rea/src/application/javascript/JavaScriptArtifactPathResol...

1247 lines
46 KiB
TypeScript

import { isBuiltin } from "node:module";
import { posix } from "node:path";
import {
normalizeJoinedLogicalPath,
logicalPathEscapesRoot,
} from "../../domain/artifactIdentity.js";
import type { JavaScriptArtifactFile } from "../../domain/javascript/javascriptArtifactFiles.js";
import {
admitsCanonicalPathSyntax,
hasScheme,
htmlUrlText,
looksExternal,
percentDecodeUrlPath,
stripQueryAndFragment,
} from "../../domain/artifactPathSyntax.js";
import {
classifyPackageImportSpecifier,
classifyPackageImportTargetString,
lookupExactPackageImport,
} from "../../domain/javascript/javascriptPackageImportAliases.js";
import { typescriptEmittedSourceCandidates } from "../TypeScriptEmittedSourceCandidates.js";
import { isRelativeJavaScriptModuleSpecifier } from "../../domain/javascript/javascriptModuleSpecifier.js";
type ArtifactPathResolutionContext =
| "package-entrypoint"
| "filesystem-expression"
| "module-specifier"
| "html-reference"
| "url-reference";
type UnresolvedArtifactPathStatus =
| "not-found"
| "unavailable"
| "external"
| "rejected";
interface ArtifactPathResolutionBase {
readonly declared_path: string;
readonly resolution_context: ArtifactPathResolutionContext;
readonly limitations: readonly string[];
}
/** Explicit outcome of artifact-confined path resolution. */
export type ArtifactPathResolution = ArtifactPathResolutionBase &
(
| {
readonly resolved_path: string;
readonly resolution_status: "resolved";
}
| {
readonly resolved_path: null;
readonly resolution_status: UnresolvedArtifactPathStatus;
}
);
/** Values needed to resolve a declaration without filesystem access. */
export interface ResolveArtifactPathInput {
readonly declaredPath: string;
readonly sourcePath: string;
readonly context: ArtifactPathResolutionContext;
readonly files: ReadonlyMap<string, JavaScriptArtifactFile>;
readonly htmlBaseHref?: string | null;
/** Import/require semantics; runtime resource paths omit this provenance. */
readonly moduleKind?: "import" | "require";
}
/** Loader selected by a source-level JavaScript module reference. */
export const moduleKindForJavaScriptReference = (
sourcePath: string,
referenceKind: "static-import" | "dynamic-import" | "require",
resolutionMode?: "import" | "require",
): "import" | "require" =>
resolutionMode ??
(referenceKind === "require" ||
(referenceKind === "static-import" &&
posix.extname(sourcePath).toLowerCase() === ".cts")
? "require"
: "import");
/** Preserve an explicitly retained loader mode on graph relationships. */
export const moduleResolutionModeProperties = (
mode: "import" | "require" | undefined,
): Readonly<Record<string, "import" | "require">> =>
mode === undefined ? {} : { resolution_mode: mode };
const EXTENSIONS = [
".js",
".cjs",
".mjs",
".ts",
".tsx",
".json",
".html",
".node",
];
// A private URL namespace for logical artifact paths, including root packages.
// Encode inventory segments so filename punctuation remains literal in the base.
const PACKAGE_ARTIFACT_ROOT_URL = new URL("file:///rea-artifact/");
const packageMetadataUrl = (path: string): URL =>
new URL(
path.split("/").map(encodeURIComponent).join("/"),
PACKAGE_ARTIFACT_ROOT_URL,
);
interface PackageFileTarget {
readonly value: string;
/** Normalized URL retained until loader-specific validation and file lookup. */
readonly url: URL;
}
type CandidateResolution =
| {
readonly resolvedPath: string;
readonly status: "resolved";
readonly limitations: readonly string[];
}
| {
readonly resolvedPath: null;
readonly status: UnresolvedArtifactPathStatus;
readonly limitations: readonly string[];
/** A truthy package main failed, so Node must not try ancestor packages. */
readonly stopsAncestorLookup?: true;
};
/** Resolve one declaration while preserving its original input identity. */
export const resolveArtifactPathByContext = (
input: ResolveArtifactPathInput,
): ArtifactPathResolution => {
const resolution = resolvePath(input);
return resolution.status === "resolved"
? {
declared_path: input.declaredPath,
resolution_context: input.context,
resolved_path: resolution.resolvedPath,
resolution_status: resolution.status,
limitations: resolution.limitations,
}
: {
declared_path: input.declaredPath,
resolution_context: input.context,
resolved_path: null,
resolution_status: resolution.status,
limitations: resolution.limitations,
};
};
const resolvePath = (input: ResolveArtifactPathInput): CandidateResolution => {
// Package import keys are literal lookup identities, not artifact paths.
