fix: tighten ePub image spacing — CSS margins, HTML cleanup for empty tags and wrappers

Made-with: Cursor

tests/test_epub_builder.py

@@ -4,7 +4,7 @@ from datetime import datetime, date
 from PIL import Image as PILImage
 
 from src.models import Article, Image
-from src.epub_builder import build_epub
+from src.epub_builder import _cleanup_html, build_epub
 
 
 def _create_test_image(path):
@@ -100,3 +100,38 @@ def test_build_epub_respects_article_order(app, db, tmp_path):
 
         assert titles[0] == "Earlier Article"
         assert titles[1] == "Later Article"
+
+
+def test_cleanup_removes_empty_paragraphs():
+    html = '<p>Real content.</p><p></p><p>&nbsp;</p><p> </p><p><br></p><p>More.</p>'
+    result = _cleanup_html(html)
+    assert "<p>Real content.</p>" in result
+    assert "<p>More.</p>" in result
+    assert result.count("<p") == 2
+
+
+def test_cleanup_removes_br_near_images():
+    html = '<p>Text</p><br><img src="test.jpg"><br/><p>More</p>'
+    result = _cleanup_html(html)
+    assert "<br" not in result
+    assert '<img src="test.jpg"' in result
+
+
+def test_cleanup_collapses_image_wrappers():
+    html = '<div><figure><img src="test.jpg"></figure></div>'
+    result = _cleanup_html(html)
+    assert '<img src="test.jpg"' in result
+    assert "<figure" not in result or "<figcaption" in result
+
+
+def test_cleanup_preserves_figcaption():
+    html = '<figure><img src="test.jpg"><figcaption>Caption</figcaption></figure>'
+    result = _cleanup_html(html)
+    assert '<img src="test.jpg"' in result
+    assert "Caption" in result
+
+
+def test_cleanup_removes_empty_divs():
+    html = '<p>Content</p><div>  </div><div></div><p>More</p>'
+    result = _cleanup_html(html)
+    assert result.count("<div") == 0