foster/denoland-deno
1
0
Fork 0
mirror of https://github.com/denoland/deno.git synced 2024-12-22 15:24:46 -05:00
Code Issues Packages 1 Wiki Activity

refactor: rewrite HTTP cache for file fetcher (#4030)

Browse source
This commit is contained in:
Bartek Iwańczuk 2020-02-19 08:17:13 -05:00 committed by GitHub
parent 046bbb2691
commit 852823fa50
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
10 changed files with 428 additions and 600 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
cli/deno_dir.rs
View file

@ -10,8 +10,6 @@ use std::path::PathBuf;
pub struct DenoDir { pub struct DenoDir {
// Example: /Users/rld/.deno/ // Example: /Users/rld/.deno/
pub root: PathBuf, pub root: PathBuf,
/// Used by SourceFileFetcher to cache remote modules.
pub deps_cache: DiskCache,
/// Used by TsCompiler to cache compiler output. /// Used by TsCompiler to cache compiler output.
pub gen_cache: DiskCache, pub gen_cache: DiskCache,
} }
@ -29,12 +27,10 @@ impl DenoDir {
.unwrap_or(fallback); .unwrap_or(fallback);
let root: PathBuf = custom_root.unwrap_or(default); let root: PathBuf = custom_root.unwrap_or(default);
let deps_path = root.join("deps");
let gen_path = root.join("gen"); let gen_path = root.join("gen");
let deno_dir = Self { let deno_dir = Self {
root, root,
deps_cache: DiskCache::new(&deps_path),
gen_cache: DiskCache::new(&gen_path), gen_cache: DiskCache::new(&gen_path),
}; };

