deno/cli/import_map.rs
2019-06-12 16:55:59 -07:00

2092 lines
62 KiB
Rust

use deno::ModuleSpecifier;
use indexmap::IndexMap;
use serde_json::Map;
use serde_json::Value;
use std::cmp::Ordering;
use std::fs;
use url::Url;
#[derive(Debug)]
pub struct ImportMapError {
pub msg: String,
}
impl ImportMapError {
pub fn new(msg: &str) -> Self {
ImportMapError {
msg: msg.to_string(),
}
}
}
// NOTE: here is difference between deno and reference implementation - deno currently
// can't resolve URL with other schemes (eg. data:, about:, blob:)
const SUPPORTED_FETCH_SCHEMES: [&str; 3] = ["http", "https", "file"];
type SpecifierMap = IndexMap<String, Vec<ModuleSpecifier>>;
type ScopesMap = IndexMap<String, SpecifierMap>;
#[derive(Debug)]
pub struct ImportMap {
base_url: String,
imports: SpecifierMap,
scopes: ScopesMap,
}
impl ImportMap {
pub fn load(base_url: &str, file_name: &str) -> Result<Self, ImportMapError> {
let cwd = std::env::current_dir().unwrap();
let resolved_path = cwd.join(file_name);
debug!(
"Attempt to load import map: {}",
resolved_path.to_str().unwrap()
);
// Load the contents of import map
match fs::read_to_string(&resolved_path) {
Ok(json_string) => ImportMap::from_json(base_url, &json_string),
_ => panic!(
"Error retrieving import map file at \"{}\"",
resolved_path.to_str().unwrap()
),
}
}
pub fn from_json(
base_url: &str,
json_string: &str,
) -> Result<Self, ImportMapError> {
let v: Value = match serde_json::from_str(json_string) {
Ok(v) => v,
Err(_) => {
return Err(ImportMapError::new("Unable to parse import map JSON"));
}
};
match v {
Value::Object(_) => {}
_ => {
return Err(ImportMapError::new("Import map JSON must be an object"));
}
}
let normalized_imports = match &v.get("imports") {
Some(imports_map) => {
if !imports_map.is_object() {
return Err(ImportMapError::new(
"Import map's 'imports' must be an object",
));
}
let imports_map = imports_map.as_object().unwrap();
ImportMap::parse_specifier_map(imports_map, base_url)
}
None => IndexMap::new(),
};
let normalized_scopes = match &v.get("scopes") {
Some(scope_map) => {
if !scope_map.is_object() {
return Err(ImportMapError::new(
"Import map's 'scopes' must be an object",
));
}
let scope_map = scope_map.as_object().unwrap();
ImportMap::parse_scope_map(scope_map, base_url)?
}
None => IndexMap::new(),
};
let import_map = ImportMap {
base_url: base_url.to_string(),
imports: normalized_imports,
scopes: normalized_scopes,
};
Ok(import_map)
}
fn try_url_like_specifier(specifier: &str, base: &str) -> Option<Url> {
// this should never fail
if specifier.starts_with('/')
|| specifier.starts_with("./")
|| specifier.starts_with("../")
{
let base_url = Url::parse(base).unwrap();
let url = base_url.join(specifier).unwrap();
return Some(url);
}
if let Ok(url) = Url::parse(specifier) {
if SUPPORTED_FETCH_SCHEMES.contains(&url.scheme()) {
return Some(url);
}
}
None
}
/// Parse provided key as import map specifier.
///
/// Specifiers must be valid URLs (eg. "https://deno.land/x/std/testing/mod.ts")
/// or "bare" specifiers (eg. "moment").
// TODO: add proper error handling: https://github.com/WICG/import-maps/issues/100
fn normalize_specifier_key(
specifier_key: &str,
base_url: &str,
) -> Option<String> {
// ignore empty keys
if specifier_key.is_empty() {
return None;
}
if let Some(url) =
ImportMap::try_url_like_specifier(specifier_key, base_url)
{
return Some(url.to_string());
}
// "bare" specifier
Some(specifier_key.to_string())
}
/// Parse provided addresses as valid URLs.
///
/// Non-valid addresses are skipped.
fn normalize_addresses(
specifier_key: &str,
base_url: &str,
potential_addresses: Vec<String>,
) -> Vec<ModuleSpecifier> {
let mut normalized_addresses: Vec<ModuleSpecifier> = vec![];
for potential_address in potential_addresses {
let url =
match ImportMap::try_url_like_specifier(&potential_address, base_url) {
Some(url) => url,
None => continue,
};
let url_string = url.to_string();
if specifier_key.ends_with('/') && !url_string.ends_with('/') {
eprintln!(
"Invalid target address {:?} for package specifier {:?}.\
Package address targets must end with \"/\".",
url_string, specifier_key
);
continue;
}
let normalized_address = ModuleSpecifier::resolve(&url_string, ".")
.expect("Address should be valid module specifier");
normalized_addresses.push(normalized_address);
}
normalized_addresses
}
/// Convert provided JSON map to valid SpecifierMap.
///
/// From specification:
/// - order of iteration must be retained
/// - SpecifierMap's keys are sorted in longest and alphabetic order
fn parse_specifier_map(
json_map: &Map<String, Value>,
base_url: &str,
) -> SpecifierMap {
let mut normalized_map: SpecifierMap = SpecifierMap::new();
// Order is preserved because of "preserve_order" feature of "serde_json".
for (specifier_key, value) in json_map.iter() {
let normalized_specifier_key =
match ImportMap::normalize_specifier_key(specifier_key, base_url) {
Some(s) => s,
None => continue,
};
let potential_addresses: Vec<String> = match value {
Value::String(address) => vec![address.to_string()],
Value::Array(address_array) => {
let mut string_addresses: Vec<String> = vec![];
for address in address_array {
match address {
Value::String(address) => {
string_addresses.push(address.to_string())
}
_ => continue,
}
}
string_addresses
}
Value::Null => vec![],
_ => vec![],
};
let normalized_address_array = ImportMap::normalize_addresses(
&normalized_specifier_key,
base_url,
potential_addresses,
);
debug!(
"normalized specifier {:?}; {:?}",
normalized_specifier_key, normalized_address_array
);
normalized_map.insert(normalized_specifier_key, normalized_address_array);
}
// Sort in longest and alphabetical order.
normalized_map.sort_by(|k1, _v1, k2, _v2| {
if k1.len() > k2.len() {
return Ordering::Less;
} else if k2.len() > k1.len() {
return Ordering::Greater;
}
k2.cmp(k1)
});
normalized_map
}
/// Convert provided JSON map to valid ScopeMap.
