Handle Windows absolute paths for schema inputs

packages/quicktype-core/src/input/JSONSchemaInput.ts

@@ -54,6 +54,7 @@ import {
     panic,
     parseJSON,
 } from "../support/Support";
+import { fixWindowsPath } from "../support/WindowsPaths";
 import {
     type PrimitiveTypeKind,
     type TransformedStringTypeKind,
@@ -143,15 +144,15 @@ function normalizeURI(uri: string | URI): URI {
     // JSONSchemaStore should take a URI, not a string, and if it reads from
     // a file it can decode by itself.
     if (typeof uri === "string") {
-        uri = new URI(uri);
+        uri = new URI(fixWindowsPath(uri));
     }
 
     return new URI(URI.decode(uri.clone().normalize().toString()));
 }
 
 export class Ref {
     public static root(address: string | undefined): Ref {
-        const uri = definedMap(address, (a) => new URI(a));
+        const uri = definedMap(address, (a) => new URI(fixWindowsPath(a)));
         return new Ref(uri, []);
     }
 
@@ -189,7 +190,7 @@ export class Ref {
     }
 
     public static parse(ref: string): Ref {
-        return Ref.parseURI(new URI(ref), true);
+        return Ref.parseURI(new URI(fixWindowsPath(ref)), true);
     }
 
     public addressURI: URI | undefined;

packages/quicktype-core/src/input/io/NodeIO.ts

@@ -8,6 +8,7 @@ import { Readable } from "readable-stream";
 
 import { messageError } from "../../Messages";
 import { panic } from "../../support/Support";
+import { filePathFromFileURI } from "../../support/WindowsPaths";
 
 import { getStream } from "./get-stream";
 
@@ -100,7 +101,9 @@ export async function readableFromFileOrURL(
     httpHeaders?: string[],
 ): Promise<Readable> {
     try {
-        if (isURL(fileOrURL)) {
+        if (fileOrURL.startsWith("file://")) {
+            fileOrURL = filePathFromFileURI(fileOrURL);
+        } else if (isURL(fileOrURL)) {
             const response = await globalThis.fetch(fileOrURL, {
                 headers: parseHeaders(httpHeaders),
             });

packages/quicktype-core/src/support/WindowsPaths.ts

@@ -0,0 +1,44 @@
+// Windows absolute paths given as schema addresses are carried through the
+// address machinery as "file:" URIs (issue #2869). These two functions are
+// the two halves of that protocol: JSONSchemaInput converts Windows paths to
+// "file:" URIs before urijs parses them, and NodeIO converts "file:" URIs
+// back to file paths when it reads them from disk.
+
+// Windows absolute paths are not valid URIs: urijs parses the drive letter
+// of e.g. "C:\Users\me\top.schema.json" as a URI scheme (and lowercases it,
+// and treats the backslashes as an opaque path), so relative refs resolve to
+// bogus addresses (issue #2869). Convert drive-letter and UNC paths to
+// "file:" URIs, which NodeIO knows how to read.
+export function fixWindowsPath(pathOrURI: string): string {
+    if (/^[A-Za-z]:[/\\]/.test(pathOrURI)) {
+        // Drive-letter path, e.g. "C:\dir\x.json" -> "file:///C:/dir/x.json"
+        return `file:///${pathOrURI.replace(/\\/g, "/")}`;
+    }
+
+    if (pathOrURI.startsWith("\\\\")) {
+        // UNC path, e.g. "\\server\share\x.json" -> "file://server/share/x.json"
+        return `file:${pathOrURI.replace(/\\/g, "/")}`;
+    }
+
+    return pathOrURI;
+}
+
+// We don't use `url.fileURLToPath` because its result is platform-dependent:
+// drive-letter URIs must also be readable on POSIX (as paths relative to the
+// working directory), which is how the tests exercise this, and Node's fs
+// accepts forward slashes on Windows anyway. The addresses were URI-decoded
+// when they were normalized, so there's no percent-decoding to do here.
+export function filePathFromFileURI(fileURI: string): string {
+    const path = fileURI.slice("file://".length);
+    if (/^\/[A-Za-z]:\//.test(path)) {
+        // Drive-letter path, e.g. "file:///C:/dir/x.json" -> "C:/dir/x.json"
+        return path.slice(1);
+    }
+
+    if (!path.startsWith("/")) {
+        // UNC path, e.g. "file://server/share/x.json" -> "//server/share/x.json"
+        return `//${path}`;
+    }
+
+    return path;
+}