618
cli/file_fetcher.rs
View file

@ -3,11 +3,10 @@ use crate::colors;
use crate::deno_error::DenoError; use crate::deno_error::DenoError;
use crate::deno_error::ErrorKind; use crate::deno_error::ErrorKind;
use crate::deno_error::GetErrorKind; use crate::deno_error::GetErrorKind;
use crate::disk_cache::DiskCache; use crate::http_cache::HttpCache;
use crate::http_util; use crate::http_util;
use crate::http_util::create_http_client; use crate::http_util::create_http_client;
use crate::http_util::FetchOnceResult; use crate::http_util::FetchOnceResult;
use crate::http_util::ResultPayload;
use crate::msg; use crate::msg;
use deno_core::ErrBox; use deno_core::ErrBox;
use deno_core::ModuleSpecifier; use deno_core::ModuleSpecifier;
@ -15,11 +14,11 @@ use futures::future::Either;
use futures::future::FutureExt; use futures::future::FutureExt;
use regex::Regex; use regex::Regex;
use reqwest; use reqwest;
use serde_json;
use std; use std;
use std::collections::HashMap; use std::collections::HashMap;
use std::fs; use std::fs;
use std::future::Future; use std::future::Future;
use std::io::Read;
use std::path::Path; use std::path::Path;
use std::path::PathBuf; use std::path::PathBuf;
use std::pin::Pin; use std::pin::Pin;
@ -30,28 +29,6 @@ use std::sync::Mutex;
use url; use url;
use url::Url; use url::Url;
pub fn source_header_cache_failed_error(
module_name: &str,
reason: &str,
) -> ErrBox {
DenoError::new(
ErrorKind::Other,
format!(
"Source code header cache failed for '{}': {}",
module_name, reason
),
)
.into()
}
pub fn source_cache_failed_error(module_name: &str, reason: &str) -> ErrBox {
DenoError::new(
ErrorKind::Other,
format!("Source code cache failed for '{}': {}", module_name, reason),
)
.into()
}
/// Structure representing local or remote file. /// Structure representing local or remote file.
/// ///
/// In case of remote file `url` might be different than originally requested URL, if so /// In case of remote file `url` might be different than originally requested URL, if so
@ -90,22 +67,21 @@ impl SourceFileCache {
const SUPPORTED_URL_SCHEMES: [&str; 3] = ["http", "https", "file"]; const SUPPORTED_URL_SCHEMES: [&str; 3] = ["http", "https", "file"];
/// `DenoDir` serves as coordinator for multiple `DiskCache`s containing them
/// in single directory that can be controlled with `$DENO_DIR` env variable.
#[derive(Clone)] #[derive(Clone)]
pub struct SourceFileFetcher { pub struct SourceFileFetcher {
deps_cache: DiskCache,
source_file_cache: SourceFileCache, source_file_cache: SourceFileCache,
cache_blacklist: Vec<String>, cache_blacklist: Vec<String>,
use_disk_cache: bool, use_disk_cache: bool,
no_remote: bool, no_remote: bool,
cached_only: bool, cached_only: bool,
http_client: reqwest::Client, http_client: reqwest::Client,
// This field is public only to expose it's location
pub http_cache: HttpCache,
} }
impl SourceFileFetcher { impl SourceFileFetcher {
pub fn new( pub fn new(
deps_cache: DiskCache, http_cache: HttpCache,
use_disk_cache: bool, use_disk_cache: bool,
cache_blacklist: Vec<String>, cache_blacklist: Vec<String>,
no_remote: bool, no_remote: bool,
@ -113,7 +89,7 @@ impl SourceFileFetcher {
ca_file: Option<String>, ca_file: Option<String>,
) -> Result<Self, ErrBox> { ) -> Result<Self, ErrBox> {
let file_fetcher = Self { let file_fetcher = Self {
deps_cache, http_cache,
source_file_cache: SourceFileCache::default(), source_file_cache: SourceFileCache::default(),
cache_blacklist, cache_blacklist,
use_disk_cache, use_disk_cache,
@ -333,54 +309,45 @@ impl SourceFileFetcher {
&self, &self,
module_url: &Url, module_url: &Url,
) -> Result<Option<SourceFile>, ErrBox> { ) -> Result<Option<SourceFile>, ErrBox> {
let source_code_headers = self.get_source_code_headers(&module_url); let result = self.http_cache.get(&module_url);
// If source code headers says that it would redirect elsewhere, let result = match result {
// (meaning that the source file might not exist; only .headers.json is present)
// Abort reading attempts to the cached source file and and follow the redirect.
if let Some(redirect_to) = source_code_headers.redirect_to {
// E.g.
// module_name https://import-meta.now.sh/redirect.js
// filename /Users/kun/Library/Caches/deno/deps/https/import-meta.now.sh/redirect.js
// redirect_to https://import-meta.now.sh/sub/final1.js
// real_filename /Users/kun/Library/Caches/deno/deps/https/import-meta.now.sh/sub/final1.js
// real_module_name = https://import-meta.now.sh/sub/final1.js
let redirect_url = Url::parse(&redirect_to).expect("Should be valid URL");
// Recurse.
// TODO(bartlomieju): I'm pretty sure we should call `fetch_remote_source_async` here.
// Should we expect that all redirects are cached?
return self.fetch_cached_remote_source(&redirect_url);
}
// No redirect needed or end of redirects.
// We can try read the file
let filepath = self
.deps_cache
.location
.join(self.deps_cache.get_cache_filename(&module_url));
let source_code = match fs::read(filepath.clone()) {
Err(e) => { Err(e) => {
if e.kind() == std::io::ErrorKind::NotFound { if let Some(e) = e.downcast_ref::<std::io::Error>() {
return Ok(None); if e.kind() == std::io::ErrorKind::NotFound {
} else { return Ok(None);
return Err(e.into()); }
} }
return Err(e);
} }
Ok(c) => c, Ok(c) => c,
}; };
let media_type =
map_content_type(&filepath, source_code_headers.mime_type.as_deref()); let (mut source_file, headers) = result;
if let Some(redirect_to) = headers.get("location") {
let redirect_url = Url::parse(redirect_to).expect("Should be valid URL");
return self.fetch_cached_remote_source(&redirect_url);
}
let mut source_code = Vec::new();
source_file.read_to_end(&mut source_code)?;
let cache_filename = self.http_cache.get_cache_filename(module_url);
let fake_filepath = PathBuf::from(module_url.path());
let media_type = map_content_type(
&fake_filepath,
headers.get("content-type").map(|e| e.as_str()),
);
let types_url = match media_type { let types_url = match media_type {
msg::MediaType::JavaScript | msg::MediaType::JSX => get_types_url( msg::MediaType::JavaScript | msg::MediaType::JSX => get_types_url(
&module_url, &module_url,
&source_code, &source_code,
source_code_headers.x_typescript_types.as_deref(), headers.get("x-typescript-types").map(|e| e.as_str()),
), ),
_ => None, _ => None,
}; };
Ok(Some(SourceFile { Ok(Some(SourceFile {
url: module_url.clone(), url: module_url.clone(),
filename: filepath, filename: cache_filename,
media_type, media_type,
source_code, source_code,
types_url, types_url,
@ -440,8 +407,10 @@ impl SourceFileFetcher {
); );
let dir = self.clone(); let dir = self.clone();
let module_url = module_url.clone(); let module_url = module_url.clone();
let headers = self.get_source_code_headers(&module_url); let module_etag = match self.http_cache.get(&module_url) {
let module_etag = headers.etag; Ok((_, headers)) => headers.get("etag").map(String::from),
Err(_) => None,
};
let http_client = self.http_client.clone(); let http_client = self.http_client.clone();
// Single pass fetch, either yields code or yields redirect. // Single pass fetch, either yields code or yields redirect.
let f = async move { let f = async move {
@ -453,20 +422,9 @@ impl SourceFileFetcher {
Ok(source_file) Ok(source_file)
} }
FetchOnceResult::Redirect(new_module_url) => { FetchOnceResult::Redirect(new_module_url, headers) => {
// If redirects, update module_name and filename for next looped call. // If redirects, update module_name and filename for next looped call.
if let Err(e) = dir.save_source_code_headers( dir.http_cache.set(&module_url, headers, &[])?;
&module_url,
None,
Some(new_module_url.to_string()),
None,
None,
) {
return Err(source_header_cache_failed_error(
module_url.as_str(),
&e.to_string(),
));
}
// Recurse // Recurse
dir dir
@ -478,51 +436,30 @@ impl SourceFileFetcher {
) )
.await .await
} }
FetchOnceResult::Code(ResultPayload { FetchOnceResult::Code(source, headers) => {
body: source,
content_type: maybe_content_type,
etag,
x_typescript_types,
}) => {
// We land on the code. // We land on the code.
if let Err(e) = dir.save_source_code_headers( dir.http_cache.set(&module_url, headers.clone(), &source)?;
&module_url,
maybe_content_type.clone(),
None,
etag,
x_typescript_types.clone(),
) {
return Err(source_header_cache_failed_error(
module_url.as_str(),
&e.to_string(),
));
}
if let Err(e) = dir.save_source_code(&module_url, &source) { let cache_filepath = dir.http_cache.get_cache_filename(&module_url);
return Err(source_cache_failed_error( // Used to sniff out content type from file extension - probably to be removed
module_url.as_str(), let fake_filepath = PathBuf::from(module_url.path());
&e.to_string(), let media_type = map_content_type(
)); &fake_filepath,
} headers.get("content-type").map(String::as_str),
);
let filepath = dir
.deps_cache
.location
.join(dir.deps_cache.get_cache_filename(&module_url));
let media_type =
map_content_type(&filepath, maybe_content_type.as_deref());
let types_url = match media_type { let types_url = match media_type {
msg::MediaType::JavaScript | msg::MediaType::JSX => { msg::MediaType::JavaScript | msg::MediaType::JSX => get_types_url(
get_types_url(&module_url, &source, x_typescript_types.as_deref()) &module_url,
} &source,
headers.get("x-typescript-types").map(String::as_str),
),
_ => None, _ => None,
}; };
let source_file = SourceFile { let source_file = SourceFile {
url: module_url.clone(), url: module_url.clone(),
filename: filepath, filename: cache_filepath,
media_type, media_type,
source_code: source, source_code: source,
types_url, types_url,
