Url util functions (#367)

* URL util functions * Update API docs * clippy * Fix markdown
awslabs · May 8, 2024 · f796a5d · f796a5d
1 parent b2649cd
commit f796a5d
Show file tree

Hide file tree

Showing 8 changed files with 275 additions and 41 deletions.
diff --git a/API.md b/API.md
@@ -190,8 +190,6 @@ export class URL {
 }
 ```
 
-### TODO, URL see tracking [ticket](https://github.com/awslabs/llrt/issues/303):
-
 ```typescript
 // Additional utilities in the URL module
 export function domainToASCII(domain: string): string;
@@ -202,11 +200,15 @@ export function fileURLToPath(url: string | URL): string;
 
 export function pathToFileURL(path: string): URL;
 
+export function format(url: string | URL, options?: { fragment?: boolean, unicode?: boolean, auth?: boolean
+}): string;
+
 export function urlToHttpOptions(url: URL): {
   protocol?: string;
   hostname?: string;
   port?: string;
   path?: string;
+  ...
 };
 ```
 

diff --git a/Cargo.lock b/Cargo.lock
diff --git a/llrt_core/Cargo.toml b/llrt_core/Cargo.toml
@@ -75,6 +75,7 @@ rand = "0.8.5"
 uname = "0.1.1"
 flate2 = { version = "1.0.30", features = ["zlib-ng"], default-features = false }
 brotlic = "0.8.2"
+idna = "0.5.0"
 
 [build-dependencies]
 rquickjs = { version = "0.5.1", features = [

diff --git a/llrt_core/src/modules/http/url.rs b/llrt_core/src/modules/http/url.rs
@@ -1,11 +1,13 @@
+use std::{path::PathBuf, str::FromStr};
+
 // Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
 // SPDX-License-Identifier: Apache-2.0
 use rquickjs::{
     atom::PredefinedAtom,
     class::{Trace, Tracer},
     function::Opt,
     prelude::This,
-    Class, Coerced, Ctx, Exception, FromJs, Function, Result, Value,
+    Class, Coerced, Ctx, Exception, FromJs, Function, Object, Result, Value,
 };
 use url::Url;
 
@@ -90,27 +92,7 @@ impl<'js> URL<'js> {
     }
 
     pub fn to_string(&self) -> String {
-        let search = search_params_to_string(&self.search_params);
-        let hash = &self.hash;
-        let hash = if !hash.is_empty() {
-            format!("#{}", &hash)
-        } else {
-            String::from("")
-        };
-        let mut user_info = String::new();
-        if !self.username.is_empty() {
-            user_info.push_str(&self.username);
-            if !self.password.is_empty() {
-                user_info.push(':');
-                user_info.push_str(&self.password)
-            }
-            user_info.push('@')
-        }
-
-        format!(
-            "{}://{}{}{}{}{}",
-            &self.protocol, user_info, &self.host, &self.pathname, &search, &hash
-        )
+        self.format(true, true, true, false)
     }
 
     #[qjs(get)]
@@ -309,6 +291,47 @@ impl<'js> URL<'js> {
             self.host.clone_from(&self.hostname);
         }
     }
+
+    fn format(
+        &self,
+        include_auth: bool,
+        include_fragment: bool,
+        include_search: bool,
+        unicode_encode: bool,
+    ) -> String {
+        let search = if include_search {
+            search_params_to_string(&self.search_params)
+        } else {
+            String::from("")
+        };
+        let hash = &self.hash;
+        let hash = if include_fragment && !hash.is_empty() {
+            format!("#{}", &hash)
+        } else {
+            String::from("")
+        };
+
+        let mut user_info = String::new();
+        if include_auth && !self.username.is_empty() {
+            user_info.push_str(&self.username);
+            if !self.password.is_empty() {
+                user_info.push(':');
+                user_info.push_str(&self.password)
+            }
+            user_info.push('@')
+        }
+
+        let host = if unicode_encode {
+            domain_to_unicode(&self.host)
+        } else {
+            self.host.clone()
+        };
+
+        format!(
+            "{}://{}{}{}{}{}",
+            &self.protocol, user_info, host, &self.pathname, &search, &hash
+        )
+    }
 }
 
 fn filtered_port(protocol: &str, port: &str) -> Option<String> {
@@ -357,3 +380,104 @@ fn split_colon<'js>(ctx: &Ctx, s: &'js str) -> Result<(&'js str, &'js str)> {
     }
     Ok((first, second))
 }
+
+pub fn url_to_http_options<'js>(ctx: Ctx<'js>, url: Class<'js, URL<'js>>) -> Result<Object<'js>> {
+    let obj = Object::new(ctx)?;
+
+    let url = url.borrow();
+
+    let port = url.port();
+    let username = url.username();
+    let search = url.search();
+    let hash = url.hash();
+
+    obj.set("protocol", url.protocol())?;
+    obj.set("hostname", url.hostname())?;
+
+    if !hash.is_empty() {
+        obj.set("hash", url.hash())?;
+    }
+    if !search.is_empty() {
+        obj.set("search", url.search())?;
+    }
+
+    obj.set("pathname", url.pathname())?;
+    obj.set("path", format!("{}{}", url.pathname(), url.search()))?;
+    obj.set("href", url.href())?;
+
+    if !username.is_empty() {
+        obj.set("auth", format!("{}:{}", username, url.password()))?;
+    }
+
+    if !port.is_empty() {
+        obj.set("port", url.port())?;
+    }
+
+    Ok(obj)
+}
+
+pub fn domain_to_unicode(domain: &str) -> String {
+    let (url, result) = idna::domain_to_unicode(domain);
+    if result.is_err() {
+        return String::from("");
+    }
+    url
+}
+
