alamin/websurfx
alamin/websurfx
0
0
Fork 0
mirror of https://github.com/neon-mmd/websurfx.git synced 2024-12-22 12:28:21 -05:00
Code Issues Releases Activity

improve error handling by using error-stack crate

Browse Source
This commit is contained in:
xffxff 2023-06-14 20:42:30 +08:00
parent e8935bc52d
commit 9a4cf949e4
5 changed files with 133 additions and 90 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
Cargo.lock generated
View File

@ -268,6 +268,12 @@ dependencies = [
"alloc-no-stdlib",
]
[[package]]
name = "anyhow"
version = "1.0.71"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "9c7d0618f0e0b7e8ff11427422b64564d5fb0be1940354bfe2e0529b18a9d9b8"
[[package]]
name = "askama_escape"
version = "0.10.3"
@ -733,6 +739,16 @@ dependencies = [
"libc",
]
[[package]]
name = "error-stack"
version = "0.3.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "5f00447f331c7f726db5b8532ebc9163519eed03c6d7c8b73c90b3ff5646ac85"
dependencies = [
"anyhow",
"rustc_version 0.4.0",
]
[[package]]
name = "failure"
version = "0.1.8"
@ -3373,6 +3389,7 @@ dependencies = [
"actix-files",
"actix-web",
"env_logger",
"error-stack",
"fake-useragent",
"handlebars",
"log",

1
Cargo.toml
View File

@ -23,6 +23,7 @@ redis = {version="*"}
md5 = {version="*"}
rand={version="*"}
once_cell = {version="*"}
error-stack = "0.3.1"
[dev-dependencies]
rusty-hook = "^0.11.2"

70
src/engines/duckduckgo.rs
View File

@ -9,7 +9,9 @@ use scraper::{Html, Selector};
use crate::search_results_handler::aggregation_models::RawSearchResult;
use super::engine_models::EngineErrorKind;
use super::engine_models::EngineError;
use error_stack::{IntoReport, Report, Result, ResultExt};
/// This function scrapes results from the upstream engine duckduckgo and puts all the scraped
/// results like title, visiting_url (href in html),engine (from which engine it was fetched from)
@ -32,7 +34,7 @@ pub async fn results(
query: &str,
page: u32,
user_agent: &str,
) -> Result<HashMap<String, RawSearchResult>, EngineErrorKind> {
) -> 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 {
@ -51,33 +53,71 @@ pub async fn results(
// initializing HeaderMap and adding appropriate headers.
let mut header_map = HeaderMap::new();
header_map.insert(USER_AGENT, user_agent.parse()?);
header_map.insert(REFERER, "https://google.com/".parse()?);
header_map.insert(CONTENT_TYPE, "application/x-www-form-urlencoded".parse()?);
header_map.insert(COOKIE, "kl=wt-wt".parse()?);
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)?,
);
// fetch the html from upstream duckduckgo engine
let results: String = reqwest::Client::new()
.get(url)
.timeout(Duration::from_secs(30))
.timeout(Duration::from_secs(5))
.headers(header_map) // add spoofed headers to emulate human behaviour
.send()
.await?
.await
.into_report()
.change_context(EngineError::RequestError)?
.text()
.await?;
.await
.into_report()
.change_context(EngineError::RequestError)?;
let document: Html = Html::parse_document(&results);
let no_result: Selector = Selector::parse(".no-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(EngineErrorKind::EmptyResultSet);
return Err(Report::new(EngineError::EmptyResultSet));
}
let results: Selector = Selector::parse(".result")?;
let result_title: Selector = Selector::parse(".result__a")?;
let result_url: Selector = Selector::parse(".result__url")?;
let result_desc: Selector = Selector::parse(".result__snippet")?;
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