///
/// From specification:
/// - order of iteration must be retained
/// - ScopeMap's keys are sorted in longest and alphabetic order
fn parse_scope_map(
scope_map: &Map<String, Value>,
base_url: &str,
) -> Result<ScopesMap, ImportMapError> {
let mut normalized_map: ScopesMap = ScopesMap::new();
// Order is preserved because of "preserve_order" feature of "serde_json".
for (scope_prefix, potential_specifier_map) in scope_map.iter() {
if !potential_specifier_map.is_object() {
return Err(ImportMapError::new(&format!(
"The value for the {:?} scope prefix must be an object",
scope_prefix
)));
}
let potential_specifier_map =
potential_specifier_map.as_object().unwrap();
let scope_prefix_url =
match Url::parse(base_url).unwrap().join(scope_prefix) {
Ok(url) => {
if !SUPPORTED_FETCH_SCHEMES.contains(&url.scheme()) {
eprintln!(
"Invalid scope {:?}. Scope URLs must have a valid fetch scheme.",
url.to_string()
);
continue;
}
url.to_string()
}
_ => continue,
};
let norm_map =
ImportMap::parse_specifier_map(potential_specifier_map, base_url);
normalized_map.insert(scope_prefix_url, norm_map);
}
// Sort in longest and alphabetical order.
normalized_map.sort_by(|k1, _v1, k2, _v2| {
if k1.len() > k2.len() {
return Ordering::Less;
} else if k2.len() > k1.len() {
return Ordering::Greater;
}
k2.cmp(k1)
});
Ok(normalized_map)
}
pub fn resolve_scopes_match(
scopes: &ScopesMap,
normalized_specifier: &str,
referrer: &str,
) -> Result<Option<ModuleSpecifier>, ImportMapError> {
// exact-match
if let Some(scope_imports) = scopes.get(referrer) {
if let Ok(scope_match) =
ImportMap::resolve_imports_match(scope_imports, normalized_specifier)
{
// Return only if there was actual match (not None).
if scope_match.is_some() {
return Ok(scope_match);
}
}
}
for (normalized_scope_key, scope_imports) in scopes.iter() {
if normalized_scope_key.ends_with('/')
&& referrer.starts_with(normalized_scope_key)
{
if let Ok(scope_match) =
ImportMap::resolve_imports_match(scope_imports, normalized_specifier)
{
// Return only if there was actual match (not None).
if scope_match.is_some() {
return Ok(scope_match);
}
}
}
}
Ok(None)
}
// TODO: https://github.com/WICG/import-maps/issues/73#issuecomment-439327758
// for some more optimized candidate implementations.
pub fn resolve_imports_match(
imports: &SpecifierMap,
normalized_specifier: &str,
) -> Result<Option<ModuleSpecifier>, ImportMapError> {
// exact-match
if let Some(address_vec) = imports.get(normalized_specifier) {
if address_vec.is_empty() {
return Err(ImportMapError::new(&format!(
"Specifier {:?} was mapped to no addresses.",
normalized_specifier
)));
} else if address_vec.len() == 1 {
let address = address_vec.first().unwrap();
debug!(
"Specifier {:?} was mapped to {:?}.",
normalized_specifier, address
);
return Ok(Some(address.clone()));
} else {
return Err(ImportMapError::new(
"Multi-address mappings are not yet supported",
));
}
}
// package-prefix match
// "most-specific wins", i.e. when there are multiple matching keys,
// choose the longest.
// https://github.com/WICG/import-maps/issues/102
for (specifier_key, address_vec) in imports.iter() {
if specifier_key.ends_with('/')
&& normalized_specifier.starts_with(specifier_key)
{
if address_vec.is_empty() {
return Err(ImportMapError::new(&format!("Specifier {:?} was mapped to no addresses (via prefix specifier key {:?}).", normalized_specifier, specifier_key)));
} else if address_vec.len() == 1 {
let address = address_vec.first().unwrap();
let after_prefix = &normalized_specifier[specifier_key.len()..];
let base_url = address.to_url();
if let Ok(url) = base_url.join(after_prefix) {
debug!("Specifier {:?} was mapped to {:?} (via prefix specifier key {:?}).", normalized_specifier, url, address);
return Ok(Some(ModuleSpecifier::from(url)));
}
unreachable!();
} else {
return Err(ImportMapError::new(
"Multi-address mappings are not yet supported",
));
}
}
}
debug!(
"Specifier {:?} was not mapped in import map.",
normalized_specifier
);
Ok(None)
}
// TODO: add support for built-in modules
/// Currently we support two types of specifiers: URL (http://, https://, file://)
/// and "bare" (moment, jquery, lodash)
///
/// Scenarios:
/// 1. import resolved using import map -> String
/// 2. import restricted by import map -> ImportMapError
/// 3. import not mapped -> None
pub fn resolve(
&self,
specifier: &str,
referrer: &str,
) -> Result<Option<ModuleSpecifier>, ImportMapError> {
let resolved_url: Option<Url> =
ImportMap::try_url_like_specifier(specifier, referrer);
let normalized_specifier = match &resolved_url {
Some(url) => url.to_string(),
None => specifier.to_string(),
};
let scopes_match = ImportMap::resolve_scopes_match(
&self.scopes,
&normalized_specifier,
&referrer.to_string(),
)?;
// match found in scopes map
if scopes_match.is_some() {
return Ok(scopes_match);
}
let imports_match =
ImportMap::resolve_imports_match(&self.imports, &normalized_specifier)?;
// match found in import map
if imports_match.is_some() {
return Ok(imports_match);
}
// no match in import map but we got resolvable URL
if let Some(resolved_url) = resolved_url {
return Ok(Some(ModuleSpecifier::from(resolved_url)));
}
Err(ImportMapError::new(&format!(
"Unmapped bare specifier {:?}",
normalized_specifier
)))
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn from_json_1() {
let base_url = "https://deno.land";
// empty JSON
assert!(ImportMap::from_json(base_url, "{}").is_ok());
let non_object_strings = vec!["null", "true", "1", "\"foo\"", "[]"];
// invalid JSON
for non_object in non_object_strings.to_vec() {
assert!(ImportMap::from_json(base_url, non_object).is_err());
}
// invalid schema: 'imports' is non-object
for non_object in non_object_strings.to_vec() {
assert!(
ImportMap::from_json(
base_url,
&format!("{{\"imports\": {}}}", non_object),
).is_err()
);
}
// invalid schema: 'scopes' is non-object
for non_object in non_object_strings.to_vec() {
assert!(
ImportMap::from_json(
base_url,
&format!("{{\"scopes\": {}}}", non_object),
).is_err()
);
}
}
#[test]
fn from_json_2() {
let json_map = r#"{
"imports": {
"foo": "https://example.com/1",
"bar": ["https://example.com/2"],
"fizz": null
}
}"#;
let result = ImportMap::from_json("https://deno.land", json_map);
assert!(result.is_ok());
}
#[test]
fn parse_specifier_keys_relative() {
// Should absolutize strings prefixed with ./, ../, or / into the corresponding URLs..
let json_map = r#"{
"imports": {
"./foo": "/dotslash",
"../foo": "/dotdotslash",
"/foo": "/slash"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map
.imports
.get("https://base.example/path1/path2/foo")
.unwrap()[0],
"https://base.example/dotslash".to_string()
);
assert_eq!(
import_map
.imports
.get("https://base.example/path1/foo")
.unwrap()[0],
"https://base.example/dotdotslash".to_string()
);
assert_eq!(
import_map.imports.get("https://base.example/foo").unwrap()[0],
"https://base.example/slash".to_string()
);
// Should absolutize the literal strings ./, ../, or / with no suffix..
