Infer date-times only for strict RFC 3339 strings

packages/quicktype-core/src/DateTime.ts

@@ -1,5 +1,12 @@
 /* eslint-disable */
+// Adapted from ajv:
 // https://github.com/epoberezkin/ajv/blob/4d76c6fb813b136b6ec4fe74990bc97233d75dea/lib/compile/formats.js
+//
+// We deviate from ajv in that we recognize only strict RFC 3339 date-times:
+// the date/time separator must be "T" (or "t"), not a space, and times must
+// carry a timezone offset ("Z"/"z" or "+hh:mm"/"-hh:mm").  ajv accepted the
+// lenient forms, which meant we inferred "date-time" for strings like
+// "2013-06-15 21:10:28" that most target languages' date parsers reject.
 
 /*
 The MIT License (MIT)
@@ -27,15 +34,17 @@ SOFTWARE.
 
 const DATE = /^(\d\d\d\d)-(\d\d)-(\d\d)$/;
 const DAYS = [0, 31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31];
-const TIME = /^(\d\d):(\d\d):(\d\d)(\.\d+)?(z|[+-]\d\d:\d\d)?$/i;
+// RFC 3339 full-time: the timezone offset is mandatory.
+const TIME = /^(\d\d):(\d\d):(\d\d)(\.\d+)?(z|[+-]\d\d:\d\d)$/i;
 
 export interface DateTimeRecognizer {
     isDate: (s: string) => boolean;
     isDateTime: (s: string) => boolean;
     isTime: (s: string) => boolean;
 }
 
-const DATE_TIME_SEPARATOR = /t|\s/i;
+// RFC 3339 date-time: the separator must be "T" or "t", never a space.
+const DATE_TIME_SEPARATOR = /t/i;
 
 export class DefaultDateTimeRecognizer implements DateTimeRecognizer {
     isDate(str: string) {

test/languages.ts

@@ -91,7 +91,7 @@ export const CSharpLanguage: Language = {
         return `dotnet run -p:CheckEolTargetFramework=false -- "${sample}"`;
     },
     diffViaSchema: true,
-    skipDiffViaSchema: ["34702.json", "437e7.json"],
+    skipDiffViaSchema: ["34702.json"],
     allowMissingNull: false,
     features: [
         "enum",
@@ -134,7 +134,7 @@ export const CSharpLanguageSystemTextJson: Language = {
         return `dotnet run -p:CheckEolTargetFramework=false -- "${sample}"`;
     },
     diffViaSchema: true,
-    skipDiffViaSchema: ["34702.json", "437e7.json"],
+    skipDiffViaSchema: ["34702.json"],
     allowMissingNull: false,
     features: [
         "enum",
@@ -469,15 +469,6 @@ export const GoLanguage: Language = {
         // can't differenciate empty array and nothing for optional empty array
         // (omitempty).
         "github-events.json",
-        // files contains datetime filed with contain non rfc3339 format value
-        "f6a65.json",
-        "e0ac7.json",
-        "c3303.json",
-        "7681c.json",
-        "437e7.json",
-        "127a1.json",
-        "26b49.json",
-        "0cffa.json",
     ],
     skipMiscJSON: false,
     skipSchema: [
@@ -592,15 +583,9 @@ export const CPlusPlusLanguage: Language = {
         "bug427.json",
         "keywords.json",
         "0a91a.json",
-        "0cffa.json",
-        "127a1.json",
-        "26b49.json",
         "34702.json",
-        "7681c.json",
         "7f568.json",
-        "c3303.json",
         "e8b04.json",
-        "f6a65.json",
         "fcca3.json",
     ],
     allowMissingNull: false,
@@ -663,9 +648,7 @@ export const ElmLanguage: Language = {
         "github-events.json",
         "nbl-stats.json",
         "0a91a.json",
-        "0cffa.json",
         "0e0c2.json",
-        "127a1.json",
         "29f47.json",
         "2df80.json",
         "27332.json",
@@ -676,13 +659,10 @@ export const ElmLanguage: Language = {
         "6de06.json",
         "76ae1.json",
         "7eb30.json",
-        "7681c.json",
         "ae9ca.json",
         "af2d1.json",
         "be234.json",
-        "c3303.json",
         "e8b04.json",
-        "f6a65.json",
     ],
     allowMissingNull: false,
     features: ["enum", "union", "no-defaults"],
@@ -731,17 +711,12 @@ export const SwiftLanguage: Language = {
         "keywords.json",
         "0a358.json", // date-time issues
         "0a91a.json",
-        "0cffa.json", // date-time issues
-        "127a1.json", // date-time issues
-        "26b49.json", // date-time issues
         "26c9c.json", // uri/string confusion
         "32d5c.json", // date-time issues
         "337ed.json",
         "34702.json",
-        "437e7.json", // date-time issues
         "54d32.json", // date-time issues
         "5eae5.json", // date-time issues
-        "7681c.json", // date-time issues
         "77392.json", // date-time issues
         "7f568.json",
         "734ad.json",
@@ -750,13 +725,10 @@ export const SwiftLanguage: Language = {
         "9ac3b.json", // date-time issues
         "a0496.json", // date-time issues
         "b4865.json", // date-time issues
-        "c3303.json", // date-time issues
         "c8c7e.json",
         "d23d5.json", // date-time issues
-        "e0ac7.json", // date-time issues
         "e53b5.json",
         "e8b04.json",
-        "f6a65.json", // date-time issues
         "fcca3.json",
         "f82d9.json",
         "bug863.json", // Unable to resolve reserved keyword use, "description"
@@ -872,7 +844,6 @@ export const TypeScriptLanguage: Language = {
     output: "TopLevel.ts",
     topLevel: "TopLevel",
     skipJSON: [
-        "7681c.json", // year 0 is out of range
     ],
     skipMiscJSON: false,
     skipSchema: ["keyword-unions.schema"], // can't handle "constructor" property
@@ -904,7 +875,6 @@ export const JavaScriptLanguage: Language = {
     output: "TopLevel.js",
     topLevel: "TopLevel",
     skipJSON: [
-        "7681c.json", // year 0 is out of range
     ],
     skipMiscJSON: false,
     skipSchema: ["keyword-unions.schema"], // can't handle "constructor" property
@@ -962,7 +932,6 @@ export const FlowLanguage: Language = {
     output: "TopLevel.js",
     topLevel: "TopLevel",
     skipJSON: [
-        "7681c.json", // year 0 is out of range
     ],
     skipMiscJSON: false,
     skipSchema: [
@@ -1612,7 +1581,6 @@ export const TypeScriptZodLanguage: Language = {
         "nst-test-suite.json",
         "keywords.json",
         "ed095.json",
-        "7681c.json",
         "32d5c.json",
     ],
     skipMiscJSON: false,