test/unit/windows-paths.test.ts

@@ -0,0 +1,85 @@
+// Unit tests for the Windows-paths-as-"file:"-URIs protocol (issue #2869):
+// fixWindowsPath converts Windows absolute schema addresses to "file:" URIs
+// before urijs parses them, and filePathFromFileURI converts them back to
+// file paths when NodeIO reads them from disk. The end-to-end pipeline is
+// covered by windows-schema-paths.test.ts; these tests pin down the string
+// mapping itself, including the inputs that must pass through untouched.
+
+import {
+    filePathFromFileURI,
+    fixWindowsPath,
+} from "quicktype-core/dist/support/WindowsPaths";
+import { describe, expect, test } from "vitest";
+
+describe("fixWindowsPath", () => {
+    test.each([
+        // Drive-letter paths become "file:///<drive>:/..." URIs.
+        ["C:\\Users\\me\\top.schema.json", "file:///C:/Users/me/top.schema.json"],
+        ["C:/Users/me/top.schema.json", "file:///C:/Users/me/top.schema.json"],
+        ["c:\\dir\\x.json", "file:///c:/dir/x.json"],
+        ["Z:/x.json", "file:///Z:/x.json"],
+        // Mixed separators are normalized to forward slashes.
+        ["C:\\dir/sub\\x.json", "file:///C:/dir/sub/x.json"],
+        // Spaces survive: the address machinery works with decoded URIs.
+        ["C:\\My Dir\\x.json", "file:///C:/My Dir/x.json"],
+        // UNC paths keep the server as the URI host.
+        ["\\\\server\\share\\x.json", "file://server/share/x.json"],
+        ["\\\\server\\share\\dir\\x.json", "file://server/share/dir/x.json"],
+    ])("converts %j to %j", (input, expected) => {
+        expect(fixWindowsPath(input)).toBe(expected);
+    });
+
+    test.each([
+        // POSIX and relative paths are not Windows paths.
+        "/home/me/top.schema.json",
+        "dir/top.schema.json",
+        "./top.schema.json",
+        "../top.schema.json",
+        "top.schema.json",
+        "",
+        // Real URIs must never be re-wrapped.
+        "http://example.com/x.schema.json",
+        "https://example.com/x.schema.json",
+        "file:///C:/dir/x.json",
+        "file://server/share/x.json",
+        // Drive-letter lookalikes: more than one letter is a URI scheme,
+        // digits are not drive letters, and "C:x" (no separator after the
+        // colon, a drive-relative path) is not supported.
+        "CC:\\x.json",
+        "1:\\x.json",
+        "C:",
+        "C:x.json",
+        // A single leading backslash is not UNC.
+        "\\dir\\x.json",
+    ])("passes %j through unchanged", (input) => {
+        expect(fixWindowsPath(input)).toBe(input);
+    });
+});
+
+describe("filePathFromFileURI", () => {
+    test.each([
+        // Drive-letter URIs lose the leading slash so fs sees "C:/...".
+        ["file:///C:/dir/x.json", "C:/dir/x.json"],
+        ["file:///c:/dir/x.json", "c:/dir/x.json"],
+        ["file:///C:/My Dir/x.json", "C:/My Dir/x.json"],
+        // UNC URIs turn the host back into a "//server/..." path.
+        ["file://server/share/x.json", "//server/share/x.json"],
+        // POSIX file URIs keep their absolute path.
+        ["file:///home/me/x.json", "/home/me/x.json"],
+        ["file:///x.json", "/x.json"],
+    ])("converts %j to %j", (input, expected) => {
+        expect(filePathFromFileURI(input)).toBe(expected);
+    });
+});
+
+describe("round-trip", () => {
+    test.each([
+        // filePathFromFileURI(fixWindowsPath(p)) yields the forward-slash
+        // form of p, which is what Node's fs accepts on every platform.
+        ["C:\\Users\\me\\top.schema.json", "C:/Users/me/top.schema.json"],
+        ["C:/Users/me/top.schema.json", "C:/Users/me/top.schema.json"],
+        ["\\\\server\\share\\x.json", "//server/share/x.json"],
+    ])("%j reads back as %j", (input, expected) => {
+        expect(filePathFromFileURI(fixWindowsPath(input))).toBe(expected);
+    });
+});

