refactoring, user agent customization, title limit, body streaming & limit, improved efficiency

Cargo.lock

@@ -142,9 +142,9 @@ checksum = "79296716171880943b8470b5f8d03aa55eb2e645a4874bdbb28adb49162e012c"
 
 [[package]]
 name = "bytes"
-version = "1.6.0"
+version = "1.7.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "514de17de45fdb8dc022b1a7975556c53c86f9f0aa5f534b98977b171857c2c9"
+checksum = "8318a53db07bb3f8dca91a600466bdb3f2eaadeedfdbcf02e1accbad9271ba50"
 
 [[package]]
 name = "cc"
@@ -1232,6 +1232,7 @@ name = "speedboat"
 version = "0.1.0"
 dependencies = [
  "base64",
+ "bytes",
  "chrono",
  "clap",
  "colored",

Cargo.toml

@@ -7,6 +7,7 @@ edition = "2021"
 
 [dependencies]
 base64 = "0.22.1"
+bytes = "1.7.1"
 chrono = "0.4.38"
 clap = { version = "4.5.4", features = ["derive"] }
 colored = "2.1.0"

src/common/conf.rs

@@ -1,10 +1,15 @@
 use super::console::fatal;
 use clap::Parser;
 
-pub const VERSION: &str = "1.0.0";
+pub const VERSION: &str = "1.2";
+
 pub struct Params {
     pub statcodes: Vec<u16>,
     pub exclude: bool,
+    pub titles: bool,
+    pub favicon: bool,
+    pub timestamps: bool,
+    pub bodysize: usize
 }
 
 #[derive(Parser, Default)]
@@ -22,6 +27,10 @@ pub struct Config {
     /// concurrent workers
     pub threads: usize,
 
+    #[clap(long = "ua")]
+    /// custom user agent to operate with
+    pub useragent: Option<String>,
+
     #[clap(long = "mc")]
     /// status codes to match, comma separated
     pub matchcodes: Option<String>,
@@ -77,5 +86,6 @@ pub fn setparams(c: &Config) -> Params {
         exclude = true;
     }
 
-    Params { statcodes, exclude }
+    Params { statcodes, exclude, titles: c.pulltitles, favicon: c.favicon,
+        timestamps: c.timestamps, bodysize: c.bodysize }
 }

src/common/console.rs

@@ -9,6 +9,16 @@ pub fn fatal(msg: &str) -> ! {
     process::exit(-1);
 }
 
+pub fn trunc(s: String, lim: usize) -> String {
+    let truncatedstr = s.char_indices()
+            .take_while(|(i, _)| *i < lim)
+            .map(|(_, c)| c)
+            .collect();
+
+    truncatedstr
+}
+
+// strips whitespace while maintaining legibility
 pub fn fmtwhitespace(s: String) -> String {
     let mut out = String::with_capacity(s.len());
     s.split_whitespace().for_each(|w| {
@@ -27,7 +37,7 @@ pub fn fmtcode(code: u16) -> ColoredString {
         300..=399 => code.to_string().yellow(),
         400..=499 => code.to_string().bright_red(),
         500..=599 => code.to_string().red().bold(),
-        _ => code.to_string().black(),
+        _ => code.to_string().bright_black(),
     }
 }
 
@@ -39,7 +49,7 @@ pub fn tstamp() -> String {
 }
 
 pub fn parsetitle(s: String) -> String {
-    let title = fmtwhitespace(s);
+    let title: String = trunc(fmtwhitespace(s), 1024);
 
     format!("{}{}{}",
     "title[".bright_black().bold(),
@@ -74,10 +84,11 @@ pub fn parsehit(sc: u16, url: &String) -> String {
 }
 
 pub fn banner() {
-    eprintln!(
+    eprintln!(r#"
-        "{}{} {}",
+  {}{}    |\___..--"/
-        "speed".bright_cyan().bold(),
+        __..--`""       /
-        "boat".bright_magenta().bold(),
+'-._.'._:._'-._____..--' {}
-        VERSION.bright_black()
+"#,         "speed".bright_cyan().bold(),
-    );
+"boat".bright_magenta().bold(),
+VERSION.bright_black());
 }

src/common/exec.rs

@@ -5,35 +5,36 @@ use std::{
 };
 
 use super::{
-    conf::{Config, Params},
+    conf::{Config, setparams},
     console::fatal,
     net::{mkclient, query},
 };
 
