feat(sources): add regex (#13)

Add source type regex

Search through a webpage for the version string that fits a given regex.
The first match of the regex is returned.

---------

Co-authored-by: Adam Perkowski <[email protected]>

Cargo.toml

@@ -21,10 +21,11 @@ include = [
 
 [features]
 nvrs_cli = ["clap", "colored", "futures"]
-default = ["aur", "github", "gitlab"]
+default = ["aur", "github", "gitlab", "regex"]
 aur = []
 github = []
 gitlab = []
+regex = ["dep:regex"]
 
 [[bin]]
 name = "nvrs"
@@ -40,6 +41,7 @@ required-features = ["nvrs_tui"]
 clap = { version = "4.5.22", features = ["derive", "color", "error-context", "help", "std", "usage"], default-features = false , optional = true }
 colored = { version = "2.1.0", optional = true }
 futures = { version = "0.3.31", optional = true }
+regex = { version = "1.11.1", optional = true }
 reqwest = { version = "0.12.9", features = ["__tls", "charset", "default-tls", "h2", "http2", "json"], default-features = false }
 serde = { version = "1.0.215", features = ["derive"], default-features = false }
 serde_json = "1.0.132"

README.md

@@ -28,6 +28,7 @@ check the [release notes](https://github.com/adamperkowski/nvrs/releases) for co
 - `aur`
 - `github`
 - `gitlab` (with custom hosts)
+- `website` (regex)
 
 ### QOL improvements
 - `ALL` argument for the `--take` command
@@ -143,6 +144,8 @@ package entries are custom entries in the main config file. they contain values
 | `host`        | domain name the source is hosted on                                       | string | ❌       | ❌     |
 | `prefix`      | the prefix used in releases / tags<br>example: `v` for tags like `v0.1.0` | string | ❌       | ❌     |
 | `use_max_tag` | use max git tag instead of the latest release                             | bool   | ❌       | ❌     |
+| `url`         | url to check for source type `regex`                                      | string | ❌       | ❌     |
+| `regex`       | regex to search url for source type `regex`                               | bool   | ❌       | ❌     |
 
 ### Keyfile structure
 this file contains API keys for various [sources](#sources). example can be found [here](/n_keyfile.toml).

nvrs.toml

@@ -28,3 +28,15 @@ prefix = "v"
 source = "github"
 github = "rust-lang/rustup"
 use_max_tag = true
+
+[rustrover]
+source = 'regex'
+url = 'https://data.services.jetbrains.com/products?code=RR&release.type=release'
+encoding = 'utf8'
+regex = 'RustRover-([\d.]+).tar.gz'
+
+[linux]
+source = 'regex'
+url = 'https://www.kernel.org/'
+encoding = 'utf8'
+regex = '<td><strong>([\d.]+)</strong></td>'

src/api/mod.rs

@@ -6,6 +6,8 @@ mod aur;
 mod github;
 #[cfg(feature = "gitlab")]
 mod gitlab;
+#[cfg(feature = "regex")]
+mod regex;
 
 /// struct containing the API name & a pointer to API's `get_latest` function
 pub struct Api {
@@ -78,6 +80,11 @@ pub const API_LIST: &[Api] = &[
         name: "gitlab",
         func: gitlab::get_latest,
     },
+    #[cfg(feature = "regex")]
+    Api {
+        name: "regex",
+        func: regex::get_latest,
+    },
 ];
 
 #[test]

src/api/regex.rs

@@ -0,0 +1,47 @@
+use crate::{api, error};
+use regex::Regex;
+
+/// get a version string from a webpage
+pub fn get_latest(args: api::ApiArgs) -> api::ReleaseFuture {
+    Box::pin(async move {
+        let url = args.args[0].clone();
+        let client = args.request_client;
+
+        let result = client
+            .get(&url)
+            .headers(api::setup_headers())
+            .send()
+            .await?;
+        api::match_statuscode(&result.status(), args.package.clone())?;
+
+        let body = result.text().await?;
+
+        let re = Regex::new(&args.args[1]).unwrap();
+        if let Some(caps) = re.captures(&body) {
+            Ok(api::Release {
+                name: caps.get(1).unwrap().as_str().to_owned(),
+                tag: None,
+                url,
+            })
+        } else {
+            Err(error::Error::NoVersion(args.package))
+        }
+    })
+}
+
+#[tokio::test]
+async fn request_test() {
+    let package = "rustrover".to_string();
+    let args = api::ApiArgs {
+        request_client: reqwest::Client::new(),
+        package: package.clone(),
+        use_max_tag: None,
+        args: vec![
+            "https://data.services.jetbrains.com/products?code=RR&release.type=release".to_string(),
+            r"RustRover-([\d.]+).tar.gz".to_string(),
+        ],
+        api_key: String::new(),
+    };
+
+    assert!(crate::api::regex::get_latest(args).await.is_ok());
+}

src/config.rs

@@ -59,6 +59,12 @@ pub struct Package {
     #[serde(default)]
     #[serde(skip_serializing_if = "is_empty_string")]
     gitlab: String,
+    #[cfg(feature = "regex")]
+    #[serde(default)]
+    url: String,
+    #[cfg(feature = "regex")]
+    #[serde(default)]
+    regex: String,
 
     /// whether to use the latest tag instead of the latest release
     #[serde(default)]
@@ -95,6 +101,11 @@ impl Package {
                 package.gitlab = target;
                 Ok(())
             }
+            #[cfg(feature = "regex")]
+            "regex" => {
+                package.url = target;
+                Ok(())
+            }
             _ => Err(error::Error::SourceNotFound(source.clone())),
         }?;
 
@@ -115,6 +126,10 @@ impl Package {
             github: String::new(),
             #[cfg(feature = "gitlab")]
             gitlab: String::new(),
+            #[cfg(feature = "regex")]
+            url: String::new(),
+            #[cfg(feature = "regex")]
+            regex: String::new(),
             use_max_tag: None,
             prefix: String::new(),
         }
@@ -137,6 +152,8 @@ impl Package {
             "github" => vec![self.github.clone()],
             #[cfg(feature = "gitlab")]
             "gitlab" => vec![self.gitlab.clone(), self.host.clone()],
+            #[cfg(feature = "regex")]
+            "regex" => vec![self.url.clone(), self.regex.clone()],
             _ => vec![],
         };
 

src/tui.rs

@@ -0,0 +1 @@
+