✨ `Bing` for the search engine (#473) (#473)

fix: add closing curly brace

- accidentally removed from previous PR. i hope i did not ruin some
build

* feat: implement mojeek engine

* Merge branch 'rolling' into FEAT_316_bing_search_engine

* ✨ feat: implement bing engine

* Fix: include cookie string to header_map

* Fix: include <strong> tags from upstream search

* Merge branch 'rolling' into FEAT_316_bing_search_engine

Co-authored-by: neon-mmd <[email protected]>

src/engines/bing.rs

@@ -0,0 +1,124 @@
+//! The `bing` module handles the scraping of results from the bing search engine
+//! by querying the upstream bing search engine with user provided query and with a page
+//! number if provided.
+
+use std::collections::HashMap;
+
+use regex::Regex;
+use reqwest::header::HeaderMap;
+use reqwest::Client;
+use scraper::Html;
+
+use crate::models::aggregation_models::SearchResult;
+
+use crate::models::engine_models::{EngineError, SearchEngine};
+
+use error_stack::{Report, Result, ResultExt};
+
+use super::search_result_parser::SearchResultParser;
+
+/// A new Bing engine type defined in-order to implement the `SearchEngine` trait which allows to
+/// reduce code duplication as well as allows to create vector of different search engines easily.
+pub struct Bing {
+    /// The parser, used to interpret the search result.
+    parser: SearchResultParser,
+}
+
+impl Bing {
+    /// Creates the Bing parser.
+    pub fn new() -> Result<Self, EngineError> {
+        Ok(Self {
+            parser: SearchResultParser::new(
+                ".b_results",
+                ".b_algo",
+                "h2 a",
+                ".tpcn a.tilk",
+                ".b_caption p",
+            )?,
+        })
+    }
+}
+
+#[async_trait::async_trait]
+impl SearchEngine for Bing {
+    async fn results(
+        &self,
+        query: &str,
+        page: u32,
+        user_agent: &str,
+        client: &Client,
+        _safe_search: u8,
+    ) -> Result<HashMap<String, SearchResult>, EngineError> {
+        // Bing uses `start results from this number` convention
+        // So, for 10 results per page, page 0 starts at 1, page 1
+        // starts at 11, and so on.
+        let results_per_page = 10;
+        let start_result = results_per_page * page + 1;
+
+        let url: String = match page {
+            0 => {
+                format!("https://www.bing.com/search?q={query}")
+            }
+            _ => {
+                format!("https://www.bing.com/search?q={query}&first={start_result}")
+            }
+        };
+
+        let query_params: Vec<(&str, &str)> = vec![
+            ("_EDGE_V", "1"),
+            ("SRCHD=AF", "NOFORM"),
+            ("_Rwho=u", "d"),
+            ("bngps=s", "0"),
+            ("_UR=QS=0&TQS", "0"),
+            ("_UR=QS=0&TQS", "0"),
+        ];
+
+        let mut cookie_string = String::new();
+        for (k, v) in &query_params {
+            cookie_string.push_str(&format!("{k}={v}; "));
+        }
+
+        let header_map = HeaderMap::try_from(&HashMap::from([
+            ("USER_AGENT".to_string(), user_agent.to_string()),
+            ("REFERER".to_string(), "https://google.com/".to_string()),
+            (
+                "CONTENT_TYPE".to_string(),
+                "application/x-www-form-urlencoded".to_string(),
+            ),
+            ("COOKIE".to_string(), cookie_string),
+        ]))
+        .change_context(EngineError::UnexpectedError)?;
+
+        let document: Html = Html::parse_document(
+            &Bing::fetch_html_from_upstream(self, &url, header_map, client).await?,
+        );
+
+        // Bing is very aggressive in finding matches
+        // even with the most absurd of queries. ".b_algo" is the
+        // class for the list item of results
+        if let Some(no_result_msg) = self.parser.parse_for_no_results(&document).nth(0) {
+            if no_result_msg
+                .value()
+                .attr("class")
+                .map(|classes| classes.contains("b_algo"))
+                .unwrap_or(false)
+            {
+                return Err(Report::new(EngineError::EmptyResultSet));
+            }
+        }
+
+        let re_span = Regex::new(r#"<span.*?>.*?(?:</span>&nbsp;·|</span>)"#).unwrap();
+        let re_strong = Regex::new(r#"(<strong>|</strong>)"#).unwrap();
+
+        // scrape all the results from the html
+        self.parser
+            .parse_for_results(&document, |title, url, desc| {
+                Some(SearchResult::new(
+                    &re_strong.replace_all(title.inner_html().trim(), ""),
+                    url.value().attr("href").unwrap(),
+                    &re_span.replace_all(desc.inner_html().trim(), ""),
+                    &["bing"],
+                ))
+            })
+    }
+}

src/engines/mod.rs

@@ -3,6 +3,7 @@
 //! provide a standard functions to be implemented for all the upstream search engine handling
 //! code. Moreover, it also provides a custom error for the upstream search engine handling code.
 
+pub mod bing;
 pub mod brave;
 pub mod duckduckgo;
 pub mod librex;

src/models/engine_models.rs

@@ -166,6 +166,10 @@ impl EngineHandler {
                     let engine = crate::engines::mojeek::Mojeek::new()?;
                     ("mojeek", Box::new(engine))
                 }
+                "bing" => {
+                    let engine = crate::engines::bing::Bing::new()?;
+                    ("bing", Box::new(engine))
+                }
                 _ => {
                     return Err(Report::from(EngineError::NoSuchEngineFound(
                         engine_name.to_string(),

websurfx/config.lua

@@ -65,4 +65,5 @@ upstream_search_engines = {
     Startpage = false,
     LibreX = false,
     Mojeek = false,
+    Bing = false,
 } -- select the upstream search engines from which the results should be fetched.