|
| 1 | +use std::collections::HashMap; |
| 2 | + |
| 3 | +use scraper::{ElementRef, Html, Selector}; |
| 4 | +use serde::Deserialize; |
| 5 | +use url::Url; |
| 6 | + |
| 7 | +use super::Editor; |
| 8 | +use crate::logger; |
| 9 | + |
| 10 | +#[derive(Debug, Clone, Deserialize)] |
| 11 | +pub struct CssSort { |
| 12 | + #[serde(deserialize_with = "super::deserialize_selector")] |
| 13 | + pub selector: Selector, |
| 14 | + |
| 15 | + #[serde(default)] |
| 16 | + pub reverse: bool, |
| 17 | + |
| 18 | + #[serde(default, skip_serializing_if = "Vec::is_empty")] |
| 19 | + pub sort_by: Vec<Editor>, |
| 20 | +} |
| 21 | + |
| 22 | +impl CssSort { |
| 23 | + pub fn apply(&self, url: &Url, html: &str) -> String { |
| 24 | + let mut html = Html::parse_document(html); |
| 25 | + let selected = html.select(&self.selector).collect::<Vec<_>>(); |
| 26 | + |
| 27 | + let mut grouped_by_parent: HashMap<_, Vec<ElementRef>> = HashMap::new(); |
| 28 | + for element in selected { |
| 29 | + if let Some(key) = element.parent().map(|parent| parent.id()) { |
| 30 | + grouped_by_parent.entry(key).or_default().push(element); |
| 31 | + } |
| 32 | + } |
| 33 | + |
| 34 | + if grouped_by_parent.is_empty() { |
| 35 | + logger::warn(&format!("css_sort selector selected nothing to sort {url}")); |
| 36 | + } |
| 37 | + |
| 38 | + // Get the order of the elements as ids |
| 39 | + // This removes the reference to html allowing to take mut references later on |
| 40 | + let sorted = grouped_by_parent |
| 41 | + .into_iter() |
| 42 | + .map(|(parent, mut elements)| { |
| 43 | + elements.sort_by_cached_key(|element| self.get_sort_key_from_element(url, element)); |
| 44 | + if self.reverse { |
| 45 | + elements.reverse(); |
| 46 | + } |
| 47 | + let elements = elements |
| 48 | + .iter() |
| 49 | + .map(|element| element.id()) |
| 50 | + .collect::<Vec<_>>(); |
| 51 | + (parent, elements) |
| 52 | + }) |
| 53 | + .collect::<HashMap<_, _>>(); |
| 54 | + |
| 55 | + for (parent, sorted) in sorted { |
| 56 | + for id in &sorted { |
| 57 | + html.tree.get_mut(*id).unwrap().detach(); |
| 58 | + } |
| 59 | + |
| 60 | + // Insert them at the beginning of the parents children |
| 61 | + // This destroyes the order with the other elements in there but its way simpler to do for now |
| 62 | + let mut parent_mut = html.tree.get_mut(parent).unwrap(); |
| 63 | + for id in sorted.into_iter().rev() { |
| 64 | + parent_mut.prepend_id(id); |
| 65 | + } |
| 66 | + } |
| 67 | + |
| 68 | + html.html() |
| 69 | + } |
| 70 | + |
| 71 | + fn get_sort_key_from_element(&self, url: &Url, element: &ElementRef) -> String { |
| 72 | + let content = super::Content { |
| 73 | + extension: Some("html"), |
| 74 | + text: element.html(), |
| 75 | + }; |
| 76 | + Editor::apply_many(&self.sort_by, url, content).map_or_else( |
| 77 | + |error| { |
| 78 | + logger::error(&format!("css_sort sort_by failed {error}")); |
| 79 | + String::new() |
| 80 | + }, |
| 81 | + |content| content.text, |
| 82 | + ) |
| 83 | + } |
| 84 | +} |
| 85 | + |
| 86 | +#[cfg(test)] |
| 87 | +mod tests { |
| 88 | + use super::*; |
| 89 | + |
| 90 | + #[track_caller] |
| 91 | + fn case(css_sort: &CssSort, input: &str, expected: &str) { |
| 92 | + const PREFIX: &str = "<html><head></head><body>"; |
| 93 | + const SUFFIX: &str = "</body></html>"; |
| 94 | + |
| 95 | + let url = Url::parse("https://edjopato.de/").unwrap(); |
| 96 | + let html = css_sort.apply(&url, input); |
| 97 | + |
| 98 | + assert!(html.starts_with(PREFIX)); |
| 99 | + assert!(html.ends_with(SUFFIX)); |
| 100 | + let end_index = html.len() - SUFFIX.len(); |
| 101 | + let html = html.get(PREFIX.len()..end_index).unwrap(); |
