accept ISO-8601 dates, and rework dateTime2ms

Author: alexcjohnson

src/lib/dates.js

@@ -10,7 +10,6 @@
 'use strict';
 
 var d3 = require('d3');
-var isNumeric = require('fast-isnumeric');
 
 var logError = require('./loggers').error;
 
@@ -21,6 +20,11 @@ var ONEHOUR = constants.ONEHOUR;
 var ONEMIN = constants.ONEMIN;
 var ONESEC = constants.ONESEC;
 
+var DATETIME_REGEXP = /^\s*(-?\d\d\d\d|\d\d)(-(0?[1-9]|1[012])(-([0-3]?\d)([ Tt]([01]?\d|2[0-3])(:([0-5]\d)(:([0-5]\d(\.\d+)?))?(Z|z|[+\-]\d\d:?\d\d)?)?)?)?)?\s*$/m;
+
+// for 2-digit years, the first year we map them onto
+var YFIRST = new Date().getFullYear() - 70;
+
 // is an object a javascript date?
 exports.isJSDate = function(v) {
     return typeof v === 'object' && v !== null && typeof v.getTime === 'function';
@@ -32,20 +36,33 @@ exports.isJSDate = function(v) {
 var MIN_MS, MAX_MS;
 
 /**
- * dateTime2ms - turn a date object or string s of the form
- * YYYY-mm-dd HH:MM:SS.sss into milliseconds (relative to 1970-01-01,
- * per javascript standard)
- * may truncate after any full field, and sss can be any length
- * even >3 digits, though javascript dates truncate to milliseconds
- * returns BADNUM if it doesn't find a date
+ * dateTime2ms - turn a date object or string s into milliseconds
+ * (relative to 1970-01-01, per javascript standard)
+ *
+ * Returns BADNUM if it doesn't find a date
+ *
+ * strings should have the form:
+ *
+ *    -?YYYY-mm-dd<sep>HH:MM:SS.sss<tzInfo>?
+ *
+ * <sep>: space (our normal standard) or T or t (ISO-8601)
+ * <tzInfo>: Z, z, or [+\-]HH:?MM and we THROW IT AWAY
+ * this format comes from https://tools.ietf.org/html/rfc3339#section-5.6
+ * but we allow it even with a space as the separator
+ *
+ * May truncate after any full field, and sss can be any length
+ * even >3 digits, though javascript dates truncate to milliseconds,
+ * we keep as much as javascript numeric precision can hold, but we only
+ * report back up to 100 microsecond precision, because most dates support
+ * this precision (close to 1970 support more, very far away support less)
  *
  * Expanded to support negative years to -9999 but you must always
  * give 4 digits, except for 2-digit positive years which we assume are
  * near the present time.
  * Note that we follow ISO 8601:2004: there *is* a year 0, which
  * is 1BC/BCE, and -1===2BC etc.
  *
- * 2-digit to 4-digit year conversion, where to cut off?
+ * Where to cut off 2-digit years between 1900s and 2000s?
  * from http://support.microsoft.com/kb/244664:
  *   1930-2029 (the most retro of all...)
  * but in my mac chrome from eg. d=new Date(Date.parse('8/19/50')):
@@ -77,89 +94,31 @@ exports.dateTime2ms = function(s) {
     // otherwise only accept strings and numbers
     if(typeof s !== 'string' && typeof s !== 'number') return BADNUM;
 
-    var y, m, d, h;
-    // split date and time parts
-    // TODO: we strip leading/trailing whitespace but not other
-    // characters like we do for numbers - do we want to?
-    var datetime = String(s).trim().split(' ');
-    if(datetime.length > 2) return BADNUM;
-
-    var p = datetime[0].split('-'); // date part
-
-    var CE = true; // common era, ie positive year
-    if(p[0] === '') {
-        // first part is blank: year starts with a minus sign
-        CE = false;
-        p.splice(0, 1);
-    }
-
-    var plen = p.length;
-    if(plen > 3 || (plen !== 3 && datetime[1]) || !plen) return BADNUM;
-
-    // year
-    if(p[0].length === 4) y = Number(p[0]);
-    else if(p[0].length === 2) {
-        if(!CE) return BADNUM;
-        var yNow = new Date().getFullYear();
-        y = ((Number(p[0]) - yNow + 70) % 100 + 200) % 100 + yNow - 70;
+    var match = String(s).match(DATETIME_REGEXP);
+    if(!match) return BADNUM;
+    var y = match[1],
+        m = Number(match[3] || 1),
+        d = Number(match[5] || 1),
+        H = Number(match[7] || 0),
+        M = Number(match[9] || 0),
+        S = Number(match[11] || 0);
+    if(y.length === 2) {
+        y = (Number(y) + 2000 - YFIRST) % 100 + YFIRST;
     }
-    else return BADNUM;
-    if(!isNumeric(y)) return BADNUM;
+    else y = Number(y);
 
     // javascript takes new Date(0..99,m,d) to mean 1900-1999, so
     // to support years 0-99 we need to use setFullYear explicitly
-    var baseDate = new Date(0, 0, 1);
-    baseDate.setFullYear(CE ? y : -y);
-    if(p.length > 1) {
+    var date = new Date(2000, m - 1, d, H, M);
+    date.setFullYear(y);
 
-        // month - may be 1 or 2 digits
-        m = Number(p[1]) - 1; // new Date() uses zero-based months
-        if(p[1].length > 2 || !(m >= 0 && m <= 11)) return BADNUM;
-        baseDate.setMonth(m);
+    if(date.getDate() !== d) return BADNUM;
 