-pub async fn takeoff(args: Config, params: Params) {
+pub async fn takeoff(args: Config) {
-    let c = mkclient(args.follow).unwrap_or_else(|_| fatal("error instantiating http client"));
+    let params = setparams(&args);
+    
+    let mut ua: String = "Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Googlebot/2.1; +http://www.google.com/bot.html) Chrome/126.0.0.0 Safari/537.36".to_string();
+    if let Some(custom_ua) = args.useragent {
+        ua = custom_ua;
+    }
 
-    let file = File::open(args.list)
+    let client = mkclient(args.follow, ua).unwrap_or_else(|_| fatal("error instantiating http client"));
+
+    let file = File::open(&args.list)
         .unwrap_or_else(|e| fatal(format!("unable to read file: {e}").as_str()));
 
-    // Create a buffered reader.
     let buf = BufReader::new(file);
 
     stream::iter(buf.lines())
         .for_each_concurrent(args.threads, |line| {
-            let wc = c.clone();
+            // workers using the same client ref > each worker getting their own? we'll see 
-            let scodes = params.statcodes.clone();
+            let wc = &client;
+            let qparams = &params;
             async move {
                 let _ = query(
                     wc,
-                    line.unwrap_or_else(|_| fatal("error attempting buffered read"))
+                    line.unwrap_or_else(|_| fatal("error attempting buffered read")).trim(),
-                        .trim(),
+                    qparams
-                    scodes,
-                    params.exclude,
-                    args.pulltitles,
-                    args.favicon,
-                    args.timestamps,
-                    args.bodysize
                 )
                 .await;
             }

src/common/mod.rs

@@ -2,4 +2,4 @@ pub mod conf;
 pub mod console;
 pub mod exec;
 pub mod net;
-pub mod modules;
+pub mod modules;

src/common/modules.rs

@@ -1,14 +1,14 @@
 use select::{document::Document, predicate::Name};
-use reqwest::Url;
+use reqwest::{Response, Url};
 use base64::{Engine as _, engine::general_purpose};
+use bytes::{Bytes, BytesMut};
 use murmur3::murmur3_32;
-use std::{error::Error,
+use std::{error::Error, io::Cursor};
-io::Cursor};
 
-use super::console::{parsetitle, parsebody, parsehash, fmtwhitespace};
+use super::console::{parsetitle, parsebody, parsehash, fmtwhitespace, trunc};
 
-pub fn get_title(body: &String) -> String {
+pub fn get_title(body: &str) -> String {
-    let document = Document::from(body.as_str());
+    let document = Document::from(body);
 
     let title = document
     .find(Name("title"))
@@ -19,18 +19,30 @@ pub fn get_title(body: &String) -> String {
     parsetitle(title)
 }
 
-pub fn read_body(body: &String, lim: usize) -> String {
+pub async fn readnbody(mut resp: Response, n: usize) -> Result<Bytes, reqwest::Error> {
-    let document = Document::from(body.as_str());
+    let mut body = BytesMut::new();
+    while let Some(chunk) = resp.chunk().await? {
+        if !(body.len() + chunk.len()) > n {
+            body.extend_from_slice(&chunk); 
+        }   
+    }
+
+    Ok(body.freeze())
+}
+
+pub fn body_str(body: Bytes) -> Result<String, Box<dyn Error>> {
+    String::from_utf8(body.to_vec())
+        .map_err(|e| format!("error reading body into a string: {}", e).into())
+}
+
+pub fn body_contents(body: &str, lim: usize) -> String {
+    let document = Document::from(body);
 
     let mut bodytext = document.find(Name("body")).next().map(|n| n.text()).unwrap_or_else(|| "".to_string());
 
     bodytext = fmtwhitespace(bodytext);
     if bodytext.len() > lim {
-        bodytext = bodytext
+        bodytext = trunc(bodytext, lim);
-            .char_indices()
-            .take_while(|(i, _)| *i < lim)
-            .map(|(_, c)| c)
-            .collect();
     }
 
     parsebody(bodytext)
@@ -76,8 +88,8 @@ async fn dl_favicon(url: String) -> Result<Vec<u8>, Box<dyn Error>> {
     Ok(data)
 }
 