@ -535,75 +472,6 @@ impl SourceFileFetcher {
f.boxed() f.boxed()
} }
/// Get header metadata associated with a remote file.
///
/// NOTE: chances are that the source file was downloaded due to redirects.
/// In this case, the headers file provides info about where we should go and get
/// the file that redirect eventually points to.
fn get_source_code_headers(&self, url: &Url) -> SourceCodeHeaders {
let cache_key = self
.deps_cache
.get_cache_filename_with_extension(url, "headers.json");
if let Ok(bytes) = self.deps_cache.get(&cache_key) {
if let Ok(json_string) = std::str::from_utf8(&bytes) {
return SourceCodeHeaders::from_json_string(json_string.to_string());
}
}
SourceCodeHeaders::default()
}
/// Save contents of downloaded remote file in on-disk cache for subsequent access.
fn save_source_code(&self, url: &Url, source: &[u8]) -> std::io::Result<()> {
let cache_key = self.deps_cache.get_cache_filename(url);
// May not exist. DON'T unwrap.
let _ = self.deps_cache.remove(&cache_key);
self.deps_cache.set(&cache_key, source)
}
/// Save headers related to source file to {filename}.headers.json file,
/// only when there is actually something necessary to save.
///
/// For example, if the extension ".js" already mean JS file and we have
/// content type of "text/javascript", then we would not save the mime type.
///
/// If nothing needs to be saved, the headers file is not created.
fn save_source_code_headers(
&self,
url: &Url,
mime_type: Option<String>,
redirect_to: Option<String>,
etag: Option<String>,
x_typescript_types: Option<String>,
) -> std::io::Result<()> {
let cache_key = self
.deps_cache
.get_cache_filename_with_extension(url, "headers.json");
// Remove possibly existing stale .headers.json file.
// May not exist. DON'T unwrap.
let _ = self.deps_cache.remove(&cache_key);
let headers = SourceCodeHeaders {
mime_type,
redirect_to,
etag,
x_typescript_types,
};
let cache_filename = self.deps_cache.get_cache_filename(url);
if let Ok(maybe_json_string) = headers.to_json_string(&cache_filename) {
if let Some(json_string) = maybe_json_string {
return self.deps_cache.set(&cache_key, json_string.as_bytes());
}
}
Ok(())
}
} }
fn map_file_extension(path: &Path) -> msg::MediaType { fn map_file_extension(path: &Path) -> msg::MediaType {
@ -755,92 +623,14 @@ pub struct SourceCodeHeaders {
pub x_typescript_types: Option<String>, pub x_typescript_types: Option<String>,
} }
static MIME_TYPE: &str = "mime_type";
static REDIRECT_TO: &str = "redirect_to";
static ETAG: &str = "etag";
static X_TYPESCRIPT_TYPES: &str = "x_typescript_types";
impl SourceCodeHeaders {
pub fn from_json_string(headers_string: String) -> Self {
// TODO: use serde for deserialization
let maybe_headers_json: serde_json::Result<serde_json::Value> =
serde_json::from_str(&headers_string);
if let Ok(headers_json) = maybe_headers_json {
let mime_type = headers_json[MIME_TYPE].as_str().map(String::from);
let redirect_to = headers_json[REDIRECT_TO].as_str().map(String::from);
let etag = headers_json[ETAG].as_str().map(String::from);
let x_typescript_types =
headers_json[X_TYPESCRIPT_TYPES].as_str().map(String::from);
return SourceCodeHeaders {
mime_type,
redirect_to,
etag,
x_typescript_types,
};
}
SourceCodeHeaders::default()
}
// TODO: remove this nonsense `cache_filename` param, this should be
// done when instantiating SourceCodeHeaders
pub fn to_json_string(
&self,
cache_filename: &Path,
) -> Result<Option<String>, serde_json::Error> {
// TODO(kevinkassimo): consider introduce serde::Deserialize to make things simpler.
// This is super ugly at this moment...
// Had trouble to make serde_derive work: I'm unable to build proc-macro2.
let mut value_map = serde_json::map::Map::new();
if let Some(mime_type) = &self.mime_type {
let resolved_mime_type =
map_content_type(Path::new(""), Some(mime_type.clone().as_str()));
// TODO: fix this
let ext_based_mime_type = map_file_extension(cache_filename);
// Add mime to headers only when content type is different from extension.
if ext_based_mime_type == msg::MediaType::Unknown
|| resolved_mime_type != ext_based_mime_type
{
value_map.insert(MIME_TYPE.to_string(), json!(mime_type));
}
}
if let Some(redirect_to) = &self.redirect_to {
value_map.insert(REDIRECT_TO.to_string(), json!(redirect_to));
}
if let Some(etag) = &self.etag {
value_map.insert(ETAG.to_string(), json!(etag));
}
if let Some(x_typescript_types) = &self.x_typescript_types {
value_map
.insert(X_TYPESCRIPT_TYPES.to_string(), json!(x_typescript_types));
}
if value_map.is_empty() {
return Ok(None);
}
serde_json::to_string(&value_map)
.and_then(|serialized| Ok(Some(serialized)))
}
}
#[cfg(test)] #[cfg(test)]
mod tests { mod tests {
use super::*; use super::*;
use crate::fs as deno_fs;
use tempfile::TempDir; use tempfile::TempDir;
fn setup_file_fetcher(dir_path: &Path) -> SourceFileFetcher { fn setup_file_fetcher(dir_path: &Path) -> SourceFileFetcher {
SourceFileFetcher::new( SourceFileFetcher::new(
DiskCache::new(&dir_path.to_path_buf().join("deps")), HttpCache::new(&dir_path.to_path_buf().join("deps")).unwrap(),
true, true,
vec![], vec![],
false, false,
@ -925,49 +715,6 @@ mod tests {
assert_eq!(check_cache_blacklist(&u, &args), true); assert_eq!(check_cache_blacklist(&u, &args), true);
} }
#[test]
fn test_source_code_headers_get_and_save() {
let (_temp_dir, fetcher) = test_setup();
let url = Url::parse("http://example.com/f.js").unwrap();
let headers_filepath = fetcher.deps_cache.location.join(
fetcher
.deps_cache
.get_cache_filename_with_extension(&url, "headers.json"),
);
if let Some(ref parent) = headers_filepath.parent() {
fs::create_dir_all(parent).unwrap();
};
let _ = deno_fs::write_file(
headers_filepath.as_path(),
"{\"mime_type\":\"text/javascript\",\"redirect_to\":\"http://example.com/a.js\"}",
0o666,
);
let headers = fetcher.get_source_code_headers(&url);
assert_eq!(headers.mime_type.clone().unwrap(), "text/javascript");
assert_eq!(headers.redirect_to.unwrap(), "http://example.com/a.js");
assert_eq!(headers.etag, None);
assert_eq!(headers.x_typescript_types, None);
let _ = fetcher.save_source_code_headers(
&url,
Some("text/typescript".to_owned()),
Some("http://deno.land/a.js".to_owned()),
Some("W/\"04572f4749af993f4961a7e5daa1e4d5\"".to_owned()),
Some("./a.d.ts".to_owned()),
);
let headers2 = fetcher.get_source_code_headers(&url);
assert_eq!(headers2.mime_type.clone().unwrap(), "text/typescript");
assert_eq!(headers2.redirect_to.unwrap(), "http://deno.land/a.js");
assert_eq!(
headers2.etag.unwrap(),
"W/\"04572f4749af993f4961a7e5daa1e4d5\""
);
assert_eq!(headers2.x_typescript_types.unwrap(), "./a.d.ts")
}
#[test] #[test]
fn test_fetch_local_file_no_panic() { fn test_fetch_local_file_no_panic() {
let (_temp_dir, fetcher) = test_setup(); let (_temp_dir, fetcher) = test_setup();
@ -992,14 +739,12 @@ mod tests {
Url::parse("http://localhost:4545/cli/tests/subdir/mod2.ts").unwrap(); Url::parse("http://localhost:4545/cli/tests/subdir/mod2.ts").unwrap();
let module_url_1 = module_url.clone(); let module_url_1 = module_url.clone();
let module_url_2 = module_url.clone(); let module_url_2 = module_url.clone();
let headers_file_name = fetcher.deps_cache.location.join( let headers_file_name = fetcher
fetcher .http_cache
.deps_cache .get_cache_filename(&module_url)
.get_cache_filename_with_extension(&module_url, "headers.json"), .with_extension("headers.json");
);
let headers_file_name_1 = headers_file_name.clone(); let headers_file_name_1 = headers_file_name.clone();
let headers_file_name_2 = headers_file_name.clone(); let headers_file_name_2 = headers_file_name.clone();
let headers_file_name_3 = headers_file_name;
let result = fetcher let result = fetcher
.get_source_file_async(&module_url, true, false, false) .get_source_file_async(&module_url, true, false, false)
@ -1011,13 +756,12 @@ mod tests {
&b"export { printHello } from \"./print_hello.ts\";\n"[..] &b"export { printHello } from \"./print_hello.ts\";\n"[..]
); );
assert_eq!(&(r.media_type), &msg::MediaType::TypeScript); assert_eq!(&(r.media_type), &msg::MediaType::TypeScript);
// Should not create .headers.json file due to matching ext assert!(fs::read_to_string(&headers_file_name_1).is_ok());
assert!(fs::read_to_string(&headers_file_name_1).is_err());
// Modify .headers.json, write using fs write and read using save_source_code_headers // Modify .headers.json, write using fs write
let _ = fs::write( let _ = fs::write(
&headers_file_name_1, &headers_file_name_1,
"{ \"mime_type\": \"text/javascript\" }", "{ \"content-type\": \"text/javascript\" }",
); );
let result2 = fetcher_1 let result2 = fetcher_1
.get_source_file_async(&module_url, true, false, false) .get_source_file_async(&module_url, true, false, false)
@ -1031,21 +775,14 @@ mod tests {
// If get_source_file_async does not call remote, this should be JavaScript // If get_source_file_async does not call remote, this should be JavaScript
// as we modified before! (we do not overwrite .headers.json due to no http fetch) // as we modified before! (we do not overwrite .headers.json due to no http fetch)
assert_eq!(&(r2.media_type), &msg::MediaType::JavaScript); assert_eq!(&(r2.media_type), &msg::MediaType::JavaScript);
assert_eq!( let (_, headers) = fetcher_2.http_cache.get(&module_url_1).unwrap();
fetcher_2
.get_source_code_headers(&module_url_1) assert_eq!(headers.get("content-type").unwrap(), "text/javascript");
.mime_type
.unwrap(),
"text/javascript"
);
// Modify .headers.json again, but the other way around // Modify .headers.json again, but the other way around
let _ = fetcher_2.save_source_code_headers( let _ = fs::write(