let json_map = r#"{
"imports": {
"./": "/dotslash/",
"../": "/dotdotslash/",
"/": "/slash/"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map
.imports
.get("https://base.example/path1/path2/")
.unwrap()[0],
"https://base.example/dotslash/".to_string()
);
assert_eq!(
import_map
.imports
.get("https://base.example/path1/")
.unwrap()[0],
"https://base.example/dotdotslash/".to_string()
);
assert_eq!(
import_map.imports.get("https://base.example/").unwrap()[0],
"https://base.example/slash/".to_string()
);
// Should treat percent-encoded variants of ./, ../, or / as bare specifiers..
let json_map = r#"{
"imports": {
"%2E/": "/dotSlash1/",
"%2E%2E/": "/dotDotSlash1/",
".%2F": "/dotSlash2",
"..%2F": "/dotDotSlash2",
"%2F": "/slash2",
"%2E%2F": "/dotSlash3",
"%2E%2E%2F": "/dotDotSlash3"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("%2E/").unwrap()[0],
"https://base.example/dotSlash1/".to_string()
);
assert_eq!(
import_map.imports.get("%2E%2E/").unwrap()[0],
"https://base.example/dotDotSlash1/".to_string()
);
assert_eq!(
import_map.imports.get(".%2F").unwrap()[0],
"https://base.example/dotSlash2".to_string()
);
assert_eq!(
import_map.imports.get("..%2F").unwrap()[0],
"https://base.example/dotDotSlash2".to_string()
);
assert_eq!(
import_map.imports.get("%2F").unwrap()[0],
"https://base.example/slash2".to_string()
);
assert_eq!(
import_map.imports.get("%2E%2F").unwrap()[0],
"https://base.example/dotSlash3".to_string()
);
assert_eq!(
import_map.imports.get("%2E%2E%2F").unwrap()[0],
"https://base.example/dotDotSlash3".to_string()
);
}
#[test]
fn parse_specifier_keys_absolute() {
// Should only accept absolute URL specifier keys with fetch schemes,.
// treating others as bare specifiers.
let json_map = r#"{
"imports": {
"file:///good": "/file",
"http://good/": "/http/",
"https://good/": "/https/",
"about:bad": "/about",
"blob:bad": "/blob",
"data:bad": "/data",
"filesystem:bad": "/filesystem",
"ftp://bad/": "/ftp/",
"import:bad": "/import",
"mailto:bad": "/mailto",
"javascript:bad": "/javascript",
"wss:bad": "/wss"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("http://good/").unwrap()[0],
"https://base.example/http/".to_string()
);
assert_eq!(
import_map.imports.get("https://good/").unwrap()[0],
"https://base.example/https/".to_string()
);
assert_eq!(
import_map.imports.get("file:///good").unwrap()[0],
"https://base.example/file".to_string()
);
assert_eq!(
import_map.imports.get("http://good/").unwrap()[0],
"https://base.example/http/".to_string()
);
assert_eq!(
import_map.imports.get("import:bad").unwrap()[0],
"https://base.example/import".to_string()
);
assert_eq!(
import_map.imports.get("mailto:bad").unwrap()[0],
"https://base.example/mailto".to_string()
);
assert_eq!(
import_map.imports.get("javascript:bad").unwrap()[0],
"https://base.example/javascript".to_string()
);
assert_eq!(
import_map.imports.get("wss:bad").unwrap()[0],
"https://base.example/wss".to_string()
);
assert_eq!(
import_map.imports.get("about:bad").unwrap()[0],
"https://base.example/about".to_string()
);
assert_eq!(
import_map.imports.get("blob:bad").unwrap()[0],
"https://base.example/blob".to_string()
);
assert_eq!(
import_map.imports.get("data:bad").unwrap()[0],
"https://base.example/data".to_string()
);
// Should parse absolute URLs, treating unparseable ones as bare specifiers..
let json_map = r#"{
"imports": {
"https://ex ample.org/": "/unparseable1/",
"https://example.com:demo": "/unparseable2",
"http://[www.example.com]/": "/unparseable3/",
"https:example.org": "/invalidButParseable1/",
"https://///example.com///": "/invalidButParseable2/",
"https://example.net": "/prettyNormal/",
"https://ex%41mple.com/": "/percentDecoding/",
"https://example.com/%41": "/noPercentDecoding"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("https://ex ample.org/").unwrap()[0],
"https://base.example/unparseable1/".to_string()
);
assert_eq!(
import_map.imports.get("https://example.com:demo").unwrap()[0],
"https://base.example/unparseable2".to_string()
);
assert_eq!(
import_map.imports.get("http://[www.example.com]/").unwrap()[0],
"https://base.example/unparseable3/".to_string()
);
assert_eq!(
import_map.imports.get("https://example.org/").unwrap()[0],
"https://base.example/invalidButParseable1/".to_string()
);
assert_eq!(
import_map.imports.get("https://example.com///").unwrap()[0],
"https://base.example/invalidButParseable2/".to_string()
);
assert_eq!(
import_map.imports.get("https://example.net/").unwrap()[0],
"https://base.example/prettyNormal/".to_string()
);
assert_eq!(
import_map.imports.get("https://example.com/").unwrap()[0],
"https://base.example/percentDecoding/".to_string()
);
assert_eq!(
import_map.imports.get("https://example.com/%41").unwrap()[0],
"https://base.example/noPercentDecoding".to_string()
);
}
#[test]
fn parse_scope_keys_relative() {
// Should work with no prefix..
let json_map = r#"{
"scopes": {
"foo": {}
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/path2/foo")
);
// Should work with ./, ../, and / prefixes..
let json_map = r#"{
"scopes": {
"./foo": {},
"../foo": {},
"/foo": {}
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/path2/foo")
);
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/foo")
);
assert!(import_map.scopes.contains_key("https://base.example/foo"));
// Should work with /s, ?s, and #s..
let json_map = r#"{
"scopes": {
"foo/bar?baz#qux": {}
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/path2/foo/bar?baz#qux")
);
// Should work with an empty string scope key..
let json_map = r#"{
"scopes": {
"": {}
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/path2/path3")
);
// Should work with / suffixes..
let json_map = r#"{
"scopes": {
"foo/": {},
"./foo/": {},
"../foo/": {},
"/foo/": {},
"/foo//": {}
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/path2/foo/")
);
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/path2/foo/")
);
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/foo/")
);
assert!(import_map.scopes.contains_key("https://base.example/foo/"));
assert!(import_map.scopes.contains_key("https://base.example/foo//"));
// Should deduplicate based on URL parsing rules..
let json_map = r#"{
"scopes": {
"foo/\\": {},
"foo//": {},
"foo\\\\": {}
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/path2/foo//")
);
assert_eq!(import_map.scopes.len(), 1);
}
#[test]
fn parse_scope_keys_absolute() {
// Should only accept absolute URL scope keys with fetch schemes..
let json_map = r#"{
"scopes": {
"http://good/": {},
"https://good/": {},
"file:///good": {},
"about:bad": {},
"blob:bad": {},
"data:bad": {},
"filesystem:bad": {},
"ftp://bad/": {},
"import:bad": {},
"mailto:bad": {},
"javascript:bad": {},
"wss:bad": {}
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(import_map.scopes.contains_key("http://good/"));
assert!(import_map.scopes.contains_key("https://good/"));
assert!(import_map.scopes.contains_key("file:///good"));
assert_eq!(import_map.scopes.len(), 3);
// Should parse absolute URL scope keys, ignoring unparseable ones..
