Add XXXX humanization by thomas-topway-it · Pull Request #88 · ProfessionalWiki/EDTF

thomas-topway-it · 2023-01-16T11:13:20Z

thomas-topway-it · 2023-01-16T11:17:16Z

if ( $unspecifiedDigit->getYear() === 4

(in InternationalizedHumanizer class)
could be moved to UnspecifiedDigit class, for instance method yearIsFullyUnspecified

JeroenDeDauw · 2023-01-16T12:35:46Z

src/PackagePrivate/Humanizer/InternationalizedHumanizer.php

+
+		// *** if the expression $date->getDay() === null && $date->getMonth() === null
+		// is superfluous as it seems the function can be restored to its
+		// original declaration ( int $year, UnspecifiedDigit $unspecifiedDigit )


We'll, find out by removing the code. We have the tests, and you can add more tests if there are cases you are not sure about.

JeroenDeDauw · 2023-01-16T12:36:49Z

src/PackagePrivate/Humanizer/InternationalizedHumanizer.php

+		// is superfluous as it seems the function can be restored to its
+		// original declaration ( int $year, UnspecifiedDigit $unspecifiedDigit )
+
+		if ( $unspecifiedDigit->getYear() === 4 && $date->getDay() === null && $date->getMonth() === null ) {


What is this 4 about?

If all 3 of these conditions are needed, then extracting into isUnspecifiedYear(): bool would be nice. Remember: one level of abstraction.

Also, I think that -5XXXX is possible, which might break your === 4 check.

mzeinstra · 2023-01-16T14:34:43Z

tests/Functional/EnglishHumanizationTest.php

 		yield 'Day only' => [ 'XXXX-XX-12', '12th' ];
+
+		// https://github.com/ProfessionalWiki/EDTF/issues/80
+		yield 'Some year' => [ 'XXXX', 'some year' ];


Maybe add
yield 'Some year' => [ 'X', 'some year' ];
yield 'Some year' => [ 'XXX', 'some year' ];

Curious as to what "-X" would yield.

And check what -5XXXX results into, unless already tested

And check what -5XXXX results into, unless already tested

50000 BC

this is nice, but still doesn't account for the unspecified digits

it should perhaps be:

50000 BC (millennium, century and year in decem millennium unspecified)

That does not scale. Because what is -5XXXXXXX ? How about 50000 BC (year unspecified)

-5XXXXXXX would be "50 millions years BC" and -5XXXX " 5 decem millennium BC",
so (given that your solution is the only viable at the moment) in general for unspecified digits we could use always the literal expression, that is "5 millenniums BC" instead than "5000 BC (year unspecified)" (with the first notation it is implicit)

That means implementing scales (https://en.wikipedia.org/wiki/Long_and_short_scales#Short_scale) e.g. the short scale to billion? As (according to Google) the universe is 13 billion years old and will likely exists for 22 billion years and this is about dates. :)

I was checking further the specification here https://www.loc.gov/standards/datetime/
(Unspecified Digit) and the parser,
but I'm not sure that the parser is checking whether
the unspecified digit is placed incrementally from the right of the date component.
e.g.
the following should throw an error:

-XXXXXXX4 (4th year of any millennium BC)

JeroenDeDauw · 2023-01-17T23:44:32Z

tests/Functional/EnglishHumanizationTest.php

+		yield 'Scales (8 digits)' => [ '5XXXXXXXX', '5 hundreds of millions' ];
+		yield 'Scales (9 digits)' => [ '5XXXXXXXXX', '5 billions' ];
+		yield 'Scales (10 digits)' => [ '5XXXXXXXXXX', '5 tens of billions' ];
+		yield 'Scales (11 digits)' => [ '5XXXXXXXXXXX', '5 hundreds of billions' ];


"5 decem millennia"? What does decem mean here?

"5 hundreds of thousands" is just an amount; should it not also have "years" in there somewhere? Is this even a valid EDTF date to begin with? @mzeinstra

yes, sometimes the word "year" is missing, I have still to add it where necessary

"5 decem millennia"? What does decem mean here?

"Decem millennium" theoretically means "Ten thousand years" it could be also used the expression "tens of thousands"

tens of thousands is indeed better, but we can also leave that up to the translators.

Cosmology seems to me a valid border of amounts of years that we humanise. Maybe a fallback to just presenting the years in numbers?

The most years I can find in the sciences is the half-life of Xenon, which is 18 billion trillion years :)

As we should not forget the objective here, that is to make these edtf-string human readable. If 99% of the use cases can be captured than that is ok with me.

Previously I also posted this link

https://asistdl.onlinelibrary.wiley.com/doi/epdf/10.1002/pra2.552

in the element.io SemanticMediawiki group because somebody asked about the support for "geological and historical eras".

