Templates: various questions not worth their own thread

[dunhill]

mistake in formula
I have a column with date read finished
I want to know what month it is in another column. So I created a column that depended on another and in the template I put it like this:
{#column name:'test($, strcat(format_date($,'mmm')),'Without date')'}

As a result I get Without date, 000 or Bad date.
How could I get the month that is in this way Jan, Feb, Mar

[dunhill]

Originally Posted by dunhill

mistake in formula
I have a column with date read finished
I want to know what month it is in another column. So I created a column that depended on another and in the template I put it like this:
{#column name:'test($, strcat(format_date($,'mmm')),'Without date')'}

As a result I get Without date, 000 or Bad date.
How could I get the month that is in this way Jan, Feb, Mar

I answer myself
The month must be capital letter (MMM) but I did not discover why I get Bad Date

[chaley]

You don't say what a "bad date" looks like when showing in the booklist or book details. Because you are using a printable date (the value of '#column name') the month names will be localized. The date functions in calibre are localized to accept 3-letter month names in English, French, and German. See replace_months() in calibre's source calibre.utils.date.

I expect in your case the month abbreviations are in Spanish. In that language many 3-letter month abbreviations are the same as in English, but several differ from their English equivalents such as "abr" instead of "Apr"

You avoid this problem by using raw_field to get the date in ISO-format instead of localized. Something like this, replacing '#mydate' with the real lookup name:

{:'ifempty(format_date(raw_field('#mydate', ''), 'MMM'), 'Without date')'}

or this, which is more like your original.

{:'d = raw_field('#mydate', ''); test(d, strcat(format_date(d,'MMM')),'Without date')'}

Note: if this is the only template in the composite column then I suggest you use General Program Mode, as that will be faster because of template caching. Something like this:

program: ifempty(format_date(raw_field('#mydate', ''), 'MMM'), 'Without date')

Or even faster: a Python template:

python:
def evaluate(book, context):
	d = book.get('#mydate')
	if d is None:
		return 'Without date'
	# Localize these as you wish
	months = ['ene', 'feb', 'mar', 'abr', 'may', 'jun', 
		  'jul', 'ago', 'sep', 'oct', 'nov', 'dic']
	return months[d.month-1]

[ownedbycats]

Is this the only way to check that #timedatecolumn1 is newer than #timedatecolumn2?

days_between($$#timedatecolumn1, $$#timedatecolumn2) >#0

[dunhill]

Originally Posted by chaley

You don't say what a "bad date" looks like when showing in the booklist or book details. Because you are using a printable date (the value of '#column name') the month names will be in localized. The date functions in calibre are localized to accept 3-letter month names in English, French, and German. See replace_months() in calibre's source calibre.utils.date.

I expect in your case the month abbreviations are in Spanish. In that language many 3-letter month abbreviations are the same as in English, but several differ from their English equivalents such as "abr" instead of "Apr"

You avoid this problem by using raw_field to get the date in ISO-format instead of localized. Something like this, replacing '#mydate' with the real lookup name:

{:'ifempty(format_date(raw_field('#mydate', ''), 'MMM'), 'Without date')'}

or this, which is more like your original.

{:'d = raw_field('#mydate', ''); test(d, strcat(format_date(d,'MMM')),'Without date')'}

Note: if this is the only template in the composite column then I suggest you use General Program Mode, as that will be faster because of template caching. Something like this:

program: ifempty(format_date(raw_field('#mydate', ''), 'MMM'), 'Without date')

Or even faster: a Python template:

python:
def evaluate(book, context):
	d = book.get('#mydate')
	if d is None:
		return 'Without date'
	# Localize these as you wish
	months = ['ene', 'feb', 'mar', 'abr', 'may', 'jun', 
		  'jul', 'ago', 'sep', 'oct', 'nov', 'dic']
	return months[d.month-1]

Brilliant !! thank you so much !!

[chaley]

Originally Posted by ownedbycats

Is this the only way to check that #timedatecolumn1 is newer than #timedatecolumn2?

days_between($$#timedatecolumn1, $$#timedatecolumn2) >#0

No.

I assume either date can be undefined.

This one works because ISO formatted dates are comparable as strings as long as the dates are in the same timezone.

program:
	d1 = $$#mydate;
	d2 = $$date;
	if d1 == 'None' || d2 == 'None' then return '' fi;
	if d1 < d2 then 'yes' fi

An alternate way that uses timestamps, which should work with differing timezones but might have problems with dates previous to "the epoch" (1970 IIRC):

program:
	d1 = $$#mydate;
	d2 = $$date;
	if d1 == 'None' || d2 == 'None' then return '' fi;
	d1 = format_date($#mydate, 'to_number');
	d2 = format_date($date, 'to_number');
	if d1 <# d2 then 'yes' fi

Or a Python template that works with timezones:

python:
def evaluate(book, context):
	d1 = book.get('#mydate')
	d2 = book.get('timestamp')
	if d1 is None or d2 is None:
		return '' # return no if either date isn't defined
	return 'Yes' if d1 < d2 else ''

[ownedbycats]

If I have no book selected when I go into template functions, I get a generic 'title' book. What are the values of this book?