&module_url_1, &headers_file_name_1,
Some("application/json".to_owned()), "{ \"content-type\": \"application/json\" }",
None,
None,
None,
); );
let result3 = fetcher_2 let result3 = fetcher_2
.get_source_file_async(&module_url_1, true, false, false) .get_source_file_async(&module_url_1, true, false, false)
@ -1073,9 +810,8 @@ mod tests {
let r4 = result4.unwrap(); let r4 = result4.unwrap();
let expected4 = &b"export { printHello } from \"./print_hello.ts\";\n"[..]; let expected4 = &b"export { printHello } from \"./print_hello.ts\";\n"[..];
assert_eq!(r4.source_code, expected4); assert_eq!(r4.source_code, expected4);
// Now the old .headers.json file should have gone! Resolved back to TypeScript // Resolved back to TypeScript
assert_eq!(&(r4.media_type), &msg::MediaType::TypeScript); assert_eq!(&(r4.media_type), &msg::MediaType::TypeScript);
assert!(fs::read_to_string(&headers_file_name_3).is_err());
drop(http_server_guard); drop(http_server_guard);
} }
@ -1084,17 +820,14 @@ mod tests {
async fn test_get_source_code_2() { async fn test_get_source_code_2() {
let http_server_guard = crate::test_util::http_server(); let http_server_guard = crate::test_util::http_server();
let (temp_dir, fetcher) = test_setup(); let (temp_dir, fetcher) = test_setup();
let fetcher_1 = fetcher.clone();
let module_url = let module_url =
Url::parse("http://localhost:4545/cli/tests/subdir/mismatch_ext.ts") Url::parse("http://localhost:4545/cli/tests/subdir/mismatch_ext.ts")
.unwrap(); .unwrap();
let module_url_1 = module_url.clone(); let module_url_1 = module_url.clone();
let module_url_2 = module_url.clone(); let headers_file_name = fetcher
let headers_file_name = fetcher.deps_cache.location.join( .http_cache
fetcher .get_cache_filename(&module_url)
.deps_cache .with_extension("headers.json");
.get_cache_filename_with_extension(&module_url, "headers.json"),
);
let result = fetcher let result = fetcher
.get_source_file_async(&module_url, true, false, false) .get_source_file_async(&module_url, true, false, false)
@ -1103,23 +836,14 @@ mod tests {
let r = result.unwrap(); let r = result.unwrap();
let expected = b"export const loaded = true;\n"; let expected = b"export const loaded = true;\n";
assert_eq!(r.source_code, expected); assert_eq!(r.source_code, expected);
// Mismatch ext with content type, create .headers.json
assert_eq!(&(r.media_type), &msg::MediaType::JavaScript); assert_eq!(&(r.media_type), &msg::MediaType::JavaScript);
assert_eq!( let (_, headers) = fetcher.http_cache.get(&module_url).unwrap();
fetcher assert_eq!(headers.get("content-type").unwrap(), "text/javascript");
.get_source_code_headers(&module_url)
.mime_type
.unwrap(),
"text/javascript"
);
// Modify .headers.json // Modify .headers.json
let _ = fetcher.save_source_code_headers( let _ = fs::write(
&module_url, &headers_file_name,
Some("text/typescript".to_owned()), "{ \"content-type\": \"text/typescript\" }",
None,
None,
None,
); );
let result2 = fetcher let result2 = fetcher
.get_source_file_async(&module_url, true, false, false) .get_source_file_async(&module_url, true, false, false)
@ -1132,9 +856,12 @@ mod tests {
// as we modified before! (we do not overwrite .headers.json due to no http // as we modified before! (we do not overwrite .headers.json due to no http
// fetch) // fetch)
assert_eq!(&(r2.media_type), &msg::MediaType::TypeScript); assert_eq!(&(r2.media_type), &msg::MediaType::TypeScript);
assert!(fs::read_to_string(&headers_file_name).is_err()); assert_eq!(
fs::read_to_string(&headers_file_name).unwrap(),
"{ \"content-type\": \"text/typescript\" }",
);
// let's create fresh instance of DenoDir (simulating another freshh Deno // let's create fresh instance of DenoDir (simulating another fresh Deno
// process) and don't use cache // process) and don't use cache
let fetcher = setup_file_fetcher(temp_dir.path()); let fetcher = setup_file_fetcher(temp_dir.path());
let result3 = fetcher let result3 = fetcher
@ -1147,13 +874,8 @@ mod tests {
// Now the old .headers.json file should be overwritten back to JavaScript! // Now the old .headers.json file should be overwritten back to JavaScript!
// (due to http fetch) // (due to http fetch)
assert_eq!(&(r3.media_type), &msg::MediaType::JavaScript); assert_eq!(&(r3.media_type), &msg::MediaType::JavaScript);
assert_eq!( let (_, headers) = fetcher.http_cache.get(&module_url).unwrap();
fetcher_1 assert_eq!(headers.get("content-type").unwrap(), "text/javascript");
.get_source_code_headers(&module_url_2)
.mime_type
.unwrap(),
"text/javascript"
);
drop(http_server_guard); drop(http_server_guard);
} }
@ -1166,11 +888,10 @@ mod tests {
"http://localhost:4545/cli/tests/subdir/mismatch_ext.ts", "http://localhost:4545/cli/tests/subdir/mismatch_ext.ts",
) )
.unwrap(); .unwrap();
let headers_file_name = fetcher.deps_cache.location.join( let headers_file_name = fetcher
fetcher .http_cache
.deps_cache .get_cache_filename(specifier.as_url())
.get_cache_filename_with_extension(specifier.as_url(), "headers.json"), .with_extension("headers.json");
);
// first download // first download
let r = fetcher.fetch_source_file_async(&specifier, None).await; let r = fetcher.fetch_source_file_async(&specifier, None).await;
@ -1210,20 +931,16 @@ mod tests {
"http://localhost:4546/cli/tests/subdir/redirects/redirect1.js", "http://localhost:4546/cli/tests/subdir/redirects/redirect1.js",
) )
.unwrap(); .unwrap();
let redirect_source_filepath = fetcher let redirect_source_filepath =
.deps_cache fetcher.http_cache.get_cache_filename(&redirect_module_url);
.location
.join("http/localhost_PORT4546/cli/tests/subdir/redirects/redirect1.js");
let redirect_source_filename = let redirect_source_filename =
redirect_source_filepath.to_str().unwrap().to_string(); redirect_source_filepath.to_str().unwrap().to_string();
let target_module_url = Url::parse( let target_module_url = Url::parse(
"http://localhost:4545/cli/tests/subdir/redirects/redirect1.js", "http://localhost:4545/cli/tests/subdir/redirects/redirect1.js",
) )
.unwrap(); .unwrap();
let redirect_target_filepath = fetcher let redirect_target_filepath =
.deps_cache fetcher.http_cache.get_cache_filename(&target_module_url);
.location
.join("http/localhost_PORT4545/cli/tests/subdir/redirects/redirect1.js");
let redirect_target_filename = let redirect_target_filename =
redirect_target_filepath.to_str().unwrap().to_string(); redirect_target_filepath.to_str().unwrap().to_string();
@ -1233,13 +950,11 @@ mod tests {
.await; .await;
assert!(result.is_ok()); assert!(result.is_ok());
let mod_meta = result.unwrap(); let mod_meta = result.unwrap();
// File that requires redirection is not downloaded. // File that requires redirection should be empty file.
assert!(fs::read_to_string(&redirect_source_filename).is_err()); assert_eq!(fs::read_to_string(&redirect_source_filename).unwrap(), "");
// ... but its .headers.json is created. let (_, headers) = fetcher.http_cache.get(&redirect_module_url).unwrap();
let redirect_source_headers =
fetcher.get_source_code_headers(&redirect_module_url);
assert_eq!( assert_eq!(
redirect_source_headers.redirect_to.unwrap(), headers.get("location").unwrap(),
"http://localhost:4545/cli/tests/subdir/redirects/redirect1.js" "http://localhost:4545/cli/tests/subdir/redirects/redirect1.js"
); );
// The target of redirection is downloaded instead. // The target of redirection is downloaded instead.
@ -1247,10 +962,8 @@ mod tests {
fs::read_to_string(&redirect_target_filename).unwrap(), fs::read_to_string(&redirect_target_filename).unwrap(),
"export const redirect = 1;\n" "export const redirect = 1;\n"
); );
let redirect_target_headers = let (_, headers) = fetcher.http_cache.get(&target_module_url).unwrap();
fetcher.get_source_code_headers(&target_module_url); assert!(headers.get("location").is_none());
assert!(redirect_target_headers.redirect_to.is_none());
// Examine the meta result. // Examine the meta result.
assert_eq!(mod_meta.url, target_module_url); assert_eq!(mod_meta.url, target_module_url);
@ -1265,28 +978,20 @@ mod tests {
"http://localhost:4548/cli/tests/subdir/redirects/redirect1.js", "http://localhost:4548/cli/tests/subdir/redirects/redirect1.js",
) )
.unwrap(); .unwrap();
let double_redirect_path = fetcher let double_redirect_path =
.deps_cache fetcher.http_cache.get_cache_filename(&double_redirect_url);
.location
.join("http/localhost_PORT4548/cli/tests/subdir/redirects/redirect1.js");
let redirect_url = Url::parse( let redirect_url = Url::parse(
"http://localhost:4546/cli/tests/subdir/redirects/redirect1.js", "http://localhost:4546/cli/tests/subdir/redirects/redirect1.js",
) )
.unwrap(); .unwrap();
let redirect_path = fetcher let redirect_path = fetcher.http_cache.get_cache_filename(&redirect_url);
.deps_cache
.location
.join("http/localhost_PORT4546/cli/tests/subdir/redirects/redirect1.js");
let target_url = Url::parse( let target_url = Url::parse(
"http://localhost:4545/cli/tests/subdir/redirects/redirect1.js", "http://localhost:4545/cli/tests/subdir/redirects/redirect1.js",
) )
.unwrap(); .unwrap();
let target_path = fetcher let target_path = fetcher.http_cache.get_cache_filename(&target_url);
.deps_cache
.location
.join("http/localhost_PORT4545/cli/tests/subdir/redirects/redirect1.js");
// Test double redirects and headers recording // Test double redirects and headers recording
let result = fetcher let result = fetcher
@ -1294,28 +999,22 @@ mod tests {
.await; .await;
assert!(result.is_ok()); assert!(result.is_ok());
let mod_meta = result.unwrap(); let mod_meta = result.unwrap();
