diff --git a/crates/programs/aichatbot-sidecar/src/workers/news_stories/news_story_audio_preprocessing/process_single_item.rs b/crates/programs/aichatbot-sidecar/src/workers/news_stories/news_story_audio_preprocessing/process_single_item.rs index 00968e2bfc..91016b392b 100644 --- a/crates/programs/aichatbot-sidecar/src/workers/news_stories/news_story_audio_preprocessing/process_single_item.rs +++ b/crates/programs/aichatbot-sidecar/src/workers/news_stories/news_story_audio_preprocessing/process_single_item.rs @@ -90,7 +90,7 @@ fn split_into_manageable_portions(text: &str) -> Vec { if trimmed_paragraph.len() > 250 { speakable_portions.extend(split_paragraph(&trimmed_paragraph)); } else { - speakable_portions.push(paragraph.to_string()); + speakable_portions.push(trimmed_paragraph.to_string()); } } diff --git a/crates/programs/aichatbot-sidecar/src/workers/news_stories/news_story_greenlighting/site_handlers/cnn_greenlighting.rs b/crates/programs/aichatbot-sidecar/src/workers/news_stories/news_story_greenlighting/site_handlers/cnn_greenlighting.rs index d70abf4d58..edbcbca995 100644 --- a/crates/programs/aichatbot-sidecar/src/workers/news_stories/news_story_greenlighting/site_handlers/cnn_greenlighting.rs +++ b/crates/programs/aichatbot-sidecar/src/workers/news_stories/news_story_greenlighting/site_handlers/cnn_greenlighting.rs @@ -3,9 +3,17 @@ use sqlite_queries::queries::by_table::news_story_productions::list::news_story_ pub fn cnn_greenlighting(target: &NewsStoryProductionItem) -> Option { - if target.original_news_canonical_url.contains("/deals/") { + if target.original_news_canonical_url.contains("/deals/") + || target.original_news_canonical_url.contains("cnn-underscored") { + // We got a 3-minute long ad for Levi's with the "/deals/" URL. + // CNN underscored is nothing but ads: https://www.cnn.com/cnn-underscored return Some(SkipReason::Advertisement); } + if target.original_news_canonical_url.contains("/videos/") { + // Video content tends to have little text + return Some(SkipReason::VideoContent); + } + None } \ No newline at end of file