @@ bool stringWithinDistance(string const& _name1, string const& _name2, size_t _ma
 	{
 		for (size_t i2 = 0; i2 <= n2; ++i2)
 		{
-			if (min(i1, i2) == 0)
-				// Base case
-				dp[i1][i2] = max(i1, i2);
-			else
-			{
-				dp[i1][i2] = min(dp[i1 - 1][i2] + 1, dp[i1][i2 - 1] + 1);
-				// Deletion and insertion
-				if (_name1[i1 - 1] == _name2[i2 - 1])
-					// Same chars, can skip
-					dp[i1][i2] = min(dp[i1][i2], dp[i1 - 1][i2 - 1]);
+				if (min(i1, i2) == 0) // base case
+					dp[i1 % 3][i2] = max(i1, i2);
 				else
-					// Different chars so try substitution
-					dp[i1][i2] = min(dp[i1][i2], dp[i1 - 1][i2 - 1] + 1);
+				{
+					dp[i1 % 3][i2] = min(dp[(i1-1) % 3][i2] + 1, dp[i1 % 3][i2-1] + 1); // deletion and insertion
+					if (_str1[i1-1] == _str2[i2-1])  // same chars, can skip
+						dp[i1 % 3][i2] = min(dp[i1 % 3][i2], dp[(i1-1) % 3][i2-1]);
+					else // different chars so try substitution
+						dp[i1 % 3][i2] = min(dp[i1 % 3][i2], dp[(i1-1) % 3][i2-1] + 1);
 
-				if (i1 > 1 && i2 > 1 && _name1[i1 - 1] == _name2[i2 - 2] && _name1[i1 - 2] == _name2[i2 - 1])
-					// Try transposing
-					dp[i1][i2] = min(dp[i1][i2], dp[i1 - 2][i2 - 2] + 1);
-			}
+					if (i1 > 1 && i2 > 1 && _str1[i1-1] == _str2[i2-2] && _str1[i1-2] == _str2[i2-1]) // Try transposing
+						dp[i1 % 3][i2] = min(dp[i1 % 3][i2], dp[(i1-2) % 3][i2-2] + 1);
+				}
 		}
 	}
 
-	size_t distance = dp[n1][n2];
-
-	// if distance is not greater than _maxDistance, and distance is strictly less than length of both names,
-	// they can be considered similar this is to avoid irrelevant suggestions
-	return distance <=  _maxDistance && distance < n1 && distance < n2;
+	return dp[n1 % 3][n2];
 }
 
+string dev::quotedAlternativesList(vector<string> const& suggestions) {
+	if (suggestions.empty())
+		return "";
+	if (suggestions.size() == 1)
+		return "\"" + suggestions.front() + "\"";
+
+	string choices = "\"" + suggestions.front() + "\"";
+	for (size_t i = 1; i + 1 < suggestions.size(); ++i)
+		choices += ", \"" + suggestions[i] + "\"";
+
+	choices += " or \"" + suggestions.back() + "\"";
+
+	return choices;
 }
+
diff --git a/libdevcore/StringUtils.h b/libdevcore/StringUtils.h
index d3f9f8d9..acd93e32 100644
--- a/libdevcore/StringUtils.h
+++ b/libdevcore/StringUtils.h
@@ -24,12 +24,16 @@
 #pragma once
 
 #include <string>
+#include <vector>
 
 namespace dev
 {
 
-/// Calculates the Damerau–Levenshtein distance between @a _name1 and @a _name2 and
-/// @returns true if that distance is not greater than @a _maxDistance
-bool stringWithinDistance(std::string const& _name1, std::string const& _name2, size_t _maxDistance);
+// Calculates the Damerau–Levenshtein distance between _str1 and _str2 and returns true if that distance is not greater than _maxDistance
+bool stringWithinDistance(std::string const& _str1, std::string const& _str2, size_t _maxDistance);
+// Calculates the Damerau–Levenshtein distance between _str1 and _str2
+size_t stringDistance(std::string const& _str1, std::string const& _str2);
+// Return a string having elements of suggestions as quoted, alternative suggestions. e.g. "a", "b" or "c"
+std::string quotedAlternativesList(std::vector<std::string> const& suggestions);
 
 }
diff --git a/libsolidity/analysis/DeclarationContainer.cpp b/libsolidity/analysis/DeclarationContainer.cpp
index f9a52dc6..7508ad9e 100644
--- a/libsolidity/analysis/DeclarationContainer.cpp
+++ b/libsolidity/analysis/DeclarationContainer.cpp
@@ -119,20 +119,19 @@ vector<Declaration const*> DeclarationContainer::resolveName(ASTString const& _n
 
 vector<ASTString> DeclarationContainer::similarNames(ASTString const& _name) const
 {
+	static size_t const MAXIMUM_EDIT_DISTANCE = 2;
+
 	vector<ASTString> similar;
 
 	for (auto const& declaration: m_declarations)
 	{
 		string const& declarationName = declaration.first;
-		if (stringWithinDistance(_name, declarationName, MAXIMUM_DISTANCE))
+		if (stringWithinDistance(_name, declarationName, MAXIMUM_EDIT_DISTANCE))
 			similar.push_back(declarationName);
 	}
 