+pub fn domain_to_ascii(domain: &str) -> String {
+    idna::domain_to_ascii(domain).unwrap_or_default()
+}
+
+//options are ignored, no windows support yet
+pub fn path_to_file_url<'js>(ctx: Ctx<'js>, path: String, _: Opt<Value>) -> Result<URL<'js>> {
+    let url = Url::from_file_path(path).unwrap();
+    URL::create(ctx, url)
+}
+
+//options are ignored, no windows support yet
+pub fn file_url_to_path<'js>(ctx: Ctx<'js>, url: Value<'js>) -> Result<String> {
+    let url_string = if let Ok(url) = Class::<URL>::from_value(url.clone()) {
+        url.borrow().to_string()
+    } else {
+        url.get::<Coerced<String>>()?.to_string()
+    };
+
+    let path = if let Some(path) = &url_string.strip_prefix("file://") {
+        path.to_string()
+    } else {
+        url_string
+    };
+
+    Ok(PathBuf::from_str(&path)
+        .or_throw(&ctx)?
+        .to_string_lossy()
+        .to_string())
+}
+
+pub fn url_format<'js>(url: Class<'js, URL<'js>>, options: Opt<Value<'js>>) -> Result<String> {
+    let mut fragment = true;
+    let mut unicode = false;
+    let mut auth = true;
+    let mut search = true;
+
+    // Parse options if provided
+    if let Some(options) = options.0 {
+        if options.is_object() {
+            let options = options.as_object().unwrap();
+            if let Some(value) = options.get("fragment")? {
+                fragment = value;
+            }
+            if let Ok(value) = options.get("unicode") {
+                unicode = value;
+            }
+            if let Ok(value) = options.get("auth") {
+                auth = value;
+            }
+            if let Ok(value) = options.get("search") {
+                search = value
+            }
+        }
+    }
+
+    Ok(url.borrow().format(auth, fragment, search, unicode))
+}
diff --git a/llrt_core/src/modules/http/url_search_params.rs b/llrt_core/src/modules/http/url_search_params.rs
@@ -125,8 +125,11 @@ impl URLSearchParams {
         let mut string = String::with_capacity(self.params.len() * 2);
         for (i, (key, value)) in self.params.iter().enumerate() {
             string.push_str(&escape(key));
-            string.push('=');
-            string.push_str(&escape(value));
+            if !value.is_empty() {
+                string.push('=');
+                string.push_str(&escape(value));
+            }
+
             if i < length - 1 {
                 string.push('&');
             }

diff --git a/llrt_core/src/modules/url.rs b/llrt_core/src/modules/url.rs
@@ -1,19 +1,29 @@
 // Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
 // SPDX-License-Identifier: Apache-2.0
 use rquickjs::{
-    function::Constructor,
+    function::{Constructor, Func},
     module::{Declarations, Exports, ModuleDef},
     Ctx, Result,
 };
 
 use crate::{module_builder::ModuleInfo, modules::module::export_default};
+
+use super::http::url::{
+    domain_to_ascii, domain_to_unicode, file_url_to_path, path_to_file_url, url_format,
+    url_to_http_options,
+};
 pub struct UrlModule;
 
 impl ModuleDef for UrlModule {
     fn declare(declare: &mut Declarations) -> Result<()> {
         declare.declare(stringify!(URL))?;
         declare.declare(stringify!(URLSearchParams))?;
-
+        declare.declare("urlToHttpOptions")?;
+        declare.declare("domainToUnicode")?;
+        declare.declare("domainToASCII")?;
+        declare.declare("fileURLToPath")?;
+        declare.declare("pathToFileURL")?;
+        declare.declare("format")?;
         declare.declare("default")?;
         Ok(())
     }
@@ -26,6 +36,18 @@ impl ModuleDef for UrlModule {
         export_default(ctx, exports, |default| {
             default.set(stringify!(URL), url)?;
             default.set(stringify!(URLSearchParams), url_search_params)?;
+            default.set("urlToHttpOptions", Func::from(url_to_http_options))?;
+            default.set(
+                "domainToUnicode",
+                Func::from(|domain: String| domain_to_unicode(&domain)),
+            )?;
+            default.set(
+                "domainToASCII",
+                Func::from(|domain: String| domain_to_ascii(&domain)),
+            )?;
+            default.set("fileURLToPath", Func::from(file_url_to_path))?;
+            default.set("pathToFileURL", Func::from(path_to_file_url))?;
+            default.set("format", Func::from(url_format))?;
             Ok(())
         })?;
 

diff --git a/llrt_core/src/vm.rs b/llrt_core/src/vm.rs
@@ -619,10 +619,11 @@ fn init(ctx: &Ctx<'_>, module_names: HashSet<&'static str>) -> Result<()> {
         "require",
         Func::from(move |ctx, specifier: String| -> Result<Value> {
             let LifetimeArgs(ctx) = LifetimeArgs(ctx);
-            let specifier: String = specifier
-                .strip_prefix("node:")
-                .unwrap_or(specifier.as_str())
-                .into();
+            let specifier = if let Some(striped_specifier) = &specifier.strip_prefix("node:") {
+                striped_specifier.to_string()
+            } else {
+                specifier
+            };
             let import_name = if module_names.contains(specifier.as_str())
                 || BYTECODE_CACHE.contains_key(&specifier)
                 || specifier.starts_with('/')