link_preview/providers/
schema.rs

1use scraper::{Html, Selector};
2
3/// Schema.org meta tags.
4pub enum SchemaMetaTag {
5    Name,
6    Description,
7    Image,
8}
9
10impl SchemaMetaTag {
11    fn str(&self) -> &str {
12        match self {
13            SchemaMetaTag::Name => "name",
14            SchemaMetaTag::Description => "description",
15            SchemaMetaTag::Image => "image",
16        }
17    }
18}
19
20/// Finds the Schema.org tag specified in the provided `Html` instance
21pub fn find_schema_tag(html: &Html, tag: SchemaMetaTag) -> Option<String> {
22    let selector = Selector::parse(&format!("meta[itemprop=\"{}\"]", tag.str())).unwrap();
23
24    if let Some(element) = html.select(&selector).next() {
25        if let Some(value) = element.value().attr("content") {
26            return Some(value.to_string());
27        }
28    }
29
30    None
31}
32
33#[cfg(test)]
34mod tests {
35    use crate::html_from_bytes;
36    use crate::tests::SCHEMA_COMPLIANT_HTML;
37
38    use super::{find_schema_tag, SchemaMetaTag};
39
40    #[test]
41    fn retrieves_schema_name() {
42        let html = html_from_bytes(SCHEMA_COMPLIANT_HTML).unwrap();
43        let value = find_schema_tag(&html, SchemaMetaTag::Name).unwrap();
44
45        assert_eq!(value, "Schema.org tags are awesome");
46    }
47
48    #[test]
49    fn retrieves_schema_description() {
50        let html = html_from_bytes(SCHEMA_COMPLIANT_HTML).unwrap();
51        let value = find_schema_tag(&html, SchemaMetaTag::Description).unwrap();
52
53        assert_eq!(
54            value,
55            "Tips to set Schema.org tags like you have been doing it since year 1"
56        );
57    }
58
59    #[test]
60    fn retrieves_schema_image() {
61        let html = html_from_bytes(SCHEMA_COMPLIANT_HTML).unwrap();
62        let value = find_schema_tag(&html, SchemaMetaTag::Image).unwrap();
63
64        assert_eq!(value, "https://www.example.com/image.jpg");
65    }
66}