test/unit/windows-schema-paths.test.ts

@@ -0,0 +1,123 @@
+// Regression test for issue #2869: schema inputs given as Windows absolute
+// paths must work.
+//
+// urijs parses the drive letter of a Windows absolute path such as
+// "C:\Users\me\top.schema.json" as a URI *scheme* ("c:"), so the address is
+// mangled (lowercased scheme, backslashes kept as an opaque path) and
+// relative $refs resolve to bogus addresses like "c:///item.schema.json",
+// which NodeIO then tries to fetch as HTTP URLs. The reported failures are
+// "Could not fetch schema ..." and "Internal error: Defined value expected".
+//
+// The fix converts Windows absolute paths (drive-letter and UNC) to "file:"
+// URIs before urijs sees them, and teaches NodeIO to read "file:" URIs from
+// disk. On POSIX the resulting drive-letter file path ("C:/Users/me/...") is
+// read relative to the working directory, which is what lets us test the
+// whole pipeline on Linux CI: we create a literal "C:/Users/me/" directory
+// tree in a temp dir, chdir into it, and run quicktype with the Windows-style
+// path the issue reported. The fixture harness cannot cover this because its
+// inputs are always plain POSIX paths.
+
+import * as fs from "node:fs";
+import * as os from "node:os";
+import * as path from "node:path";
+
+import {
+    FetchingJSONSchemaStore,
+    InputData,
+    JSONSchemaInput,
+    quicktype,
+} from "quicktype-core";
+import { describe, test } from "vitest";
+
+const topLevelSchema = JSON.stringify({
+    $schema: "http://json-schema.org/draft-07/schema#",
+    type: "array",
+    items: { $ref: "item.schema.json" },
+});
+
+const itemSchema = JSON.stringify({
+    $schema: "http://json-schema.org/draft-07/schema#",
+    type: "object",
+    properties: { name: { type: "string" } },
+    required: ["name"],
+});
+
+interface SchemaPathCase {
+    description: string;
+    // The path passed to quicktype, as a user would type it.
+    schemaArg: (tempDir: string) => string;
+    // Where to create the schema files, relative to the temp dir.
+    schemaDir: string;
+}
+
+const cases: SchemaPathCase[] = [
+    {
+        description: "Windows absolute path with backslashes",
+        schemaDir: "C:/Users/quicktype",
+        schemaArg: () => "C:\\Users\\quicktype\\top.schema.json",
+    },
+    {
+        description: "Windows absolute path with forward slashes",
+        schemaDir: "C:/Users/quicktype",
+        schemaArg: () => "C:/Users/quicktype/top.schema.json",
+    },
+    {
+        // Must keep working exactly as before the fix.
+        description: "POSIX absolute path",
+        schemaDir: "posix",
+        schemaArg: (tempDir) => path.join(tempDir, "posix", "top.schema.json"),
+    },
+];
+
+async function generateTypeScript(schemaURI: string): Promise<string> {
+    // The same setup the CLI uses for `-s schema <path>`.
+    const schemaInput = new JSONSchemaInput(new FetchingJSONSchemaStore());
+    await schemaInput.addSource({ name: "TopLevel", uris: [schemaURI] });
+    const inputData = new InputData();
+    inputData.addInput(schemaInput);
+
+    const result = await quicktype({ inputData, lang: "typescript" });
+    return result.lines.join("\n");
+}
+
+describe("schema inputs given as absolute paths (issue #2869)", () => {
+    // Sequential, not concurrent: the cases chdir into their temp dirs.
+    for (const c of cases) {
+        test(c.description, async () => {
+            const tempDir = fs.mkdtempSync(
+                path.join(os.tmpdir(), "quicktype-windows-paths-"),
+            );
+            const previousCwd = process.cwd();
+            try {
+                const schemaDir = path.join(tempDir, c.schemaDir);
+                fs.mkdirSync(schemaDir, { recursive: true });
+                fs.writeFileSync(
+                    path.join(schemaDir, "top.schema.json"),
+                    topLevelSchema,
+                );
+                fs.writeFileSync(
+                    path.join(schemaDir, "item.schema.json"),
+                    itemSchema,
+                );
+
+                // On POSIX the drive-letter path is read relative to the
+                // working directory, so the "C:/Users/quicktype" tree must
+                // be under the cwd.
+                process.chdir(tempDir);
+                const output = await generateTypeScript(c.schemaArg(tempDir));
+
+                // The item schema is only reachable through the relative
+                // $ref, so this asserts that $ref resolution against the
+                // address works, too.
+                if (!/name\s*:\s*string/.test(output)) {
+                    throw new Error(
+                        `generated output does not contain the type from the $ref'd schema:\n${output}`,
+                    );
+                }
+            } finally {
+                process.chdir(previousCwd);
+                fs.rmSync(tempDir, { recursive: true, force: true });
+            }
+        });
+    }
+});