assert!(fs::read_to_string(&double_redirect_path).is_err()); assert_eq!(fs::read_to_string(&double_redirect_path).unwrap(), "");
assert!(fs::read_to_string(&redirect_path).is_err()); assert_eq!(fs::read_to_string(&redirect_path).unwrap(), "");
let double_redirect_headers = let (_, headers) = fetcher.http_cache.get(&double_redirect_url).unwrap();
fetcher.get_source_code_headers(&double_redirect_url); assert_eq!(headers.get("location").unwrap(), &redirect_url.to_string());
assert_eq!(
double_redirect_headers.redirect_to.unwrap(), let (_, headers) = fetcher.http_cache.get(&redirect_url).unwrap();
redirect_url.to_string() assert_eq!(headers.get("location").unwrap(), &target_url.to_string());
);
let redirect_headers = fetcher.get_source_code_headers(&redirect_url);
assert_eq!(
redirect_headers.redirect_to.unwrap(),
target_url.to_string()
);
// The target of redirection is downloaded instead. // The target of redirection is downloaded instead.
assert_eq!( assert_eq!(
fs::read_to_string(&target_path).unwrap(), fs::read_to_string(&target_path).unwrap(),
"export const redirect = 1;\n" "export const redirect = 1;\n"
); );
let redirect_target_headers = fetcher.get_source_code_headers(&target_url); let (_, headers) = fetcher.http_cache.get(&target_url).unwrap();
assert!(redirect_target_headers.redirect_to.is_none()); assert!(headers.get("location").is_none());
// Examine the meta result. // Examine the meta result.
assert_eq!(mod_meta.url, target_url); assert_eq!(mod_meta.url, target_url);
@ -1338,10 +1037,7 @@ mod tests {
) )
.unwrap(); .unwrap();
let target_path = fetcher let target_path = fetcher.http_cache.get_cache_filename(&redirect_url);
.deps_cache
.location
.join("http/localhost_PORT4545/cli/tests/subdir/redirects/redirect1.js");
let target_path_ = target_path.clone(); let target_path_ = target_path.clone();
// Test that redirect target is not downloaded twice for different redirect source. // Test that redirect target is not downloaded twice for different redirect source.
@ -1459,12 +1155,10 @@ mod tests {
let module_url = let module_url =
Url::parse("http://127.0.0.1:4545/cli/tests/subdir/mt_video_mp2t.t3.ts") Url::parse("http://127.0.0.1:4545/cli/tests/subdir/mt_video_mp2t.t3.ts")
.unwrap(); .unwrap();
let headers_file_name = fetcher.deps_cache.location.join( let headers_file_name = fetcher
fetcher .http_cache
.deps_cache .get_cache_filename(&module_url)
.get_cache_filename_with_extension(&module_url, "headers.json"), .with_extension("headers.json");
);
let result = fetcher let result = fetcher
.fetch_remote_source_async(&module_url, false, false, 10) .fetch_remote_source_async(&module_url, false, false, 10)
.await; .await;
@ -1472,15 +1166,10 @@ mod tests {
let r = result.unwrap(); let r = result.unwrap();
assert_eq!(r.source_code, b"export const loaded = true;\n"); assert_eq!(r.source_code, b"export const loaded = true;\n");
assert_eq!(&(r.media_type), &msg::MediaType::TypeScript); assert_eq!(&(r.media_type), &msg::MediaType::TypeScript);
// matching ext, no .headers.json file created
assert!(fs::read_to_string(&headers_file_name).is_err());
// Modify .headers.json, make sure read from local // Modify .headers.json, make sure read from local
let _ = fetcher.save_source_code_headers( let _ = fs::write(
&module_url, &headers_file_name,
Some("text/javascript".to_owned()), "{ \"content-type\": \"text/javascript\" }",
None,
None,
None,
); );
let result2 = fetcher.fetch_cached_remote_source(&module_url); let result2 = fetcher.fetch_cached_remote_source(&module_url);
assert!(result2.is_ok()); assert!(result2.is_ok());
@ -1500,11 +1189,6 @@ mod tests {
let module_url = let module_url =
Url::parse("http://localhost:4545/cli/tests/subdir/mt_video_mp2t.t3.ts") Url::parse("http://localhost:4545/cli/tests/subdir/mt_video_mp2t.t3.ts")
.unwrap(); .unwrap();
let headers_file_name = fetcher.deps_cache.location.join(
fetcher
.deps_cache
.get_cache_filename_with_extension(&module_url, "headers.json"),
);
let result = fetcher let result = fetcher
.fetch_remote_source_async(&module_url, false, false, 10) .fetch_remote_source_async(&module_url, false, false, 10)
@ -1513,16 +1197,14 @@ mod tests {
let r = result.unwrap(); let r = result.unwrap();
assert_eq!(r.source_code, b"export const loaded = true;\n"); assert_eq!(r.source_code, b"export const loaded = true;\n");
assert_eq!(&(r.media_type), &msg::MediaType::TypeScript); assert_eq!(&(r.media_type), &msg::MediaType::TypeScript);
// matching ext, no .headers.json file created
assert!(fs::read_to_string(&headers_file_name).is_err());
// Modify .headers.json, make sure read from local // Modify .headers.json, make sure read from local
let _ = fetcher.save_source_code_headers( let headers_file_name = fetcher
&module_url, .http_cache
Some("text/javascript".to_owned()), .get_cache_filename(&module_url)
None, .with_extension("headers.json");
None, let _ = fs::write(
None, &headers_file_name,
"{ \"content-type\": \"text/javascript\" }",
); );
let result2 = fetcher.fetch_cached_remote_source(&module_url); let result2 = fetcher.fetch_cached_remote_source(&module_url);
assert!(result2.is_ok()); assert!(result2.is_ok());
@ -1540,7 +1222,6 @@ mod tests {
let (_temp_dir, fetcher) = test_setup(); let (_temp_dir, fetcher) = test_setup();
let fetcher_1 = fetcher.clone(); let fetcher_1 = fetcher.clone();
let fetcher_2 = fetcher.clone(); let fetcher_2 = fetcher.clone();
let fetcher_3 = fetcher.clone();
let module_url = let module_url =
Url::parse("http://localhost:4545/cli/tests/subdir/no_ext").unwrap(); Url::parse("http://localhost:4545/cli/tests/subdir/no_ext").unwrap();
let module_url_2 = let module_url_2 =
@ -1559,14 +1240,8 @@ mod tests {
let r = result.unwrap(); let r = result.unwrap();
assert_eq!(r.source_code, b"export const loaded = true;\n"); assert_eq!(r.source_code, b"export const loaded = true;\n");
assert_eq!(&(r.media_type), &msg::MediaType::TypeScript); assert_eq!(&(r.media_type), &msg::MediaType::TypeScript);
// no ext, should create .headers.json file let (_, headers) = fetcher.http_cache.get(&module_url).unwrap();
assert_eq!( assert_eq!(headers.get("content-type").unwrap(), "text/typescript");
fetcher_1
.get_source_code_headers(&module_url)
.mime_type
.unwrap(),
"text/typescript"
);
let result = fetcher_1 let result = fetcher_1
.fetch_remote_source_async(&module_url_2, false, false, 10) .fetch_remote_source_async(&module_url_2, false, false, 10)
.await; .await;
@ -1574,14 +1249,9 @@ mod tests {
let r2 = result.unwrap(); let r2 = result.unwrap();
assert_eq!(r2.source_code, b"export const loaded = true;\n"); assert_eq!(r2.source_code, b"export const loaded = true;\n");
assert_eq!(&(r2.media_type), &msg::MediaType::JavaScript); assert_eq!(&(r2.media_type), &msg::MediaType::JavaScript);
// mismatch ext, should create .headers.json file let (_, headers) = fetcher.http_cache.get(&module_url_2_).unwrap();
assert_eq!( assert_eq!(headers.get("content-type").unwrap(), "text/javascript");
fetcher_2
.get_source_code_headers(&module_url_2_)
.mime_type
.unwrap(),
"text/javascript"
);
// test unknown extension // test unknown extension
let result = fetcher_2 let result = fetcher_2
.fetch_remote_source_async(&module_url_3, false, false, 10) .fetch_remote_source_async(&module_url_3, false, false, 10)
@ -1590,14 +1260,8 @@ mod tests {
let r3 = result.unwrap(); let r3 = result.unwrap();
assert_eq!(r3.source_code, b"export const loaded = true;\n"); assert_eq!(r3.source_code, b"export const loaded = true;\n");
assert_eq!(&(r3.media_type), &msg::MediaType::TypeScript); assert_eq!(&(r3.media_type), &msg::MediaType::TypeScript);
// unknown ext, should create .headers.json file let (_, headers) = fetcher.http_cache.get(&module_url_3_).unwrap();
assert_eq!( assert_eq!(headers.get("content-type").unwrap(), "text/typescript");
fetcher_3
.get_source_code_headers(&module_url_3_)
.mime_type
.unwrap(),
"text/typescript"
);
drop(http_server_guard); drop(http_server_guard);
} }
@ -1904,23 +1568,21 @@ mod tests {
assert_eq!(source.source_code, b"console.log('etag')"); assert_eq!(source.source_code, b"console.log('etag')");
assert_eq!(&(source.media_type), &msg::MediaType::TypeScript); assert_eq!(&(source.media_type), &msg::MediaType::TypeScript);
let headers = fetcher.get_source_code_headers(&module_url); let (_, headers) = fetcher.http_cache.get(&module_url).unwrap();
assert_eq!(headers.etag, Some("33a64df551425fcc55e".to_string())); assert_eq!(headers.get("etag").unwrap(), "33a64df551425fcc55e");
let header_path = fetcher.deps_cache.location.join( let header_path = fetcher
fetcher .http_cache
.deps_cache .get_cache_filename(&module_url)
.get_cache_filename_with_extension(&module_url, "headers.json"), .with_extension("headers.json");
);
let modified1 = header_path.metadata().unwrap().modified().unwrap(); let modified1 = header_path.metadata().unwrap().modified().unwrap();
// Forcibly change the contents of the cache file and request // Forcibly change the contents of the cache file and request
// it again with the cache parameters turned off. // it again with the cache parameters turned off.
// If the fetched content changes, the cached content is used. // If the fetched content changes, the cached content is used.
fetcher let file_name = fetcher.http_cache.get_cache_filename(&module_url);
.save_source_code(&module_url, b"changed content") let _ = fs::write(&file_name, "changed content");
.unwrap();
let cached_source = fetcher let cached_source = fetcher
.fetch_remote_source_async(&module_url, false, false, 1) .fetch_remote_source_async(&module_url, false, false, 1)
.await .await