Regarding the following

Cosmology seems to me a valid border of amounts of years that we humanise. Maybe a fallback to just presenting the years in numbers?

in my opinion if the user registers a date with unspecified digits and the library outputs a human readable format, and additionally computes correctly comparisons and intervals, that's really useful.

Still in my opinion the approach to present a year with mixed unspecified digits using numbers for the specified digits and literals for the unspecified (scaled) part, seems also formally rigorous. So using the example above -5XXXXXXX is not to be precise "50 millions years BC" but "5 tens of millions years BC" (in the output the number represents the specified digits, and the literal part the unspecified digits)

So what about, when this is more refined, to publish a demo (which could query the library using Ajax) and to propose this approach to the authors of the format, the Washington' Library of Congress, as far as I know ?

Afaik there is no standard for the humanisation of EDTF, humanised examples are used on the library of congres website but they are not the owner of the standard.

So your best efforts here will de facto help set the humanisation standard.

@mnhn-paul you might be able to provide us with a standard of humanising eras in EDTF. Eg. What would the date: -5XXXXXXX be humanised to?

Paul is from our museum of natural history and might have more insight into this.

JeroenDeDauw · 2023-01-17T23:45:42Z

src/Model/ExtDate.php

+
+			return $ret + 1;
+		}
+		return 0;


Can apply https://www.entropywins.wtf/blog/2019/01/14/readable-functions-guard-clause/

$ret is not a good var name

JeroenDeDauw · 2023-01-17T23:46:20Z

src/PackagePrivate/Humanizer/InternationalizedHumanizer.php

 	}

-	private function humanizeYear( int $year, UnspecifiedDigit $unspecifiedDigit ): string {
+	private function scaleToMessageKey( int $scale ): string {


Yay for putting this in its own method!

JeroenDeDauw · 2023-01-17T23:47:05Z

src/PackagePrivate/Humanizer/InternationalizedHumanizer.php

+
+		// FIXME: reuse recursively the scale with trillions
+		// e.g. tens-of-trillions etc., 
+		return 'edtf-tens-of-trillions';


Let's first make sure we actually need this. I am dubious that these are valid EDTF dates.

sure, the only sensible use-case is in cosmology

Let's first make sure we actually need this. I am dubious that these are valid EDTF dates.

actually they are required only for testing purposes, e.g. by an user who wonder what is the output entering something like that 56XXXXXXXXXXXXXXXXXXX

JeroenDeDauw · 2023-01-17T23:48:51Z

src/PackagePrivate/Humanizer/InternationalizedHumanizer.php

+
+		if ( $date->isBC() ) {
+			$ret .= " " . $this->message( "edtf-date-BC" );
+		}


Concatenating messages like this bakes in an assumption about the order of these messages, which might not hold in all languages. That is why for instance we have messages such as "edtf-qualified-date": "$1 ($2)".

yes, I guessed something like that, so that should be taken into account and enhanced

JeroenDeDauw · 2023-01-17T23:49:34Z

src/PackagePrivate/Humanizer/InternationalizedHumanizer.php

+			$ret .= " " . $this->message( "edtf-date-BC" );
+		}
+
+		return $ret;


$humanizedYear

JeroenDeDauw reviewed Jan 16, 2023

View reviewed changes

mzeinstra reviewed Jan 16, 2023

View reviewed changes

thomas-topway-it added 19 commits January 18, 2023 00:39

fix XXXX humanization

72e9cc4

fix XXXX humanization

5ffc831

XXXX humanization

6199d54

implements scales

80bee4e

implement scales

57252e1

implements scales

35ee9fc

added comment

ee0bc9b

psalm error

95e812b

fix static error

b5ca5fc

test scales

35ad309

fix scales

7d40066

fix scales test

bf5dbb4

fix scales plural

bd86be4

fix BC value

2daa0a3

fix BC value (revert previous commit)

0e58e88

fix BC

234ebb4

fix static

d569714

fix tests

0d6a4e6

fix static

edb98e4

JeroenDeDauw force-pushed the KMA-XXXX-humanization branch from e008d42 to edb98e4 Compare January 17, 2023 23:40

JeroenDeDauw reviewed Jan 17, 2023

View reviewed changes

JeroenDeDauw changed the title ~~Kma XXXX humanization~~ Add XXXX humanization Jan 17, 2023

Uh oh!

Conversation

thomas-topway-it commented Jan 16, 2023

Uh oh!

thomas-topway-it commented Jan 16, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

thomas-topway-it Jan 17, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

thomas-topway-it commented Jan 16, 2023 •

edited

Loading

thomas-topway-it Jan 17, 2023 •

edited

Loading