// The selected target is admitted separately by resolveExportTarget.
if (
input.context === "module-specifier" &&
input.declaredPath.startsWith("#")
)
return resolvePackageImportAlias(input, input.declaredPath);
const candidate = contextualCandidate(input);
if (typeof candidate !== "string") return candidate;
const confined = confineCandidate(candidate);
if (typeof confined !== "string") return confined;
if (input.context === "html-reference" || input.context === "url-reference")
return (
resolveFileCandidates(input, [confined]) ??
unresolvedCandidate("not-found", [
`The exact ${input.context === "html-reference" ? "HTML" : "URL"} resource ${confined} was not found among the selected application files.`,
])
);
const isModuleUrlPath =
input.context === "module-specifier" &&
(isRelativeJavaScriptModuleSpecifier(input.declaredPath) ||
input.declaredPath.startsWith("/") ||
fileUrlPath(input.declaredPath) !== undefined);
const emittedSourceCandidates =
isModuleUrlPath && input.moduleKind !== undefined
? typescriptEmittedSourceCandidates(input.sourcePath, confined)
: [];
if (isModuleUrlPath && input.moduleKind === "import")
return (
resolveFileCandidates(input, [confined, ...emittedSourceCandidates]) ??
notFoundCandidate()
);
return resolveCandidate(input, confined, new Set(), emittedSourceCandidates);
};
const rejectDeclaration = (declared: string): CandidateResolution | null => {
if (declared.length === 0)
return unresolvedCandidate("rejected", ["The declared path is empty."]);
if (declared.includes("\0") || declared.includes("\\"))
return unresolvedCandidate("rejected", [
"NUL and backslash path syntax are not admitted for canonical artifact paths.",
]);
if (!admitsCanonicalPathSyntax(declared))
return unresolvedCandidate("rejected", [
"Encoded dot or separator bytes are rejected before artifact path resolution.",
]);
return null;
};
const contextualCandidate = (
input: ResolveArtifactPathInput,
): string | CandidateResolution => {
const { context } = input;
// URL suffixes are evidence, not pathname bytes. Admit only the path while
// retaining the complete declaration on the resolution result.
let declared =
context === "url-reference"
? stripQueryAndFragment(input.declaredPath)
: input.declaredPath;
if (context === "module-specifier" && /^file:/iu.test(declared)) {
const fileUrl = fileUrlPath(declared);
if (fileUrl === null)
return unresolvedCandidate("rejected", [
"The file URL path contains syntax that cannot identify a canonical artifact path.",
]);
if (fileUrl !== undefined)
return input.moduleKind === "require"
? unresolvedCandidate("external", [
"Node CommonJS require specifiers do not resolve file URLs as local artifacts.",
])
: fileUrl;
return unresolvedCandidate("external", [
"File URL authorities or invalid file URL forms cannot be mapped to artifact-local paths.",
]);
}
const rejected = rejectDeclaration(declared);
if (rejected !== null) return rejected;
if (context === "html-reference") return htmlCandidate(input);
if (context === "module-specifier") {
if (hasScheme(declared))
return unresolvedCandidate("external", [
"URL and Node builtin schemes are outside static artifact module resolution.",
]);
if (
!isRelativeJavaScriptModuleSpecifier(declared) &&
!declared.startsWith("/")
)
return bareModuleCandidate(input);
if (input.moduleKind !== "require") {
const decoded = decodeModuleUrlPath(stripQueryAndFragment(declared));
if (typeof decoded !== "string") return decoded;
declared = decoded;
}
} else if (looksExternal(declared))
return unresolvedCandidate("external", [
"URL schemes and protocol-relative URLs are outside this local artifact path context.",
]);
if (context === "url-reference") {
// A file base gives raw drive-letter segments special meaning, including
// after dot segments reach the root. Parse before percent decoding so an
// encoded drive-like filename stays literal. The synthetic base establishes
// only this ambiguity; it does not establish the source's runtime scheme.
const sourceUrlPath = input.sourcePath
.split("/")
.map(encodeURIComponent)
.join("/");
const fileReference = URL.parse(declared, `file:///${sourceUrlPath}`);
if (fileReference === null)
return unresolvedCandidate("rejected", [
"The URL reference cannot be parsed as a URL.",
]);
if (/^\/[A-Za-z]:(?:\/|$)/u.test(fileReference.pathname))
return unresolvedCandidate("unavailable", [
"Drive-letter URL paths cannot be mapped without the source URL scheme and filesystem root.",
]);
const decoded = percentDecodeUrlPath(declared);
if (decoded === null)
return unresolvedCandidate("rejected", [
"The URL reference path percent-decodes to NUL or to bytes that are not UTF-8.",
]);
// Empty URL segments are significant; canonical filesystem paths cannot
// preserve them, including their interaction with following dot segments.
if (decoded.includes("//"))
return unresolvedCandidate("unavailable", [
"Repeated URL path separators cannot be mapped to an exact canonical artifact file.",
]);
return exactUrlPathCandidate(
decoded,
relativeCandidate(input.sourcePath, decoded),
"URL",
);
}
const candidate = relativeCandidate(input.sourcePath, declared);
return context === "module-specifier" && input.moduleKind === "import"
? exactUrlPathCandidate(declared, candidate, "module URL")
: candidate;
};
const decodeModuleUrlPath = (
declared: string,
): string | CandidateResolution => {
let decoded: string;
try {
decoded = decodeURIComponent(declared);
} catch (cause: unknown) {
void cause;
return unresolvedCandidate("rejected", [
"The module URL path contains malformed percent encoding.",
]);
}
return decoded.includes("\0")
? unresolvedCandidate("rejected", [
"The decoded module URL path contains NUL.",
])
: decoded;
};
const relativeCandidate = (sourcePath: string, declared: string): string =>
declared.startsWith("/")
? declared.slice(1)
: posix.join(posix.dirname(sourcePath), declared);
const bareModuleCandidate = (
input: ResolveArtifactPathInput,
): CandidateResolution => {
const request = parseBarePackageRequest(input.declaredPath, input.moduleKind);
if (request === null || isBuiltin(input.declaredPath))
return unresolvedCandidate("external", [
"The bare specifier is a Node builtin or invalid package name.",
]);
if (
input.moduleKind === "import" &&
(request.packageName.startsWith(".") || request.packageName.includes("%"))
)
return unresolvedCandidate("rejected", [
"ES module package names cannot start with a dot or contain percent signs.",
]);
const self = packageSelfCandidate(input, request);
if (self !== null) return self;
// ESM constructs a package.json URL; its parsed identity is not a literal
// dependency directory. Self resolution above still uses the exact name.
