fix: fix mikan web extractors

apps/recorder/src/extract/html/mod.rs

@@ -1,3 +1,11 @@
 pub mod styles;
 
-pub use styles::parse_style_attr;
+use html_escape::decode_html_entities;
+use itertools::Itertools;
+use scraper::ElementRef;
+pub use styles::{extract_background_image_src_from_style_attr, extract_style_from_attr};
+
+pub fn extract_inner_text_from_element_ref(el: ElementRef<'_>) -> String {
+    let raw_text = el.text().collect_vec().join(",");
+    decode_html_entities(&raw_text).trim().to_string()
+}

apps/recorder/src/extract/html/styles.rs

@@ -1,6 +1,45 @@
-use lightningcss::declaration::DeclarationBlock;
+use lightningcss::{
+    declaration::DeclarationBlock, properties::Property, values::image::Image as CSSImage,
+};
+use url::Url;
 
-pub fn parse_style_attr(style_attr: &str) -> Option<DeclarationBlock> {
+use crate::extract::media::extract_image_src_from_str;
+
+pub fn extract_style_from_attr(style_attr: &str) -> Option<DeclarationBlock> {
     let result = DeclarationBlock::parse_string(style_attr, Default::default()).ok()?;
     Some(result)
 }
+
+pub fn extract_background_image_src_from_style_attr(
+    style_attr: &str,
+    base_url: &Url,
+) -> Option<Url> {
+    extract_style_from_attr(style_attr).and_then(|style| {
+        style.iter().find_map(|(prop, _)| {
+            match prop {
+                Property::BackgroundImage(images) => {
+                    for img in images {
+                        if let CSSImage::Url(path) = img {
+                            if let Some(url) = extract_image_src_from_str(path.url.trim(), base_url)
+                            {
+                                return Some(url);
+                            }
+                        }
+                    }
+                }
+                Property::Background(backgrounds) => {
+                    for bg in backgrounds {
+                        if let CSSImage::Url(path) = &bg.image {
+                            if let Some(url) = extract_image_src_from_str(path.url.trim(), base_url)
+                            {
+                                return Some(url);
+                            }
+                        }
+                    }
+                }
+                _ => {}
+            }
+            None
+        })
+    })
+}