From 27c2a8ba951a4cf219b0445df780794d0dbd1d88 Mon Sep 17 00:00:00 2001
From: Tim Quatmann <tim.quatmann@cs.rwth-aachen.de>
Date: Tue, 19 Mar 2019 16:04:46 +0100
Subject: [PATCH] Added string utility functions to find similar strings.

---
 src/storm/utility/string.cpp | 73 ++++++++++++++++++++++++++++++++++++
 src/storm/utility/string.h   | 53 ++++++++++++++++++++++++++
 2 files changed, 126 insertions(+)
 create mode 100644 src/storm/utility/string.cpp
 create mode 100644 src/storm/utility/string.h

diff --git a/src/storm/utility/string.cpp b/src/storm/utility/string.cpp
new file mode 100644
index 000000000..536bacf24
--- /dev/null
+++ b/src/storm/utility/string.cpp
@@ -0,0 +1,73 @@
+#include "storm/utility/string.h"
+#include <vector>
+#include <boost/algorithm/string/join.hpp>
+
+namespace storm {
+    namespace utility {
+        namespace string {
+
+            SimilarStrings::SimilarStrings(std::string reference, double similarityFactor, bool caseSensitive) : reference(reference), similarityFactor(similarityFactor), caseSensitive(caseSensitive), cmp([](std::pair<uint64_t, std::string> const& lhs, std::pair<uint64_t, std::string> const& rhs) { return lhs.first > rhs.first; }), distances(cmp) {
+                // intentionally left empty.
+            }
+            
+            bool SimilarStrings::add(std::string const& string) {
+                double distance = levenshteinDistance(reference, string, caseSensitive);
+                if (distance <= static_cast<double>(std::max(reference.size(), string.size())) * (1.0 - similarityFactor)) {
+                    distances.emplace(storm::utility::string::levenshteinDistance(reference, string, caseSensitive), string);
+                    return true;
+                }
+                return false;
+            }
+            
+            std::vector<std::string> SimilarStrings::toList() const {
+                auto distancesCopy = distances;
+                std::vector<std::string> result;
+                while (!distancesCopy.empty()) {
+                    result.push_back(distancesCopy.top().second);
+                    distancesCopy.pop();
+                }
+                return result;
+            }
+            
+            std::string SimilarStrings::toDidYouMeanString() const {
+                uint64_t size = distances.size();
+                std::string result = boost::algorithm::join(toList(), ", ");
+                if (size == 0) {
+                    return "";
+                } else if (size == 1) {
+                    return "Did you mean " + result + "?";
+                } else {
+                    return "Did you mean any of [" + result + "]?";
+                }
+            }
+            
+            
+            uint64_t levenshteinDistance(std::string const& lhs, std::string const& rhs, bool caseSensitive) {
+                std::vector<std::vector<uint64_t>> d(lhs.size() + 1, std::vector<uint64_t>(rhs.size() + 1, 0ull));
+                for (uint64_t row = 1; row < d.size(); ++row) {
+                    d[row].front() = row;
+                }
+                for (uint64_t col = 1; col < d.front().size(); ++col) {
+                    d.front()[col] = col;
+                }
+                
+                for (uint64_t row = 1; row < d.size(); ++row) {
+                    for (uint64_t col = 1; col < d[row].size(); ++col) {
+                        uint64_t cost = 1;
+                        if (caseSensitive) {
+                            if (tolower(lhs[row-1]) == tolower(rhs[col-1])) {
+                                cost = 0;
+                            }
+                        } else {
+                            if (lhs[row-1] == rhs[col-1]) {
+                                cost = 0;
+                            }
+                        }
+                        d[row][col] = std::min( { d[row-1][col] + 1, d[row][col - 1] + 1, d[row-1][col-1] + cost } );
+                    }
+                }
+                return d.back().back();
+            }
+        }
+    }
+}
\ No newline at end of file
diff --git a/src/storm/utility/string.h b/src/storm/utility/string.h
new file mode 100644
index 000000000..6b44992c5
--- /dev/null
+++ b/src/storm/utility/string.h
@@ -0,0 +1,53 @@
+#pragma once
+
+#include <string>
+#include <functional>
+#include <queue>
+
+namespace storm {
+    namespace utility {
+        namespace string {
+            
+            class SimilarStrings {
+            public:
+                /*!
+                 * Gathers strings that are similar to the given reference string
+                 * @param reference
+                 * @param similarityFactor controls how similar the strings need to be (0 means any string is similar, 1 means only the reference string is similar)
+                 * @param caseSensitive if false, lower/upper case is ignored
+                 */
+                SimilarStrings(std::string reference, double similarityFactor = 0.6, bool caseSensitive = true);
+                
+                /*!
+                 * Adds the given string to the set of similar strings (if it is similar)
+                 * @return true, if the given string is considered similar.
+                 */
+                bool add(std::string const& string);
+                
+                /*!
+                 * Gets a list of all added strings that are similar to the reference string.
+                 * Erases all strings gathered so far.
+                 */
+                std::vector<std::string> toList() const;
+                
+                /*!
+                 * Returns a "Did you mean abc?" string
+                 * @return
+                 */
+                std::string toDidYouMeanString() const;
+
+            private:
+                std::string reference;
+                double similarityFactor;
+                bool caseSensitive;
+                std::function<bool (std::pair<uint64_t, std::string> const&, std::pair<uint64_t, std::string> const&)> cmp;
+                std::priority_queue<std::pair<uint64_t, std::string>, std::vector<std::pair<uint64_t, std::string>>, decltype(cmp)> distances;
+            };
+            
+            /*!
+             * Levenstein distance to find similar strings
+             */
+            uint64_t levenshteinDistance(std::string const& lhs, std::string const& rhs, bool caseSensitive = true);
+        }
+    }
+}
\ No newline at end of file