-pub async fn hash_favicon(body: &String, url: String) -> String {
+pub async fn hash_favicon(body: &str, url: String) -> String {
-    let document = Document::from(body.as_str());
+    let document = Document::from(body);
 
     if let Ok(favurl) = faviconurl(document, url) {
         if let Ok(data) = dl_favicon(favurl).await {
@@ -90,4 +102,4 @@ pub async fn hash_favicon(body: &String, url: String) -> String {
     }
 
     "".into()
-}
+}

src/common/net.rs

@@ -2,17 +2,18 @@ use reqwest::{redirect::Policy, Client};
 use std::time::Duration;
 
 use super::console::{parsehit, tstamp};
+use super::conf::Params;
 use super::modules::*;
 
-pub fn mkclient(redir: bool) -> Result<Client, reqwest::Error> {
+pub fn mkclient(redir: bool, ua: String) -> Result<Client, reqwest::Error> {
     let rpolicy: Policy = if redir {
-        Policy::limited(5)
+        Policy::limited(3)
     } else {
         Policy::none()
     };
 
     Client::builder()
-        .user_agent("buttplug/1.0")
+        .user_agent(ua)
         .redirect(rpolicy)
         .timeout(Duration::from_secs(2))
         .connect_timeout(Duration::from_millis(500))
@@ -25,57 +26,60 @@ async fn sendreq(c: &Client, use_https: bool, url: &str) -> Result<reqwest::Resp
 }
 
 pub async fn query(
-    c: Client,
+    c: &Client,
     url: &str,
-    codes: Vec<u16>,
+    params: &Params
-    exclude: bool,
-    titles: bool,
-    favicon: bool,
-    timestamps: bool,
-    bodysize: usize,
 ) -> Result<(), reqwest::Error> {
     let response: reqwest::Response;
-    if let Ok(res) = sendreq(&c, true, url).await {
-        response = res;
-    } else {
-        response = sendreq(&c, false, url).await?;
-    }
 
-    let statcode = response.status().as_u16();
+    match sendreq(c, true, url).await {
-
+        Ok(res) => response = res,
-    if codes.len() > 0 {
+        Err(e) => {
-        if codes.contains(&statcode) {
+            if e.is_request() {
-            if exclude {
+                response = sendreq(c, false, url).await?;
-                return Ok(());
+            } else {
+                return Err(e);
             }
-        } else if !exclude {
-            return Ok(());
         }
     }
 
     let sc = response.status().as_u16();
+
+    if !params.statcodes.is_empty() {
+        if params.statcodes.contains(&sc) {
+            if params.exclude {
+                return Ok(());
+            }
+        } else if !params.exclude {
+            return Ok(());
+        }
+    }
+
     let url: String = response.url().to_string();
-    let body = response.text().await?;
+    
+    // 10mb response body limit
+    let body_raw = readnbody(response, 10485760).await?;
+    let body = body_str(body_raw).unwrap_or("error parsing response body".to_string());
 
     let mut out = parsehit(sc, &url);
 
-    if timestamps {
+    if params.timestamps {
         out = format!("{} {}", tstamp(), out);
     }
 
-    if titles {
+    if params.titles {
         out = format!("{} {}", out, get_title(&body));
     }
 
-    if favicon {
+    if params.favicon {
         let hash = hash_favicon(&body, url).await;
         if !hash.is_empty() {
             out = format!("{} {}", out, hash);
         }
     }
 
-    if bodysize > 0 {
+    if params.bodysize > 0 {
-        out = format!("{} {}", out, read_body(&body, bodysize));
+        out = format!("{} {}", out, body_contents(&body, params.bodysize));
     }
 
     println!("{}", out);

src/main.rs

@@ -2,9 +2,6 @@ mod common;
 
 #[tokio::main]
 async fn main() {
-    let args = common::conf::load();
-    let scanparams = common::conf::setparams(&args);
-
     common::console::banner();
-    common::exec::takeoff(args, scanparams).await;
+    common::exec::takeoff(common::conf::load()).await;
-}
+}