fuzz token sort ratio

Author: lxian

Fuzzywuzzy_swift.xcodeproj/project.pbxproj

@@ -14,6 +14,7 @@
 		3078FD661D7495D20028F477 /* String_Fuzzywuzzy.swift in Sources */ = {isa = PBXBuildFile; fileRef = 3078FD651D7495D20028F477 /* String_Fuzzywuzzy.swift */; };
 		3078FD681D74AF5D0028F477 /* StringMatcher.swift in Sources */ = {isa = PBXBuildFile; fileRef = 3078FD671D74AF5D0028F477 /* StringMatcher.swift */; };
 		3078FD6A1D74B3FE0028F477 /* CommonSubstrings.swift in Sources */ = {isa = PBXBuildFile; fileRef = 3078FD691D74B3FE0028F477 /* CommonSubstrings.swift */; };
+		308B74501D75EFE800FC5C2D /* StringProcessor.swift in Sources */ = {isa = PBXBuildFile; fileRef = 308B744F1D75EFE800FC5C2D /* StringProcessor.swift */; };
 /* End PBXBuildFile section */
 
 /* Begin PBXContainerItemProxy section */
@@ -37,6 +38,7 @@
 		3078FD651D7495D20028F477 /* String_Fuzzywuzzy.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = String_Fuzzywuzzy.swift; sourceTree = "<group>"; };
 		3078FD671D74AF5D0028F477 /* StringMatcher.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = StringMatcher.swift; sourceTree = "<group>"; };
 		3078FD691D74B3FE0028F477 /* CommonSubstrings.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = CommonSubstrings.swift; sourceTree = "<group>"; };
+		308B744F1D75EFE800FC5C2D /* StringProcessor.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = StringProcessor.swift; sourceTree = "<group>"; };
 /* End PBXFileReference section */
 
 /* Begin PBXFrameworksBuildPhase section */
@@ -85,6 +87,7 @@
 				3078FD651D7495D20028F477 /* String_Fuzzywuzzy.swift */,
 				3078FD671D74AF5D0028F477 /* StringMatcher.swift */,
 				3078FD691D74B3FE0028F477 /* CommonSubstrings.swift */,
+				308B744F1D75EFE800FC5C2D /* StringProcessor.swift */,
 			);
 			path = Fuzzywuzzy_swift;
 			sourceTree = "<group>";
@@ -207,6 +210,7 @@
 			buildActionMask = 2147483647;
 			files = (
 				3078FD641D7495B80028F477 /* LevenshteinDistance.swift in Sources */,
+				308B74501D75EFE800FC5C2D /* StringProcessor.swift in Sources */,
 				3078FD661D7495D20028F477 /* String_Fuzzywuzzy.swift in Sources */,
 				3078FD6A1D74B3FE0028F477 /* CommonSubstrings.swift in Sources */,
 				3078FD681D74AF5D0028F477 /* StringMatcher.swift in Sources */,
@@ -397,6 +401,7 @@
 				3078FD5F1D74959A0028F477 /* Release */,
 			);
 			defaultConfigurationIsVisible = 0;
+			defaultConfigurationName = Release;
 		};
 		3078FD601D74959A0028F477 /* Build configuration list for PBXNativeTarget "Fuzzywuzzy_swiftTests" */ = {
 			isa = XCConfigurationList;
@@ -405,6 +410,7 @@
 				3078FD621D74959A0028F477 /* Release */,
 			);
 			defaultConfigurationIsVisible = 0;
+			defaultConfigurationName = Release;
 		};
 /* End XCConfigurationList section */
 	};

Fuzzywuzzy_swift.xcodeproj/project.xcworkspace/xcuserdata/xianli.xcuserdatad/UserInterfaceState.xcuserstate

@@ -0,0 +1,32 @@
+//
+//  StringProcessor.swift
+//  Fuzzywuzzy_swift
+//
+//  Created by XianLi on 31/8/2016.
+//  Copyright © 2016 LiXian. All rights reserved.
+//
+
+import UIKit
+
+class StringProcessor: NSObject {
+    /// Process string by
+    /// removing all but letters and numbers
+    /// trim whitespace
+    /// force to lower case
+    class func process(str: String) -> String {
+        /// lower case
+        var str = str.lowercaseString
+        
+        /// replace other charcters in to white space
+        let regex = try! NSRegularExpression(pattern: "\\W+",
+                                             options: NSRegularExpressionOptions.CaseInsensitive)
+        let range = NSMakeRange(0, str.characters.count)
+        str = regex.stringByReplacingMatchesInString(str,
+                                                         options: [],
+                                                         range: range,
+                                                         withTemplate: " ")
+        str = str.stringByTrimmingCharactersInSet(NSCharacterSet.init(charactersInString: " "))
+        
+        return str
+    }
+}

Fuzzywuzzy_swift/StringProcessor.swift

@@ -53,4 +53,32 @@ public extension String {
         }
         return Int((scores.maxElement() ?? 0) * 100)
     }
+    
+    static private func _fuzzProcessAndSort(str: String, fullProcess: Bool = true) -> String {
+        var str = str
+        if fullProcess {
+            str = StringProcessor.process(str)
+        }
+        let tokens = Array(str.componentsSeparatedByString(" "))
+        return tokens.sort().joinWithSeparator(" ").stringByTrimmingCharactersInSet(NSCharacterSet.init(charactersInString: ""))
+    }
+    
+    static private func _fuzzTokenSort(str1 str1: String, str2: String, partial: Bool = true, fullProcess: Bool = true) -> Int {
+        let sorted1 = _fuzzProcessAndSort(str1, fullProcess: fullProcess)
+        let sorted2 = _fuzzProcessAndSort(str2, fullProcess: fullProcess)
+        
+        if partial {
+            return fuzzPartialRatio(str1: sorted1, str2: sorted2)
+        } else {
+            return fuzzRatio(str1: sorted1, str2: sorted2)
+        }
+    }
+    
+    static public func fuzzTokenSortRatio(str1 str1: String, str2: String, fullProcess: Bool = true) -> Int {
+        return _fuzzTokenSort(str1: str1, str2: str2, partial: false, fullProcess: fullProcess)
+    }
+    
+    static public func fuzzPartialTokenSortRatio(str1 str1: String, str2: String, fullProcess: Bool = true) -> Int {
+        return _fuzzTokenSort(str1: str1, str2: str2, partial: true, fullProcess: fullProcess)
+    }
 }

Fuzzywuzzy_swift/String_Fuzzywuzzy.swift

@@ -21,7 +21,17 @@ class Fuzzywuzzy_swiftTests: XCTestCase {
         super.tearDown()
     }
 
-    func testpartialRatio() {
+    func testTokenSortRatio() {
+        let strPairs = [("some", ""), ("", "some"), ("", ""), ("fuzzy wuzzy was a bear", "wuzzy fuzzy was a bear"), ("fuzzy$*#&)$#(wuzzy*@()#*()!<><>was a bear", "wuzzy fuzzy was a bear")]
+        for (str1, str2) in strPairs {
+            print("STR1: \(str1)")
+            print("STR2: \(str2)")
+            print("TOKEN RATIO: \(String.fuzzTokenSortRatio(str1: str1, str2: str2))")
+            print("-----------------")
+        }
+    }
+    
+    func testPartialRatio() {
         let strPairs = [("some", ""), ("", "some"), ("", ""), ("abcd", "XXXbcdeEEE"), ("what a wonderful 世界", "wonderful 世"), ("this is a test", "this is a test!")]
         for (str1, str2) in strPairs {
             print("STR1: \(str1)")