feat: add markdown chunker with sliding window and section strategies

tests/test_chunker.py

@@ -0,0 +1,41 @@
+from companion.rag.chunker import sliding_window_chunks
+
+
+def test_sliding_window_basic():
+    text = "word " * 100
+    chunks = sliding_window_chunks(text, chunk_size=20, chunk_overlap=5)
+    assert len(chunks) > 1
+    assert len(chunks[0].split()) == 20
+    # overlap check: last 5 words of chunk 0 should appear in chunk 1
+    last_five = chunks[0].split()[-5:]
+    first_chunk1 = chunks[1].split()[:5]
+    assert last_five == first_chunk1
+from companion.rag.chunker import section_based_chunks, chunk_file, ChunkingRule
+import tempfile
+from pathlib import Path
+
+
+def test_section_based_chunks_splits_on_tags():
+    text = "#DayInShort: good day\n#mentalhealth: stressed\n#work: busy"
+    chunks = section_based_chunks(text, ["#DayInShort", "#mentalhealth", "#work"], chunk_size=10, chunk_overlap=2)
+    assert len(chunks) == 3
+    assert chunks[0][1] == "#DayInShort"
+    assert chunks[1][1] == "#mentalhealth"
+    assert chunks[2][1] == "#work"
+
+
+def test_chunk_file_extracts_metadata():
+    with tempfile.TemporaryDirectory() as tmp:
+        vault = Path(tmp)
+        journal = vault / "Journal" / "2026" / "04" / "2026-04-12.md"
+        journal.parent.mkdir(parents=True)
+        journal.write_text("#DayInShort: good day\n#Relations: [[Person/Vinay]] visited.", encoding="utf-8")
+        rules = {
+            "default": ChunkingRule(strategy="sliding_window", chunk_size=500, chunk_overlap=100),
+            "Journal/**": ChunkingRule(strategy="section", chunk_size=300, chunk_overlap=50, section_tags=["#DayInShort", "#Relations"]),
+        }
+        chunks = chunk_file(journal, vault, rules, modified_at=1234567890.0)
+        assert len(chunks) == 2
+        assert chunks[0].source_directory == "Journal"
+        assert chunks[0].date == "2026-04-12"
+        assert "Person/Vinay" in (chunks[1].tags or [])