Updated Regx Options, updated timespan, examples, tests

sganz · Jun 20, 2018 · ba3477b · ba3477b
1 parent b115dd5
commit ba3477b
Show file tree

Hide file tree

Showing 4 changed files with 62 additions and 13 deletions.
diff --git a/Examples/Example/Program.cs b/Examples/Example/Program.cs
@@ -108,6  108,13 @@ private static void Main(string[] args)
             translated = st.RegxDefined("ScriptTags", string.Empty).RegxDefined("TagsSimple", string.Empty).ToString();
 
             Console.WriteLine("Strip Script and Tags   : >{0}<", translated);
 
             // reset and set up a predefined match pattern and set regx case sensitivity
             st.Set("wtf does RemoveWTF do? Is WtF Case SeNsItIvE?");
             st.RegxMatchesDefined.Add("RemoveWTF", @"(wtf)|(what the)\s (hell|$hit)");
 
             translated = st.RegxIgnoreCase().RegxDefined("RemoveWTF", "XXX").ToString();
             Console.WriteLine("New Pre-defined Match   : >{0}<", translated);
         }
     }
 }
diff --git a/Source/Scrubbie/Scrub.cs b/Source/Scrubbie/Scrub.cs
@@ -14,15  14,16 @@ public class Scrub
         public List<(string, string)> RegxTuples { private set; get; }
         public Dictionary<char, char> CharTransDict { private set; get; }
         private string _translatedStr;
-        private double _tkoSeconds;
         private TimeSpan _tkoSeconds;
         private RegexOptions _regxOptions;
 
         /// <summary>
         /// Sets the MatchTimeout value for all regx calls
         /// </summary>
         public double TkoSeconds
         {
-            set => _tkoSeconds = value <= 0.0 ? DefaultTkoSeconds : value;
-            get => _tkoSeconds;
             set => _tkoSeconds = value <= 0.0 ? TimeSpan.FromSeconds(DefaultTkoSeconds) : TimeSpan.FromSeconds(value);
             get => _tkoSeconds.TotalSeconds;
         }
 
         /// <summary>
@@ -52,6  53,9 @@ public int CacheSize
             { "NonAscii", @"[^\x00-\x7F] \ *(?:[^\x00-\x7F]| )*" }, // removes all non-Ascii
             { "TagsSimple" , @"\<[^\>]*\>" },     // strip tags, simple version
             { "ScriptTags" , @"<script[^>]*>[\s\S]*?</script>" },
             { "ENNumber", @"[ -]?([0-9] ([.][0-9]*)?|[.][0-9] )"}, // format with a decimal as a period, no commas for 1000's
             { "EUumber", @"[ -]?([0-9] ([,][0-9]*)?|[,][0-9] )"}, // format with a decimal as a comma, no period for 1000's
             { "UniNumber", @"[ -]?([0-9] ([.,][0-9]*)?|[,.][0-9] )"}, // picks up numbers with either comma, period in either place. May not be valid numbers
         };
 
         /// <summary>
@@ -73,7  77,11 @@ public Scrub(string origString)
 
             // set local time out (TKO) for all regx's
 
-            TkoSeconds = DefaultTkoSeconds;
             _tkoSeconds = TimeSpan.FromSeconds(DefaultTkoSeconds);
 
             // set to match case (not not ignore it)
 
             _regxOptions = RegexOptions.None;
         }
 
         /// <summary>
@@ -145,6  153,30 @@ public void SetRegxTranslator(List<(string, string)> regxTuplesList = null)
             }
         }
 
         /// <summary>
         /// Sets the Regx pattern matcher to Ignore case. This can
         /// be used prior to any regx call. It does not affect any Map
         /// function as those typically require the dictionay to be
         /// setup prior. So be warned this is ONLY for REGX's not MAP
         /// </summary>
         /// <param name="ignoreCase"></param>
         /// <returns></returns>
         public Scrub RegxIgnoreCase(bool ignoreCase = true)
         {
             if (ignoreCase)
             {
                 // add flag
                 _regxOptions |= RegexOptions.IgnoreCase;
             }
             else
             {
                 // remove flag (Think bitwise)
                 _regxOptions &= ~RegexOptions.IgnoreCase;
             }
 
             return this;
         }
 
         /// <summary>
         /// Translates given string based on the the characters in the dictionary. If character is
         /// not in the dictionay, it is pass thru untouched. Size of string is not changed.
@@ -178,7  210,7 @@ public Scrub Strip(string matchRegx)
         {
             // Call static replace method, strip and save
 
