Select only raw-strings for regex fuzzing and filter blank strings (#2301)

Author: tamarinvs19

utbot-python/src/main/kotlin/org/utbot/python/fuzzing/provider/StrValueProvider.kt

@@ -22,16 +22,22 @@ object StrValueProvider : ValueProvider<Type, PythonFuzzedValue, PythonMethodDes
         return type.pythonTypeName() == pythonStrClassId.canonicalName
     }
 
-    private fun getStrConstants(concreteValues: Collection<PythonFuzzedConcreteValue>): List<String> {
+    private fun getConstants(concreteValues: Collection<PythonFuzzedConcreteValue>): List<String> {
         return concreteValues
             .filter { accept(it.type) }
-            .map {
-                val value = it.value as String
-                value.transformRawString()
-            }
-            .map {
-                it.transformQuotationMarks()
-            }
+            .map { it.value as String }
+    }
+
+    private fun getStrConstants(concreteValues: Collection<PythonFuzzedConcreteValue>): List<String> {
+        return getConstants(concreteValues)
+            .filterNot { it.isPattern() }
+            .map { it.transformQuotationMarks() }
+    }
+
+    private fun getRegexConstants(concreteValues: Collection<PythonFuzzedConcreteValue>): List<String> {
+        return getConstants(concreteValues)
+            .filter { it.isPattern() }
+            .map { it.transformRawString().transformQuotationMarks() }
     }
 
     override fun generate(description: PythonMethodDescription, type: Type) = sequence {
@@ -42,13 +48,10 @@ object StrValueProvider : ValueProvider<Type, PythonFuzzedValue, PythonMethodDes
         )
         strConstants.forEach { yieldStrings(StringValue(it)) { value } }
 
-        strConstants
-            .filter {
-                it.isPattern()
-            }
-            .forEach {
-                yieldStrings(RegexValue(it, description.random), StringValue::value)
-            }
+        val regexConstants = getRegexConstants(description.concreteValues)
+        regexConstants.forEach {
+            yieldStrings(RegexValue(it, description.random), StringValue::value)
+        }
     }
 
     private suspend fun <T : KnownValue<T>> SequenceScope<Seed<Type, PythonFuzzedValue>>.yieldStrings(value: T, block: T.() -> Any) {

utbot-python/src/main/kotlin/org/utbot/python/fuzzing/provider/utils/StringUtils.kt

@@ -23,19 +23,28 @@ fun String.transformQuotationMarks(): String {
 }
 
 fun String.transformRawString(): String {
-    val rawStringWithDoubleQuotationMarks = this.startsWith("r\"") && this.endsWith("\"")
-    val rawStringWithOneQuotationMarks = this.startsWith("r'") && this.endsWith("'")
-    return if (rawStringWithOneQuotationMarks || rawStringWithDoubleQuotationMarks) {
+    return if (this.isRawString()) {
         this.substring(2, this.length-1)
     } else {
         this
     }
 }
 
+fun String.isRawString(): Boolean {
+    val rawStringWithDoubleQuotationMarks = this.startsWith("r\"") && this.endsWith("\"")
+    val rawStringWithOneQuotationMarks = this.startsWith("r'") && this.endsWith("'")
+    return rawStringWithOneQuotationMarks || rawStringWithDoubleQuotationMarks
+}
+
 fun String.isPattern(): Boolean {
-    return try {
-        Pattern.compile(this); true
-    } catch (_: PatternSyntaxException) {
-        false
-    }
+    return if (this.isRawString()) {
+        val stringContent = this.transformRawString()
+        if (stringContent.isNotBlank()) {
+            try {
+                Pattern.compile(stringContent); true
+            } catch (_: PatternSyntaxException) {
+                false
+            }
+        } else false
+    } else false
 }