continuing experiments

Author: zebdelk

schemas-analyze/src/main/java/org/cedar/schemas/analyze/DateInfo.java

@@ -26,6 +26,7 @@ public class DateInfo implements Comparable<DateInfo> {
   public final boolean indexable;
   public final String zoneSpecified;
   public final String utcDateTimeString;
+  public final String endUtcDateTimeString;
   public final Long year;
   public Integer dayOfYear; // values 1 - 366 // TODO temp remove final from these fields, just to sanity check things
   public Integer dayOfMonth; // values 1 - 31
@@ -43,6 +44,7 @@ public DateInfo(String dateString, boolean start) {
       indexable = true;
       zoneSpecified = null;
       utcDateTimeString = null;
+      endUtcDateTimeString = null;
       year = null;
       dayOfYear = null;
       dayOfMonth = null;
@@ -68,20 +70,21 @@ public DateInfo(String dateString, boolean start) {
       endDayOfMonth = ((YearMonth)parsedDate).lengthOfMonth();
       dayOfYear = ((YearMonth)parsedDate).atDay(1).getDayOfYear();
       endDayOfYear = ((YearMonth)parsedDate).atEndOfMonth().getDayOfYear();
-      System.out.println("year month - setting endDayOf* fields..." + endDayOfMonth);
-      System.out.println("year month - setting endDayOf* fields..." + dayOfYear);
-      System.out.println("year month - setting endDayOf* fields..." + endDayOfYear);
+      // endUtcDateTimeString = utcDateTimeString(((YearMonth)parsedDate).atEndOfMonth(), false);
     } else if (parsedDate != null && month == null && year != null) { // TODO or instanceof Year?
       dayOfMonth = 1;
       dayOfYear = 1;
       endDayOfYear = ((Year)parsedDate).length(); // number of days in the year, including leap years
       endDayOfMonth = 31;
       month = 1;
       endMonth = 12;
+
+      // endUtcDateTimeString = utcDateTimeString(((Year)parsedDate).atDay(endDayOfYear), false);
     } else {
       endMonth = month;
       endDayOfMonth = dayOfMonth;
       endDayOfYear = dayOfYear;
+      // endUtcDateTimeString = utcDateTimeString;
     }
 
     if (longDate != null && !indexable(longDate)) {
@@ -90,20 +93,23 @@ public DateInfo(String dateString, boolean start) {
       indexable = indexable(longDate);
       zoneSpecified = timezone(longDate);
       utcDateTimeString = utcDateTimeString(longDate, start);
+      endUtcDateTimeString = utcDateTimeString(longDate, false);
     }
     else if (parsedDate != null) {
       descriptor = ValidDescriptor.VALID;
       precision = precision(parsedDate);
       indexable = true;
       zoneSpecified = timezone(parsedDate);
       utcDateTimeString = utcDateTimeString(parsedDate, start);
+      endUtcDateTimeString = utcDateTimeString(parsedDate, false);
     }
     else {
       descriptor = ValidDescriptor.INVALID;
       precision = null;
       indexable = false;
       zoneSpecified = null;
       utcDateTimeString = null;
+      endUtcDateTimeString = null;
     }
   }
 

schemas-analyze/src/main/java/org/cedar/schemas/analyze/Temporal.java

@@ -98,15 +98,16 @@ public static TemporalBoundingAnalysis analyzeBounding(Discovery metadata) {
       builder.setEndZoneSpecified(endInfo.zoneSpecified);
       builder.setEndUtcDateTimeString(endInfo.utcDateTimeString);
       builder.setEndYear(endInfo.year);
-      builder.setEndDayOfYear(endInfo.dayOfYear);
-      builder.setEndDayOfMonth(endInfo.dayOfMonth);
-      builder.setEndMonth(endInfo.month);
+      builder.setEndDayOfYear(endInfo.endDayOfYear == null ? endInfo.dayOfYear : endInfo.endDayOfYear);
+      builder.setEndDayOfMonth(endInfo.endDayOfMonth == null ? endInfo.dayOfMonth : endInfo.endDayOfMonth);
+      builder.setEndMonth(endInfo.endMonth == null ? endInfo.month : endInfo.endMonth);
 
       builder.setInstantDescriptor(instantInfo.descriptor);
       builder.setInstantPrecision(instantInfo.precision);
       builder.setInstantIndexable(instantInfo.indexable);
       builder.setInstantZoneSpecified(instantInfo.zoneSpecified);
       builder.setInstantUtcDateTimeString(instantInfo.utcDateTimeString);
+      builder.setInstantEndUtcDateTimeString(instantInfo.endUtcDateTimeString);
       builder.setInstantYear(instantInfo.year);
       builder.setInstantDayOfYear(instantInfo.dayOfYear);
       builder.setInstantDayOfMonth(instantInfo.dayOfMonth);