let dependencyDirectory = posix.join("node_modules", request.packageName);
if (input.moduleKind === "import") {
const metadataUrl = new URL(
`./node_modules/${request.packageName}/package.json`,
PACKAGE_ARTIFACT_ROOT_URL,
);
if (!metadataUrl.pathname.endsWith("/package.json"))
return unresolvedCandidate("external", [
`ESM package name ${JSON.stringify(request.packageName)} produces a dependency URL whose pathname does not name package.json; no literal or suffix-stripped package directory was searched.`,
]);
const decoded = decodeModuleUrlPath(
metadataUrl.pathname.slice(PACKAGE_ARTIFACT_ROOT_URL.pathname.length),
);
if (typeof decoded !== "string") return decoded;
const confined = confineCandidate(decoded);
if (typeof confined !== "string") return confined;
dependencyDirectory = posix.dirname(confined);
}
const source = input.files.get(input.sourcePath);
let directory = posix.dirname(input.sourcePath);
let nearestNotFound: CandidateResolution | null = null;
while (true) {
const candidate = posix.join(directory, dependencyDirectory);
const metadata = readPackageMetadata(input, candidate);
const continueCommonJsLookup =
input.moduleKind === "require" &&
(metadata.kind === "absent" ||
(metadata.kind === "available" &&
(!request.dependencyExportsEligible ||
metadata.exports === undefined)));
const packageResolution =
metadata.kind !== "absent" ||
hasContainerDirectory(input.files, candidate, source?.container_sha256)
? resolvePackageDependency(input, candidate, request, metadata)
: request.packageSubpath === "." && input.moduleKind !== "import"
? resolveFileCandidates(
input,
fileCandidates(candidate, input.moduleKind),
)
: null;
if (packageResolution !== null) {
if (
continueCommonJsLookup &&
packageResolution.status === "not-found" &&
packageResolution.stopsAncestorLookup !== true
) {
nearestNotFound ??= packageResolution;
} else return packageResolution;
}
if (directory === "." || directory === "") break;
directory = posix.dirname(directory);
}
if (nearestNotFound !== null) return nearestNotFound;
return unresolvedCandidate("external", [
"No matching bare package was inventoried in an enclosing node_modules directory.",
]);
};
/**
* Nearest same-container package.json, stopping at a `node_modules` directory
* so a nested package cannot inherit an ancestor's map.
*/
const enclosingPackageMetadata = (
input: ResolveArtifactPathInput,
):
| Extract<PackageMetadata, { readonly kind: "available" | "unavailable" }>
| { readonly kind: "none" } => {
let directory = posix.dirname(input.sourcePath);
while (posix.basename(directory) !== "node_modules") {
const metadata = readPackageMetadata(input, directory);
if (metadata.kind !== "absent") return metadata;
if (directory === "." || directory === "") break;
directory = posix.dirname(directory);
}
return { kind: "none" };
};
/**
* Resolve an exact `#alias` against the importer's enclosing `imports` map.
* Selected relative targets are exact files; patterns, recursive aliases, and
* external-package targets stay unresolved without falling through.
*/
const resolvePackageImportAlias = (
input: ResolveArtifactPathInput,
specifier: string,
): CandidateResolution => {
const classified = classifyPackageImportSpecifier(specifier);
if (classified.kind === "invalid")
return unresolvedCandidate("rejected", [
`The package import specifier ${JSON.stringify(specifier)} is invalid; "#" and specifiers that end with "/" are not valid import aliases.`,
]);
if (classified.kind === "pattern")
return unresolvedCandidate("external", [
`Wildcard and directory package import aliases remain unresolved; ${JSON.stringify(specifier)} is not an exact imports key.`,
]);
const metadata = enclosingPackageMetadata(input);
if (metadata.kind === "none")
return unresolvedCandidate("external", [
`No enclosing package.json defines package import alias ${specifier}; package import maps do not cross package scopes.`,
]);
if (metadata.kind === "unavailable")
return unresolvedCandidate("unavailable", [metadata.limitation]);
const lookup = lookupExactPackageImport(metadata.imports, specifier);
if (lookup.kind === "undefined-map")
return unresolvedCandidate("external", [
`Package metadata ${metadata.path} does not define package import alias ${specifier}.`,
]);
if (lookup.kind === "missing-key")
return unresolvedCandidate("external", [
`Package metadata ${metadata.path} does not define package import alias ${specifier}; it declares ${lookup.declared.join(", ") || "no imports keys"}.`,
]);
if (lookup.kind === "pattern-unresolved")
return unresolvedCandidate("external", [
`Wildcard package import aliases remain unresolved; package metadata ${metadata.path} has no exact imports key ${specifier}.`,
]);
const selected = selectExportTarget(
lookup.target,
packageExportConditions(input.moduleKind),
packageMetadataUrl(metadata.path),
"imports",
);
if (selected.kind === "invalid" || selected.kind === "invalid-config")
return unresolvedCandidate("rejected", [
`Package metadata ${metadata.path}: ${selected.limitation}`,
]);
if (selected.kind === "unmatched")
return unresolvedCandidate("external", [
`Package metadata ${metadata.path} declares no imports target for alias ${specifier} under the active conditions; it declares ${selected.declared.join(", ") || "no conditions"}.`,
]);
if (selected.kind === "blocked")
return unresolvedCandidate("external", [
`Package metadata ${metadata.path} blocks import alias ${specifier} with ${selected.cause === "null" ? "an explicit null imports target" : "an empty imports target array"}.`,
]);
if (selected.kind === "non-relative") {