let json_map = r#"{
"scopes": {
"https://ex ample.org/": {},
"https://example.com:demo": {},
"http://[www.example.com]/": {},
"https:example.org": {},
"https://///example.com///": {},
"https://example.net": {},
"https://ex%41mple.com/foo/": {},
"https://example.com/%41": {}
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
// tricky case! remember we have a base URL
assert!(
import_map
.scopes
.contains_key("https://base.example/path1/path2/example.org")
);
assert!(import_map.scopes.contains_key("https://example.com///"));
assert!(import_map.scopes.contains_key("https://example.net/"));
assert!(import_map.scopes.contains_key("https://example.com/foo/"));
assert!(import_map.scopes.contains_key("https://example.com/%41"));
assert_eq!(import_map.scopes.len(), 5);
}
#[test]
fn parse_addresses_relative_url_like() {
// Should accept strings prefixed with ./, ../, or /..
let json_map = r#"{
"imports": {
"dotSlash": "./foo",
"dotDotSlash": "../foo",
"slash": "/foo"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("dotSlash").unwrap(),
&vec!["https://base.example/path1/path2/foo".to_string()]
);
assert_eq!(
import_map.imports.get("dotDotSlash").unwrap(),
&vec!["https://base.example/path1/foo".to_string()]
);
assert_eq!(
import_map.imports.get("slash").unwrap(),
&vec!["https://base.example/foo".to_string()]
);
// Should accept the literal strings ./, ../, or / with no suffix..
let json_map = r#"{
"imports": {
"dotSlash": "./",
"dotDotSlash": "../",
"slash": "/"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("dotSlash").unwrap(),
&vec!["https://base.example/path1/path2/".to_string()]
);
assert_eq!(
import_map.imports.get("dotDotSlash").unwrap(),
&vec!["https://base.example/path1/".to_string()]
);
assert_eq!(
import_map.imports.get("slash").unwrap(),
&vec!["https://base.example/".to_string()]
);
// Should ignore percent-encoded variants of ./, ../, or /..
let json_map = r#"{
"imports": {
"dotSlash1": "%2E/",
"dotDotSlash1": "%2E%2E/",
"dotSlash2": ".%2F",
"dotDotSlash2": "..%2F",
"slash2": "%2F",
"dotSlash3": "%2E%2F",
"dotDotSlash3": "%2E%2E%2F"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(import_map.imports.get("dotSlash1").unwrap().is_empty());
assert!(import_map.imports.get("dotDotSlash1").unwrap().is_empty());
assert!(import_map.imports.get("dotSlash2").unwrap().is_empty());
assert!(import_map.imports.get("dotDotSlash2").unwrap().is_empty());
assert!(import_map.imports.get("slash2").unwrap().is_empty());
assert!(import_map.imports.get("dotSlash3").unwrap().is_empty());
assert!(import_map.imports.get("dotDotSlash3").unwrap().is_empty());
}
#[test]
fn parse_addresses_absolute_with_fetch_schemes() {
// Should only accept absolute URL addresses with fetch schemes..
let json_map = r#"{
"imports": {
"http": "http://good/",
"https": "https://good/",
"file": "file:///good",
"about": "about:bad",
"blob": "blob:bad",
"data": "data:bad",
"filesystem": "filesystem:bad",
"ftp": "ftp://good/",
"import": "import:bad",
"mailto": "mailto:bad",
"javascript": "javascript:bad",
"wss": "wss:bad"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("file").unwrap(),
&vec!["file:///good".to_string()]
);
assert_eq!(
import_map.imports.get("http").unwrap(),
&vec!["http://good/".to_string()]
);
assert_eq!(
import_map.imports.get("https").unwrap(),
&vec!["https://good/".to_string()]
);
assert!(import_map.imports.get("about").unwrap().is_empty());
assert!(import_map.imports.get("blob").unwrap().is_empty());
assert!(import_map.imports.get("data").unwrap().is_empty());
assert!(import_map.imports.get("filesystem").unwrap().is_empty());
assert!(import_map.imports.get("ftp").unwrap().is_empty());
assert!(import_map.imports.get("import").unwrap().is_empty());
assert!(import_map.imports.get("mailto").unwrap().is_empty());
assert!(import_map.imports.get("javascript").unwrap().is_empty());
assert!(import_map.imports.get("wss").unwrap().is_empty());
}
#[test]
fn parse_addresses_absolute_with_fetch_schemes_arrays() {
// Should only accept absolute URL addresses with fetch schemes inside arrays..
let json_map = r#"{
"imports": {
"http": ["http://good/"],
"https": ["https://good/"],
"file": ["file:///good"],
"about": ["about:bad"],
"blob": ["blob:bad"],
"data": ["data:bad"],
"filesystem": ["filesystem:bad"],
"ftp": ["ftp://good/"],
"import": ["import:bad"],
"mailto": ["mailto:bad"],
"javascript": ["javascript:bad"],
"wss": ["wss:bad"]
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("file").unwrap(),
&vec!["file:///good".to_string()]
);
assert_eq!(
import_map.imports.get("http").unwrap(),
&vec!["http://good/".to_string()]
);
assert_eq!(
import_map.imports.get("https").unwrap(),
&vec!["https://good/".to_string()]
);
assert!(import_map.imports.get("about").unwrap().is_empty());
assert!(import_map.imports.get("blob").unwrap().is_empty());
assert!(import_map.imports.get("data").unwrap().is_empty());
assert!(import_map.imports.get("filesystem").unwrap().is_empty());
assert!(import_map.imports.get("ftp").unwrap().is_empty());
assert!(import_map.imports.get("import").unwrap().is_empty());
assert!(import_map.imports.get("mailto").unwrap().is_empty());
assert!(import_map.imports.get("javascript").unwrap().is_empty());
assert!(import_map.imports.get("wss").unwrap().is_empty());
}
#[test]
fn parse_addresses_unparseable() {
// Should parse absolute URLs, ignoring unparseable ones..
let json_map = r#"{
"imports": {
"unparseable1": "https://ex ample.org/",
"unparseable2": "https://example.com:demo",
"unparseable3": "http://[www.example.com]/",
"invalidButParseable1": "https:example.org",
"invalidButParseable2": "https://///example.com///",
"prettyNormal": "https://example.net",
"percentDecoding": "https://ex%41mple.com/",
"noPercentDecoding": "https://example.com/%41"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("invalidButParseable1").unwrap(),
&vec!["https://example.org/".to_string()]
);
assert_eq!(
import_map.imports.get("invalidButParseable2").unwrap(),
&vec!["https://example.com///".to_string()]
);
assert_eq!(
import_map.imports.get("prettyNormal").unwrap(),
&vec!["https://example.net/".to_string()]
);
assert_eq!(
import_map.imports.get("percentDecoding").unwrap(),
&vec!["https://example.com/".to_string()]
);
assert_eq!(
import_map.imports.get("noPercentDecoding").unwrap(),
&vec!["https://example.com/%41".to_string()]
);
assert!(import_map.imports.get("unparseable1").unwrap().is_empty());
assert!(import_map.imports.get("unparseable2").unwrap().is_empty());
assert!(import_map.imports.get("unparseable3").unwrap().is_empty());
}
#[test]
fn parse_addresses_unparseable_arrays() {
// Should parse absolute URLs, ignoring unparseable ones inside arrays..
