0
0
mirror of https://github.com/neon-mmd/websurfx.git synced 2024-12-22 20:38:22 -05:00

feat: rewrite code by implementing common engine trait SearchEngine

This commit is contained in:
neon_arch 2023-07-11 19:42:17 +03:00
parent b72af01e0e
commit f9b9e87a0e
2 changed files with 205 additions and 213 deletions

View File

@ -9,7 +9,7 @@ use scraper::{Html, Selector};
use crate::search_results_handler::aggregation_models::RawSearchResult;
use super::engine_models::EngineError;
use super::engine_models::{EngineError, SearchEngine};
use error_stack::{IntoReport, Report, Result, ResultExt};
@ -30,126 +30,121 @@ use error_stack::{IntoReport, Report, Result, ResultExt};
/// reach the above `upstream search engine` page or if the `upstream search engine` is unable to
/// provide results for the requested search query and also returns error if the scraping selector
/// or HeaderMap fails to initialize.
pub async fn results(
query: &str,
page: u32,
user_agent: &str,
) -> Result<HashMap<String, RawSearchResult>, EngineError> {
// Page number can be missing or empty string and so appropriate handling is required
// so that upstream server recieves valid page number.
let url: String = match page {
1 => {
format!("https://html.duckduckgo.com/html/?q={query}&s=&dc=&v=1&o=json&api=/d.js")
}
_ => {
format!(
"https://duckduckgo.com/html/?q={}&s={}&dc={}&v=1&o=json&api=/d.js",
query,
(page / 2 + (page % 2)) * 30,
(page / 2 + (page % 2)) * 30 + 1
)
}
};
// initializing HeaderMap and adding appropriate headers.
let mut header_map = HeaderMap::new();
header_map.insert(
USER_AGENT,
user_agent
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
REFERER,
"https://google.com/"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
CONTENT_TYPE,
"application/x-www-form-urlencoded"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
COOKIE,
"kl=wt-wt"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
pub struct DuckDuckGo;
// fetch the html from upstream duckduckgo engine
let results: String = reqwest::Client::new()
.get(url)
.timeout(Duration::from_secs(5))
.headers(header_map) // add spoofed headers to emulate human behaviour
.send()
.await
.into_report()
.change_context(EngineError::RequestError)?
.text()
.await
.into_report()
.change_context(EngineError::RequestError)?;
let document: Html = Html::parse_document(&results);
let no_result: Selector = Selector::parse(".no-results")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".no-results"))?;
if document.select(&no_result).next().is_some() {
return Err(Report::new(EngineError::EmptyResultSet));
}
let results: Selector = Selector::parse(".result")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result"))?;
let result_title: Selector = Selector::parse(".result__a")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result__a"))?;
let result_url: Selector = Selector::parse(".result__url")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result__url"))?;
let result_desc: Selector = Selector::parse(".result__snippet")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result__snippet"))?;
// scrape all the results from the html
Ok(document
.select(&results)
.map(|result| {
RawSearchResult::new(
result
.select(&result_title)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
#[async_trait::async_trait]
impl SearchEngine for DuckDuckGo {
async fn results(
&self,
query: String,
page: u32,
user_agent: String,
) -> Result<HashMap<String, RawSearchResult>, EngineError> {
// Page number can be missing or empty string and so appropriate handling is required
// so that upstream server recieves valid page number.
let url: String = match page {
1 => {
format!("https://html.duckduckgo.com/html/?q={query}&s=&dc=&v=1&o=json&api=/d.js")
}
_ => {
format!(
"https://{}",
"https://duckduckgo.com/html/?q={}&s={}&dc={}&v=1&o=json&api=/d.js",
query,
(page / 2 + (page % 2)) * 30,
(page / 2 + (page % 2)) * 30 + 1
)
}
};
// initializing HeaderMap and adding appropriate headers.
let mut header_map = HeaderMap::new();
header_map.insert(
USER_AGENT,
user_agent
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
REFERER,
"https://google.com/"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
CONTENT_TYPE,
"application/x-www-form-urlencoded"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
COOKIE,
"kl=wt-wt"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
let document: Html = Html::parse_document(
&DuckDuckGo::fetch_html_from_upstream(&self, url, header_map).await?,
);
let no_result: Selector = Selector::parse(".no-results")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".no-results"))?;
if document.select(&no_result).next().is_some() {
return Err(Report::new(EngineError::EmptyResultSet));
}
let results: Selector = Selector::parse(".result")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result"))?;
let result_title: Selector = Selector::parse(".result__a")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result__a"))?;
let result_url: Selector = Selector::parse(".result__url")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result__url"))?;
let result_desc: Selector = Selector::parse(".result__snippet")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result__snippet"))?;
// scrape all the results from the html
Ok(document
.select(&results)
.map(|result| {
RawSearchResult::new(
result
.select(&result_url)
.select(&result_title)
.next()
.unwrap()
.inner_html()
.trim()
),
result
.select(&result_desc)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
vec!["duckduckgo".to_string()],
)
})
.map(|search_result| (search_result.visiting_url.clone(), search_result))
.collect())
.to_string(),
format!(
"https://{}",
result
.select(&result_url)
.next()
.unwrap()
.inner_html()
.trim()
),
result
.select(&result_desc)
.next()
.unwrap()
.inner_html()
.trim()
.to_string(),
vec!["duckduckgo".to_string()],
)
})
.map(|search_result| (search_result.visiting_url.clone(), search_result))
.collect())
}
}

