fix: regex faster (#172)

* fix: regex faster

* fix: refactor

* feat: restrict hashing of .min.js files

Author: yaronskaya

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

@@ -14,6 +14,7 @@ class Extractor : ExtractorInterface {
         val TYPE_KEYWORD = 3
         val TYPE_SYNTAX = 4
         val SEPARATOR = ">"
+        val RESTRICTED_EXTS = listOf(".min.js")
     }
 
     fun create(extension: String): ExtractorInterface {
@@ -36,6 +37,7 @@ class Extractor : ExtractorInterface {
 
     override fun extract(files: List<DiffFile>): List<CommitStats> {
         return files.groupBy { file -> file.extension }
+            .filter { (extension, _) -> !RESTRICTED_EXTS.contains(extension) }
             .map { (extension, files) -> create(extension).extract(files) }
             .fold(mutableListOf()) { accStats, stats ->
                 accStats.addAll(stats)

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

@@ -23,22 +23,19 @@ class JavascriptExtractor : ExtractorInterface {
     }
 
     override fun extractImports(fileContent: List<String>): List<String> {
-        val imports = mutableSetOf<String>()
-
         val splitRegex =
             Regex("""\s+|,|;|:|\*|\n|\(|\)|\\[|]|\{|}|\+|=|\.|>|<|#|@|\$""")
-        val twoOrMoreWordsRegex = Regex("""(".+?\s.+?[^"]*"|'.+?\s.+?[^']*')""")
-        val fileTokens = twoOrMoreWordsRegex.replace(fileContent.joinToString(separator = " ").toLowerCase(), "")
-            .split(splitRegex)
+        val twoOrMoreWordsRegex = Regex("""(".+?\s.+?"|'.+?\s.+?')""")
 
-        imports.addAll(fileTokens.filter { token -> token in LIBRARIES })
+        val line = fileContent.joinToString(separator = " ").toLowerCase()
+        val fileTokens = twoOrMoreWordsRegex.replace(line, "").split(splitRegex)
 
-        return imports.toList()
+        return fileTokens.filter { token -> token in LIBRARIES }.distinct()
     }
 
     override fun getLineLibraries(line: String,
                                   fileLibraries: List<String>): List<String> {
-
-        return super.getLineLibraries(line, fileLibraries, evaluator, LANGUAGE_NAME)
+        return super.getLineLibraries(line, fileLibraries, evaluator,
+            LANGUAGE_NAME)
     }
 }

src/main/kotlin/app/utils/FileHelper.kt

@@ -17,6 +17,8 @@ object FileHelper {
     private val jarPath = getJarPath()
     private val settingsPath = jarPath.resolve(dirName)
 
+    private val specificExts = listOf(".min.js")
+
     fun getPath(name: String, vararg parts: String): Path {
         val path = settingsPath.resolve(Paths.get("", *parts))
         if (Files.notExists(path)) {
@@ -34,7 +36,12 @@ object FileHelper {
     }
 
     fun getFileExtension(path: String): String {
-        val fileName = Paths.get(path).fileName.toString()
+        val fileName = Paths.get(path).fileName.toString().toLowerCase()
+        for (ext in specificExts) {
+            if (fileName.endsWith(ext)) {
+                return ext
+            }
+        }
         return fileName.substringAfterLast(delimiter = '.',
                                            missingDelimiterValue = "")
     }