[GUI Plugin] Action Chains

[capink]

Originally Posted by chaley

You can do this with a python template. It would probably be better as an Action Chains python action but I didn't feel like doing that.

python:
def evaluate(book, context):
	import os
	from shutil import copy
	from calibre.db.constants import DATA_DIR_NAME

	db = context.db
	cache = db.new_api
	# Get the normalized library path
	library_path = cache.backend.library_path
	
    # Construct the full path to the book folder
	path = os.path.join(library_path, cache.field_for('path', book.id).replace('/', os.sep))

	# Ensure the data directory exists
	data_dir = os.path.join(path, DATA_DIR_NAME)
	if not os.path.exists(data_dir):
		os.mkdir(data_dir)

	cover_file = os.path.join(path, 'cover.jpg')
	if os.path.exists(cover_file):
		# It does. Copy it to the data directory. The 'copy' method takes a
		# directory as a target.
		copy(cover_file, data_dir)
		return 'copied'
	return 'not copied'

I added a link to this in the templates section in the Action Chains Resources thread.

[ownedbycats]

Thanks

[ownedbycats]

Question: Where should I put it? I tried a SFE and ended up accidentally changing the title of a book to 'copied.' Oops.

[chaley]

Originally Posted by ownedbycats

Question: Where should I put it? I tried a SFE and ended up accidentally changing the title of a book to 'copied.' Oops.

I would put it in an SFE cover action. Return the empty string so the cover isn't changed.

Or leave it in a title SFE and return the current title so nothing changes.

[ownedbycats]

That did it.

[chaley]

@capink: A question ...

I sometimes want to run a (python) template over the selected books. This isn't the easiest thing to do.

Solutions I have found:

• Select the books then use the template tester. Although this works, it isn't an obvious or particularly efficient solution. For example it doesn't support keyboard shortcuts. It does have the advantage that I can see error output.
• Write a Python code action. This works, but I must implement in the template iteration over the currently selected books. In addition, there isn't a good way to show errors.
• Write a Formulas action. Here I again must implement my own selection iteration. However, it has the advantage that I can display errors. However, I can't find a way to show nothing if there is no output.
• Use a Single Field Edit. This solution takes care of iteration. However, it really wants to modify some column, which in this case I don't want to do. The work-around: pick a column like Title and return that value so nothing changes. In addition, it isn't easy to see errors without using an "error column" (long comments) and writing the errors into that.

To solve this problem I have considered making a calibre action that runs a template over the selected books. If the template produces no output then the action doesn't show a dialog. If the template does produce output for any/all books then it opens a dialog like the middle part of the template tester, showing the output per book.

My problem with this solution is that it feels like I am reinventing a part of Action Chains, which seems a bit silly to me. A Formulas-like action that iterates over the selection and shows output only if there is any is more-or-less the same thing.

What are your thoughts? How would you solve this problem?

[capink]

Originally Posted by chaley

@capink: A question ...

I sometimes want to run a (python) template over the selected books. This isn't the easiest thing to do.

Solutions I have found:

• Select the books then use the template tester. Although this works, it isn't an obvious or particularly efficient solution. For example it doesn't support keyboard shortcuts. It does have the advantage that I can see error output.
• Write a Python code action. This works, but I must implement in the template iteration over the currently selected books. In addition, there isn't a good way to show errors.
• Write a Formulas action. Here I again must implement my own selection iteration. However, it has the advantage that I can display errors. However, I can't find a way to show nothing if there is no output.
• Use a Single Field Edit. This solution takes care of iteration. However, it really wants to modify some column, which in this case I don't want to do. The work-around: pick a column like Title and return that value so nothing changes. In addition, it isn't easy to see errors without using an "error column" (long comments) and writing the errors into that.

To solve this problem I have considered making a calibre action that runs a template over the selected books. If the template produces no output then the action doesn't show a dialog. If the template does produce output for any/all books then it opens a dialog like the middle part of the template tester, showing the output per book.

My problem with this solution is that it feels like I am reinventing a part of Action Chains, which seems a bit silly to me. A Formulas-like action that iterates over the selection and shows output only if there is any is more-or-less the same thing.

What are your thoughts? How would you solve this problem?

If you don't want to modify the values it is better to steer away from SFE. Two solutions spring to mind:

1. An iterate checkbox was introduced to the Chains Variables Action. This will run the template for all the books in the current scope (default is selected books).
2. The Second solution is using a Formulas Action with a combination of from_selection() and book_field()
   

   program:
       for book_id in from_selection('id'):
           title = book_field(book_id, 'title')
       rof;

   This might not be convenient as it does not allow the usage of a short form like $title.