test/unit/date-time-recognizer.test.ts

@@ -0,0 +1,85 @@
+// The default date/time recognizer decides whether a JSON string sample is
+// inferred as "date", "time", or "date-time". It must accept only strict
+// RFC 3339 forms: "date-time" requires the "T" (or "t") separator — a space
+// is not allowed — and times must carry a timezone offset. Before this was
+// tightened, strings like "2013-06-15 21:10:28" were inferred as date-times,
+// producing generated code whose strict date parsers (Go, Swift, java.time,
+// ...) could not read the very samples the types were inferred from.
+import { DefaultDateTimeRecognizer } from "quicktype-core/dist/DateTime.js";
+import { describe, expect, test } from "vitest";
+
+const recognizer = new DefaultDateTimeRecognizer();
+
+describe("isDateTime", () => {
+    test("accepts RFC 3339 date-times with offset", () => {
+        expect(recognizer.isDateTime("2018-08-14T02:45:50Z")).toBe(true);
+        expect(recognizer.isDateTime("2018-08-14T02:45:50+01:00")).toBe(true);
+        expect(recognizer.isDateTime("2018-08-14T02:45:50-11:30")).toBe(true);
+    });
+
+    test("accepts lowercase t and z", () => {
+        expect(recognizer.isDateTime("2018-08-14t02:45:50z")).toBe(true);
+    });
+
+    test("accepts fractional seconds", () => {
+        expect(recognizer.isDateTime("2015-04-24T01:46:50.342496Z")).toBe(true);
+        expect(recognizer.isDateTime("2010-01-12T00:00:00.000Z")).toBe(true);
+        expect(recognizer.isDateTime("2015-07-15T23:25:21.541+02:00")).toBe(
+            true,
+        );
+    });
+
+    test("rejects a space separator", () => {
+        expect(recognizer.isDateTime("2013-06-15 21:10:28")).toBe(false);
+        expect(recognizer.isDateTime("1970-01-01 00:00:00")).toBe(false);
+        expect(recognizer.isDateTime("2015-04-24 01:46:50.342496Z")).toBe(
+            false,
+        );
+    });
+
+    test("rejects a missing timezone offset", () => {
+        expect(recognizer.isDateTime("2018-08-14T02:45:50")).toBe(false);
+        expect(recognizer.isDateTime("2018-08-14T02:45:50.123")).toBe(false);
+    });
+
+    test("rejects invalid dates and times", () => {
+        expect(recognizer.isDateTime("0000-00-00T00:00:00Z")).toBe(false);
+        expect(recognizer.isDateTime("2018-13-01T00:00:00Z")).toBe(false);
+        expect(recognizer.isDateTime("2018-08-14T24:00:00Z")).toBe(false);
+        expect(recognizer.isDateTime("2018-08-14X02:45:50Z")).toBe(false);
+    });
+});
+
+describe("isDate", () => {
+    test("accepts RFC 3339 full-date", () => {
+        expect(recognizer.isDate("2018-08-14")).toBe(true);
+        expect(recognizer.isDate("1970-01-01")).toBe(true);
+    });
+
+    test("rejects invalid dates", () => {
+        expect(recognizer.isDate("0000-00-00")).toBe(false);
+        expect(recognizer.isDate("2018-13-01")).toBe(false);
+        expect(recognizer.isDate("2018-02-30")).toBe(false);
+        expect(recognizer.isDate("2018-8-14")).toBe(false);
+    });
+});
+
+describe("isTime", () => {
+    test("accepts RFC 3339 full-time (offset required)", () => {
+        expect(recognizer.isTime("02:45:50Z")).toBe(true);
+        expect(recognizer.isTime("02:45:50z")).toBe(true);
+        expect(recognizer.isTime("02:45:50.123+01:00")).toBe(true);
+        expect(recognizer.isTime("23:59:59-11:30")).toBe(true);
+    });
+
+    test("rejects a missing timezone offset", () => {
+        expect(recognizer.isTime("02:45:50")).toBe(false);
+        expect(recognizer.isTime("02:45:50.123")).toBe(false);
+    });
+
+    test("rejects invalid times", () => {
+        expect(recognizer.isTime("24:00:00Z")).toBe(false);
+        expect(recognizer.isTime("02:60:00Z")).toBe(false);
+        expect(recognizer.isTime("02:45:60Z")).toBe(false);
+    });
+});