From 625a0f33c9550cce715f51dc026e2d496bf76311 Mon Sep 17 00:00:00 2001
From: Roman Babenko <babenek@gmail.com>
Date: Mon, 16 Sep 2024 15:26:36 +0300
Subject: [PATCH] Tests for coverage

---
 tests/ml_model/test_features.py | 47 ++++++++++++++++++++++++++-------
 tests/utils/test_util.py        |  1 +
 2 files changed, 39 insertions(+), 9 deletions(-)

diff --git a/tests/ml_model/test_features.py b/tests/ml_model/test_features.py
index 55ab26838..bccc2ffcb 100644
--- a/tests/ml_model/test_features.py
+++ b/tests/ml_model/test_features.py
@@ -3,7 +3,7 @@
 
 from credsweeper.common.constants import Severity
 from credsweeper.credentials import Candidate, LineData
-from credsweeper.ml_model.features import SearchInAttribute, CharSet
+from credsweeper.ml_model.features import SearchInAttribute, CharSet, WordInPath
 from credsweeper.ml_model.features.has_html_tag import HasHtmlTag
 from credsweeper.ml_model.features.is_secret_numeric import IsSecretNumeric
 from credsweeper.ml_model.features.reny_entropy import RenyiEntropy
@@ -21,7 +21,7 @@ def setUp(self):
                                   line=AZ_STRING,
                                   line_pos=0,
                                   line_num=1,
-                                  path="path",
+                                  path="path.ext",
                                   file_type="type",
                                   info="info",
                                   pattern=RE_TEST_PATTERN)
@@ -57,30 +57,59 @@ def test_estimate_entropy_p(self):
         probabilities = test_entropy.get_probabilities(AZ_STRING)
         self.assertEqual(4.754887502163468, test_entropy.estimate_entropy(probabilities))
 
+    def test_word_in_path_n(self):
+        self.assertListEqual([[0, 0, 0, 0]],
+                             WordInPath(["dog", "lazy", "small",
+                                         "the"])([Candidate([self.line_data], [], "rule", Severity.MEDIUM)]).tolist())
+
+    def test_word_in_path_p(self):
+        self.assertListEqual([[1, 0, 0, 0]],
+                             WordInPath([".ext", "lazy", "small",
+                                         "the"])(
+                                 [Candidate([self.line_data], [], "rule", Severity.MEDIUM)]).tolist())
+
+    def test_word_in_value_n(self):
+        self.assertListEqual([[0, 0, 0, 0]],
+                             WordInPath(["aaa", "bbb", "ccc",
+                                         "ddd"])([Candidate([self.line_data], [], "rule", Severity.MEDIUM)]).tolist())
+
+
     def test_word_in_value_p(self):
         self.assertListEqual([[1, 1, 0, 1]],
                              WordInValue(["dog", "lazy", "small",
                                           "the"]).extract(Candidate([self.line_data], [], "rule",
                                                                     Severity.MEDIUM)).tolist())
 
-    def test_word_in_value_n(self):
-        self.assertListEqual([[0, 0]],
-                             WordInValue(["pink",
-                                          "quick"]).extract(Candidate([self.line_data], [], "rule",
-                                                                      Severity.MEDIUM)).tolist())
+    def test_word_in_line_dup_n(self):
+        with self.assertRaises(Exception):
+            WordInLine(["fox", "fox"])
 
     def test_word_in_line_n(self):
-        test = WordInLine(["text"])
-        self.assertListEqual([[0]], test.extract(Candidate([self.line_data], [], "rule", Severity.MEDIUM)).tolist())
+        test = WordInLine(["dummy", "text"])
+        self.assertListEqual([[0, 0]], test.extract(Candidate([self.line_data], [], "rule", Severity.MEDIUM)).tolist())
+
+    def test_word_in_line_p(self):
+        test = WordInLine(["bear", "brown"])
+        self.assertListEqual([[0, 1]], test.extract(Candidate([self.line_data], [], "rule", Severity.MEDIUM)).tolist())
 
     def test_has_html_tag_n(self):
         test = HasHtmlTag()
         self.assertFalse(test.extract(Candidate([self.line_data], [], "rule", Severity.MEDIUM)))
 
+    def test_has_html_tag_p(self):
+        test = HasHtmlTag()
+        self.line_data.line = f"<p>{self.line_data.line}</p>"
+        self.assertTrue(test.extract(Candidate([self.line_data], [], "rule", Severity.MEDIUM)))
+
     def test_is_secret_numeric_n(self):
         test = IsSecretNumeric()
         self.assertFalse(test.extract(Candidate([self.line_data], [], "rule", Severity.MEDIUM)))
 
+    def test_is_secret_numeric_p(self):
+        test = IsSecretNumeric()
+        self.line_data.value = "2.718281828"
+        self.assertTrue(test.extract(Candidate([self.line_data], [], "rule", Severity.MEDIUM)))
+
     def test_search_in_attribute_n(self):
         self.assertFalse(
             SearchInAttribute("^the lazy dog$", "line").extract(Candidate([self.line_data], [], "rule",
diff --git a/tests/utils/test_util.py b/tests/utils/test_util.py
index b2cf2cca0..cdcaef7eb 100644
--- a/tests/utils/test_util.py
+++ b/tests/utils/test_util.py
@@ -322,6 +322,7 @@ def test_is_binary_p(self):
     def test_is_binary_n(self):
         self.assertFalse(Util.is_binary("Üben von Xylophon und Querflöte ist ja zweckmäßig".encode(LATIN_1)))
         self.assertFalse(Util.is_binary(b"\x7Ffew unprintable letters\x00"))
+        self.assertFalse(Util.is_binary(b""))
 
     def test_is_ascii_entropy_validate_p(self):
         self.assertTrue(Util.is_ascii_entropy_validate(b''))