Also, is it intended that it fails on functions that only work in GUI? approximate_formats is one.

[ownedbycats]

Additionally, I updated my #kobocoll generation template again to resolve a Kobo glitch (tl;dr: too many books in a collection causes sync to stall):

program:
	genre = $#subjects;
	colls = strcat(	
##		$#booktype, ',',
##		if $$#purchasecost == '0.00' then 'Freebies,' fi,
		if 'Omnibus' in $#admintags then 'Omnibuses,' fi,
		if 'overdrive' in approximate_formats() && !'formerlibrary' in $#admintags then 'Loans,' fi, 
		if $#booktype=='Fanfiction' then list_re($tags, ',', '^Fanfiction.(.*)$', 'Fanfiction: \1,') fi,
		if '^Cozy Mystery$' inlist genre then 'Cozy Mysteries,' fi,
		if '^Horses$' inlist genre then 'Horses,' fi,
		if '^(Science Fiction)' inlist genre then 'Science Fiction,' fi,
		if '^(Fantasy|Paranormal)$' inlist genre then 'Fantasy & Paranormal,' fi,
);

	if colls == '' then colls == strcat($#booktype, ',') fi;

	new_colls = list_union($#kobocoll, colls, ',')

I'm trying to make it so that

a) Define whatever the strcat returns as 'colls'
b) if strcat returns nothing and 'colls' is empty, define #booktype and a comma as 'colls'
c) Merge colls with the existing column (taglike).

It's not working for me though; I just get a blank result. I also tried !colls with the same result. I thought of enclosing it in a switch_if but wasn't sure what to put as the first test_expression.

[ownedbycats]

Third question: Is there some sort of recursion protection in stored templates? I added columnupdate_kobocoll() (the name of the stored template I'm editing) to the strcat just
because I like trying to break things to see what would happen and got the same results.

[isarl]

Originally Posted by ownedbycats

if colls == '' then colls == strcat($#booktype, ',') fi;

Replace the second double equals sign in this line with a single equals sign. Does your template work now?

[ownedbycats]

Yes. Silly mistake.

[ownedbycats]

When short-circuiting is accounted for, would it maybe be faster to do two separate exact checks:

($#columnname=='foo'||$#columnname=='bar')

Than one regex match?

'^(foo|bar)$' in $#columnname

[chaley]

Originally Posted by ownedbycats

When short-circuiting is accounted for, would it maybe be faster to do two separate exact checks:

($#columnname=='foo'||$#columnname=='bar')

Than one regex match?

'^(foo|bar)$' in $#columnname

In this case and assuming that the first check is true a reasonable number of times, then yes because the == operator is much faster than the 'in' operator.

[ownedbycats]

New question: I have this template that populates my Kobo collections column. (Non-composite, Action Chains single-field edit).

program:
	genre = $#subjects;
	colls = strcat(	
##		if $$#purchasecost == '0.00' then 'Freebies,' fi,
		if 'Omnibus' in $#admintags then 'Omnibuses,' fi,
		if 'overdrive' in approximate_formats() && !$purchasesource then 'Loans,' fi, 
		if $#booktype=='Fanfiction' then list_re($tags, ',', '^Fanfiction.(.*)$', 'Fanfiction: \1,') fi,
		if 'public domain' inlist $#admintags then 'Public Domain,' fi,
		if '^Cozy Mystery$' inlist genre then 'Cozy Mysteries,' fi,
		if '^Horses$' inlist genre then 'Horses,' fi,
		if 'Folktales & Mythology' inlist genre then 'Folktales & Mythology,' fi,
		if '^Science Fiction$' inlist genre then 'Science Fiction,' fi,
		if '^(Fantasy|Paranormal)$' inlist genre then 'Fantasy & Paranormal,' fi,
	);

	if colls == '' then colls = strcat($#booktype, ',') fi;

	new_colls = list_union($#kobocoll, colls, ',')

The strcat lines generates various entries, line 16 fallsback to #booktype if there's nothing generated, and line 18 merges with the existing entries. (Being able to add additional entries for individual books is why I don't use a composite column here.)

Sometimes this merging results in leftover #booktype entries, due to my workflow (running the column update when there's a #booktype but no #subjects). Less common, but also happens when adding a new entry to the strcat and bulk-updating.

A real example:
A book with 'Fiction' in #booktype, 'Science Fiction' in #subjects, and an existing #kobocoll value of 'Fiction'
colls returns 'Science Fiction'
new_colls then produces 'Fiction, Science Fiction'

What is the best way to remove 'Fiction' from this without it getting merged back in? A list_difference, but I'm having trouble figuring out what order to put things in

[chaley]

Originally Posted by ownedbycats

New question: I have this template that populates my Kobo collections column. (Non-composite, Action Chains single-field edit).

It is very hard to know how to "fix" this without knowing the universe of possibilities.

The following works for your example. I don't know if it will work in general. It removes any value in #booktype from #kobocol before merging, which seems to be what you are looking for.

new_colls = list_union(list_difference($#kobocol, $#booktype, ','), colls, ',')