const target = classifyPackageImportTargetString(selected.value);
return unresolvedCandidate("external", [
target.kind === "recursive"
? `Recursive package import aliases remain unresolved; package metadata ${metadata.path} maps ${specifier} to ${JSON.stringify(target.target)}.`
: `External-package import targets remain unresolved; package metadata ${metadata.path} maps ${specifier} to ${JSON.stringify(target.target)}.`,
]);
}
return resolveExportTarget(input, metadata.path, selected, "imports");
};
/** Self references use the nearest package scope before dependency lookup. */
const packageSelfCandidate = (
input: ResolveArtifactPathInput,
request: BarePackageRequest,
): CandidateResolution | null => {
if (input.moduleKind === undefined || !input.files.has(input.sourcePath))
return null;
const metadata = enclosingPackageMetadata(input);
if (metadata.kind === "none") return null;
if (metadata.kind === "unavailable")
return unresolvedCandidate("unavailable", [metadata.limitation]);
if (metadata.name === undefined || metadata.exports === undefined)
return null;
let subpath: string;
if (input.moduleKind === "import") {
if (metadata.name !== request.packageName) return null;
subpath = request.packageSubpath;
} else if (input.declaredPath === metadata.name) {
subpath = ".";
} else if (input.declaredPath.startsWith(`${metadata.name}/`)) {
subpath = `.${input.declaredPath.slice(metadata.name.length)}`;
} else return null;
// Self resolution does not apply the CommonJS dependency-name matcher.
return resolvePackageExports(input, metadata.path, subpath, metadata.exports);
};
/** Resolve an installed dependency using one parsed request and its metadata. */
const resolvePackageDependency = (
input: ResolveArtifactPathInput,
directory: string,
request: BarePackageRequest,
metadata: PackageMetadata,
): CandidateResolution => {
const subpath = request.packageSubpath;
if (request.dependencyExportsEligible && metadata.kind === "unavailable")
return unresolvedCandidate("unavailable", [metadata.limitation]);
const rawExports =
request.dependencyExportsEligible && metadata.kind === "available"
? metadata.exports
: undefined;
if (rawExports === undefined) {
if (subpath === ".")
return (
(input.moduleKind === "import"
? null
: resolveFileCandidates(
input,
fileCandidates(directory, input.moduleKind),
)) ?? resolveDirectory(input, directory, metadata, new Set())
);
// Resolve legacy ESM subpaths as URLs: suffixes identify the loaded module
// but do not belong to its filesystem path. Keep the declared specifier on
// the relationship; only derive the inventory lookup path here.
const fileSubpath =
input.moduleKind === "import"
? decodeModuleUrlPath(stripQueryAndFragment(subpath.slice(2)))
: subpath.slice(2);
if (typeof fileSubpath !== "string") return fileSubpath;
const candidate = confineCandidate(`${directory}/${fileSubpath}`);
if (input.moduleKind === "import" && typeof candidate === "string")
return resolveLegacyEsmSubpath(input, candidate);
return typeof candidate === "string"
? resolveCandidate(input, candidate, new Set())
: candidate;
}
return resolvePackageExports(
input,
posix.join(directory, "package.json"),
subpath,
rawExports,
);
};
/** Legacy ESM package subpaths are exact URLs, never extension or directory lookups. */
const resolveLegacyEsmSubpath = (
input: ResolveArtifactPathInput,
candidate: string,
): CandidateResolution => {
const file = resolveFileCandidates(input, [candidate]);
if (file !== null) return file;
const source = input.files.get(input.sourcePath);
if (hasContainerDirectory(input.files, candidate, source?.container_sha256))
return unresolvedCandidate("rejected", [
`ESM package subpath ${candidate} names a directory; directory imports are unsupported and package main or index lookup was not attempted.`,
]);
return exactPackageTargetNotFound(candidate, "ESM subpath");
};
const resolvePackageExports = (
input: ResolveArtifactPathInput,
packagePath: string,
subpath: string,
rawExports: string | object,
): CandidateResolution => {
const exported = selectPackageExport(
rawExports,
subpath,
input.moduleKind,
packagePath,
);
if (exported.status === "rejected")
return unresolvedCandidate("rejected", [
`Package metadata ${packagePath}: ${exported.limitation}`,
]);
if (exported.status === "unmatched")
return unresolvedCandidate("external", [
`Directory package metadata ${packagePath} declares no exports target ${subpath === "." ? "for the active conditions" : `for subpath ${subpath}`}; exports encapsulation permits no file or index lookup. It declares ${exported.declared.join(", ") || "a root target only"}.`,
]);
if (exported.status === "blocked")
return unresolvedCandidate("external", [
`Directory package metadata ${packagePath} blocks subpath ${subpath} with ${exported.cause === "null" ? "an explicit null exports target" : "an empty exports target array"}.`,
]);
return resolveExportTarget(input, packagePath, exported);
};
/** An exports or import target is a URL naming one exact file for both Node loaders. */
const resolveExportTarget = (
input: ResolveArtifactPathInput,
packagePath: string,
target: PackageFileTarget,
mapKind: "exports" | "imports" = "exports",
): CandidateResolution => {
const withTargetSource = (
outcome: CandidateResolution,
): CandidateResolution => ({
...outcome,
limitations: outcome.limitations.map(
(reason) =>
`Package metadata ${packagePath}, ${mapKind} target ${JSON.stringify(target.value)}: ${reason}`,
),
});
// Node's CommonJS finalizer checks the serialized URL, including suffixes;
// its ESM finalizer checks only the pathname. Neither error retries arrays.
