feat: add fsharp support (#223) (#267)

* Added fsharp support, #223

* wip: fsharp libraries classifier to be the same as csharp

Author: Thorium

src/main/kotlin/app/extractors/Extractor.kt

@@ -23,6 +23,7 @@ class Extractor : ExtractorInterface {
                 CppExtractor.FILE_EXTS +
                 CSharpExtractor.FILE_EXTS +
                 CssExtractor.FILE_EXTS +
+                FSharpExtractor.FILE_EXTS +
                 GoExtractor.FILE_EXTS +
                 JavaExtractor.FILE_EXTS +
                 JavascriptExtractor.FILE_EXTS +
@@ -47,6 +48,7 @@ class Extractor : ExtractorInterface {
             in CExtractor.FILE_EXTS -> CExtractor()
             in CppExtractor.FILE_EXTS -> CppExtractor()
             in CSharpExtractor.FILE_EXTS -> CSharpExtractor()
+            in FSharpExtractor.FILE_EXTS -> FSharpExtractor()
             in GoExtractor.FILE_EXTS -> GoExtractor()
             in ObjectiveCExtractor.FILE_EXTS -> ObjectiveCExtractor()
             in SwiftExtractor.FILE_EXTS -> SwiftExtractor()

src/main/kotlin/app/extractors/FSharpExtractor.kt

@@ -0,0 +1,61 @@
+// Copyright 2018 Sourcerer Inc. All Rights Reserved.
+// Author: Tuomas Hietanen
+// Author: Liubov Yaronskaya ([email protected])
+
+package app.extractors
+
+import app.model.CommitStats
+import app.model.DiffFile
+
+class FSharpExtractor : ExtractorInterface {
+    companion object {
+        val LANGUAGE_NAME = "fsharp"
+        val FILE_EXTS = listOf("fs", "fsx")
+        // The behaviour of csharp library classifier is the same as for csharp.
+        val LIBRARIES = ExtractorInterface.getLibraries("cs")
+        val evaluator by lazy {
+            ExtractorInterface.getLibraryClassifier("csharp")
+        }
+        val importRegex = Regex("""^.*open\s+(\w+[.\w+]*)""")
+        val commentRegex = Regex("""^([^\n]*//)[^\n]*""")
+        val extractImportRegex = Regex("""open\s+(\w+[.\w+]*)""")
+    }
+
+    override fun extract(files: List<DiffFile>): List<CommitStats> {
+        files.map { file -> file.language = LANGUAGE_NAME }
+        return super.extract(files)
+    }
+
+    override fun extractImports(fileContent: List<String>): List<String> {
+        val imports = mutableSetOf<String>()
+
+        fileContent.forEach {
+            val res = extractImportRegex.find(it)
+            if (res != null) {
+                val importedName = res.groupValues[1]
+                LIBRARIES.forEach { library ->
+                    if (importedName.startsWith(library)) {
+                        imports.add(library)
+                    }
+                }
+            }
+        }
+
+        return imports.toList()
+    }
+
+    override fun tokenize(line: String): List<String> {
+        val importRegex = Regex("""^.*open\s+(\w+[.\w+]*)""")
+        val commentRegex = Regex("""^([^\n]*//)[^\n]*""")
+        var newLine = importRegex.replace(line, "")
+        newLine = commentRegex.replace(newLine, "")
+        return super.tokenize(newLine)
+    }
+
+    override fun getLineLibraries(line: String,
+                                  fileLibraries: List<String>): List<String> {
+
+        return super.getLineLibraries(line, fileLibraries, evaluator,
+            LANGUAGE_NAME)
+    }
+}

src/test/kotlin/test/tests/extractors/ExtractorTest.kt

@@ -93,6 +93,12 @@ class ExtractorTest : Spek({
                     line, CSharpExtractor())
         }
 
+        it("fsharp extractor extracts the library") {
+            val line = "Algorithm = fun (h, v, i) -> ContrastiveDivergenceLearning(h, v)"
+            assertExtractsLineLibraries("Accord",
+                    line, FSharpExtractor())
+        }
+
         it("php extractor extracts the library") {
             val line = "public function listRepos(string \$user, int \$limit): Call;"
             assertExtractsLineLibraries("Tebru\\Retrofit",
@@ -164,6 +170,11 @@ class ExtractorTest : Spek({
             assertExtractsNoLibraries(line, CSharpExtractor())
         }
 
+        it("fsharp extractor returns empty list") {
+            val line = "let main()"
+            assertExtractsNoLibraries(line, FSharpExtractor())
+        }
+
         it("cpp extractor returns empty list") {
             val line = "std::cerr << status.ToString() << std::endl;"
             assertExtractsNoLibraries(line, CppExtractor())