Merge pull request #2137 from OpenRefine/issue-2136-wikidata-calendar

Improved support for Wikidata dates
This commit is contained in:
Antonin Delpeuch 2019-09-11 17:16:49 +01:00 committed by GitHub
commit 29f6d1d14b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
6 changed files with 145 additions and 5 deletions

View File

@ -154,5 +154,7 @@
"warnings-messages/no-unit-provided/title": "Unit missing for {property_entity}",
"warnings-messages/no-unit-provided/body": "Values such as <span class=\"wb-issue-preformat\">{example_value}</span> on {example_item_entity} are expected to have units.",
"warnings-messages/invalid-entity-type/title": "{property_entity} used on items",
"warnings-messages/invalid-entity-type/body": "Uses of {property_entity} on items such as {example_entity} are invalid."
"warnings-messages/invalid-entity-type/body": "Uses of {property_entity} on items such as {example_entity} are invalid.",
"warnings-messages/early-gregorian-date/title": "Early dates in the Gregorian calendar",
"warnings-messages/early-gregorian-date/body": "Dates earlier than October 1582 (such as in year {example_year}) are unlikely to be expressed using the Gregorian calendar. See the <a href=\"https://www.wikidata.org/wiki/Wikidata:Tools/OpenRefine/Editing/Schema_alignment#Dates\" target=\"_blank\">manual</a> to specify the appropriate calendar for your dates."
}

View File

@ -28,6 +28,7 @@ import java.util.List;
import java.util.Map;
import java.util.stream.Collectors;
import org.openrefine.wikidata.qa.scrutinizers.CalendarScrutinizer;
import org.openrefine.wikidata.qa.scrutinizers.DistinctValuesScrutinizer;
import org.openrefine.wikidata.qa.scrutinizers.EditScrutinizer;
import org.openrefine.wikidata.qa.scrutinizers.EntityTypeScrutinizer;
@ -79,6 +80,7 @@ public class EditInspector {
register(new QuantityScrutinizer());
register(new RestrictedValuesScrutinizer());
register(new EntityTypeScrutinizer());
register(new CalendarScrutinizer());
}
/**

View File

@ -0,0 +1,31 @@
package org.openrefine.wikidata.qa.scrutinizers;
import org.openrefine.wikidata.qa.QAWarning;
import org.wikidata.wdtk.datamodel.helpers.Datamodel;
import org.wikidata.wdtk.datamodel.interfaces.TimeValue;
import org.wikidata.wdtk.datamodel.interfaces.Value;
public class CalendarScrutinizer extends ValueScrutinizer {
public static final String earlyGregorianDateType = "early-gregorian-date";
public static final TimeValue earliestGregorian = Datamodel.makeTimeValue(
1582, (byte)10, (byte)15, (byte)0, (byte)0, (byte)0, (byte)11, 0, 0, 0, TimeValue.CM_GREGORIAN_PRO);
@Override
public void scrutinize(Value value) {
if(TimeValue.class.isInstance(value)) {
TimeValue time = (TimeValue)value;
if(time.getPreferredCalendarModel().equals(earliestGregorian.getPreferredCalendarModel()) &&
time.getPrecision() >= 10 &&
(time.getYear() < earliestGregorian.getYear() ||
time.getYear() == earliestGregorian.getYear() && time.getMonth() < earliestGregorian.getMonth() ||
time.getYear() == earliestGregorian.getYear() && time.getMonth() == earliestGregorian.getMonth() && time.getDay() < earliestGregorian.getDay())) {
QAWarning warning = new QAWarning(earlyGregorianDateType, null, QAWarning.Severity.WARNING, 1);
warning.setProperty("example_year", Long.toString(time.getYear()));
addIssue(warning);
}
}
}
}