if (input.moduleKind === "require" && /%(?:2f|5c)/iu.test(target.url.href))
return withTargetSource(
unresolvedCandidate("rejected", [
"The selected package URL contains encoded separator bytes.",
]),
);
const encodedPath = target.url.pathname.slice(
PACKAGE_ARTIFACT_ROOT_URL.pathname.length,
);
const rejected = rejectDeclaration(encodedPath);
if (rejected !== null) return withTargetSource(rejected);
const decoded = decodeModuleUrlPath(encodedPath);
if (typeof decoded !== "string") return withTargetSource(decoded);
const candidate = confineCandidate(decoded);
if (typeof candidate !== "string") return withTargetSource(candidate);
return (
resolveFileCandidates(input, [candidate]) ??
withTargetSource(exactPackageTargetNotFound(candidate, mapKind))
);
};
/** Derived package identity and dependency policy; input retains the raw request. */
interface BarePackageRequest {
readonly packageName: string;
readonly packageSubpath: string;
readonly dependencyExportsEligible: boolean;
}
const parseBarePackageRequest = (
specifier: string,
moduleKind: ResolveArtifactPathInput["moduleKind"],
): BarePackageRequest | null => {
const segments = specifier.split("/");
let packageName = segments[0] ?? "";
if (packageName === "") return null;
if (specifier.startsWith("@")) {
const packageSegment = segments[1];
if (packageSegment === undefined || packageSegment === "") return null;
packageName = `${packageName}/${packageSegment}`;
}
// Node's CommonJS exports matcher excludes percent-containing package names.
// Compare the captured name so a scoped name cannot fall back to its scope.
const dependencyExportsEligible =
moduleKind !== "require" ||
/^((?:@[^/\\%]+\/)?[^./\\%][^/\\%]*)(\/.*)?$/.exec(specifier)?.[1] ===
packageName;
return {
packageName,
packageSubpath: `.${specifier.slice(packageName.length)}`,
dependencyExportsEligible,
};
};
const hasContainerDirectory = (
files: ReadonlyMap<string, JavaScriptArtifactFile>,
candidate: string,
containerSha256: string | undefined,
): boolean => {
if (
[...indexCandidates(candidate), posix.join(candidate, "package.json")].some(
(path) => {
const file = files.get(path);
return (
file !== undefined &&
(containerSha256 === undefined ||
file.container_sha256 === containerSha256)
);
},
)
)
return true;
// Inventory has files rather than directory entries. Any same-container
// descendant establishes a package directory even without package.json.
const prefix = `${candidate}/`;
for (const [path, file] of files)
if (
path.startsWith(prefix) &&
(containerSha256 === undefined ||
file.container_sha256 === containerSha256)
)
return true;
return false;
};
const htmlCandidate = (
input: ResolveArtifactPathInput,
): string | CandidateResolution => {
const declared = stripQueryAndFragment(htmlUrlText(input.declaredPath));
if (looksExternal(declared))
return unresolvedCandidate("external", [
"External HTML references are not mapped to local artifact assets.",
]);
const rawBase = input.htmlBaseHref;
const base =
rawBase === undefined || rawBase === null
? rawBase
: stripQueryAndFragment(htmlUrlText(rawBase));
if (base !== undefined && base !== null && looksExternal(base))
return unresolvedCandidate("external", [
"The document base href is external, so its script reference is not a local artifact path.",
]);
// Removing URL tabs and newlines can join an encoded dot or separator that
// the raw declaration split, so admit the parsed URL text again.
if (!admitsCanonicalPathSyntax(declared))
return unresolvedCandidate("rejected", [
"Encoded dot or separator bytes are rejected before artifact path resolution.",
]);
const declaredPath = percentDecodeUrlPath(declared);
if (declaredPath === null)
return unresolvedCandidate("rejected", [
"The HTML reference path percent-decodes to NUL or to bytes that are not UTF-8.",
]);
if (declaredPath.startsWith("/")) return declaredPath.slice(1);
if (base === undefined || base === null || base === "")
return declaredPath === ""
? input.sourcePath
: posix.join(posix.dirname(input.sourcePath), declaredPath);
// A local base href is a second untrusted path input; apply the same
// admission rules a declared path gets so it cannot smuggle traversal or
// separator syntax past canonicalization.