let json_map = r#"{
"imports": {
"unparseable1": ["https://ex ample.org/"],
"unparseable2": ["https://example.com:demo"],
"unparseable3": ["http://[www.example.com]/"],
"invalidButParseable1": ["https:example.org"],
"invalidButParseable2": ["https://///example.com///"],
"prettyNormal": ["https://example.net"],
"percentDecoding": ["https://ex%41mple.com/"],
"noPercentDecoding": ["https://example.com/%41"]
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("invalidButParseable1").unwrap(),
&vec!["https://example.org/".to_string()]
);
assert_eq!(
import_map.imports.get("invalidButParseable2").unwrap(),
&vec!["https://example.com///".to_string()]
);
assert_eq!(
import_map.imports.get("prettyNormal").unwrap(),
&vec!["https://example.net/".to_string()]
);
assert_eq!(
import_map.imports.get("percentDecoding").unwrap(),
&vec!["https://example.com/".to_string()]
);
assert_eq!(
import_map.imports.get("noPercentDecoding").unwrap(),
&vec!["https://example.com/%41".to_string()]
);
assert!(import_map.imports.get("unparseable1").unwrap().is_empty());
assert!(import_map.imports.get("unparseable2").unwrap().is_empty());
assert!(import_map.imports.get("unparseable3").unwrap().is_empty());
}
#[test]
fn parse_addresses_mismatched_trailing_slashes() {
// Should parse absolute URLs, ignoring unparseable ones inside arrays..
let json_map = r#"{
"imports": {
"trailer/": "/notrailer"
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(import_map.imports.get("trailer/").unwrap().is_empty());
// TODO: I'd be good to assert that warning was shown
}
#[test]
fn parse_addresses_mismatched_trailing_slashes_array() {
// Should warn for a mismatch alone in an array..
let json_map = r#"{
"imports": {
"trailer/": ["/notrailer"]
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert!(import_map.imports.get("trailer/").unwrap().is_empty());
// TODO: I'd be good to assert that warning was shown
}
#[test]
fn parse_addresses_mismatched_trailing_slashes_with_nonmismatched_array() {
// Should warn for a mismatch alone in an array..
let json_map = r#"{
"imports": {
"trailer/": ["/atrailer/", "/notrailer"]
}
}"#;
let import_map =
ImportMap::from_json("https://base.example/path1/path2/path3", json_map)
.unwrap();
assert_eq!(
import_map.imports.get("trailer/").unwrap(),
&vec!["https://base.example/atrailer/".to_string()]
);
// TODO: I'd be good to assert that warning was shown
}
#[test]
fn parse_addresses_other_invalid() {
// Should ignore unprefixed strings that are not absolute URLs.
for bad in &["bar", "\\bar", "~bar", "#bar", "?bar"] {
let json_map = json!({
"imports": {
"foo": bad
}
});
let import_map = ImportMap::from_json(
"https://base.example/path1/path2/path3",
&json_map.to_string(),
).unwrap();
assert!(import_map.imports.get("foo").unwrap().is_empty());
}
}
fn get_empty_import_map() -> ImportMap {
ImportMap {
base_url: "https://example.com/app/main.ts".to_string(),
imports: IndexMap::new(),
scopes: IndexMap::new(),
}
}
fn assert_resolve(
result: Result<Option<ModuleSpecifier>, ImportMapError>,
expected_url: &str,
) {
let maybe_url = result
.unwrap_or_else(|err| panic!("ImportMap::resolve failed: {:?}", err));
let resolved_url =
maybe_url.unwrap_or_else(|| panic!("Unexpected None resolved URL"));
assert_eq!(resolved_url, expected_url.to_string());
}
#[test]
fn resolve_unmapped_relative_specifiers() {
let referrer_url = "https://example.com/js/script.ts";
let import_map = get_empty_import_map();
// Should resolve ./ specifiers as URLs.
assert_resolve(
import_map.resolve("./foo", referrer_url),
"https://example.com/js/foo",
);
assert_resolve(
import_map.resolve("./foo/bar", referrer_url),
"https://example.com/js/foo/bar",
);
assert_resolve(
import_map.resolve("./foo/../bar", referrer_url),
"https://example.com/js/bar",
);
assert_resolve(
import_map.resolve("./foo/../../bar", referrer_url),
"https://example.com/bar",
);
// Should resolve ../ specifiers as URLs.
assert_resolve(
import_map.resolve("../foo", referrer_url),
"https://example.com/foo",
);
assert_resolve(
import_map.resolve("../foo/bar", referrer_url),
"https://example.com/foo/bar",
);
assert_resolve(
import_map.resolve("../../../foo/bar", referrer_url),
"https://example.com/foo/bar",
);
}
#[test]
fn resolve_unmapped_absolute_specifiers() {
let referrer_url = "https://example.com/js/script.ts";
let import_map = get_empty_import_map();
// Should resolve / specifiers as URLs.
assert_resolve(
import_map.resolve("/foo", referrer_url),
"https://example.com/foo",
);
assert_resolve(
import_map.resolve("/foo/bar", referrer_url),
"https://example.com/foo/bar",
);
assert_resolve(
import_map.resolve("../../foo/bar", referrer_url),
"https://example.com/foo/bar",
);
assert_resolve(
import_map.resolve("/../foo/../bar", referrer_url),
"https://example.com/bar",
);
// Should parse absolute fetch-scheme URLs.
assert_resolve(
import_map.resolve("https://example.net", referrer_url),
"https://example.net/",
);
assert_resolve(
import_map.resolve("https://ex%41mple.com/", referrer_url),
"https://example.com/",
);
assert_resolve(
import_map.resolve("https:example.org", referrer_url),
"https://example.org/",
);
assert_resolve(
import_map.resolve("https://///example.com///", referrer_url),
"https://example.com///",
);
}
#[test]
fn resolve_unmapped_bad_specifiers() {
let referrer_url = "https://example.com/js/script.ts";
let import_map = get_empty_import_map();
// Should fail for absolute non-fetch-scheme URLs.
assert!(import_map.resolve("about:good", referrer_url).is_err());
assert!(import_map.resolve("mailto:bad", referrer_url).is_err());
assert!(import_map.resolve("import:bad", referrer_url).is_err());
assert!(import_map.resolve("javascript:bad", referrer_url).is_err());
assert!(import_map.resolve("wss:bad", referrer_url).is_err());
// Should fail for string not parseable as absolute URLs and not starting with ./, ../ or /.