View File

@ -8,7 +8,7 @@ use std::collections::HashMap;
use crate::search_results_handler::aggregation_models::RawSearchResult;
use super::engine_models::EngineError;
use super::engine_models::{EngineError, SearchEngine};
use error_stack::{IntoReport, Report, Result, ResultExt};
/// This function scrapes results from the upstream engine duckduckgo and puts all the scraped
@ -28,111 +28,108 @@ use error_stack::{IntoReport, Report, Result, ResultExt};
/// reach the above `upstream search engine` page or if the `upstream search engine` is unable to
/// provide results for the requested search query and also returns error if the scraping selector
/// or HeaderMap fails to initialize.
pub async fn results(
query: &str,
page: u32,
user_agent: &str,
) -> Result<HashMap<String, RawSearchResult>, EngineError> {
// Page number can be missing or empty string and so appropriate handling is required
// so that upstream server recieves valid page number.
let url: String = format!("https://searx.work/search?q={query}&pageno={page}");
// initializing headers and adding appropriate headers.
let mut header_map = HeaderMap::new();
header_map.insert(
USER_AGENT,
user_agent
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
REFERER,
"https://google.com/"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
CONTENT_TYPE,
"application/x-www-form-urlencoded"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(COOKIE, "categories=general; language=auto; locale=en; autocomplete=duckduckgo; image_proxy=1; method=POST; safesearch=2; theme=simple; results_on_new_tab=1; doi_resolver=oadoi.org; simple_style=auto; center_alignment=1; query_in_title=1; infinite_scroll=0; disabled_engines=; enabled_engines=\"archive is__general\\054yep__general\\054curlie__general\\054currency__general\\054ddg definitions__general\\054wikidata__general\\054duckduckgo__general\\054tineye__general\\054lingva__general\\054startpage__general\\054yahoo__general\\054wiby__general\\054marginalia__general\\054alexandria__general\\054wikibooks__general\\054wikiquote__general\\054wikisource__general\\054wikiversity__general\\054wikivoyage__general\\054dictzone__general\\054seznam__general\\054mojeek__general\\054naver__general\\054wikimini__general\\054brave__general\\054petalsearch__general\\054goo__general\"; disabled_plugins=; enabled_plugins=\"searx.plugins.hostname_replace\\054searx.plugins.oa_doi_rewrite\\054searx.plugins.vim_hotkeys\"; tokens=; maintab=on; enginetab=on".parse().into_report().change_context(EngineError::UnexpectedError)?);
pub struct Searx;
// fetch the html from upstream searx instance engine
let results: String = reqwest::Client::new()
.get(url)
.headers(header_map) // add spoofed headers to emulate human behaviours.
.send()
.await
.into_report()
.change_context(EngineError::RequestError)?
.text()
.await
.into_report()
.change_context(EngineError::RequestError)?;
#[async_trait::async_trait]
impl SearchEngine for Searx {
async fn results(
&self,
query: String,
page: u32,
user_agent: String,
) -> Result<HashMap<String, RawSearchResult>, EngineError> {
// Page number can be missing or empty string and so appropriate handling is required
// so that upstream server recieves valid page number.
let url: String = format!("https://searx.work/search?q={query}&pageno={page}");
let document: Html = Html::parse_document(&results);
// initializing headers and adding appropriate headers.
let mut header_map = HeaderMap::new();