if (!admitsCanonicalPathSyntax(base))
return unresolvedCandidate("rejected", [
"The document base href uses NUL, backslash, or encoded dot and separator bytes that are not admitted for canonical artifact paths.",
]);
const decodedBase = percentDecodeUrlPath(base);
if (decodedBase === null)
return unresolvedCandidate("rejected", [
"The document base href path percent-decodes to NUL or to bytes that are not UTF-8.",
]);
const basePath = decodedBase.startsWith("/")
? decodedBase.slice(1)
: posix.join(posix.dirname(input.sourcePath), decodedBase);
if (declaredPath === "")
return exactUrlPathCandidate(decodedBase, basePath, "HTML");
return posix.join(htmlBaseDirectory(decodedBase, basePath), declaredPath);
};
const exactUrlPathCandidate = (
declared: string,
candidate: string,
kind: "HTML" | "URL" | "module URL",
): string | CandidateResolution => {
if (!urlPathIsDirectory(declared)) return candidate;
const confined = confineCandidate(candidate);
if (typeof confined !== "string") return confined;
return unresolvedCandidate("not-found", [
`The ${kind} reference resolves to directory ${confined || "."}, not an exact selected application file.`,
]);
};
/** Whether a URL path retains directory form after dot-segment normalization. */
const urlPathIsDirectory = (path: string): boolean =>
path.endsWith("/") ||
path.endsWith("/.") ||
path.endsWith("/..") ||
path === "." ||
path === "..";
const htmlBaseDirectory = (base: string, basePath: string): string =>
urlPathIsDirectory(base) ? basePath : posix.dirname(basePath);
const confineCandidate = (candidate: string): string | CandidateResolution => {
const normalized = normalizeJoinedLogicalPath(candidate);
if (logicalPathEscapesRoot(normalized))
return unresolvedCandidate("rejected", [
"The resolved candidate escapes the canonical artifact root.",
]);
return normalized === "." ? "" : normalized;
};
type PackageMetadata =
| {
readonly kind: "available";
readonly path: string;
readonly name: string | undefined;
readonly exports: string | object | undefined;
readonly imports: object | undefined;
readonly main: unknown;
readonly module: unknown;
}
| { readonly kind: "absent" }
| { readonly kind: "unavailable"; readonly limitation: string };
const readPackageMetadata = (
input: ResolveArtifactPathInput,
directory: string,
): PackageMetadata => {
const path = posix.join(directory, "package.json");
const file = input.files.get(path);
const source = input.files.get(input.sourcePath);
if (
file === undefined ||
(source !== undefined && file.container_sha256 !== source.container_sha256)
)
return { kind: "absent" };
if (!file.text.included)
return {
kind: "unavailable",
limitation: `Directory package metadata ${path} was inventoried but its text is unavailable: ${file.text.reason}.`,
};
let value: unknown;
try {
value = JSON.parse(file.text.value);
} catch (cause: unknown) {
void cause;
return {
kind: "unavailable",
limitation: `Directory package metadata ${path} is not valid package JSON.`,
};
}
if (typeof value !== "object" || value === null)
return {
kind: "unavailable",
limitation: `Directory package metadata ${path} is not valid package JSON.`,
};
const name = Reflect.get(value, "name");
const exports = Reflect.get(value, "exports");
const imports = Reflect.get(value, "imports");
return {
kind: "available",
path,
name: typeof name === "string" ? name : undefined,
// Node's package reader ignores scalar root exports other than strings.
// Nested targets retain their raw values for exports validation.
exports:
typeof exports === "string" ||
(typeof exports === "object" && exports !== null)
? exports
: undefined,
// Import maps must be non-array objects; other shapes are not a map.
imports:
typeof imports === "object" && imports !== null && !Array.isArray(imports)
? imports
: undefined,
main: Reflect.get(value, "main"),
module: Reflect.get(value, "module"),
};
};
/** File and directory lookup never interprets a nested directory's exports. */
const resolveCandidate = (
input: ResolveArtifactPathInput,
candidate: string,
packageChain: ReadonlySet<string>,
emittedSourceCandidates: readonly string[] = [],
): CandidateResolution =>
resolveFileCandidates(input, [
...fileCandidates(candidate, input.moduleKind),
...emittedSourceCandidates,
]) ??
resolveDirectory(
input,
candidate,
readPackageMetadata(input, candidate),
packageChain,
);
const resolveDirectory = (
input: ResolveArtifactPathInput,
directory: string,
metadata: PackageMetadata,
packageChain: ReadonlySet<string>,
): CandidateResolution => {
if (metadata.kind === "absent")
return (
resolveFileCandidates(
input,
indexCandidates(directory, input.moduleKind),
) ?? notFoundCandidate()
);
if (metadata.kind === "unavailable")
return unresolvedCandidate("unavailable", [metadata.limitation]);
if (packageChain.has(metadata.path))
return unresolvedCandidate("unavailable", [
`Directory package entrypoint cycle includes ${metadata.path}.`,
]);
const entry =
metadata.main === ""
? undefined
: metadata.main !== undefined
? metadata.main
: input.moduleKind === undefined
? metadata.module
: undefined;
if (entry === undefined)
return (
resolveFileCandidates(
input,
indexCandidates(directory, input.moduleKind),
) ?? notFoundCandidate()
);
if (typeof entry !== "string" || entry.length === 0)
return unresolvedCandidate("unavailable", [
`Directory package metadata ${metadata.path} is not valid package JSON.`,
]);
const entryInput: ResolveArtifactPathInput = {
declaredPath: entry,
sourcePath: metadata.path,
context: "package-entrypoint",
files: input.files,
...(input.moduleKind === undefined ? {} : { moduleKind: input.moduleKind }),
};
const candidate = contextualCandidate(entryInput);
if (typeof candidate !== "string") return candidate;
const confined = confineCandidate(candidate);
if (typeof confined !== "string") return confined;
const selected =
resolveFileCandidates(entryInput, [
...fileCandidates(confined, entryInput.moduleKind),
...indexCandidates(confined, entryInput.moduleKind),
]) ??