-            _translatedStr = Regex.Replace(_translatedStr, matchRegx, String.Empty, RegexOptions.None, TimeSpan.FromSeconds(TkoSeconds));
             _translatedStr = Regex.Replace(_translatedStr, matchRegx, String.Empty, _regxOptions, _tkoSeconds);
 
             return this;
         }
@@ -270,7  302,7 @@ public Scrub RegxTranslate()
             {
                 // static will compile and cache the regx for each one
 
-                _translatedStr = Regex.Replace(_translatedStr, regxTuple.Item1, regxTuple.Item2, RegexOptions.None, TimeSpan.FromSeconds(TkoSeconds));
                 _translatedStr = Regex.Replace(_translatedStr, regxTuple.Item1, regxTuple.Item2, _regxOptions, _tkoSeconds);
             }
 
             return this;
@@ -297,7  329,7 @@ public Scrub RegxDefined(string preDefined, string replacement = "")
 
             // static will compile and cache the regx for each one
 
-            _translatedStr = Regex.Replace(_translatedStr, RegxMatchesDefined[preDefined], replacement, RegexOptions.None, TimeSpan.FromSeconds(TkoSeconds));
             _translatedStr = Regex.Replace(_translatedStr, RegxMatchesDefined[preDefined], replacement, _regxOptions, _tkoSeconds);
 
             return this;
         }

diff --git a/Tests/IntegrationTests/ScrubbieIntegrationTests.cs b/Tests/IntegrationTests/ScrubbieIntegrationTests.cs
@@ -17,8  17,10 @@ public void Predefined_CompactWhitespace_Compacted()
             Scrub st = new Scrub(sentence);
 
             // Compact whitespaces to one space, note does not imply trim!
             // overides default empty string replace to replace with single space
             // note trailing space at end of string
 
-            st.RegxDefined("CompactWhitespace");
             st.RegxDefined("WhitespaceCompact", " ");
 
             Assert.AreEqual(expectedSentance, st.ToString());
         }
@@ -27,15  29,13 @@ public void Predefined_CompactWhitespace_Compacted()
         public void Predefined_InvalidName_Untouched()
         {
             string sentence = "¿¡Señor, the Chevrolet guys don't like     Dodge     guys, and and no one like MaZdA, Ola Senor?!    ";
-            string expectedSentance = "¿¡Señor, the Chevrolet guys don't like     Dodge     guys, and and no one like MaZdA, Ola Senor?!    ";
             string expectedSentance = "¿¡Señor,theChevroletguysdon'tlikeDodgeguys,andandnoonelikeMaZdA,OlaSenor?!";
 
             Scrub st = new Scrub(sentence);
 
-            // Compact whitespaces to one space, note does not imply trim!
-
-            st.RegxDefined("NotInTheListOfDefined");
             // Invalid pre-defined patter, should throw
 
-            Assert.AreEqual(expectedSentance, st.ToString());
             Assert.ThrowsException<KeyNotFoundException>(() => st.RegxDefined("NotInTheListOfDefined"));
         }
 
         [TestMethod]

diff --git a/Tests/UnitTests/ScrubbieTests.cs b/Tests/UnitTests/ScrubbieTests.cs
@@ -211,5  211,15 @@ public void SetRegxTimeOut_MatchTimeOut_Matches()
 
             Assert.AreEqual(expectedTKO, st.TkoSeconds);
         }
 
         [TestMethod]
         public void SetRegxOptions_MatchSetting_Matches()
         {
             Scrub st = new Scrub("");
 
             // TBD
 
         }
 
     }
 }