refactor(tests): use omniread public API instead of internal module imports

- Replace deep imports with top-level omniread exports in tests
- Ensure tests validate only the supported public API surface
- Align HTML and PDF tests with documented library usage

tests/conftest.py

@@ -4,10 +4,17 @@ import httpx
 from pathlib import Path
 from jinja2 import Environment, BaseLoader
 
-from omniread.core.content import ContentType
+from omniread import (
-from omniread.html.scraper import HTMLScraper
+    # core
-from omniread.pdf.client import FileSystemPDFClient
+    ContentType,
-from omniread.pdf.scraper import PDFScraper
+
+    # html
+    HTMLScraper,
+
+    # pdf
+    FileSystemPDFClient,
+    PDFScraper,
+)
 
 
 MOCK_HTML_DIR = Path(__file__).parent / "mocks" / "html"

tests/test_html_simple.py

@@ -3,8 +3,13 @@ from typing import Optional
 from pydantic import BaseModel
 from bs4 import Tag
 
-from omniread.html.parser import HTMLParser
+from omniread import (
-from omniread.core.content import Content
+    # core
+    Content,
+
+    # html
+    HTMLParser,
+)
 
 
 class ParsedSimpleHTML(BaseModel):

tests/test_html_table.py

@@ -2,8 +2,13 @@ from typing import Optional
 
 from pydantic import BaseModel
 
-from omniread.html.parser import HTMLParser
+from omniread import (
-from omniread.core.content import Content
+    # core
+    Content,
+
+    # html
+    HTMLParser,
+)
 
 
 class ParsedTableHTML(BaseModel):

tests/test_pdf_simple.py

@@ -1,9 +1,13 @@
 from typing import Literal
 from pydantic import BaseModel
 
-from omniread.pdf import PDFParser
+from omniread import (
-from omniread.core.content import Content
+    # core
+    Content,
 
+    # pdf
+    PDFParser,
+)
 
 class ParsedPDF(BaseModel):
     size_bytes: int