resolveFileCandidates(input, indexCandidates(directory, input.moduleKind));
if (selected !== null) return selected;
if (input.moduleKind === "require")
return { ...notFoundCandidate(), stopsAncestorLookup: true };
if (input.moduleKind !== undefined) return notFoundCandidate();
return resolveCandidate(
entryInput,
confined,
new Set([...packageChain, metadata.path]),
);
};
const resolveFileCandidates = (
input: ResolveArtifactPathInput,
candidates: readonly string[],
): CandidateResolution | null => {
const source = input.files.get(input.sourcePath);
for (const path of candidates) {
const target = input.files.get(path);
if (
target !== undefined &&
(source === undefined ||
target.container_sha256 === source.container_sha256)
)
return { resolvedPath: path, status: "resolved", limitations: [] };
}
return null;
};
const NODE_MODULE_EXTENSIONS = [".js", ".json", ".node"] as const;
const fileCandidates = (
candidate: string,
moduleKind?: ResolveArtifactPathInput["moduleKind"],
): readonly string[] => [
candidate,
...(moduleKind === undefined ? EXTENSIONS : NODE_MODULE_EXTENSIONS).map(
(extension) => `${candidate}${extension}`,
),
];
const indexCandidates = (
candidate: string,
moduleKind?: ResolveArtifactPathInput["moduleKind"],
): readonly string[] =>
(moduleKind === undefined ? EXTENSIONS : NODE_MODULE_EXTENSIONS).map(
(extension) => posix.join(candidate, `index${extension}`),
);
type PackageTargetBlockCause = "null" | "empty-array";
type PackageExportOutcome =
| ({ readonly status: "value" } & PackageFileTarget)
| { readonly status: "rejected"; readonly limitation: string }
| { readonly status: "blocked"; readonly cause: PackageTargetBlockCause }
| { readonly status: "unmatched"; readonly declared: readonly string[] };
/** Only exact non-directory keys participate; wildcard expansion is unsupported. */
const selectPackageExport = (
value: unknown,
subpath: string,
moduleKind: ResolveArtifactPathInput["moduleKind"],
packagePath: string,
): PackageExportOutcome => {
let target = value;
let declared: string[] = [];
if (typeof value === "object" && value !== null && !Array.isArray(value)) {
const keys = Object.keys(value);
const hasSubpaths = keys.some((key) => key.startsWith("."));
if (hasSubpaths && keys.some((key) => !key.startsWith(".")))
return {
status: "rejected",
limitation:
"The package exports configuration mixes subpath keys and condition keys.",
};
declared = keys;
if (hasSubpaths) {
if (
subpath.endsWith("/") ||
subpath.includes("*") ||
!Object.hasOwn(value, subpath)
)
return { status: "unmatched", declared };
target = Reflect.get(value, subpath);
} else if (subpath !== ".") return { status: "unmatched", declared };
} else if (subpath !== ".") return { status: "unmatched", declared };
const selected = selectExportTarget(
target,
packageExportConditions(moduleKind),
packageMetadataUrl(packagePath),
);
if (selected.kind === "invalid" || selected.kind === "invalid-config")
return { status: "rejected", limitation: selected.limitation };
if (selected.kind === "unmatched")
return {
status: "unmatched",
declared: selected.declared,
};
if (selected.kind === "blocked")
return { status: "blocked", cause: selected.cause };
if (selected.kind === "non-relative")
return {
status: "rejected",
limitation: `The package exports target ${JSON.stringify(selected.value)} must start with "./".`,
};
return {
status: "value",
value: selected.value,
url: selected.url,
};
};
type ExportTargetOutcome =
| ({ readonly kind: "value" } & PackageFileTarget)
| { readonly kind: "non-relative"; readonly value: string }
| { readonly kind: "blocked"; readonly cause: PackageTargetBlockCause }
| { readonly kind: "unmatched"; readonly declared: readonly string[] }
| { readonly kind: "invalid"; readonly limitation: string }
| { readonly kind: "invalid-config"; readonly limitation: string };
const invalidExportTarget = (
value: unknown,
reason: string,
mapKind: "exports" | "imports" = "exports",
): ExportTargetOutcome => ({
kind: "invalid",
limitation: `The package ${mapKind === "imports" ? "import" : "exports"} target ${JSON.stringify(value) ?? String(value)} ${reason}.`,
});
// Node checks raw target segments before URL normalization. Compare encoded
// segment names here; full URL decoding follows array target selection.
const forbiddenExportSegment = (value: string): string | undefined =>
value
.slice(2)
.split(/[\\/]/u)
.find((raw) => {
const segment = raw.replace(/%[0-9a-f]{2}/giu, (encoded) =>
String.fromCharCode(Number.parseInt(encoded.slice(1), 16)),
);
return (
segment === "." ||
segment === ".." ||
segment.toLowerCase() === "node_modules"
);
});
const selectStringPackageTarget = (
value: string,
packageUrl: URL,
mapKind: "exports" | "imports",
): ExportTargetOutcome => {
if (mapKind === "imports") {
const target = classifyPackageImportTargetString(value);
if (target.kind === "invalid")
return invalidExportTarget(
value,
"cannot be parent-relative, absolute, or a URL",
mapKind,
);
if (target.kind !== "relative") return { kind: "non-relative", value };
} else if (!value.startsWith("./"))
return invalidExportTarget(value, 'must start with "./"', mapKind);
const forbidden = forbiddenExportSegment(value);
if (forbidden !== undefined)
return invalidExportTarget(
value,
`contains forbidden path segment ${JSON.stringify(forbidden)} before URL normalization`,
mapKind,
);
// Node constructs the URL before deciding whether this array entry is an
// invalid package target. Tabs/newlines and dot segments normalize here;
// percent decoding and existence checks belong after target selection.