header_map.insert(
USER_AGENT,
user_agent
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
REFERER,
"https://google.com/"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(
CONTENT_TYPE,
"application/x-www-form-urlencoded"
.parse()
.into_report()
.change_context(EngineError::UnexpectedError)?,
);
header_map.insert(COOKIE, "categories=general; language=auto; locale=en; autocomplete=duckduckgo; image_proxy=1; method=POST; safesearch=2; theme=simple; results_on_new_tab=1; doi_resolver=oadoi.org; simple_style=auto; center_alignment=1; query_in_title=1; infinite_scroll=0; disabled_engines=; enabled_engines=\"archive is__general\\054yep__general\\054curlie__general\\054currency__general\\054ddg definitions__general\\054wikidata__general\\054duckduckgo__general\\054tineye__general\\054lingva__general\\054startpage__general\\054yahoo__general\\054wiby__general\\054marginalia__general\\054alexandria__general\\054wikibooks__general\\054wikiquote__general\\054wikisource__general\\054wikiversity__general\\054wikivoyage__general\\054dictzone__general\\054seznam__general\\054mojeek__general\\054naver__general\\054wikimini__general\\054brave__general\\054petalsearch__general\\054goo__general\"; disabled_plugins=; enabled_plugins=\"searx.plugins.hostname_replace\\054searx.plugins.oa_doi_rewrite\\054searx.plugins.vim_hotkeys\"; tokens=; maintab=on; enginetab=on".parse().into_report().change_context(EngineError::UnexpectedError)?);
let no_result: Selector = Selector::parse("#urls>.dialog-error>p")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", "#urls>.dialog-error>p"))?;
let document: Html =
Html::parse_document(&Searx::fetch_html_from_upstream(&self, url, header_map).await?);
if let Some(no_result_msg) = document.select(&no_result).nth(1) {
if no_result_msg.inner_html()
let no_result: Selector = Selector::parse("#urls>.dialog-error>p")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| {
format!("invalid CSS selector: {}", "#urls>.dialog-error>p")
})?;
if let Some(no_result_msg) = document.select(&no_result).nth(1) {
if no_result_msg.inner_html()
== "we didn't find any results. Please use another query or search in more categories"
{
return Err(Report::new(EngineError::EmptyResultSet));
}
}
let results: Selector = Selector::parse(".result")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result"))?;
let result_title: Selector = Selector::parse("h3>a")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", "h3>a"))?;
let result_url: Selector = Selector::parse("h3>a")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", "h3>a"))?;
let result_desc: Selector = Selector::parse(".content")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".content"))?;
// scrape all the results from the html
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())
}
let results: Selector = Selector::parse(".result")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".result"))?;
let result_title: Selector = Selector::parse("h3>a")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", "h3>a"))?;
let result_url: Selector = Selector::parse("h3>a")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", "h3>a"))?;
let result_desc: Selector = Selector::parse(".content")
.map_err(|_| Report::new(EngineError::UnexpectedError))
.attach_printable_lazy(|| format!("invalid CSS selector: {}", ".content"))?;
// scrape all the results from the html
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())
}