schemas-analyze/src/test/groovy/org/cedar/schemas/analyze/AnalyzersSpec.groovy

@@ -93,6 +93,7 @@ class AnalyzersSpec extends Specification {
             instantIndexable        : true,
             instantZoneSpecified    : null,
             instantUtcDateTimeString: null,
+            instantEndUtcDateTimeString: null,
             instantDayOfYear        : null,
             instantDayOfMonth       : null,
             instantYear             : null,

schemas-analyze/src/test/groovy/org/cedar/schemas/analyze/TemporalSpec.groovy

@@ -17,6 +17,29 @@ class TemporalSpec extends Specification {
 
   final String analysisAvro = ClassLoader.systemClassLoader.getResourceAsStream('avro/psi/analysis.avsc').text
 
+  def "end date #description (#endDate) correctly extrapolates date"() {
+    given:
+    def bounding = TemporalBounding.newBuilder()
+        .setEndDate(endDate)
+        .build()
+    def discovery = Discovery.newBuilder().setTemporalBounding(bounding).build()
+
+    when:
+    def result = Temporal.analyzeBounding(discovery)
+
+    then:
+    result.endMonth == month
+    result.endDayOfMonth == dayOfMonth
+    result.endDayOfYear == dayOfYear
+
+    where:
+    description | endDate | month | dayOfMonth | dayOfYear
+    "month precision" | '2003-02' | 2 | 28 | 59
+    "month precision leap year" | '2004-02' | 2 | 29 | 60
+    "year precision leap year" | '2004' | 12 | 31 | 366
+    "year precision" | '2003' | 12 | 31 | 365
+  }
+
   def "instant #description (#instant) correctly extrapolates date range"() {
     given:
     def bounding = TemporalBounding.newBuilder()
@@ -28,21 +51,21 @@ class TemporalSpec extends Specification {
     def result = Temporal.analyzeBounding(discovery)
 
     then:
-    println("????"+result)
     result.instantMonth == month
     result.instantEndMonth == endMonth
     result.instantDayOfMonth == dayOfMonth
     result.instantEndDayOfMonth == endDayOfMonth
     result.instantDayOfYear == dayOfYear
     result.instantEndDayOfYear == endDayOfYear
+    result.instantEndUtcDateTimeString == endString
 
     where:
 
-    description | instant | year | dayOfYear | dayOfMonth | month | endYear | endDayOfYear | endDayOfMonth | endMonth
-    "instant with month precision" | '2003-02' | 2003 | 32 | 1 | 2 | 2003 | 59 | 28 | 2
-    "instant on leapyear with month precision" | '2004-02' | 2004 | 32 | 1 | 2 | 2004 | 60 | 29 | 2
-    "instant with day precision" | '2001-06-22' | 2001 | 173 | 22 | 6 | 2001 | 173 | 22 | 6
-    "instant with day precision on leapyear" | '2020-06-22' | 2020 | 174 | 22 | 6 | 2020 | 174 | 22 | 6
+    description | instant | year | dayOfYear | dayOfMonth | month | endYear | endDayOfYear | endDayOfMonth | endMonth | endString
+    "instant with month precision" | '2003-02' | 2003 | 32 | 1 | 2 | 2003 | 59 | 28 | 2 | '2003-02-28T23:59:59.999Z'
+    "instant on leapyear with month precision" | '2004-02' | 2004 | 32 | 1 | 2 | 2004 | 60 | 29 | 2 | '2004-02-29T23:59:59.999Z'
+    "instant with day precision" | '2001-06-22' | 2001 | 173 | 22 | 6 | 2001 | 173 | 22 | 6 | '2001-06-22T23:59:59.999Z'
+    "instant with day precision on leapyear" | '2020-06-22' | 2020 | 174 | 22 | 6 | 2020 | 174 | 22 | 6 | '2020-06-22T23:59:59.999Z'
 
   }
 

schemas-core/src/main/resources/avro/psi/analysis.avsc

@@ -488,6 +488,15 @@
               "default": null,
               "doc": "If possible, a full ISO-8601 formatted date time string in the UTC time zone for the analyzed instant value"
             },
+            {
+              "name": "instantEndUtcDateTimeString",
+              "type": [
+                "null",
+                "string"
+              ],
+              "default": null,
+              "doc": "If possible, a full ISO-8601 formatted date time string in the UTC time zone for the analyzed instant value. If instantPrecision does not include time, day, or month, the value is the assumed date time the instant ends, e.g. the last day of a month or year"
+            },
             {
               "name": "instantYear",
               "type": [

schemas-core/src/test/resources/example-record-avro.json

@@ -578,6 +578,7 @@
           },
           "instantZoneSpecified": null,
           "instantUtcDateTimeString": null,
+          "instantEndUtcDateTimeString": null,
           "instantYear": null,
           "instantMonth": null,
           "instantDayOfMonth": null,