fix: update tests for image path changes and removed init dirs

Author: rejojer

tests/test_cli.py

@@ -23,13 +23,11 @@ def test_init_creates_structure(tmp_path):
         assert (cwd / "wiki" / "sources" / "images").is_dir()
         assert (cwd / "wiki" / "summaries").is_dir()
         assert (cwd / "wiki" / "concepts").is_dir()
-        assert (cwd / "wiki" / "reports").is_dir()
         assert (cwd / ".openkb").is_dir()
 
         # Files
         assert (cwd / "wiki" / "AGENTS.md").is_file()
         assert (cwd / "wiki" / "log.md").is_file()
-        assert (cwd / "wiki" / "explorations").is_dir()
         assert (cwd / "wiki" / "index.md").is_file()
         assert (cwd / ".openkb" / "config.yaml").is_file()
         assert (cwd / ".openkb" / "hashes.json").is_file()

tests/test_images.py

@@ -44,7 +44,7 @@ def test_single_base64_image_extracted(self, tmp_path):
 
         # Result should reference a saved file, not the raw base64
         assert "data:image/png;base64," not in result
-        assert "![alt text](images/doc/img_001.png)" == result
+        assert "![alt text](sources/images/doc/img_001.png)" == result
 
         # File should exist on disk
         saved = images_dir / "img_001.png"
@@ -62,8 +62,8 @@ def test_multiple_base64_images_numbered_sequentially(self, tmp_path):
         )
         result = extract_base64_images(md, "doc", images_dir)
 
-        assert "![fig1](images/doc/img_001.png)" in result
-        assert "![fig2](images/doc/img_002.jpeg)" in result
+        assert "![fig1](sources/images/doc/img_001.png)" in result
+        assert "![fig2](sources/images/doc/img_002.jpeg)" in result
         assert (images_dir / "img_001.png").exists()
         assert (images_dir / "img_002.jpeg").exists()
 
@@ -92,7 +92,7 @@ def test_mixed_valid_invalid_base64(self, tmp_path, caplog):
         import logging
         with caplog.at_level(logging.WARNING, logger="openkb.images"):
             result = extract_base64_images(md, "doc", images_dir)
-        assert "![good](images/doc/img_001.png)" in result
+        assert "![good](sources/images/doc/img_001.png)" in result
         assert f"data:image/png;base64,{bad}" in result
 
 
@@ -114,7 +114,7 @@ def test_existing_relative_image_copied_and_rewritten(self, tmp_path):
         md = "![diagram](diagram.png)"
         result = copy_relative_images(md, source_dir, "doc", images_dir)
 
-        assert "![diagram](images/doc/diagram.png)" == result
+        assert "![diagram](sources/images/doc/diagram.png)" == result
         assert (images_dir / "diagram.png").read_bytes() == FAKE_PNG
 
     def test_missing_relative_image_leaves_original(self, tmp_path, caplog):
@@ -163,7 +163,7 @@ def test_multiple_relative_images_all_copied(self, tmp_path):
         md = "![a](a.png)\n![b](b.jpg)"
         result = copy_relative_images(md, source_dir, "doc", images_dir)
 
-        assert "![a](images/doc/a.png)" in result
-        assert "![b](images/doc/b.jpg)" in result
+        assert "![a](sources/images/doc/a.png)" in result
+        assert "![b](sources/images/doc/b.jpg)" in result
         assert (images_dir / "a.png").exists()
         assert (images_dir / "b.jpg").exists()

tests/test_indexer.py

@@ -28,6 +28,12 @@ def _make_fake_collection(self, doc_id: str, sample_tree: dict):
         col.get_page_content.return_value = []
         return col
 
+    def _fake_pages(self):
+        return [
+            {"page": 1, "content": "Page one text.", "images": []},
+            {"page": 2, "content": "Page two text.", "images": []},
+        ]
+
     def test_returns_index_result(self, kb_dir, sample_tree, tmp_path):
         doc_id = "abc-123"
         fake_col = self._make_fake_collection(doc_id, sample_tree)
@@ -38,7 +44,8 @@ def test_returns_index_result(self, kb_dir, sample_tree, tmp_path):
         pdf_path = tmp_path / "sample.pdf"
         pdf_path.write_bytes(b"%PDF-1.4 fake")
 
-        with patch("openkb.indexer.PageIndexClient", return_value=fake_client):
+        with patch("openkb.indexer.PageIndexClient", return_value=fake_client), \
+             patch("openkb.images.convert_pdf_to_pages", return_value=self._fake_pages()):
             result = index_long_document(pdf_path, kb_dir)
 
         assert isinstance(result, IndexResult)
@@ -63,7 +70,8 @@ def test_source_page_written_as_json(self, kb_dir, sample_tree, tmp_path):
         pdf_path = tmp_path / "sample.pdf"
         pdf_path.write_bytes(b"%PDF-1.4 fake")
 
-        with patch("openkb.indexer.PageIndexClient", return_value=fake_client):
+        with patch("openkb.indexer.PageIndexClient", return_value=fake_client), \
+             patch("openkb.images.convert_pdf_to_pages", return_value=self._fake_pages()):
             index_long_document(pdf_path, kb_dir)
 
         json_file = kb_dir / "wiki" / "sources" / "sample.json"
@@ -84,7 +92,8 @@ def test_summary_page_written(self, kb_dir, sample_tree, tmp_path):
         pdf_path = tmp_path / "sample.pdf"
         pdf_path.write_bytes(b"%PDF-1.4 fake")
 
-        with patch("openkb.indexer.PageIndexClient", return_value=fake_client):
+        with patch("openkb.indexer.PageIndexClient", return_value=fake_client), \
+             patch("openkb.images.convert_pdf_to_pages", return_value=self._fake_pages()):
             index_long_document(pdf_path, kb_dir)
 
         summary_file = kb_dir / "wiki" / "summaries" / "sample.md"
@@ -104,7 +113,8 @@ def test_localclient_called_with_index_config(self, kb_dir, sample_tree, tmp_pat
         pdf_path = tmp_path / "report.pdf"
         pdf_path.write_bytes(b"%PDF-1.4 fake")
 
-        with patch("openkb.indexer.PageIndexClient", return_value=fake_client) as mock_cls:
+        with patch("openkb.indexer.PageIndexClient", return_value=fake_client) as mock_cls, \
+             patch("openkb.images.convert_pdf_to_pages", return_value=self._fake_pages()):
             index_long_document(pdf_path, kb_dir)
 
         # Verify PageIndexClient was instantiated with correct IndexConfig