Ensure image captions and authors are extracted correctly

bresslem · bresslem · commit 7aafd6c293f6 · 2025-10-23T23:03:18.000+02:00
diff --git a/src/fundus/publishers/de/der_freitag.py b/src/fundus/publishers/de/der_freitag.py
@@ -52,4 +52,6 @@ def images(self) -> List[Image]:
                 upper_boundary_selector=CSSSelector("header.bc-article-intro"),
                 lower_boundary_selector=CSSSelector("span.freitag-article-end"),
                 image_selector=CSSSelector("figure img,div[role='figure'] img"),
+                caption_selector=XPath("./ancestor::figure//figcaption//span[@class='bo-image__caption__desc']"),
+                author_selector=XPath("./ancestor::figure//figcaption//span[@class='bo-image__caption__credit']"),
             )

Original file line number	Diff line number	Diff line change
`@@ -52,4 +52,6 @@ def images(self) -> List[Image]:`
`52`	`52`	`upper_boundary_selector=CSSSelector("header.bc-article-intro"),`
`53`	`53`	`lower_boundary_selector=CSSSelector("span.freitag-article-end"),`
`54`	`54`	`image_selector=CSSSelector("figure img,div[role='figure'] img"),`
	`55`	`+ caption_selector=XPath("./ancestor::figure//figcaption//span[@class='bo-image__caption__desc']"),`
	`56`	`+ author_selector=XPath("./ancestor::figure//figcaption//span[@class='bo-image__caption__credit']"),`
`55`	`57`	`)`