added a few tests

src/utils.rs

@@ -712,4 +712,46 @@ mod tests {
 
         assert!(!should_deny_url(&tested_url, handles).unwrap());
     }
+
+    #[test]
+    /// provide a denier where the tested url is matched against a regular expression in the path
+    /// of the url
+    fn should_deny_url_blocks_urls_based_on_regex_in_path() {
+        let scan_url = "https://testdomain.com/";
+        let deny_pattern = "/deni.*";
+        let tested_url = Url::parse("https://testdomain.com/denied/").unwrap();
+
+        let scans = Arc::new(FeroxScans::default());
+        scans.add_directory_scan(scan_url, ScanOrder::Initial);
+
+        let mut config = Configuration::new().unwrap();
+        config.regex_denylist = vec![Regex::new(deny_pattern).unwrap()];
+        let config = Arc::new(config);
+
+        let handles = Arc::new(Handles::for_testing(Some(scans), Some(config)).0);
+
+        assert!(should_deny_url(&tested_url, handles).unwrap());
+    }
+
+    #[test]
+    /// provide a denier where the tested url is matched against a regular expression in the scheme
+    /// of the url
+    fn should_deny_url_blocks_urls_based_on_regex_in_scheme() {
+        let scan_url = "https://testdomain.com/";
+        let deny_pattern = "http:";
+        let tested_http_url = Url::parse("http://testdomain.com/denied/").unwrap();
+        let tested_https_url = Url::parse("https://testdomain.com/denied/").unwrap();
+
+        let scans = Arc::new(FeroxScans::default());
+        scans.add_directory_scan(scan_url, ScanOrder::Initial);
+
+        let mut config = Configuration::new().unwrap();
+        config.regex_denylist = vec![Regex::new(deny_pattern).unwrap()];
+        let config = Arc::new(config);
+
+        let handles = Arc::new(Handles::for_testing(Some(scans), Some(config)).0);
+
+        assert!(!should_deny_url(&tested_https_url, handles.clone()).unwrap());
+        assert!(should_deny_url(&tested_http_url, handles).unwrap());
+    }
 }

tests/test_deny_list.rs

@@ -210,3 +210,73 @@ fn deny_list_works_during_recursion_with_inverted_parents() {
 
     teardown_tmp_directory(tmp_dir);
 }
+
+#[test]
+/// test that a regex that prevents the base url from being scanned results in an early exit
+fn deny_list_prevents_regex_that_denies_base_url() {
+    let srv = MockServer::start();
+    let (tmp_dir, file) = setup_tmp_directory(&["LICENSE".to_string()], "wordlist").unwrap();
+
+    let mock = srv.mock(|when, then| {
+        when.method(GET).path("/LICENSE");
+        then.status(200).body("this is a test");
+    });
+
+    let cmd = Command::cargo_bin("feroxbuster")
+        .unwrap()
+        .arg("--url")
+        .arg(srv.url("/"))
+        .arg("--wordlist")
+        .arg(file.as_os_str())
+        .arg("--dont-scan")
+        .arg("/")
+        .unwrap();
+
+    teardown_tmp_directory(tmp_dir);
+
+    let err_msg = format!(
+        "Could not determine initial targets: The regex '/' matches {}/; the scan will never start",
+        srv.base_url()
+    );
+    cmd.assert()
+        .success()
+        .stderr(predicate::str::contains(err_msg));
+
+    assert_eq!(mock.hits(), 0);
+}
+
+#[test]
+/// test that a url that prevents the base url from being scanned results in an early exit
+fn deny_list_prevents_url_that_denies_base_url() {
+    let srv = MockServer::start();
+    let (tmp_dir, file) = setup_tmp_directory(&["LICENSE".to_string()], "wordlist").unwrap();
+
+    let mock = srv.mock(|when, then| {
+        when.method(GET).path("/LICENSE");
+        then.status(200).body("this is a test");
+    });
+
+    let cmd = Command::cargo_bin("feroxbuster")
+        .unwrap()
+        .arg("--url")
+        .arg(srv.url("/"))
+        .arg("--wordlist")
+        .arg(file.as_os_str())
+        .arg("--dont-scan")
+        .arg(srv.base_url())
+        .unwrap();
+
+    teardown_tmp_directory(tmp_dir);
+
+    let err_msg = format!(
+        "Could not determine initial targets: The url '{}/' matches {}/; the scan will never start",
+        srv.base_url(),
+        srv.base_url()
+    );
+
+    cmd.assert()
+        .success()
+        .stderr(predicate::str::contains(err_msg));
+
+    assert_eq!(mock.hits(), 0);
+}