From 7178279586749a802337b2426b95218b13712775 Mon Sep 17 00:00:00 2001
From: haosenwang1018 <1293965075@qq.com>
Date: Sun, 15 Mar 2026 17:27:55 +0800
Subject: [PATCH] test: add coverage for TextHandler regex, clean, and
 TextHandlers.re()

Several critical code paths in custom_types.py lacked test coverage:

- TextHandler.re(check_match=True): returns bool, not TextHandlers
- TextHandler.re(replace_entities=False): entity preservation path
- TextHandler.re() with capture groups: flatten behavior
- TextHandler.re_first() default value when no match
- TextHandler.clean(remove_entities=True): entity replacement path
- TextHandler.json() valid and invalid input
- TextHandlers.re(): list-level regex with result flattening
- TextHandlers.extract()/get_all(): identity return

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 tests/parser/test_parser_advanced.py | 93 ++++++++++++++++++++++++++++
 1 file changed, 93 insertions(+)

diff --git a/tests/parser/test_parser_advanced.py b/tests/parser/test_parser_advanced.py
index f34ba5c..969ccc3 100644
--- a/tests/parser/test_parser_advanced.py
+++ b/tests/parser/test_parser_advanced.py
@@ -250,6 +250,68 @@ class TestTextHandlerAdvanced:
         matches = text3.re(r"He l lo", clean_match=True, case_sensitive=False)
         assert len(matches) == 1
 
+    def test_text_handler_regex_check_match(self):
+        """Test TextHandler.re() with check_match=True returns bool"""
+        text = TextHandler("Price: $10.99")
+        assert text.re(r"\$[\d.]+", check_match=True) is True
+        assert text.re(r"no-match-pattern", check_match=True) is False
+
+    def test_text_handler_regex_replace_entities_false(self):
+        """Test TextHandler.re() with replace_entities=False preserves entities"""
+        text = TextHandler("Hello &amp; World")
+        results = text.re(r"&amp;", replace_entities=False)
+        assert len(results) == 1
+        assert results[0] == "&amp;"
+
+    def test_text_handler_regex_with_groups(self):
+        """Test TextHandler.re() with capture groups flattens results"""
+        text = TextHandler("name=Alice age=30 name=Bob age=25")
+        results = text.re(r"name=(\w+) age=(\d+)")
+        assert len(results) == 4
+        assert "Alice" in results
+        assert "30" in results
+
+    def test_text_handler_re_first_with_default(self):
+        """Test TextHandler.re_first() returns default when no match"""
+        text = TextHandler("no numbers here")
+        result = text.re_first(r"\d+", default="N/A")
+        assert result == "N/A"
+
+    def test_text_handler_re_first_returns_first_match(self):
+        """Test TextHandler.re_first() returns first match"""
+        text = TextHandler("a1 b2 c3")
+        result = text.re_first(r"\d")
+        assert result == "1"
+        assert isinstance(result, TextHandler)
+
+    def test_text_handler_clean_with_entities(self):
+        """Test TextHandler.clean() with remove_entities=True"""
+        text = TextHandler("Hello\t&amp;\nWorld")
+        cleaned = text.clean(remove_entities=True)
+        assert "&amp;" not in cleaned
+        assert "&" in cleaned
+        assert "\t" not in cleaned
+        assert "\n" not in cleaned
+
+    def test_text_handler_clean_without_entities(self):
+        """Test TextHandler.clean() preserves entities by default"""
+        text = TextHandler("Hello\t&amp;\nWorld")
+        cleaned = text.clean(remove_entities=False)
+        assert "&amp;" in cleaned
+
+    def test_text_handler_json_valid(self):
+        """Test TextHandler.json() with valid JSON"""
+        text = TextHandler('{"key": "value", "num": 42}')
+        data = text.json()
+        assert data["key"] == "value"
+        assert data["num"] == 42
+
+    def test_text_handler_json_invalid(self):
+        """Test TextHandler.json() raises on invalid JSON"""
+        text = TextHandler("not json")
+        with pytest.raises(Exception):
+            text.json()
+
     def test_text_handlers_operations(self):
         """Test TextHandlers list operations"""
         handlers = TextHandlers([
@@ -266,6 +328,37 @@ class TestTextHandlerAdvanced:
         assert handlers.get("default") == "First"
         assert TextHandlers([]).get("default") == "default"
 
+    def test_text_handlers_re(self):
+        """Test TextHandlers.re() flattens results across all elements"""
+        handlers = TextHandlers([
+            TextHandler("a1 b2"),
+            TextHandler("c3 d4"),
+        ])
+        results = handlers.re(r"[a-z]\d")
+        assert isinstance(results, TextHandlers)
+        assert len(results) == 4
+        assert results[0] == "a1"
+        assert results[3] == "d4"
+
+    def test_text_handlers_re_empty(self):
+        """Test TextHandlers.re() on empty list"""
+        handlers = TextHandlers([])
+        results = handlers.re(r"\d+")
+        assert isinstance(results, TextHandlers)
+        assert len(results) == 0
+
+    def test_text_handlers_re_no_matches(self):
+        """Test TextHandlers.re() when no element matches"""
+        handlers = TextHandlers([TextHandler("abc"), TextHandler("def")])
+        results = handlers.re(r"\d+")
+        assert len(results) == 0
+
+    def test_text_handlers_extract(self):
+        """Test TextHandlers.extract() returns self"""
+        handlers = TextHandlers([TextHandler("a"), TextHandler("b")])
+        assert handlers.extract() is handlers
+        assert handlers.get_all() is handlers
+
 
 class TestSelectorsAdvanced:
     """Test advanced Selectors functionality"""