morrigan/crabbysearch
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Refactoring code and separating code into files for better maintainability

Browse source
This commit is contained in:
neon_arch 2023-04-25 16:30:04 +03:00
parent bb61ee3afe
commit f94ac503e2
7 changed files with 159 additions and 123 deletions
Download patch file Download diff file Expand all files Collapse all files

56
src/engines/duckduckgo.rs
View file

@ -48,49 +48,37 @@ pub async fn results(
let result_url: Selector = Selector::parse(".result__url")?;
let result_desc: Selector = Selector::parse(".result__snippet")?;
let mut search_results: HashMap<String, RawSearchResult> = HashMap::new();
// scrape all the results from the html
for result in document.select(&results) {
let search_result: RawSearchResult = RawSearchResult {
title: result
.select(&result_title)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
visiting_url: format!(
"https://{}",
Ok(document
.select(&results)
.map(|result| {
RawSearchResult::new(
result
.select(&result_url)
.select(&result_title)
.next()
.unwrap()
.inner_html()
.trim()
),
description: result
.select(&result_desc)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
engine: vec!["duckduckgo".to_string()],
};
search_results.insert(
format!(
"https://{}",
.to_string(),
format!(
"https://{}",
result
.select(&result_url)
.next()
.unwrap()
.inner_html()
.trim()
),
result
.select(&result_url)
.select(&result_desc)
.next()
.unwrap()
.inner_html()
.trim()
),
search_result,
);
}
Ok(search_results)
.to_string(),
vec!["duckduckgo".to_string()],
)
})
.map(|search_result| (search_result.visiting_url.clone(), search_result))
.collect())
}

73
src/engines/searx.rs
View file

@ -43,47 +43,36 @@ pub async fn results(
let result_url: Selector = Selector::parse("h3>a")?;
let result_desc: Selector = Selector::parse(".content")?;
let mut search_results: HashMap<String, RawSearchResult> = HashMap::new();
// scrape all the results from the html
for result in document.select(&results) {
let search_result: RawSearchResult = RawSearchResult {
title: result
.select(&result_title)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
visiting_url: result
.select(&result_url)
.next()
.unwrap()
.value()
.attr("href")
.unwrap()
.to_string(),
description: result
.select(&result_desc)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
engine: vec!["searx".to_string()],
};
search_results.insert(
result
.select(&result_url)
.next()
.unwrap()
.value()
.attr("href")
.unwrap()
.to_string(),
search_result,
);
}
Ok(search_results)
Ok(document
.select(&results)
.map(|result| {
RawSearchResult::new(
result
.select(&result_title)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
result
.select(&result_url)
.next()
.unwrap()
.value()
.attr("href")
.unwrap()
.to_string(),
result
.select(&result_desc)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
vec!["searx".to_string()],
)
})
.map(|search_result| (search_result.visiting_url.clone(), search_result))
.collect())
}