5
cli/global_state.rs
View file

@ -9,6 +9,7 @@ use crate::deno_dir;
use crate::deno_error::permission_denied; use crate::deno_error::permission_denied;
use crate::file_fetcher::SourceFileFetcher; use crate::file_fetcher::SourceFileFetcher;
use crate::flags; use crate::flags;
use crate::http_cache;
use crate::lockfile::Lockfile; use crate::lockfile::Lockfile;
use crate::msg; use crate::msg;
use crate::permissions::DenoPermissions; use crate::permissions::DenoPermissions;
@ -58,9 +59,11 @@ impl GlobalState {
pub fn new(flags: flags::DenoFlags) -> Result<Self, ErrBox> { pub fn new(flags: flags::DenoFlags) -> Result<Self, ErrBox> {
let custom_root = env::var("DENO_DIR").map(String::into).ok(); let custom_root = env::var("DENO_DIR").map(String::into).ok();
let dir = deno_dir::DenoDir::new(custom_root)?; let dir = deno_dir::DenoDir::new(custom_root)?;
let deps_cache_location = dir.root.join("deps");
let http_cache = http_cache::HttpCache::new(&deps_cache_location)?;
let file_fetcher = SourceFileFetcher::new( let file_fetcher = SourceFileFetcher::new(
dir.deps_cache.clone(), http_cache,
!flags.reload, !flags.reload,
flags.cache_blacklist.clone(), flags.cache_blacklist.clone(),
flags.no_remote, flags.no_remote,

195
cli/http_cache.rs Normal file
View file

@ -0,0 +1,195 @@
// Copyright 2018-2020 the Deno authors. All rights reserved. MIT license.
/// This module is meant to eventually implement HTTP cache
/// as defined in RFC 7234 (https://tools.ietf.org/html/rfc7234).
/// Currently it's a very simplified version to fulfill Deno needs
/// at hand.
use crate::fs as deno_fs;
use crate::http_util::HeadersMap;
use deno_core::ErrBox;
use std::fs;
use std::fs::File;
use std::path::Path;
use std::path::PathBuf;
use url::Url;
/// Turn base of url (scheme, hostname, port) into a valid filename.
/// This method replaces port part with a special string token (because
/// ":" cannot be used in filename on some platforms).
pub fn base_url_to_filename(url: &Url) -> PathBuf {
let mut out = PathBuf::new();
let scheme = url.scheme();
out.push(scheme);
match scheme {
"http" | "https" => {
let host = url.host_str().unwrap();
let host_port = match url.port() {
Some(port) => format!("{}_PORT{}", host, port),
None => host.to_string(),
};
out.push(host_port);
}
scheme => {
unimplemented!(
"Don't know how to create cache name for scheme: {}",
scheme
);
}
};
out
}
/// Turn provided `url` into a hashed filename.
/// URLs can contain a lot of characters that cannot be used
/// in filenames (like "?", "#", ":"), so in order to cache
/// them properly they are deterministically hashed into ASCII
/// strings.
///
/// NOTE: this method is `pub` because it's used in integration_tests
pub fn url_to_filename(url: &Url) -> PathBuf {
let mut cache_filename = base_url_to_filename(url);
let mut rest_str = url.path().to_string();
if let Some(query) = url.query() {
rest_str.push_str("?");
rest_str.push_str(query);
}
// NOTE: fragment is omitted on purpose - it's not taken into
// account when caching - it denotes parts of webpage, which
// in case of static resources doesn't make much sense
let hashed_filename = crate::checksum::gen(vec![rest_str.as_bytes()]);
cache_filename.push(hashed_filename);
cache_filename
}
#[derive(Clone)]
pub struct HttpCache {
pub location: PathBuf,
}
impl HttpCache {
/// Returns error if unable to create directory
/// at specified location.
pub fn new(location: &Path) -> Result<Self, ErrBox> {
fs::create_dir_all(&location)?;
Ok(Self {
location: location.to_owned(),
})
}
pub(crate) fn get_cache_filename(&self, url: &Url) -> PathBuf {
self.location.join(url_to_filename(url))
}
// TODO(bartlomieju): this method should check headers file
// and validate against ETAG/Last-modified-as headers.
// ETAG check is currently done in `cli/file_fetcher.rs`.
pub fn get(&self, url: &Url) -> Result<(File, HeadersMap), ErrBox> {
let cache_filename = self.location.join(url_to_filename(url));
let headers_filename = cache_filename.with_extension("headers.json");
let file = File::open(cache_filename)?;
let headers_json = fs::read_to_string(headers_filename)?;
let headers_map: HeadersMap = serde_json::from_str(&headers_json)?;
Ok((file, headers_map))
}
pub fn set(
&self,
url: &Url,
headers_map: HeadersMap,
content: &[u8],
) -> Result<(), ErrBox> {
let cache_filename = self.location.join(url_to_filename(url));
let headers_filename = cache_filename.with_extension("headers.json");
// Create parent directory
let parent_filename = cache_filename
.parent()
.expect("Cache filename should have a parent dir");
fs::create_dir_all(parent_filename)?;
// Cache content
deno_fs::write_file(&cache_filename, content, 0o666)?;
let serialized_headers = serde_json::to_string(&headers_map)?;
// Cache headers
deno_fs::write_file(&headers_filename, serialized_headers, 0o666)?;
Ok(())
}
}
#[cfg(test)]
mod tests {
use super::*;
use std::collections::HashMap;
use std::io::Read;
use tempfile::TempDir;
#[test]
fn test_create_cache() {
let dir = TempDir::new().unwrap();
let mut cache_path = dir.path().to_owned();
cache_path.push("foobar");
let r = HttpCache::new(&cache_path);
assert!(r.is_ok());
assert!(cache_path.is_dir());
}
#[test]
fn test_get_set() {
let dir = TempDir::new().unwrap();
let cache = HttpCache::new(dir.path()).unwrap();
let url = Url::parse("https://deno.land/x/welcome.ts").unwrap();
let mut headers = HashMap::new();
headers.insert(
"content-type".to_string(),
"application/javascript".to_string(),
);
headers.insert("etag".to_string(), "as5625rqdsfb".to_string());
let content = b"Hello world";
let r = cache.set(&url, headers, content);
eprintln!("result {:?}", r);
assert!(r.is_ok());
let r = cache.get(&url);
assert!(r.is_ok());
let (mut file, headers) = r.unwrap();
let mut content = String::new();
file.read_to_string(&mut content).unwrap();
assert_eq!(content, "Hello world");
assert_eq!(
headers.get("content-type").unwrap(),
"application/javascript"
);
assert_eq!(headers.get("etag").unwrap(), "as5625rqdsfb");
assert_eq!(headers.get("foobar"), None);
drop(dir);
}
#[test]
fn test_url_to_filename() {
let test_cases = [
("https://deno.land/x/foo.ts", "https/deno.land/2c0a064891b9e3fbe386f5d4a833bce5076543f5404613656042107213a7bbc8"),
(
"https://deno.land:8080/x/foo.ts",
"https/deno.land_PORT8080/2c0a064891b9e3fbe386f5d4a833bce5076543f5404613656042107213a7bbc8",
),
("https://deno.land/", "https/deno.land/8a5edab282632443219e051e4ade2d1d5bbc671c781051bf1437897cbdfea0f1"),
(
"https://deno.land/?asdf=qwer",
"https/deno.land/e4edd1f433165141015db6a823094e6bd8f24dd16fe33f2abd99d34a0a21a3c0",
),
// should be the same as case above, fragment (#qwer) is ignored
// when hashing
(
"https://deno.land/?asdf=qwer#qwer",
"https/deno.land/e4edd1f433165141015db6a823094e6bd8f24dd16fe33f2abd99d34a0a21a3c0",
),
];
for (url, expected) in test_cases.iter() {
let u = Url::parse(url).unwrap();
let p = url_to_filename(&u);
assert_eq!(p, PathBuf::from(expected));
}
}
}