-        if(p.length > 2) {
+    // does that hour exist in this day? (Daylight time!)
+    // (TODO: remove this check when we move to UTC)
+    if(date.getHours() !== H) return BADNUM;
 
-            // day - may be 1 or 2 digits
-            d = Number(p[2]);
-            if(p[2].length > 2 || !(d >= 1 && d <= 31)) return BADNUM;
-            baseDate.setDate(d);
-
-            // does that date exist in this month?
-            if(baseDate.getDate() !== d) return BADNUM;
-
-            if(datetime[1]) {
-
-                p = datetime[1].split(':');
-                if(p.length > 3) return BADNUM;
-
-                // hour - may be 1 or 2 digits
-                h = Number(p[0]);
-                if(p[0].length > 2 || !p[0].length || !(h >= 0 && h <= 23)) return BADNUM;
-                baseDate.setHours(h);
-
-                // does that hour exist in this day? (Daylight time!)
-                // (TODO: remove this check when we move to UTC)
-                if(baseDate.getHours() !== h) return BADNUM;
-
-                if(p.length > 1) {
-                    d = baseDate.getTime();
-
-                    // minute - must be 2 digits
-                    m = Number(p[1]);
-                    if(p[1].length !== 2 || !(m >= 0 && m <= 59)) return BADNUM;
-                    d += ONEMIN * m;
-                    if(p.length === 2) return d;
-
-                    // second (and milliseconds) - must have 2-digit seconds
-                    if(p[2].split('.')[0].length !== 2) return BADNUM;
-                    s = Number(p[2]);
-                    if(!(s >= 0 && s < 60)) return BADNUM;
-                    return d + s * ONESEC;
-                }
-            }
-        }
-    }
-    return baseDate.getTime();
+    return date.getTime() + S * ONESEC;
 };
 
 MIN_MS = exports.MIN_MS = exports.dateTime2ms('-9999');

test/jasmine/tests/lib_date_test.js

@@ -29,6 +29,8 @@ describe('dates', function() {
                 ['0122-04-08 08:22', new Date(122, 3, 8, 8, 22)],
                 ['-0098-11-19 23:59:59', new Date(-98, 10, 19, 23, 59, 59)],
                 ['-9730-12-01 12:34:56.789', new Date(-9730, 11, 1, 12, 34, 56, 789)],
+                // random whitespace before and after gets stripped
+                ['\r\n\t -9730-12-01 12:34:56.789\r\n\t ', new Date(-9730, 11, 1, 12, 34, 56, 789)],
                 // first century, also allow month, day, and hour to be 1-digit, and not all
                 // three digits of milliseconds
                 ['0013-1-1 1:00:00.6', d1c],
@@ -40,9 +42,19 @@ describe('dates', function() {
                 // 2-digit years get mapped to now-70 -> now+29
                 [thisYear_2 + '-05', new Date(thisYear, 4, 1)],
                 [nowMinus70_2 + '-10-18', new Date(nowMinus70, 9, 18)],
-                [nowPlus29_2 + '-02-12 14:29:32', new Date(nowPlus29, 1, 12, 14, 29, 32)]
+                [nowPlus29_2 + '-02-12 14:29:32', new Date(nowPlus29, 1, 12, 14, 29, 32)],
+
+                // including timezone info (that we discard)
+                ['2014-03-04 08:15Z', new Date(2014, 2, 4, 8, 15)],
+                ['2014-03-04 08:15:00.00z', new Date(2014, 2, 4, 8, 15)],
+                ['2014-03-04 08:15:34+1200', new Date(2014, 2, 4, 8, 15, 34)],
+                ['2014-03-04 08:15:34.567-05:45', new Date(2014, 2, 4, 8, 15, 34, 567)],
             ].forEach(function(v) {
                 expect(Lib.dateTime2ms(v[0])).toBe(+v[1], v[0]);
+
+                // ISO-8601: all the same stuff with t or T as the separator
+                expect(Lib.dateTime2ms(v[0].trim().replace(' ', 't'))).toBe(+v[1], v[0].trim().replace(' ', 't'));
+                expect(Lib.dateTime2ms('\r\n\t ' + v[0].trim().replace(' ', 'T') + '\r\n\t ')).toBe(+v[1], v[0].trim().replace(' ', 'T'));
             });
         });
 
@@ -105,7 +117,9 @@ describe('dates', function() {
                 '2015-01-00', '2015-01-32', '2015-02-29', '2015-04-31', '2015-01-001', // bad day (incl non-leap year)
                 '2015-01-01 24:00', '2015-01-01 -1:00', '2015-01-01 001:00', // bad hour
                 '2015-01-01 12:60', '2015-01-01 12:-1', '2015-01-01 12:001', '2015-01-01 12:1', // bad minute
-                '2015-01-01 12:00:60', '2015-01-01 12:00:-1', '2015-01-01 12:00:001', '2015-01-01 12:00:1' // bad second
+                '2015-01-01 12:00:60', '2015-01-01 12:00:-1', '2015-01-01 12:00:001', '2015-01-01 12:00:1', // bad second
+                '2015-01-01T', '2015-01-01TT12:34', // bad ISO separators
+                '2015-01-01Z', '2015-01-01T12Z', '2015-01-01T12:34Z05:00', '2015-01-01 12:34+500', '2015-01-01 12:34-5:00' // bad TZ info
             ].forEach(function(v) {
                 expect(Lib.dateTime2ms(v)).toBeUndefined(v);
             });