hotfix: better platform-agnostic handling of file:// urls (#156)

crates/client/src/components/mod.rs

@@ -5,6 +5,7 @@ use yew::prelude::*;
 
 use btn::DeleteButton;
 use shared::response::{LensResult, SearchResult};
+use shared::url_to_file_path;
 use url::Url;
 
 #[derive(Clone, Debug, PartialEq)]
@@ -114,8 +115,7 @@ pub fn search_result_component(props: &SearchResultProps) -> Html {
                         .clone()
                         .unwrap_or_else(|| "example.com".to_string());
 
-                    let path = url.path();
-
+                    let path = url_to_file_path(url.path(), false);
                     let uri_icon = if url.scheme() == "file" {
                         html! {
                             <icons::DesktopComputerIcon

crates/shared/src/lib.rs

@@ -22,3 +22,21 @@ pub struct SettingOpts {
     pub form_type: FormType,
     pub help_text: Option<String>,
 }
+
+/// A platform-agnostic way to turn a URL file path into something that can
+/// be opened & crawled.
+pub fn url_to_file_path(path: &str, is_windows: bool) -> String {
+    // Unescape colons & spaces
+    let mut path = path.replace("%3A", ":").replace("%20", " ");
+    // Strip superfluous path prefix
+    if is_windows {
+        path = path
+            .strip_prefix('/')
+            .map(|s| s.to_string())
+            .unwrap_or(path);
+        // Convert path dividers into Windows specific ones.
+        path = path.replace('/', "\\");
+    }
+
+    path
+}

crates/spyglass/src/crawler/mod.rs

@@ -11,13 +11,14 @@ use url::{Host, Url};
 use entities::models::{crawl_queue, fetch_history};
 use entities::sea_orm::prelude::*;
 use entities::sea_orm::DatabaseConnection;
-
-pub mod bootstrap;
-pub mod robots;
+use shared::url_to_file_path;
 
 use crate::crawler::bootstrap::create_archive_url;
 use crate::fetch::HTTPClient;
 use crate::scraper::{html_to_text, DEFAULT_DESC_LENGTH};
+
+pub mod bootstrap;
+pub mod robots;
 use robots::check_resource_rules;
 
 // TODO: Make this configurable by domain
@@ -247,7 +248,25 @@ impl Crawler {
         _: &crawl_queue::Model,
         url: &Url,
     ) -> anyhow::Result<Option<CrawlResult>, anyhow::Error> {
-        let path = Path::new(url.path());
+        // Attempt to convert from the URL to a file path
+        #[allow(unused_assignments)]
+        let mut url_path = url
+            .to_file_path()
+            .map(|p| p.display().to_string())
+            .unwrap_or_else(|_| url.path().to_string());
+
+        #[cfg(not(target_os = "windows"))]
+        {
+            url_path = url_to_file_path(&url_path, false);
+        }
+
+        // Fixes issues handling Windows drive paths
+        #[cfg(target_os = "windows")]
+        {
+            url_path = url_to_file_path(&url.path(), true);
+        }
+
+        let path = Path::new(&url_path);
         // Is this a file and does this exist?
         if !path.exists() || !path.is_file() {
             return Ok(None);

crates/spyglass/src/main.rs

@@ -21,6 +21,7 @@ use crate::api::start_api_ipc;
 fn main() -> Result<(), Box<dyn std::error::Error>> {
     let config = Config::new();
 
+    #[cfg(not(debug_assertions))]
     let _guard = if config.user_settings.disable_telementry {
         None
     } else {

crates/spyglass/src/task.rs

@@ -86,14 +86,14 @@ pub async fn manager_task(
                 let cmd = Command::Fetch(CrawlTask { id: task.id });
                 if queue.send(cmd).await.is_err() {
                     eprintln!("unable to send command to worker");
-                    return;
                 }
             }
             // ignore everything else
-            _ => {
-                tokio::time::sleep(tokio::time::Duration::from_secs(1)).await;
-            }
+            _ => {}
         }
+
+        // Wait a little before we dequeue another URL
+        tokio::time::sleep(tokio::time::Duration::from_millis(100)).await;
     }
 }
 

crates/tauri/src/cmd.rs

@@ -48,7 +48,15 @@ pub async fn open_result(_: tauri::Window, url: &str) -> Result<(), String> {
     if let Ok(mut url) = url::Url::parse(url) {
         // treat open files as a local action.
         if url.scheme() == "file" {
-            let _ = url.set_host(Some("localhost"));
+            let _ = url.set_host(None);
+
+            #[cfg(target_os = "windows")]
+            {
+                use shared::url_to_file_path;
+                let path = url_to_file_path(url.path(), true);
+                open::that(format!("file://{}", path)).unwrap();
+                return Ok(());
+            }
         }
 
         open::that(url.to_string()).unwrap();

crates/tauri/src/main.rs

@@ -2,6 +2,7 @@
     all(not(debug_assertions), target_os = "windows"),
     windows_subsystem = "windows"
 )]
+#[allow(unused_imports)]
 use std::borrow::Cow;
 use std::io;
 use std::path::PathBuf;
@@ -42,6 +43,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
     let ctx = tauri::generate_context!();
     let config = Config::new();
 
+    #[cfg(not(debug_assertions))]
     let _guard = if config.user_settings.disable_telementry {
         None
     } else {

plugins/local-file-indexer/Cargo.toml

@@ -10,4 +10,5 @@ path = "src/main.rs"
 [dependencies]
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
-spyglass-plugin = { path = "../../crates/spyglass-plugin" }
+spyglass-plugin = { path = "../../crates/spyglass-plugin" }
+url = "2.2.2"

plugins/local-file-indexer/src/main.rs

@@ -1,6 +1,7 @@
 use spyglass_plugin::*;
 use std::collections::HashSet;
 use std::path::Path;
+use url::Url;
 
 #[derive(Default)]
 struct Plugin {
@@ -21,7 +22,11 @@ fn to_uri(path: &str) -> String {
         "home.local".into()
     };
 
-    format!("file://{}/{}", host, path)
+    let mut new_url = Url::parse("file://").expect("Base URI");
+    let _ = new_url.set_host(Some(&host));
+    // Fixes issues handling windows drive letters
+    new_url.set_path(&path.replace(':', "%3A"));
+    new_url.to_string()
 }
 
 impl Plugin {