176
cli/http_util.rs
View file

@ -12,8 +12,6 @@ use reqwest::header::HeaderMap;
use reqwest::header::HeaderValue; use reqwest::header::HeaderValue;
use reqwest::header::ACCEPT_ENCODING; use reqwest::header::ACCEPT_ENCODING;
use reqwest::header::CONTENT_ENCODING; use reqwest::header::CONTENT_ENCODING;
use reqwest::header::CONTENT_TYPE;
use reqwest::header::ETAG;
use reqwest::header::IF_NONE_MATCH; use reqwest::header::IF_NONE_MATCH;
use reqwest::header::LOCATION; use reqwest::header::LOCATION;
use reqwest::header::USER_AGENT; use reqwest::header::USER_AGENT;
@ -22,6 +20,7 @@ use reqwest::Client;
use reqwest::Response; use reqwest::Response;
use reqwest::StatusCode; use reqwest::StatusCode;
use std::cmp::min; use std::cmp::min;
use std::collections::HashMap;
use std::fs::File; use std::fs::File;
use std::future::Future; use std::future::Future;
use std::io; use std::io;
@ -86,19 +85,13 @@ fn resolve_url_from_location(base_url: &Url, location: &str) -> Url {
} }
} }
#[derive(Debug, PartialEq)] pub type HeadersMap = HashMap<String, String>;
pub struct ResultPayload {
pub body: Vec<u8>,
pub content_type: Option<String>,
pub etag: Option<String>,
pub x_typescript_types: Option<String>,
}
#[derive(Debug, PartialEq)] #[derive(Debug, PartialEq)]
pub enum FetchOnceResult { pub enum FetchOnceResult {
Code(ResultPayload), Code(Vec<u8>, HeadersMap),
NotModified, NotModified,
Redirect(Url), Redirect(Url, HeadersMap),
} }
/// Asynchronously fetches the given HTTP URL one pass only. /// Asynchronously fetches the given HTTP URL one pass only.
@ -128,6 +121,19 @@ pub fn fetch_once(
return Ok(FetchOnceResult::NotModified); return Ok(FetchOnceResult::NotModified);
} }
let mut headers_: HashMap<String, String> = HashMap::new();
let headers = response.headers();
for key in headers.keys() {
let key_str = key.to_string();
let values = headers.get_all(key);
let values_str = values
.iter()
.map(|e| e.to_str().unwrap().to_string())
.collect::<Vec<String>>()
.join(",");
headers_.insert(key_str, values_str);
}
if response.status().is_redirection() { if response.status().is_redirection() {
let location_string = response let location_string = response
.headers() .headers()
@ -138,7 +144,7 @@ pub fn fetch_once(
debug!("Redirecting to {:?}...", &location_string); debug!("Redirecting to {:?}...", &location_string);
let new_url = resolve_url_from_location(&url, location_string); let new_url = resolve_url_from_location(&url, location_string);
return Ok(FetchOnceResult::Redirect(new_url)); return Ok(FetchOnceResult::Redirect(new_url, headers_));
} }
if response.status().is_client_error() if response.status().is_client_error()
@ -151,31 +157,11 @@ pub fn fetch_once(
return Err(err.into()); return Err(err.into());
} }
let content_type = response
.headers()
.get(CONTENT_TYPE)
.map(|content_type| content_type.to_str().unwrap().to_owned());
let etag = response
.headers()
.get(ETAG)
.map(|etag| etag.to_str().unwrap().to_owned());
let content_encoding = response let content_encoding = response
.headers() .headers()
.get(CONTENT_ENCODING) .get(CONTENT_ENCODING)
.map(|content_encoding| content_encoding.to_str().unwrap().to_owned()); .map(|content_encoding| content_encoding.to_str().unwrap().to_owned());
const X_TYPESCRIPT_TYPES: &str = "X-TypeScript-Types";
let x_typescript_types =
response
.headers()
.get(X_TYPESCRIPT_TYPES)
.map(|x_typescript_types| {
x_typescript_types.to_str().unwrap().to_owned()
});
let body; let body;
if let Some(content_encoding) = content_encoding { if let Some(content_encoding) = content_encoding {
body = match content_encoding { body = match content_encoding {
@ -192,12 +178,7 @@ pub fn fetch_once(
body = response.bytes().await?.to_vec(); body = response.bytes().await?.to_vec();
} }
return Ok(FetchOnceResult::Code(ResultPayload { return Ok(FetchOnceResult::Code(body, headers_));
body,
content_type,
etag,
x_typescript_types,
}));
}; };
fut.boxed() fut.boxed()
@ -291,11 +272,11 @@ mod tests {
Url::parse("http://127.0.0.1:4545/cli/tests/fixture.json").unwrap(); Url::parse("http://127.0.0.1:4545/cli/tests/fixture.json").unwrap();
let client = create_http_client(None).unwrap(); let client = create_http_client(None).unwrap();
let result = fetch_once(client, &url, None).await; let result = fetch_once(client, &url, None).await;
if let Ok(FetchOnceResult::Code(payload)) = result { if let Ok(FetchOnceResult::Code(body, headers)) = result {
assert!(!payload.body.is_empty()); assert!(!body.is_empty());
assert_eq!(payload.content_type, Some("application/json".to_string())); assert_eq!(headers.get("content-type").unwrap(), "application/json");
assert_eq!(payload.etag, None); assert_eq!(headers.get("etag"), None);
assert_eq!(payload.x_typescript_types, None); assert_eq!(headers.get("x-typescript-types"), None);
} else { } else {
panic!(); panic!();
} }
@ -312,17 +293,14 @@ mod tests {
.unwrap(); .unwrap();
let client = create_http_client(None).unwrap(); let client = create_http_client(None).unwrap();
let result = fetch_once(client, &url, None).await; let result = fetch_once(client, &url, None).await;
if let Ok(FetchOnceResult::Code(payload)) = result { if let Ok(FetchOnceResult::Code(body, headers)) = result {
assert_eq!(String::from_utf8(body).unwrap(), "console.log('gzip')");
assert_eq!( assert_eq!(
String::from_utf8(payload.body).unwrap(), headers.get("content-type").unwrap(),
"console.log('gzip')" "application/javascript"
); );
assert_eq!( assert_eq!(headers.get("etag"), None);
payload.content_type, assert_eq!(headers.get("x-typescript-types"), None);
Some("application/javascript".to_string())
);
assert_eq!(payload.etag, None);
assert_eq!(payload.x_typescript_types, None);
} else { } else {
panic!(); panic!();
} }
@ -335,18 +313,14 @@ mod tests {
let url = Url::parse("http://127.0.0.1:4545/etag_script.ts").unwrap(); let url = Url::parse("http://127.0.0.1:4545/etag_script.ts").unwrap();
let client = create_http_client(None).unwrap(); let client = create_http_client(None).unwrap();
let result = fetch_once(client.clone(), &url, None).await; let result = fetch_once(client.clone(), &url, None).await;
if let Ok(FetchOnceResult::Code(ResultPayload { if let Ok(FetchOnceResult::Code(body, headers)) = result {
body,
content_type,
etag,
x_typescript_types,
})) = result
{
assert!(!body.is_empty()); assert!(!body.is_empty());
assert_eq!(String::from_utf8(body).unwrap(), "console.log('etag')"); assert_eq!(String::from_utf8(body).unwrap(), "console.log('etag')");
assert_eq!(content_type, Some("application/typescript".to_string())); assert_eq!(
assert_eq!(etag, Some("33a64df551425fcc55e".to_string())); headers.get("content-type").unwrap(),
assert_eq!(x_typescript_types, None); "application/typescript"
);
assert_eq!(headers.get("etag").unwrap(), "33a64df551425fcc55e");
} else { } else {
panic!(); panic!();
} }
@ -368,18 +342,15 @@ mod tests {
.unwrap(); .unwrap();
let client = create_http_client(None).unwrap(); let client = create_http_client(None).unwrap();
let result = fetch_once(client, &url, None).await; let result = fetch_once(client, &url, None).await;
if let Ok(FetchOnceResult::Code(payload)) = result { if let Ok(FetchOnceResult::Code(body, headers)) = result {
assert!(!payload.body.is_empty()); assert!(!body.is_empty());
assert_eq!(String::from_utf8(body).unwrap(), "console.log('brotli');");
assert_eq!( assert_eq!(
String::from_utf8(payload.body).unwrap(), headers.get("content-type").unwrap(),
"console.log('brotli');" "application/javascript"
); );
assert_eq!( assert_eq!(headers.get("etag"), None);
payload.content_type, assert_eq!(headers.get("x-typescript-types"), None);
Some("application/javascript".to_string())
);
assert_eq!(payload.etag, None);
assert_eq!(payload.x_typescript_types, None);
} else { } else {
panic!(); panic!();
} }
@ -397,7 +368,7 @@ mod tests {
Url::parse("http://localhost:4545/cli/tests/fixture.json").unwrap(); Url::parse("http://localhost:4545/cli/tests/fixture.json").unwrap();
let client = create_http_client(None).unwrap(); let client = create_http_client(None).unwrap();
let result = fetch_once(client, &url, None).await; let result = fetch_once(client, &url, None).await;
if let Ok(FetchOnceResult::Redirect(url)) = result { if let Ok(FetchOnceResult::Redirect(url, _)) = result {
assert_eq!(url, target_url); assert_eq!(url, target_url);
} else { } else {
panic!(); panic!();
@ -459,11 +430,11 @@ mod tests {
.unwrap(); .unwrap();
let result = fetch_once(client, &url, None).await; let result = fetch_once(client, &url, None).await;
if let Ok(FetchOnceResult::Code(payload)) = result { if let Ok(FetchOnceResult::Code(body, headers)) = result {
assert!(!payload.body.is_empty()); assert!(!body.is_empty());
assert_eq!(payload.content_type, Some("application/json".to_string())); assert_eq!(headers.get("content-type").unwrap(), "application/json");
assert_eq!(payload.etag, None); assert_eq!(headers.get("etag"), None);
assert_eq!(payload.x_typescript_types, None); assert_eq!(headers.get("x-typescript-types"), None);
} else { } else {
panic!(); panic!();
} }
@ -486,17 +457,14 @@ mod tests {
))) )))
.unwrap(); .unwrap();
let result = fetch_once(client, &url, None).await; let result = fetch_once(client, &url, None).await;
if let Ok(FetchOnceResult::Code(payload)) = result { if let Ok(FetchOnceResult::Code(body, headers)) = result {
assert_eq!(String::from_utf8(body).unwrap(), "console.log('gzip')");
assert_eq!( assert_eq!(
String::from_utf8(payload.body).unwrap(), headers.get("content-type").unwrap(),
"console.log('gzip')" "application/javascript"
); );
assert_eq!( assert_eq!(headers.get("etag"), None);
payload.content_type, assert_eq!(headers.get("x-typescript-types"), None);
Some("application/javascript".to_string())
);
assert_eq!(payload.etag, None);
assert_eq!(payload.x_typescript_types, None);
} else { } else {
panic!(); panic!();
} }
@ -515,18 +483,15 @@ mod tests {
))) )))
.unwrap(); .unwrap();
let result = fetch_once(client.clone(), &url, None).await; let result = fetch_once(client.clone(), &url, None).await;
if let Ok(FetchOnceResult::Code(ResultPayload { if let Ok(FetchOnceResult::Code(body, headers)) = result {
body,
content_type,
etag,
x_typescript_types,
})) = result
{
assert!(!body.is_empty()); assert!(!body.is_empty());
assert_eq!(String::from_utf8(body).unwrap(), "console.log('etag')"); assert_eq!(String::from_utf8(body).unwrap(), "console.log('etag')");
assert_eq!(content_type, Some("application/typescript".to_string())); assert_eq!(
assert_eq!(etag, Some("33a64df551425fcc55e".to_string())); headers.get("content-type").unwrap(),
assert_eq!(x_typescript_types, None); "application/typescript"
);
assert_eq!(headers.get("etag").unwrap(), "33a64df551425fcc55e");
assert_eq!(headers.get("x-typescript-types"), None);
} else { } else {
panic!(); panic!();
} }
@ -554,18 +519,15 @@ mod tests {
))) )))
.unwrap(); .unwrap();
let result = fetch_once(client, &url, None).await; let result = fetch_once(client, &url, None).await;
if let Ok(FetchOnceResult::Code(payload)) = result { if let Ok(FetchOnceResult::Code(body, headers)) = result {
assert!(!payload.body.is_empty()); assert!(!body.is_empty());
assert_eq!(String::from_utf8(body).unwrap(), "console.log('brotli');");
assert_eq!( assert_eq!(
String::from_utf8(payload.body).unwrap(), headers.get("content-type").unwrap(),
"console.log('brotli');" "application/javascript"
); );
assert_eq!( assert_eq!(headers.get("etag"), None);
payload.content_type, assert_eq!(headers.get("x-typescript-types"), None);
Some("application/javascript".to_string())
);
assert_eq!(payload.etag, None);
assert_eq!(payload.x_typescript_types, None);
} else { } else {
panic!(); panic!();
} }

