Refactor search results and add some actual search functionality

This commit is contained in:
Jonas Zeunert
2024-04-23 17:01:18 +02:00
parent 4c7d6143ae
commit 94d4425661
6 changed files with 109 additions and 26 deletions

4
Cargo.lock generated
View File

@@ -126,6 +126,7 @@ checksum = "f1fdabc7756949593fe60f30ec81974b613357de856987752631dea1e3394c80"
name = "awesm" name = "awesm"
version = "0.1.0" version = "0.1.0"
dependencies = [ dependencies = [
"chrono",
"clap", "clap",
"env_logger", "env_logger",
"lazy_static", "lazy_static",
@@ -239,7 +240,10 @@ checksum = "a21f936df1771bf62b77f047b726c4625ff2e8aa607c01ec06e5a05bd8463401"
dependencies = [ dependencies = [
"android-tzdata", "android-tzdata",
"iana-time-zone", "iana-time-zone",
"js-sys",
"num-traits", "num-traits",
"serde",
"wasm-bindgen",
"windows-targets 0.52.5", "windows-targets 0.52.5",
] ]

View File

@@ -6,6 +6,7 @@ edition = "2021"
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
[dependencies] [dependencies]
chrono = { version = "0.4.38", features = ["serde"] }
clap = { version = "4.5.4", features = ["derive"] } clap = { version = "4.5.4", features = ["derive"] }
env_logger = "0.11.3" env_logger = "0.11.3"
lazy_static = "1.4.0" lazy_static = "1.4.0"

View File

@@ -24,10 +24,12 @@ pub struct Source {
#[derive(Debug, Deserialize)] #[derive(Debug, Deserialize)]
pub struct File { pub struct File {
pub filename: String, pub filename: String,
pub last_modified: String,
pub filesize: usize,
} }
#[test] #[test]
fn simple() { fn simple() {
let data = "{\"hits\":{\"hits\":[{\"_source\":{\"file\":{\"filename\":\"4girls.md\"}}},{\"_source\":{\"file\":{\"filename\":\"diversity.md\"}}},{\"_source\":{\"file\":{\"filename\":\"gametalks.md\"}}},{\"_source\":{\"file\":{\"filename\":\"wagtail.md\"}}},{\"_source\":{\"file\":{\"filename\":\"pythonineducation.md\"}}},{\"_source\":{\"file\":{\"filename\":\"hyper.md\"}}},{\"_source\":{\"file\":{\"filename\":\"fantasy.md\"}}}]}}"; let data = "{\"took\":20,\"timed_out\":false,\"_shards\":{\"total\":4,\"successful\":4,\"skipped\":0,\"failed\":0},\"hits\":{\"total\":{\"value\":402,\"relation\":\"eq\"},\"max_score\":7.386201,\"hits\":[{\"_index\":\"awesm\",\"_id\":\"english.md\",\"_score\":7.386201,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:32.951+00:00\",\"filesize\":16423,\"filename\":\"english.md\"}}},{\"_index\":\"awesm\",\"_id\":\"persian.md\",\"_score\":7.2951937,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:33.171+00:00\",\"filesize\":8665,\"filename\":\"persian.md\"}}},{\"_index\":\"awesm\",\"_id\":\"nologinwebapps.md\",\"_score\":6.871194,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:33.143+00:00\",\"filesize\":30453,\"filename\":\"nologinwebapps.md\"}}},{\"_index\":\"awesm\",\"_id\":\"research.md\",\"_score\":6.614027,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:33.231+00:00\",\"filesize\":41865,\"filename\":\"research.md\"}}},{\"_index\":\"awesm\",\"_id\":\"AutoHotkey.md\",\"_score\":6.309735,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:32.807+00:00\",\"filesize\":40670,\"filename\":\"AutoHotkey.md\"}}},{\"_index\":\"awesm\",\"_id\":\"devfun.md\",\"_score\":6.2060246,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:32.927+00:00\",\"filesize\":13573,\"filename\":\"devfun.md\"}}},{\"_index\":\"awesm\",\"_id\":\"arabic.md\",\"_score\":5.7819433,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:32.795+00:00\",\"filesize\":8660,\"filename\":\"arabic.md\"}}},{\"_index\":\"awesm\",\"_id\":\"i18n.md\",\"_score\":5.737568,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:33.027+00:00\",\"filesize\":16880,\"filename\":\"i18n.md\"}}},{\"_index\":\"awesm\",\"_id\":\"ethics.md\",\"_score\":5.4912834,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:32.959+00:00\",\"filesize\":71797,\"filename\":\"ethics.md\"}}},{\"_index\":\"awesm\",\"_id\":\"fantasy.md\",\"_score\":5.2991915,\"_source\":{\"file\":{\"last_modified\":\"2024-04-22T20:43:32.959+00:00\",\"filesize\":142570,\"filename\":\"fantasy.md\"}}}]}}";
let _: ElasticResponse = serde_json::from_str(data).unwrap(); let _: ElasticResponse = serde_json::from_str(data).unwrap();
} }