View File

@ -30,6 +30,8 @@ import java.util.Calendar;
import java.util.Date;
import java.util.Map;
import java.util.Map.Entry;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import org.jsoup.helper.Validate;
import org.openrefine.wikidata.schema.exceptions.SkipSchemaExpressionException;
@ -62,6 +64,8 @@ public class WbDateConstant implements WbExpression<TimeValue> {
.put(new SimpleDateFormat("yyyy-MM-dd"), 11)
.build();
public static Pattern calendarSuffixPattern = Pattern.compile("_(Q[1-9][0-9]*)$");
private TimeValue parsed;
private String origDatestamp;
@ -100,15 +104,46 @@ public class WbDateConstant implements WbExpression<TimeValue> {
Date bestDate = null;
int precision = 0; // default precision (will be overridden if successfully parsed)
int maxLength = 0; // the maximum length parsed
String calendarIri = TimeValue.CM_GREGORIAN_PRO; // Gregorian calendar is assumed by default
String trimmedDatestamp = datestamp.trim();
if("TODAY".equals(trimmedDatestamp)) {
Calendar calendar = Calendar.getInstance();
TimeValue todaysDate = Datamodel.makeTimeValue(
calendar.get(Calendar.YEAR),
(byte)calendar.get(Calendar.MONTH),
(byte)calendar.get(Calendar.DAY_OF_MONTH),
(byte)0, (byte)0, (byte)0, (byte)11, 0,0,0, TimeValue.CM_GREGORIAN_PRO);
return todaysDate;
}
for (Entry<SimpleDateFormat, Integer> entry : acceptedFormats.entrySet()) {
ParsePosition position = new ParsePosition(0);
String trimmedDatestamp = datestamp.trim();
Date date = entry.getKey().parse(trimmedDatestamp, position);
if (date == null) {
continue;
}
// Potentially parse the calendar Qid after the date
int consumedUntil = position.getIndex();
if(consumedUntil < trimmedDatestamp.length()) {
Matcher matcher = calendarSuffixPattern.matcher(
trimmedDatestamp.subSequence(position.getIndex(), trimmedDatestamp.length()));
if(matcher.find()) {
String calendarQid = matcher.group(1);
calendarIri = Datamodel.SITE_WIKIDATA + calendarQid;
consumedUntil = trimmedDatestamp.length();
}
}
// Ignore parses which failed or do not consume all the input
if (date != null && position.getIndex() > maxLength
// only allow to partially consume the input if the precision is more than a year
&& (entry.getValue() > 9 || position.getIndex() == trimmedDatestamp.length())) {
// only allow to partially consume the input if the precision is day and followed by a T (as in ISO)
&& (consumedUntil == trimmedDatestamp.length()
|| (entry.getValue() == 11 && trimmedDatestamp.charAt(consumedUntil) == 'T'))) {
precision = entry.getValue();
bestDate = date;
maxLength = position.getIndex();
@ -123,7 +158,7 @@ public class WbDateConstant implements WbExpression<TimeValue> {
return Datamodel.makeTimeValue(calendar.get(Calendar.YEAR), (byte) (calendar.get(Calendar.MONTH) + 1),
(byte) calendar.get(Calendar.DAY_OF_MONTH), (byte) calendar.get(Calendar.HOUR_OF_DAY),
(byte) calendar.get(Calendar.MINUTE), (byte) calendar.get(Calendar.SECOND), (byte) precision, 0, 0,
0, TimeValue.CM_GREGORIAN_PRO);
0, calendarIri);
}
}

View File

@ -0,0 +1,37 @@
package org.openrefine.wikidata.qa.scrutinizers;
import org.testng.annotations.Test;
import org.wikidata.wdtk.datamodel.helpers.Datamodel;
import org.wikidata.wdtk.datamodel.interfaces.TimeValue;
public class CalendarScrutinizerTest extends ValueScrutinizerTest {
@Override
public EditScrutinizer getScrutinizer() {
return new CalendarScrutinizer();
}
@Test
public void testScrutinizeRecentValue() {
scrutinize(Datamodel.makeTimeValue(1978L, (byte)3, (byte)4, (byte)0, (byte)0, (byte)0, 11, TimeValue.CM_GREGORIAN_PRO));
assertNoWarningRaised();
}
@Test
public void testScrutinizeCloseValue() {
scrutinize(Datamodel.makeTimeValue(1582L, (byte)10, (byte)17, (byte)0, (byte)0, (byte)0, 11, TimeValue.CM_GREGORIAN_PRO));
assertNoWarningRaised();
}
@Test
public void testScrutinizeEarlyYear() {
scrutinize(Datamodel.makeTimeValue(1400L, (byte)1, (byte)1, (byte)0, (byte)0, (byte)0, (byte)9, 0, 0, 0, TimeValue.CM_GREGORIAN_PRO));
assertNoWarningRaised();
}
@Test
public void testScrutinizeEarlyDay() {
scrutinize(Datamodel.makeTimeValue(1440L, (byte)10, (byte)17, (byte)0, (byte)0, (byte)0, 11, TimeValue.CM_GREGORIAN_PRO));
assertWarningsRaised(CalendarScrutinizer.earlyGregorianDateType);
}
}