73
src/engines/engine_models.rs
View File

@ -1,8 +1,8 @@
//! This module provides the error enum to handle different errors associated while requesting data from
//! the upstream search engines with the search query provided by the user.
use reqwest::header::InvalidHeaderValue;
use scraper::error::SelectorErrorKind;
use error_stack::Context;
use std::fmt;
/// A custom error type used for handle engine associated errors.
///
@ -15,73 +15,24 @@ use scraper::error::SelectorErrorKind;
/// and are errors mostly related to failure in initialization of HeaderMap, Selector errors and
/// all other errors occuring within the code handling the `upstream search engines`.
#[derive(Debug)]
pub enum EngineErrorKind {
RequestError(reqwest::Error),
pub enum EngineError {
EmptyResultSet,
UnexpectedError {
message: String,
source: Option<Box<dyn std::error::Error>>,
},
RequestError,
UnexpectedError,
}
/// Implementing `Display` trait to make errors writable on the stdout and also providing/passing the
/// appropriate errors that should be written to the stdout when this error is raised/encountered.
impl std::fmt::Display for EngineErrorKind {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
impl fmt::Display for EngineError {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
match self {
EngineErrorKind::RequestError(request_error) => {
write!(f, "Request error: {}", request_error)
}
EngineErrorKind::EmptyResultSet => {
EngineError::EmptyResultSet => {
write!(f, "The upstream search engine returned an empty result set")
}
EngineErrorKind::UnexpectedError { message, source } => {
write!(f, "Unexpected error: {}", message)?;
if let Some(source) = source {
write!(f, "\nCaused by: {}", source)?;
}
Ok(())
EngineError::RequestError => {
write!(f, "Request error")
}
EngineError::UnexpectedError => write!(f, "Unexpected error"),
}
}
}
/// Implementing `Error` trait to make the the `EngineErrorKind` enum an error type and
/// mapping `ReqwestErrors` to `RequestError` and `UnexpectedError` errors to all other unexpected
/// errors ocurring within the code handling the upstream search engines.
impl std::error::Error for EngineErrorKind {
fn source(&self) -> Option<&(dyn std::error::Error + 'static)> {
match self {
EngineErrorKind::RequestError(request_error) => Some(request_error),
EngineErrorKind::UnexpectedError { source, .. } => source.as_deref().map(|s| s),
_ => None,
}
}
}
/// Implementing `From` trait to map the `SelectorErrorKind` to `UnexpectedError` variant.
impl From<SelectorErrorKind<'_>> for EngineErrorKind {
fn from(err: SelectorErrorKind<'_>) -> Self {
Self::UnexpectedError {
message: err.to_string(),
source: None,
}
}
}
/// Implementing `From` trait to map the `InvalidHeaderValue` to `UnexpectedError` variant.
impl From<InvalidHeaderValue> for EngineErrorKind {
fn from(err: InvalidHeaderValue) -> Self {
Self::UnexpectedError {
message: err.to_string(),
source: Some(Box::new(err)),
}
}
}
/// Implementing `From` trait to map all `reqwest::Error` to `UnexpectedError` variant.
impl From<reqwest::Error> for EngineErrorKind {
fn from(err: reqwest::Error) -> Self {
Self::RequestError(err)
}
}
impl Context for EngineError {}

62
src/engines/searx.rs
View File

@ -8,7 +8,8 @@ use std::collections::HashMap;
use crate::search_results_handler::aggregation_models::RawSearchResult;
use super::engine_models::EngineErrorKind;
use super::engine_models::EngineError;
use error_stack::{IntoReport, Report, Result, ResultExt};
/// This function scrapes results from the upstream engine duckduckgo and puts all the scraped
/// results like title, visiting_url (href in html),engine (from which engine it was fetched from)
@ -31,43 +32,76 @@ pub async fn results(
query: &str,
page: u32,
user_agent: &str,
) -> Result<HashMap<String, RawSearchResult>, EngineErrorKind> {
) -> 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()?);
header_map.insert(REFERER, "https://google.com/".parse()?);
header_map.insert(CONTENT_TYPE, "application/x-www-form-urlencoded".parse()?);
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()?);
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)?);
// 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?
.await
.into_report()
.change_context(EngineError::RequestError)?
.text()
.await?;
.await
.into_report()
.change_context(EngineError::RequestError)?;
let document: Html = Html::parse_document(&results);
let no_result: Selector = Selector::parse("#urls>.dialog-error>p")?;
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(EngineErrorKind::EmptyResultSet);
return Err(Report::new(EngineError::EmptyResultSet));
}
}
let results: Selector = Selector::parse(".result")?;
let result_title: Selector = Selector::parse("h3>a")?;
let result_url: Selector = Selector::parse("h3>a")?;
let result_desc: Selector = Selector::parse(".content")?;
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