Merge pull request #12 from CraigOates/0.7-p2

0.7 p2

WetPancake/DataServices.fs

@@ -12,7 +12,6 @@
         ListSampleFiles
         |> Array.contains filePath
 
-
     let rec GenerateMarkovChain (map: Map<string, string List>) (state:string) chain =
         if map.ContainsKey state then
             let nextChoice = map.[state] |> PickRandomItem
@@ -20,7 +19,7 @@
             else
                 let currentWords =
                     state
-                    |> SplitText @"\s+"
+                    |> SplitText "\s+"
                     |> Seq.skip 1
                     |> ConcatToString
                 GenerateMarkovChain map (CombineWords currentWords nextChoice) (nextChoice :: chain)
@@ -62,8 +61,7 @@
 
     As this NuGet package is about generating random text, this should not be a problem for the most part.
     With that said, when this happens, it is an incorrect result -- from the end-users point-of-view.
-    So, to get a more accurate result, pass the result of this function through the data-cleaner function.
-    (THE DATA-CLEANER FUNCTION HAS NOT BEEN MADE YET.)
+    So, to get a more accurate result, pass the result of this function through to the RemoveArtifactSentences function.
     *) 
     let GenerateMarkovText noOfSentences map =
         let startWords = fst(SeperateStartWords map)
@@ -73,4 +71,11 @@
                     Thread.Sleep(100)
                     yield GenerateMarkovSentence map (PickRandomItem startWords).Key
             }
-        result |> ConcatToString
+        result |> ConcatToString
+
+    let RemoveArtifactSentences noOfSentences text =
+        text
+        |> SplitText @"(?<=[\.\!\?]\s)"
+        |> Array.take noOfSentences
+        |> ConcatToString
+        |> ReplaceArtifact "  "

WetPancake/ProductServices.fs

@@ -141,4 +141,13 @@ module  Pancake =
                 ListSampleFiles
                 |> Array.toList
             return files
+        }
+
+    // Finish annotating this function when this branch (0.7-p2) is merged back into 0.7-p1.
+    let CleanResult noOfSentences text =
+        async {
+            let cleanText =
+                text
+                |> RemoveArtifactSentences noOfSentences
+            return cleanText
         }

WetPancake/Script.fsx

@@ -154,6 +154,11 @@ let dss_text =
 let dss_startwords = SeperateStartWords dss_text
 let dss_sentences = GenerateMarkovText 2 dss_text
 printfn "Text: %A" dss_sentences
+let dss_sentences2 = "This is a sentence. And, so is this. This shouldn't be here! How about this? No!"
+let dss_clean =
+    dss_sentences2
+    |> RemoveArtifactSentences 2 // Change the no. of sentences to test.
+printfn "CLEANED TEXT: %s" dss_clean
 
 
 // Product Services (Pancake)