const url = new URL(value, packageUrl);
if (!url.pathname.startsWith(new URL(".", packageUrl).pathname))
return invalidExportTarget(
value,
"escapes the package root after URL normalization",
mapKind,
);
return {
kind: "value",
value,
url,
};
};
const selectArrayPackageTarget = (
value: readonly unknown[],
conditions: ReadonlySet<string>,
packageUrl: URL,
mapKind: "exports" | "imports",
): ExportTargetOutcome => {
if (value.length === 0) return { kind: "blocked", cause: "empty-array" };
let lastRefusal:
| Extract<ExportTargetOutcome, { readonly kind: "invalid" | "blocked" }>
| undefined;
const declared = new Set<string>();
for (const entry of value) {
const nested = selectExportTarget(entry, conditions, packageUrl, mapKind);
// Node's array fallback catches invalid targets, not invalid configuration.
if (nested.kind === "invalid-config") return nested;
if (nested.kind === "unmatched") {
for (const condition of nested.declared) declared.add(condition);
continue;
}
if (nested.kind === "invalid" || nested.kind === "blocked") {
lastRefusal = nested;
continue;
}
return nested;
}
return lastRefusal ?? { kind: "unmatched", declared: [...declared] };
};
/**
* Resolve a target in Node's declared order. Explicit null blocks an active
* condition, whereas an unmatched nested object permits the next condition.
* An array skips unmatched and invalid entries, stops at its first string, and
* preserves the last invalid/null outcome if no string is selected. A nonempty
* array of unmatched conditions stays unmatched; an empty array blocks.
*/
const selectExportTarget = (
value: unknown,
conditions: ReadonlySet<string>,
packageUrl: URL,
mapKind: "exports" | "imports" = "exports",
): ExportTargetOutcome => {
if (typeof value === "string")
return selectStringPackageTarget(value, packageUrl, mapKind);
if (value === null) return { kind: "blocked", cause: "null" };
if (Array.isArray(value))
return selectArrayPackageTarget(value, conditions, packageUrl, mapKind);
if (typeof value !== "object")
return invalidExportTarget(
value,
"must be a relative string, object, array, or null",
mapKind,
);
for (const key of Object.keys(value)) {
const numeric = Number(key);
if (String(numeric) === key && numeric >= 0 && numeric < 0xffff_ffff)
return {
kind: "invalid-config",
limitation: `The package exports configuration contains numeric condition key ${JSON.stringify(key)}.`,
};
}
const declared = new Set<string>();
for (const [condition, target] of Object.entries(value)) {
declared.add(condition);
if (!conditions.has(condition)) continue;
const nested = selectExportTarget(target, conditions, packageUrl, mapKind);
if (nested.kind !== "unmatched") return nested;
for (const name of nested.declared) declared.add(name);
}
return { kind: "unmatched", declared: [...declared] };
};
/**
* Node selects an exports target by walking the declared keys in order and
* taking the first whose condition is active for the calling resolver.
* "default" is always active, and "node" plus "node-addons" are active for the
* built-in resolver that owns installed-package imports and requires. The
* supported Node runtimes also activate "module-sync" for both loaders.
*/
const packageExportConditions = (
moduleKind: ResolveArtifactPathInput["moduleKind"],
): ReadonlySet<string> =>
new Set([
"node",
"node-addons",
"module-sync",
...(moduleKind === undefined ? [] : [moduleKind]),
"default",
]);
const notFoundCandidate = (): Extract<
CandidateResolution,
{ readonly resolvedPath: null }
> => ({
resolvedPath: null,
status: "not-found",
limitations: [
"No extension, directory package, or index candidate exists in the inventoried artifact container.",
],
});
const exactPackageTargetNotFound = (
path: string,
targetKind: "exports" | "imports" | "ESM subpath",
): CandidateResolution => ({
resolvedPath: null,
status: "not-found",
limitations: [
`The selected package ${targetKind === "imports" ? "import" : targetKind} target ${path} does not exist in the inventoried artifact container; ${targetKind === "imports" ? "import" : targetKind} targets are exact files, so no extension or directory index was tried.`,
],
});
const fileUrlPath = (value: string): string | null | undefined => {
if (!/^file:\/\//iu.test(value)) return undefined;
// URL parsing removes tabs and newlines and normalizes encoded dot segments.
// Check the raw path first, excluding URL identity suffixes, then check the
// parsed pathname so parser normalization cannot introduce a separator.
const rawPath = stripQueryAndFragment(value);
if (
value.includes("\0") ||
value.includes("\\") ||
/[\t\n\r]/u.test(rawPath) ||
!admitsCanonicalPathSyntax(rawPath)
)
return null;
try {
const url = new URL(value);
if (url.protocol !== "file:" || url.hostname !== "") return undefined;
if (!admitsCanonicalPathSyntax(url.pathname)) return null;
const decoded = decodeURIComponent(url.pathname);
if (decoded.includes("\0")) return null;
return decoded.replace(/^\/+/, "");
} catch (cause: unknown) {
// Unparseable file URLs have no path to resolve.
void cause;
return undefined;
}
};
const unresolvedCandidate = (
status: UnresolvedArtifactPathStatus,
limitations: readonly string[],
): CandidateResolution => ({ resolvedPath: null, status, limitations });