feature(http): init parser

tests/unit/byteb4rb1e/utils/http/parser/test_node.py

@@ -0,0 +1,102 @@
+import pytest
+
+from byteb4rb1e.utils.http.parser import Node, TreeBuilder
+
+
+@pytest.fixture
+def sample_dom():
+    """
+    Build a small DOM tree for testing:
+
+    <div id="root" class="container">
+        <p class="text">Hello</p>
+        <span class="text highlight">World</span>
+        <div class="box">
+            <span id="inner">Inside</span>
+        </div>
+    </div>
+    """
+    html = """
+    <div id="root" class="container">
+        <p class="text">Hello</p>
+        <span class="text highlight">World</span>
+        <div class="box">
+            <span id="inner">Inside</span>
+        </div>
+    </div>
+    """
+    parser = TreeBuilder()
+    parser.feed(html)
+    return parser.root.children[0]   # the <div id="root">
+
+
+class TestGetElementsByTagName:
+    def test_find_all_spans(self, sample_dom):
+        spans = sample_dom.get_elements_by_tag_name("span")
+        assert len(spans) == 2
+        assert spans[0].tag == "span"
+        assert spans[1].tag == "span"
+
+    def test_find_no_matches(self, sample_dom):
+        assert sample_dom.get_elements_by_tag_name("table") == []
+
+
+class TestGetElementsByClassName:
+    def test_find_single_class(self, sample_dom):
+        items = sample_dom.get_elements_by_class_name("text")
+        assert len(items) == 2
+
+    def test_find_multiple_classes(self, sample_dom):
+        items = sample_dom.get_elements_by_class_name("highlight")
+        assert len(items) == 1
+        assert items[0].tag == "span"
+
+    def test_no_such_class(self, sample_dom):
+        assert sample_dom.get_elements_by_class_name("missing") == []
+
+
+class TestGetElementById:
+    def test_find_existing_id(self, sample_dom):
+        node = sample_dom.get_element_by_id("inner")
+        assert node is not None
+        assert node.tag == "span"
+        assert node.inner_content == "Inside"
+
+    def test_missing_id(self, sample_dom):
+        assert sample_dom.get_element_by_id("nope") is None
+
+
+class TestQuerySelectorAll:
+    def test_class_selector(self, sample_dom):
+        items = sample_dom.query_selector_all(".text")
+        assert len(items) == 2
+
+    def test_id_selector(self, sample_dom):
+        items = sample_dom.query_selector_all("#inner")
+        assert len(items) == 1
+        assert items[0].inner_content == "Inside"
+
+    def test_tag_selector(self, sample_dom):
+        items = sample_dom.query_selector_all("p")
+        assert len(items) == 1
+        assert items[0].inner_content == "Hello"
+
+    def test_chained_selector(self, sample_dom):
+        items = sample_dom.query_selector_all("div .highlight")
+        assert len(items) == 1
+        assert items[0].inner_content == "World"
+
+
+class TestXPath:
+    def test_simple_tag(self, sample_dom):
+        spans = sample_dom.xpath("//span")
+        assert len(spans) == 2
+
+    def test_attribute_match(self, sample_dom):
+        nodes = sample_dom.xpath('//span[@id="inner"]')
+        assert len(nodes) == 1
+        assert nodes[0].inner_content == "Inside"
+
+    def test_nested(self, sample_dom):
+        nodes = sample_dom.xpath("//div[@class='box']")
+        assert len(nodes) == 1