View File

@ -23,6 +23,8 @@
******************************************************************************/
package org.openrefine.wikidata.schema;
import java.util.Calendar;
import org.openrefine.wikidata.testing.JacksonSerializationTest;
import org.testng.annotations.Test;
import org.wikidata.wdtk.datamodel.helpers.Datamodel;
@ -40,6 +42,12 @@ public class WbDateConstantTest extends WbExpressionTest<TimeValue> {
private WbDateConstant second = new WbDateConstant("2017-01-03T04:12:45");
private WbDateConstant secondz = new WbDateConstant("2017-01-03T04:12:45Z");
private WbDateConstant julianDay = new WbDateConstant("1324-02-27_Q1985786");
private WbDateConstant julianMonth = new WbDateConstant("1324-02_Q1985786");
private WbDateConstant julianYear = new WbDateConstant("1324_Q1985786");
private WbDateConstant julianDecade = new WbDateConstant("1320D_Q1985786");
@Test
public void testSerialize() {
JacksonSerializationTest.canonicalSerialization(WbExpression.class, year,
@ -70,6 +78,26 @@ public class WbDateConstantTest extends WbExpressionTest<TimeValue> {
evaluatesTo(Datamodel.makeTimeValue(2018, (byte) 2, (byte) 27, (byte) 0, (byte) 0, (byte) 0, (byte) 11, 0, 0, 0,
TimeValue.CM_GREGORIAN_PRO), whitespace);
evaluatesTo(Datamodel.makeTimeValue(1320, (byte) 1, (byte) 1, (byte) 0, (byte) 0, (byte) 0, (byte) 8, 0, 0, 0,
TimeValue.CM_JULIAN_PRO), julianDecade);
evaluatesTo(Datamodel.makeTimeValue(1324, (byte) 1, (byte) 1, (byte) 0, (byte) 0, (byte) 0, (byte) 9, 0, 0, 0,
TimeValue.CM_JULIAN_PRO), julianYear);
evaluatesTo(Datamodel.makeTimeValue(1324, (byte) 2, (byte) 1, (byte) 0, (byte) 0, (byte) 0, (byte) 10, 0, 0, 0,
TimeValue.CM_JULIAN_PRO), julianMonth);
evaluatesTo(Datamodel.makeTimeValue(1324, (byte) 2, (byte) 27, (byte) 0, (byte) 0, (byte) 0, (byte) 11, 0, 0, 0,
TimeValue.CM_JULIAN_PRO), julianDay);
}
@Test
public void testToday() {
Calendar calendar = Calendar.getInstance();
TimeValue expectedDate = Datamodel.makeTimeValue(
calendar.get(Calendar.YEAR),
(byte)calendar.get(Calendar.MONTH),
(byte)calendar.get(Calendar.DAY_OF_MONTH),
(byte)0, (byte)0, (byte)0, (byte)11, 0,0,0, TimeValue.CM_GREGORIAN_PRO);
evaluatesTo(expectedDate, new WbDateConstant("TODAY"));
}
@Test(expectedExceptions = IllegalArgumentException.class)
@ -81,4 +109,9 @@ public class WbDateConstantTest extends WbExpressionTest<TimeValue> {
public void testPartlyValid() {
new WbDateConstant("2018-partly valid");
}
@Test(expectedExceptions = IllegalArgumentException.class)
public void testInvalidCalendar() {
new WbDateConstant("2018-01-02_P234");
}
}