3
cli/lib.rs
View file

@ -33,6 +33,7 @@ pub mod fmt_errors;
mod fs; mod fs;
mod global_state; mod global_state;
mod global_timer; mod global_timer;
pub mod http_cache;
mod http_util; mod http_util;
mod import_map; mod import_map;
pub mod installer; pub mod installer;
@ -132,7 +133,7 @@ fn print_cache_info(state: &GlobalState) {
println!( println!(
"{} {:?}", "{} {:?}",
colors::bold("Remote modules cache:".to_string()), colors::bold("Remote modules cache:".to_string()),
state.dir.deps_cache.location state.file_fetcher.http_cache.location
); );
println!( println!(
"{} {:?}", "{} {:?}",

2
cli/tests/022_info_flag_script.out
View file

@ -1,4 +1,4 @@
local: [WILDCARD]019_media_types.ts local: [WILDCARD]http[WILDCARD]127.0.0.1_PORT4545[WILDCARD]
type: TypeScript type: TypeScript
compiled: [WILDCARD].js compiled: [WILDCARD].js
map: [WILDCARD].js.map map: [WILDCARD].js.map

2
cli/tests/049_info_flag_script_jsx.out
View file

@ -1,4 +1,4 @@
local: [WILDCARD]048_media_types_jsx.ts local: [WILDCARD]http[WILDCARD]127.0.0.1_PORT4545[WILDCARD]
type: TypeScript type: TypeScript
compiled: [WILDCARD]048_media_types_jsx.ts.js compiled: [WILDCARD]048_media_types_jsx.ts.js
map: [WILDCARD]048_media_types_jsx.ts.js.map map: [WILDCARD]048_media_types_jsx.ts.js.map

2
cli/tests/cafile_info.ts.out
View file

@ -1,4 +1,4 @@
local: [WILDCARD]cafile_info.ts local: [WILDCARD]https[WILDCARD]localhost_PORT5545[WILDCARD]
type: TypeScript type: TypeScript
compiled: [WILDCARD].js compiled: [WILDCARD].js
map: [WILDCARD].js.map map: [WILDCARD].js.map

21
cli/tests/integration_tests.rs
View file

@ -23,20 +23,23 @@ fn deno_dir_test() {
#[test] #[test]
fn fetch_test() { fn fetch_test() {
use deno::http_cache::url_to_filename;
pub use deno::test_util::*; pub use deno::test_util::*;
use std::process::Command; use std::process::Command;
use tempfile::TempDir; use tempfile::TempDir;
use url::Url;
let g = util::http_server(); let g = util::http_server();
let deno_dir = TempDir::new().expect("tempdir fail"); let deno_dir = TempDir::new().expect("tempdir fail");
let t = util::root_path().join("cli/tests/006_url_imports.ts"); let module_url =
Url::parse("http://localhost:4545/cli/tests/006_url_imports.ts").unwrap();
let output = Command::new(deno_exe_path()) let output = Command::new(deno_exe_path())
.env("DENO_DIR", deno_dir.path()) .env("DENO_DIR", deno_dir.path())
.current_dir(util::root_path()) .current_dir(util::root_path())
.arg("fetch") .arg("fetch")
.arg(t) .arg(module_url.to_string())
.output() .output()
.expect("Failed to spawn script"); .expect("Failed to spawn script");
@ -48,7 +51,8 @@ fn fetch_test() {
let expected_path = deno_dir let expected_path = deno_dir
.path() .path()
.join("deps/http/localhost_PORT4545/cli/tests/subdir/mod2.ts"); .join("deps")
.join(url_to_filename(&module_url));
assert_eq!(expected_path.exists(), true); assert_eq!(expected_path.exists(), true);
drop(g); drop(g);
@ -966,14 +970,18 @@ itest!(cafile_info {
#[test] #[test]
fn cafile_fetch() { fn cafile_fetch() {
use deno::http_cache::url_to_filename;
pub use deno::test_util::*; pub use deno::test_util::*;
use std::process::Command; use std::process::Command;
use tempfile::TempDir; use tempfile::TempDir;
use url::Url;
let g = util::http_server(); let g = util::http_server();
let deno_dir = TempDir::new().expect("tempdir fail"); let deno_dir = TempDir::new().expect("tempdir fail");
let t = util::root_path().join("cli/tests/cafile_url_imports.ts"); let module_url =
Url::parse("http://localhost:4545/cli/tests/cafile_url_imports.ts")
.unwrap();
let cafile = util::root_path().join("cli/tests/tls/RootCA.pem"); let cafile = util::root_path().join("cli/tests/tls/RootCA.pem");
let output = Command::new(deno_exe_path()) let output = Command::new(deno_exe_path())
.env("DENO_DIR", deno_dir.path()) .env("DENO_DIR", deno_dir.path())
@ -981,7 +989,7 @@ fn cafile_fetch() {
.arg("fetch") .arg("fetch")
.arg("--cert") .arg("--cert")
.arg(cafile) .arg(cafile)
.arg(t) .arg(module_url.to_string())
.output() .output()
.expect("Failed to spawn script"); .expect("Failed to spawn script");
@ -993,7 +1001,8 @@ fn cafile_fetch() {
let expected_path = deno_dir let expected_path = deno_dir
.path() .path()
.join("deps/https/localhost_PORT5545/cli/tests/subdir/mod2.ts"); .join("deps")
.join(url_to_filename(&module_url));
assert_eq!(expected_path.exists(), true); assert_eq!(expected_path.exists(), true);
drop(g); drop(g);