| 102 | + |
| 103 | + assert_eq!(html, expected); |
| 104 | + } |
| 105 | + |
| 106 | + #[test] |
| 107 | + fn simple_example() { |
| 108 | + let input = "<p>A</p><p>C</p><p>B</p>"; |
| 109 | + let expected = "<p>A</p><p>B</p><p>C</p>"; |
| 110 | + let sort_by = CssSort { |
| 111 | + selector: Selector::parse("p").unwrap(), |
| 112 | + sort_by: Vec::new(), |
| 113 | + reverse: false, |
| 114 | + }; |
| 115 | + case(&sort_by, input, expected); |
| 116 | + } |
| 117 | + |
| 118 | + #[test] |
| 119 | + fn reverse() { |
| 120 | + let input = "<p>A</p><p>C</p><p>B</p>"; |
| 121 | + let expected = "<p>C</p><p>B</p><p>A</p>"; |
| 122 | + let sort_by = CssSort { |
| 123 | + selector: Selector::parse("p").unwrap(), |
| 124 | + sort_by: Vec::new(), |
| 125 | + reverse: true, |
| 126 | + }; |
| 127 | + case(&sort_by, input, expected); |
| 128 | + } |
| 129 | + |
| 130 | + #[test] |
| 131 | + fn sort_by() { |
| 132 | + let input = r#"<article><h3>A</h3><a id="Y">Bla</a></article><article><h3>B</h3><a id="X">Bla</a></article>"#; |
| 133 | + let expected = r#"<article><h3>B</h3><a id="X">Bla</a></article><article><h3>A</h3><a id="Y">Bla</a></article>"#; |
| 134 | + let sort_by = CssSort { |
| 135 | + selector: Selector::parse("article").unwrap(), |
| 136 | + sort_by: vec![Editor::CssSelect(Selector::parse("a").unwrap())], |
| 137 | + reverse: false, |
| 138 | + }; |
| 139 | + case(&sort_by, input, expected); |
| 140 | + } |
| 141 | + |
| 142 | + #[test] |
| 143 | + fn sort_by_same_key_keeps_order() { |
| 144 | + let input = r#"<article><h3>C</h3><a id="X">Bla</a></article><article><h3>A</h3><a id="X">Bla</a></article>"#; |
| 145 | + let expected = r#"<article><h3>C</h3><a id="X">Bla</a></article><article><h3>A</h3><a id="X">Bla</a></article>"#; |
| 146 | + let sort_by = CssSort { |
| 147 | + selector: Selector::parse("article").unwrap(), |
| 148 | + sort_by: vec![Editor::CssSelect(Selector::parse("a").unwrap())], |
| 149 | + reverse: false, |
| 150 | + }; |
| 151 | + case(&sort_by, input, expected); |
| 152 | + } |
| 153 | + |
| 154 | + #[test] |
| 155 | + fn sorting_toplevel_keeps_children_unsorted() { |
| 156 | + let input = "<div><p>D</p><p>A</p></div><div><p>C</p><p>B</p></div>"; |
| 157 | + let expected = "<div><p>C</p><p>B</p></div><div><p>D</p><p>A</p></div>"; |
| 158 | + let sort_by = CssSort { |
| 159 | + selector: Selector::parse("div").unwrap(), |
| 160 | + sort_by: Vec::new(), |
| 161 | + reverse: false, |
| 162 | + }; |
| 163 | + case(&sort_by, input, expected); |
| 164 | + } |
| 165 | + |
| 166 | + #[test] |
| 167 | + fn sorting_bottomlevel_keeps_parents_unsorted() { |
| 168 | + let input = "<div><p>D</p><p>A</p></div><div><p>C</p><p>B</p></div>"; |
| 169 | + let expected = "<div><p>A</p><p>D</p></div><div><p>B</p><p>C</p></div>"; |
| 170 | + let sort_by = CssSort { |
| 171 | + selector: Selector::parse("p").unwrap(), |
| 172 | + sort_by: Vec::new(), |
| 173 | + reverse: false, |
| 174 | + }; |
| 175 | + case(&sort_by, input, expected); |
| 176 | + } |
| 177 | + |
| 178 | + /// Documents current sorting order when other elements are there. |
| 179 | + /// Needs to be adapted when sorting order is improved. |
| 180 | + #[test] |
| 181 | + fn sort_with_other_elements() { |
| 182 | + let input = "<div>1</div><p>A</p><img><p>B</p>"; |
| 183 | + let expected = "<p>A</p><p>B</p><div>1</div><img>"; |
| 184 | + let sort_by = CssSort { |
| 185 | + selector: Selector::parse("p").unwrap(), |
| 186 | + sort_by: Vec::new(), |
| 187 | + reverse: false, |
| 188 | + }; |
| 189 | + case(&sort_by, input, expected); |
| 190 | + } |
| 191 | +} |
0 commit comments