assert!(import_map.resolve("foo", referrer_url).is_err());
assert!(import_map.resolve("\\foo", referrer_url).is_err());
assert!(import_map.resolve(":foo", referrer_url).is_err());
assert!(import_map.resolve("@foo", referrer_url).is_err());
assert!(import_map.resolve("%2E/foo", referrer_url).is_err());
assert!(import_map.resolve("%2E%2Efoo", referrer_url).is_err());
assert!(import_map.resolve(".%2Efoo", referrer_url).is_err());
assert!(
import_map
.resolve("https://ex ample.org", referrer_url)
.is_err()
);
assert!(
import_map
.resolve("https://example.org:deno", referrer_url)
.is_err()
);
assert!(
import_map
.resolve("https://[example.org]", referrer_url)
.is_err()
);
}
#[test]
fn resolve_imports_mapped() {
let base_url = "https://example.com/app/main.ts";
let referrer_url = "https://example.com/js/script.ts";
// Should fail when mapping is to an empty array.
let json_map = r#"{
"imports": {
"moment": null,
"lodash": []
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
assert!(import_map.resolve("moment", referrer_url).is_err());
assert!(import_map.resolve("lodash", referrer_url).is_err());
}
#[test]
fn resolve_imports_package_like_modules() {
let base_url = "https://example.com/app/main.ts";
let referrer_url = "https://example.com/js/script.ts";
let json_map = r#"{
"imports": {
"moment": "/deps/moment/src/moment.js",
"moment/": "/deps/moment/src/",
"lodash-dot": "./deps/lodash-es/lodash.js",
"lodash-dot/": "./deps/lodash-es/",
"lodash-dotdot": "../deps/lodash-es/lodash.js",
"lodash-dotdot/": "../deps/lodash-es/",
"nowhere/": []
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
// Should work for package main modules.
assert_resolve(
import_map.resolve("moment", referrer_url),
"https://example.com/deps/moment/src/moment.js",
);
assert_resolve(
import_map.resolve("lodash-dot", referrer_url),
"https://example.com/app/deps/lodash-es/lodash.js",
);
assert_resolve(
import_map.resolve("lodash-dotdot", referrer_url),
"https://example.com/deps/lodash-es/lodash.js",
);
// Should work for package submodules.
assert_resolve(
import_map.resolve("moment/foo", referrer_url),
"https://example.com/deps/moment/src/foo",
);
assert_resolve(
import_map.resolve("lodash-dot/foo", referrer_url),
"https://example.com/app/deps/lodash-es/foo",
);
assert_resolve(
import_map.resolve("lodash-dotdot/foo", referrer_url),
"https://example.com/deps/lodash-es/foo",
);
// Should work for package names that end in a slash.
assert_resolve(
import_map.resolve("moment/", referrer_url),
"https://example.com/deps/moment/src/",
);
// Should fail for package modules that are not declared.
assert!(import_map.resolve("underscore/", referrer_url).is_err());
assert!(import_map.resolve("underscore/foo", referrer_url).is_err());
// Should fail for package submodules that map to nowhere.
assert!(import_map.resolve("nowhere/foo", referrer_url).is_err());
}
#[test]
fn resolve_imports_tricky_specifiers() {
let base_url = "https://example.com/app/main.ts";
let referrer_url = "https://example.com/js/script.ts";
let json_map = r#"{
"imports": {
"package/withslash": "/deps/package-with-slash/index.mjs",
"not-a-package": "/lib/not-a-package.mjs",
".": "/lib/dot.mjs",
"..": "/lib/dotdot.mjs",
"..\\\\": "/lib/dotdotbackslash.mjs",
"%2E": "/lib/percent2e.mjs",
"%2F": "/lib/percent2f.mjs"
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
// Should work for explicitly-mapped specifiers that happen to have a slash.
assert_resolve(
import_map.resolve("package/withslash", referrer_url),
"https://example.com/deps/package-with-slash/index.mjs",
);
// Should work when the specifier has punctuation.
assert_resolve(
import_map.resolve(".", referrer_url),
"https://example.com/lib/dot.mjs",
);
assert_resolve(
import_map.resolve("..", referrer_url),
"https://example.com/lib/dotdot.mjs",
);
assert_resolve(
import_map.resolve("..\\\\", referrer_url),
"https://example.com/lib/dotdotbackslash.mjs",
);
assert_resolve(
import_map.resolve("%2E", referrer_url),
"https://example.com/lib/percent2e.mjs",
);
assert_resolve(
import_map.resolve("%2F", referrer_url),
"https://example.com/lib/percent2f.mjs",
);
// Should fail for attempting to get a submodule of something not declared with a trailing slash.
assert!(
import_map
.resolve("not-a-package/foo", referrer_url)
.is_err()
);
}
#[test]
fn resolve_imports_url_like_specifier() {
let base_url = "https://example.com/app/main.ts";
let referrer_url = "https://example.com/js/script.ts";
let json_map = r#"{
"imports": {
"/node_modules/als-polyfill/index.mjs": "std:kv-storage",
"/lib/foo.mjs": "./more/bar.mjs",
"./dotrelative/foo.mjs": "/lib/dot.mjs",
"../dotdotrelative/foo.mjs": "/lib/dotdot.mjs",
"/lib/no.mjs": null,
"./dotrelative/no.mjs": [],
"/": "/lib/slash-only/",
"./": "/lib/dotslash-only/",
"/test/": "/lib/url-trailing-slash/",
"./test/": "/lib/url-trailing-slash-dot/",
"/test": "/lib/test1.mjs",
"../test": "/lib/test2.mjs"
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
// Should remap to other URLs.
assert_resolve(
import_map.resolve("https://example.com/lib/foo.mjs", referrer_url),
"https://example.com/app/more/bar.mjs",
);
assert_resolve(
import_map.resolve("https://///example.com/lib/foo.mjs", referrer_url),
"https://example.com/app/more/bar.mjs",
);
assert_resolve(
import_map.resolve("/lib/foo.mjs", referrer_url),
"https://example.com/app/more/bar.mjs",
);
assert_resolve(
import_map
.resolve("https://example.com/app/dotrelative/foo.mjs", referrer_url),
"https://example.com/lib/dot.mjs",
);
assert_resolve(
import_map.resolve("../app/dotrelative/foo.mjs", referrer_url),
"https://example.com/lib/dot.mjs",
);
assert_resolve(
import_map
.resolve("https://example.com/dotdotrelative/foo.mjs", referrer_url),
"https://example.com/lib/dotdot.mjs",
);
assert_resolve(
import_map.resolve("../dotdotrelative/foo.mjs", referrer_url),
"https://example.com/lib/dotdot.mjs",
);
// Should fail for URLs that remap to empty arrays.
assert!(
import_map
.resolve("https://example.com/lib/no.mjs", referrer_url)
.is_err()
);
assert!(import_map.resolve("/lib/no.mjs", referrer_url).is_err());
assert!(import_map.resolve("../lib/no.mjs", referrer_url).is_err());
assert!(
import_map
.resolve("https://example.com/app/dotrelative/no.mjs", referrer_url)
.is_err()
);
assert!(
import_map
.resolve("/app/dotrelative/no.mjs", referrer_url)
.is_err()
);
assert!(
import_map
.resolve("../app/dotrelative/no.mjs", referrer_url)
.is_err()
);
// Should remap URLs that are just composed from / and ..
assert_resolve(
import_map.resolve("https://example.com/", referrer_url),
"https://example.com/lib/slash-only/",
);
assert_resolve(
import_map.resolve("/", referrer_url),
"https://example.com/lib/slash-only/",
);
assert_resolve(
import_map.resolve("../", referrer_url),
"https://example.com/lib/slash-only/",
);
assert_resolve(
import_map.resolve("https://example.com/app/", referrer_url),
"https://example.com/lib/dotslash-only/",
);
assert_resolve(
import_map.resolve("/app/", referrer_url),
"https://example.com/lib/dotslash-only/",
);
assert_resolve(
import_map.resolve("../app/", referrer_url),
"https://example.com/lib/dotslash-only/",
);
// Should remap URLs that are prefix-matched by keys with trailing slashes.
assert_resolve(
import_map.resolve("/test/foo.mjs", referrer_url),
"https://example.com/lib/url-trailing-slash/foo.mjs",
);
assert_resolve(
import_map.resolve("https://example.com/app/test/foo.mjs", referrer_url),
"https://example.com/lib/url-trailing-slash-dot/foo.mjs",
);
// Should use the last entry's address when URL-like specifiers parse to the same absolute URL.