View File

@@ -1,11 +1,13 @@
use clap::Parser; use clap::Parser;
use log::info; use log::info;
use std::net::{IpAddr, TcpListener}; use std::net::{IpAddr, TcpListener};
mod options; mod options;
mod server; mod server;
mod elastic_query; mod elastic_query;
mod elastic_response; mod elastic_response;
mod search_result;
#[derive(Parser, Debug)] #[derive(Parser, Debug)]
struct Args { struct Args {

View File

@@ -4,7 +4,11 @@ pub mod elastic_search {
lazy_static! { lazy_static! {
pub static ref URL: Url = pub static ref URL: Url =
Url::parse("http://127.0.0.1:9200/_search?filter_path=hits.hits._source").unwrap(); Url::parse("http://127.0.0.1:9200/_search?filter_path=hits.hits._source").unwrap();
pub static ref FIELDS: Vec<String> = vec!["file.filename".to_string()]; pub static ref FIELDS: Vec<String> = vec![
"file.filename".to_string(),
"file.filesize".to_string(),
"file.last_modified".to_string()
];
} }
} }

View File

@@ -1,3 +1,4 @@
use log::warn;
use log::{error, info}; use log::{error, info};
use rouille::router; use rouille::router;
use rouille::Request; use rouille::Request;
@@ -6,45 +7,84 @@ use std::error::Error;
use crate::elastic_query::ElasticQuery; use crate::elastic_query::ElasticQuery;
use crate::options; use crate::options;
use crate::search_result::SearchResult;
pub fn serve_request(req: &Request) -> Response { pub fn serve_request(req: &Request) -> Response {
router!(req, router!(req,
(GET) (/search) => { query(req) }, (GET) (/redirect) => { redirect(req) },
(GET) (/search) => { search(req) },
_ => Response::empty_404() _ => Response::empty_404()
) )
} }
fn query(req: &Request) -> Response { fn get_query(req: &Request) -> Result<String, Response> {
info!("Handling request: {:?}", req);
let Some(query) = req.get_param("q") else { let Some(query) = req.get_param("q") else {
return Response::text("Query Parameter q must be specified!").with_status_code(400); return Err(Response::text("Query Parameter q must be specified!").with_status_code(400));
}; };
match query.len() { match query.len() {
0..=options::query::MIN_LENGTH => { 0..=options::query::MIN_LENGTH => {
return Response::text(format!( return Err(Response::text(format!(
"Min query length: {:?}", "Min query length: {:?}",
options::query::MIN_LENGTH options::query::MIN_LENGTH
)) ))
.with_status_code(400); .with_status_code(400));
} }
options::query::MAX_LENGTH..=usize::MAX => { options::query::MAX_LENGTH..=usize::MAX => {
return Response::text(format!( return Err(Response::text(format!(
"Max query length: {:?}", "Max query length: {:?}",
options::query::MAX_LENGTH options::query::MAX_LENGTH
)) ))
.with_status_code(400); .with_status_code(400));
} }
_ => (), _ => Ok(query),
} }
}
fn search(req: &Request) -> Response {
info!("Searching request: {:?}", req);
let query = match get_query(req) {
Ok(v) => v,
Err(e) => {
return e;
}
};
let response = query_elastic_search(query); let response = match query_elastic_search(query) {
if let Err(e) = response { Ok(v) => v,
error!("Error searching: {}", e); Err(e) => {
return Response::text("No fitting awesome list found :(").with_status_code(404); error!("Error searching: {}", e);
} return Response::text("Could not find any awesome pages :(").with_status_code(404);
let response = response.unwrap(); }
};
let Some(name) = response.split('.').next() else { Response::json(&response).with_status_code(200)
}
fn redirect(req: &Request) -> Response {
info!("Redirecting request: {:?}", req);
let query = match get_query(req) {
Ok(v) => v,
Err(e) => {
warn!("Error with request: {:?}", e);
return e;
}
};
let response = match query_elastic_search(query) {
Ok(v) => v,
Err(e) => {
error!("Error searching: {}", e);
return Response::text("Could not find any awesome page to redirect you to :(")
.with_status_code(404);
}