 	if (m_enclosingContainer)
-	{
-		vector<ASTString> enclosingSimilar = m_enclosingContainer->similarNames(_name);
-		similar.insert(similar.end(), enclosingSimilar.begin(), enclosingSimilar.end());
-	}
+		similar += m_enclosingContainer->similarNames(_name);
 
 	return similar;
 }
diff --git a/libsolidity/analysis/DeclarationContainer.h b/libsolidity/analysis/DeclarationContainer.h
index 6c1459c7..94bbe129 100644
--- a/libsolidity/analysis/DeclarationContainer.h
+++ b/libsolidity/analysis/DeclarationContainer.h
@@ -68,7 +68,6 @@ private:
 	std::map<ASTString, std::vector<Declaration const*>> m_declarations;
 	std::map<ASTString, std::vector<Declaration const*>> m_invisibleDeclarations;
 
-	static size_t const MAXIMUM_DISTANCE = 2;
 };
 
 }
diff --git a/libsolidity/analysis/NameAndTypeResolver.cpp b/libsolidity/analysis/NameAndTypeResolver.cpp
index 21f2af5f..5e4d414b 100644
--- a/libsolidity/analysis/NameAndTypeResolver.cpp
+++ b/libsolidity/analysis/NameAndTypeResolver.cpp
@@ -25,6 +25,7 @@
 #include <libsolidity/ast/AST.h>
 #include <libsolidity/analysis/TypeChecker.h>
 #include <libsolidity/interface/ErrorReporter.h>
+#include <libdevcore/StringUtils.h>
 
 #include <boost/algorithm/string.hpp>
 
@@ -427,19 +428,7 @@ vector<_T const*> NameAndTypeResolver::cThreeMerge(list<list<_T const*>>& _toMer
 
 string NameAndTypeResolver::similarNameSuggestions(ASTString const& _name) const
 {
-	vector<ASTString> suggestions = m_currentScope->similarNames(_name);
-	if (suggestions.empty())
-		return "";
-	if (suggestions.size() == 1)
-		return "\"" + suggestions.front() + "\"";
-
-	string choices = "\"" + suggestions.front() + "\"";
-	for (size_t i = 1; i + 1 < suggestions.size(); ++i)
-		choices += ", \"" + suggestions[i] + "\"";
-
-	choices += " or \"" + suggestions.back() + "\"";
-
-	return choices;
+	return quotedAlternativesList(m_currentScope->similarNames(_name));
 }
 
 DeclarationRegistrationHelper::DeclarationRegistrationHelper(
diff --git a/test/libdevcore/StringUtils.cpp b/test/libdevcore/StringUtils.cpp
index 7a5fe26e..597457cc 100644
--- a/test/libdevcore/StringUtils.cpp
+++ b/test/libdevcore/StringUtils.cpp
@@ -50,6 +50,38 @@ BOOST_AUTO_TEST_CASE(test_similarity)
 	BOOST_CHECK_EQUAL(stringWithinDistance("", "", 2), true);
 }
 
+BOOST_AUTO_TEST_CASE(test_dldistance)
+{
+	BOOST_CHECK_EQUAL(stringDistance("hello", "hellw"), 1);
+	BOOST_CHECK_EQUAL(stringDistance("hello", "helol"), 1);
+	BOOST_CHECK_EQUAL(stringDistance("hello", "helo"), 1);
+	BOOST_CHECK_EQUAL(stringDistance("hello", "helllo"), 1);
+	BOOST_CHECK_EQUAL(stringDistance("hello", "hlllo"), 1);
+	BOOST_CHECK_EQUAL(stringDistance("hello", "hllllo"), 2);
+	BOOST_CHECK_EQUAL(stringDistance("a", ""), 1);
+	BOOST_CHECK_EQUAL(stringDistance("abc", "ba"), 2);
+	BOOST_CHECK_EQUAL(stringDistance("abc", "abcdef"), 3);
+	BOOST_CHECK_EQUAL(stringDistance("abcd", "wxyz"), 4);
+	BOOST_CHECK_EQUAL(stringDistance("", ""), 0);
+	BOOST_CHECK_EQUAL(stringDistance("abcdefghijklmnopqrstuvwxyz", "abcabcabcabcabcabcabcabca"), 23);
+
+}
+
+BOOST_AUTO_TEST_CASE(test_alternatives_list)
+{
+	vector<string> strings;
+	BOOST_CHECK_EQUAL(quotedAlternativesList(strings), "");
+	strings.push_back("a");
+	BOOST_CHECK_EQUAL(quotedAlternativesList(strings), "\"a\"");
+	strings.push_back("b");
+	BOOST_CHECK_EQUAL(quotedAlternativesList(strings), "\"a\" or \"b\"");
+	strings.push_back("c");
+	BOOST_CHECK_EQUAL(quotedAlternativesList(strings), "\"a\", \"b\" or \"c\"");
+	strings.push_back("d");
+	BOOST_CHECK_EQUAL(quotedAlternativesList(strings), "\"a\", \"b\", \"c\" or \"d\"");
+}
+
+
 BOOST_AUTO_TEST_SUITE_END()
 
 }