//
// NOTE: this works properly because of "preserve_order" feature flag to "serde_json" crate
assert_resolve(
import_map.resolve("/test", referrer_url),
"https://example.com/lib/test2.mjs",
);
}
#[test]
fn resolve_imports_overlapping_entities_with_trailing_slashes() {
let base_url = "https://example.com/app/main.ts";
let referrer_url = "https://example.com/js/script.ts";
// Should favor the most-specific key (no empty arrays).
{
let json_map = r#"{
"imports": {
"a": "/1",
"a/": "/2/",
"a/b": "/3",
"a/b/": "/4/"
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
assert_resolve(
import_map.resolve("a", referrer_url),
"https://example.com/1",
);
assert_resolve(
import_map.resolve("a/", referrer_url),
"https://example.com/2/",
);
assert_resolve(
import_map.resolve("a/b", referrer_url),
"https://example.com/3",
);
assert_resolve(
import_map.resolve("a/b/", referrer_url),
"https://example.com/4/",
);
assert_resolve(
import_map.resolve("a/b/c", referrer_url),
"https://example.com/4/c",
);
}
// Should favor the most-specific key when empty arrays are involved for less-specific keys.
{
let json_map = r#"{
"imports": {
"a": [],
"a/": [],
"a/b": "/3",
"a/b/": "/4/"
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
assert!(import_map.resolve("a", referrer_url).is_err());
assert!(import_map.resolve("a/", referrer_url).is_err());
assert!(import_map.resolve("a/x", referrer_url).is_err());
assert_resolve(
import_map.resolve("a/b", referrer_url),
"https://example.com/3",
);
assert_resolve(
import_map.resolve("a/b/", referrer_url),
"https://example.com/4/",
);
assert_resolve(
import_map.resolve("a/b/c", referrer_url),
"https://example.com/4/c",
);
assert!(import_map.resolve("a/x/c", referrer_url).is_err());
}
}
#[test]
fn resolve_scopes_map_to_empty_array() {
let base_url = "https://example.com/app/main.ts";
let referrer_url = "https://example.com/js";
let json_map = r#"{
"scopes": {
"/js/": {
"moment": "null",
"lodash": []
}
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
assert!(import_map.resolve("moment", referrer_url).is_err());
assert!(import_map.resolve("lodash", referrer_url).is_err());
}
#[test]
fn resolve_scopes_exact_vs_prefix_matching() {
let base_url = "https://example.com/app/main.ts";
let json_map = r#"{
"scopes": {
"/js": {
"moment": "/only-triggered-by-exact/moment",
"moment/": "/only-triggered-by-exact/moment/"
},
"/js/": {
"moment": "/triggered-by-any-subpath/moment",
"moment/": "/triggered-by-any-subpath/moment/"
}
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
let js_non_dir = "https://example.com/js";
let js_in_dir = "https://example.com/js/app.mjs";
let with_js_prefix = "https://example.com/jsiscool";
assert_resolve(
import_map.resolve("moment", js_non_dir),
"https://example.com/only-triggered-by-exact/moment",
);
assert_resolve(
import_map.resolve("moment/foo", js_non_dir),
"https://example.com/only-triggered-by-exact/moment/foo",
);
assert_resolve(
import_map.resolve("moment", js_in_dir),
"https://example.com/triggered-by-any-subpath/moment",
);
assert_resolve(
import_map.resolve("moment/foo", js_in_dir),
"https://example.com/triggered-by-any-subpath/moment/foo",
);
assert!(import_map.resolve("moment", with_js_prefix).is_err());
assert!(import_map.resolve("moment/foo", with_js_prefix).is_err());
}
#[test]
fn resolve_scopes_only_exact_in_map() {
let base_url = "https://example.com/app/main.ts";
let json_map = r#"{
"scopes": {
"/js": {
"moment": "/only-triggered-by-exact/moment",
"moment/": "/only-triggered-by-exact/moment/"
}
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
// Should match correctly when only an exact match is in the map.
let js_non_dir = "https://example.com/js";
let js_in_dir = "https://example.com/js/app.mjs";
let with_js_prefix = "https://example.com/jsiscool";
assert_resolve(
import_map.resolve("moment", js_non_dir),
"https://example.com/only-triggered-by-exact/moment",
);
assert_resolve(
import_map.resolve("moment/foo", js_non_dir),
"https://example.com/only-triggered-by-exact/moment/foo",
);
assert!(import_map.resolve("moment", js_in_dir).is_err());
assert!(import_map.resolve("moment/foo", js_in_dir).is_err());
assert!(import_map.resolve("moment", with_js_prefix).is_err());
assert!(import_map.resolve("moment/foo", with_js_prefix).is_err());
}
#[test]
fn resolve_scopes_only_prefix_in_map() {
let base_url = "https://example.com/app/main.ts";
let json_map = r#"{
"scopes": {
"/js/": {
"moment": "/triggered-by-any-subpath/moment",
"moment/": "/triggered-by-any-subpath/moment/"
}
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
// Should match correctly when only a prefix match is in the map.
let js_non_dir = "https://example.com/js";
let js_in_dir = "https://example.com/js/app.mjs";
let with_js_prefix = "https://example.com/jsiscool";
assert!(import_map.resolve("moment", js_non_dir).is_err());
assert!(import_map.resolve("moment/foo", js_non_dir).is_err());
assert_resolve(
import_map.resolve("moment", js_in_dir),
"https://example.com/triggered-by-any-subpath/moment",
);
assert_resolve(
import_map.resolve("moment/foo", js_in_dir),
"https://example.com/triggered-by-any-subpath/moment/foo",
);
assert!(import_map.resolve("moment", with_js_prefix).is_err());
assert!(import_map.resolve("moment/foo", with_js_prefix).is_err());
}
#[test]
fn resolve_scopes_package_like() {
let base_url = "https://example.com/app/main.ts";
let json_map = r#"{
"imports": {
"moment": "/node_modules/moment/src/moment.js",
"moment/": "/node_modules/moment/src/",
"lodash-dot": "./node_modules/lodash-es/lodash.js",
"lodash-dot/": "./node_modules/lodash-es/",
"lodash-dotdot": "../node_modules/lodash-es/lodash.js",
"lodash-dotdot/": "../node_modules/lodash-es/"
},
"scopes": {
"/": {
"moment": "/node_modules_3/moment/src/moment.js",
"vue": "/node_modules_3/vue/dist/vue.runtime.esm.js"
},
"/js/": {
"lodash-dot": "./node_modules_2/lodash-es/lodash.js",
"lodash-dot/": "./node_modules_2/lodash-es/",
"lodash-dotdot": "../node_modules_2/lodash-es/lodash.js",
"lodash-dotdot/": "../node_modules_2/lodash-es/"
}
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
// Should match correctly when only a prefix match is in the map.
