oratos_html/
page.rs

1use serde::{Deserialize, Serialize};
2
3#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
4pub struct HtmlPage {
5    pub url_or_path: String,
6    pub source: String,
7    pub title: Option<String>,
8    pub meta_description: Option<String>,
9    pub canonical_urls: Vec<String>,
10    pub language: Option<String>,
11    pub headings: Vec<Heading>,
12    pub links: Vec<LinkInfo>,
13    pub images: Vec<ImageInfo>,
14    pub open_graph: OpenGraphTags,
15    pub twitter_card: TwitterCardTags,
16    pub json_ld_blocks: Vec<JsonLdBlock>,
17    pub main_text: String,
18    pub has_main_landmark: bool,
19    pub forms: Vec<FormInfo>,
20    pub breadcrumbs_detected: bool,
21}
22
23#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
24pub struct Heading {
25    pub level: u8,
26    pub text: String,
27}
28
29#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
30pub struct LinkInfo {
31    pub href: String,
32    pub text: String,
33    pub is_internal: bool,
34}
35
36#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
37pub struct ImageInfo {
38    pub src: String,
39    pub alt: Option<String>,
40    pub width: Option<String>,
41    pub height: Option<String>,
42    pub has_caption: bool,
43}
44
45#[derive(Debug, Clone, Default, PartialEq, Serialize, Deserialize)]
46pub struct OpenGraphTags {
47    pub title: Option<String>,
48    pub description: Option<String>,
49    pub image: Option<String>,
50    pub url: Option<String>,
51    pub type_: Option<String>,
52}
53
54#[derive(Debug, Clone, Default, PartialEq, Serialize, Deserialize)]
55pub struct TwitterCardTags {
56    pub card: Option<String>,
57    pub title: Option<String>,
58    pub description: Option<String>,
59    pub image: Option<String>,
60}
61
62#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
63pub struct JsonLdBlock {
64    pub raw: String,
65    pub valid_json: bool,
66    pub types: Vec<String>,
67}
68
69#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
70pub struct FormInfo {
71    pub id: Option<String>,
72    pub inputs_without_label: Vec<String>,
73}
74
75impl HtmlPage {
76    pub fn display_name(&self) -> &str {
77        self.title
78            .as_deref()
79            .filter(|t| !t.is_empty())
80            .unwrap_or(&self.url_or_path)
81    }
82}
oratos_html/page.rs

oratos_html/
page.rs