};
let Some(first) = response.get(0) else {
error!("No search results");
return Response::text("Could not find any awesome page to redirect you to :()")
.with_status_code(404);
};
let Some(name) = first.name.split('.').next() else {
error!("Error getting name from response!"); error!("Error getting name from response!");
return Response::text("No fitting awesome list found :(").with_status_code(404); return Response::text("No fitting awesome list found :(").with_status_code(404);
}; };
@@ -53,22 +93,36 @@ fn query(req: &Request) -> Response {
Response::redirect_303(format!("/{}", name)) Response::redirect_303(format!("/{}", name))
} }
fn query_elastic_search(query: String) -> Result<String, Box<dyn Error>> { fn query_elastic_search(query: String) -> Result<Vec<SearchResult>, Box<dyn Error>> {
let query = ElasticQuery::new(options::elastic_search::FIELDS.clone(), query); let query = ElasticQuery::new(options::elastic_search::FIELDS.clone(), query);
let res = query.send(options::elastic_search::URL.clone())?; let query_res = query.send(options::elastic_search::URL.clone())?;
Ok(res.hits.hits[0].source.file.filename.clone()) let search_res = query_res
.hits
.hits
.iter()
.map(|hit| SearchResult::from(&hit.source.file))
.collect();
Ok(search_res)
} }
#[cfg(test)] #[cfg(test)]
mod test { mod test {
const REDIRECT_URL: &str = "/redirect";
const SEARCH_URL: &str = "/search"; const SEARCH_URL: &str = "/search";
use super::*; use super::*;
mod serve_request { mod serve_request {
use super::*; use super::*;
#[test] #[test]
fn test_query_route_exists() { fn test_redirect_route_exists() {
let req = Request::fake_http("GET", REDIRECT_URL, vec![], vec![]);
let res = serve_request(&req);
assert_ne!(res.status_code, 404)
}
#[test]
fn test_search_route_exists() {
let req = Request::fake_http("GET", SEARCH_URL, vec![], vec![]); let req = Request::fake_http("GET", SEARCH_URL, vec![], vec![]);
let res = serve_request(&req); let res = serve_request(&req);
assert_ne!(res.status_code, 404) assert_ne!(res.status_code, 404)
@@ -80,12 +134,24 @@ mod test {
assert_eq!(res.status_code, 404) assert_eq!(res.status_code, 404)
} }
} }
mod query { mod get_query {
use super::*; use super::*;
#[test] #[test]
fn simple() {
let q = "simple";
let req = Request::fake_http("GET", format!("/{}?q={}", SEARCH_URL, q), vec![], vec![]);
let Ok(res) = get_query(&req) else {
panic!("No Value")
};
assert_eq!(res, q);
}
#[test]
fn no_query_parameter() { fn no_query_parameter() {
let req = Request::fake_http("GET", SEARCH_URL, vec![], vec![]); let req = Request::fake_http("GET", SEARCH_URL, vec![], vec![]);
let res = query(&req); let Err(res) = get_query(&req) else {
panic!("No Error")
};
assert_eq!(res.status_code, 400) assert_eq!(res.status_code, 400)
} }
#[test] #[test]
@@ -94,7 +160,9 @@ mod test {
.take(options::query::MAX_LENGTH + 1) .take(options::query::MAX_LENGTH + 1)
.collect(); .collect();
let req = Request::fake_http("GET", format!("/{}?q={}", SEARCH_URL, q), vec![], vec![]); let req = Request::fake_http("GET", format!("/{}?q={}", SEARCH_URL, q), vec![], vec![]);
let res = query(&req); let Err(res) = get_query(&req) else {
panic!("No Error")
};
let mut body = String::new(); let mut body = String::new();
let _ = res.data.into_reader_and_size().0.read_to_string(&mut body); let _ = res.data.into_reader_and_size().0.read_to_string(&mut body);
@@ -104,7 +172,9 @@ mod test {
#[test] #[test]
fn query_too_short() { fn query_too_short() {
let req = Request::fake_http("GET", format!("/{}?q=", SEARCH_URL), vec![], vec![]); let req = Request::fake_http("GET", format!("/{}?q=", SEARCH_URL), vec![], vec![]);
let res = query(&req); let Err(res) = get_query(&req) else {
panic!("No Error")
};
assert_eq!(res.status_code, 400); assert_eq!(res.status_code, 400);