let js_in_dir = "https://example.com/js/app.mjs";
let top_level = "https://example.com/app.mjs";
// Should resolve scoped.
assert_resolve(
import_map.resolve("lodash-dot", js_in_dir),
"https://example.com/app/node_modules_2/lodash-es/lodash.js",
);
assert_resolve(
import_map.resolve("lodash-dotdot", js_in_dir),
"https://example.com/node_modules_2/lodash-es/lodash.js",
);
assert_resolve(
import_map.resolve("lodash-dot/foo", js_in_dir),
"https://example.com/app/node_modules_2/lodash-es/foo",
);
assert_resolve(
import_map.resolve("lodash-dotdot/foo", js_in_dir),
"https://example.com/node_modules_2/lodash-es/foo",
);
// Should apply best scope match.
assert_resolve(
import_map.resolve("moment", top_level),
"https://example.com/node_modules_3/moment/src/moment.js",
);
assert_resolve(
import_map.resolve("moment", js_in_dir),
"https://example.com/node_modules_3/moment/src/moment.js",
);
assert_resolve(
import_map.resolve("vue", js_in_dir),
"https://example.com/node_modules_3/vue/dist/vue.runtime.esm.js",
);
// Should fallback to "imports".
assert_resolve(
import_map.resolve("moment/foo", top_level),
"https://example.com/node_modules/moment/src/foo",
);
assert_resolve(
import_map.resolve("moment/foo", js_in_dir),
"https://example.com/node_modules/moment/src/foo",
);
assert_resolve(
import_map.resolve("lodash-dot", top_level),
"https://example.com/app/node_modules/lodash-es/lodash.js",
);
assert_resolve(
import_map.resolve("lodash-dotdot", top_level),
"https://example.com/node_modules/lodash-es/lodash.js",
);
assert_resolve(
import_map.resolve("lodash-dot/foo", top_level),
"https://example.com/app/node_modules/lodash-es/foo",
);
assert_resolve(
import_map.resolve("lodash-dotdot/foo", top_level),
"https://example.com/node_modules/lodash-es/foo",
);
// Should still fail for package-like specifiers that are not declared.
assert!(import_map.resolve("underscore/", js_in_dir).is_err());
assert!(import_map.resolve("underscore/foo", js_in_dir).is_err());
}
#[test]
fn resolve_scopes_inheritance() {
// https://github.com/WICG/import-maps#scope-inheritance
let base_url = "https://example.com/app/main.ts";
let json_map = r#"{
"imports": {
"a": "/a-1.mjs",
"b": "/b-1.mjs",
"c": "/c-1.mjs"
},
"scopes": {
"/scope2/": {
"a": "/a-2.mjs"
},
"/scope2/scope3/": {
"b": "/b-3.mjs"
}
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
let scope_1_url = "https://example.com/scope1/foo.mjs";
let scope_2_url = "https://example.com/scope2/foo.mjs";
let scope_3_url = "https://example.com/scope2/scope3/foo.mjs";
// Should fall back to "imports" when none match.
assert_resolve(
import_map.resolve("a", scope_1_url),
"https://example.com/a-1.mjs",
);
assert_resolve(
import_map.resolve("b", scope_1_url),
"https://example.com/b-1.mjs",
);
assert_resolve(
import_map.resolve("c", scope_1_url),
"https://example.com/c-1.mjs",
);
// Should use a direct scope override.
assert_resolve(
import_map.resolve("a", scope_2_url),
"https://example.com/a-2.mjs",
);
assert_resolve(
import_map.resolve("b", scope_2_url),
"https://example.com/b-1.mjs",
);
assert_resolve(
import_map.resolve("c", scope_2_url),
"https://example.com/c-1.mjs",
);
// Should use an indirect scope override.
assert_resolve(
import_map.resolve("a", scope_3_url),
"https://example.com/a-2.mjs",
);
assert_resolve(
import_map.resolve("b", scope_3_url),
"https://example.com/b-3.mjs",
);
assert_resolve(
import_map.resolve("c", scope_3_url),
"https://example.com/c-1.mjs",
);
}
#[test]
fn resolve_scopes_relative_url_keys() {
// https://github.com/WICG/import-maps#scope-inheritance
let base_url = "https://example.com/app/main.ts";
let json_map = r#"{
"imports": {
"a": "/a-1.mjs",
"b": "/b-1.mjs",
"c": "/c-1.mjs"
},
"scopes": {
"": {
"a": "/a-empty-string.mjs"
},
"./": {
"b": "/b-dot-slash.mjs"
},
"../": {
"c": "/c-dot-dot-slash.mjs"
}
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
let in_same_dir_as_map = "https://example.com/app/foo.mjs";
let in_dir_above_map = "https://example.com/foo.mjs";
// Should resolve an empty string scope using the import map URL.
assert_resolve(
import_map.resolve("a", base_url),
"https://example.com/a-empty-string.mjs",
);
assert_resolve(
import_map.resolve("a", in_same_dir_as_map),
"https://example.com/a-1.mjs",
);
// Should resolve a ./ scope using the import map URL's directory.
assert_resolve(
import_map.resolve("b", base_url),
"https://example.com/b-dot-slash.mjs",
);
assert_resolve(
import_map.resolve("b", in_same_dir_as_map),
"https://example.com/b-dot-slash.mjs",
);
// Should resolve a ../ scope using the import map URL's directory.
assert_resolve(
import_map.resolve("c", base_url),
"https://example.com/c-dot-dot-slash.mjs",
);
assert_resolve(
import_map.resolve("c", in_same_dir_as_map),
"https://example.com/c-dot-dot-slash.mjs",
);
assert_resolve(
import_map.resolve("c", in_dir_above_map),
"https://example.com/c-dot-dot-slash.mjs",
);
}
#[test]
fn cant_resolve_to_built_in() {
let base_url = "https://example.com/app/main.ts";
let import_map = ImportMap::from_json(base_url, "{}").unwrap();
assert!(import_map.resolve("std:blank", base_url).is_err());
}
#[test]
fn resolve_builtins_remap() {
let base_url = "https://example.com/app/main.ts";
let json_map = r#"{
"imports": {
"std:blank": "./blank.mjs",
"std:none": "./none.mjs"
}
}"#;
let import_map = ImportMap::from_json(base_url, json_map).unwrap();
assert_resolve(
import_map.resolve("std:blank", base_url),
"https://example.com/app/blank.mjs",
);
assert_resolve(
import_map.resolve("std:none", base_url),
"https://example.com/app/none.mjs",
);
}
}