If either of these solutions not sufficient, please let me know why, so we can work out a solution together.

A solution is not necessarily a new action, it might be a matter of a new template function. While we are at it, is there for a way for a function to know which template mode it is running in, to modify the output depending on the mode. e.g. from_selection() could return a python list instead of a string if running in a python mode.

Edit: As for errors, you can save them in a global variable, and display them using a Formula Action at the end of the chain. This works for any kind of solution you choose, whether a Formula Action, a Chains Variables Action or even a "Run Python Code" Action.

[chaley]

Originally Posted by capink

If you don't want to modify the values it is better to steer away from SFE. Two solutions spring to mind:
[snip]

I ended up using chain variables and a formula to show output, as you suggested.

A solution is not necessarily a new action, it might be a matter of a new template function. While we are at it, is there for a way for a function to know which template mode it is running in, to modify the output depending on the mode. e.g. from_selection() could return a python list instead of a string if running in a python mode.

utils.formatter_functions.function_object_type(template)

This accepts template text, a stored template, or a template function.

Edit: As for errors, you can save them in a global variable, and display them using a Formula Action at the end of the chain. This works for any kind of solution you choose, whether a Formula Action, a Chains Variables Action or even a "Run Python Code" Action.

I ended up building an "example" action chain that I can use as the starting point. It contains two actions: a chain variables action that is run over the selected books, and a formulas action that displays any output.

The chain variables action puts its output into a list stored in globals. It also sets the chain variable "has_output" if there is anything in the list. The formulas action has a condition ensuring the action is run only if "has_output" is set to the non-empty string. The export of the prototype action chain is attached.

Here is the prototype chain variables action:

python:
def evaluate(book, context):
	# Run the desired template. It must produce a string
	output = run_template(book, context)

	# if the template produced some output, store it in globals to be printed later
	all_output = context.globals.get('output', [])
	if output:
		all_output.append(f'{book.title} ({book.id}): {output}')
		context.globals['output'] = all_output
	return 'yes' if all_output else ''

def run_template(book, context):
	# you can directly enter a python template here, or alternatively call
	# a stored template (GPM or python) or formatter function using
	# context.funcs.name.
	# Example: context.funcs.template('{series:|| [}{series_index:||]}')
	# runs the SFM template to return the series and series index
	return context.funcs.template('{series:|| [}{series_index:||]}')

Here is the formula action. It formats the list of "errors" from the chain vars action into a single string.

python:
def evaluate(book, context):
	return '\n'.join(context.globals.get('output', ''))

The condition check is equally simple:

program:
	return globals(has_output) != ''

Here is an example that uses @ownedbycats request, to copy the cover into the data directory for all selected books. At the end it shows a dialog listing the books and whether or not a cover was copied.

python:
def evaluate(book, context):
	# Run the desired template. It must produce a string
	output = run_template(book, context)

	# if the template produced some output, store it in globals to be printed later
	all_output = context.globals.get('output', [])
	if output:
		all_output.append(f'{book.title} ({book.id}): {output}')
		context.globals['output'] = all_output
	return 'yes' if all_output else ''

def run_template(book, context):
	import os
	from shutil import copy
	from calibre.db.constants import DATA_DIR_NAME

	db = context.db
	cache = db.new_api
	# Get the normalized library path
	library_path = cache.backend.library_path
	
    # Construct the full path to the book folder
	path = os.path.join(library_path, cache.field_for('path', book.id).replace('/', os.sep))

	# Ensure the data directory exists
	data_dir = os.path.join(path, DATA_DIR_NAME)
	if not os.path.exists(data_dir):
		os.mkdir(data_dir)

	cover_file = os.path.join(path, 'cover.jpg')
	if os.path.exists(cover_file):
		# It does. Copy it to the data directory. The 'copy' method takes a
		# directory as a target.
		copy(cover_file, data_dir)
		return 'cover copied'
	return 'cover not copied'

[chaley]

@capink: I noticed that when I run a chain with several quick-acting actions I get a "screen flash" because of the action chains progress bar. Looking at the code I saw that you already didn't show the progress bar for Chain Caller.

To avoid that flash, using the Chain Caller code as an example I added a checkbox to the chain dialog to explicitly enable/disable the progress bar. This eliminated the flash. The changes are compatible with existing chains and the import/export process.

In case you are interested in seeing or incorporating these changes I am attaching the plugin zip and a patch file. I didn't change the version number.

EDIT: removed the plugin zip because it has been incorporated into a release.

[ownedbycats]

I would use this if implemented.

[capink]

Originally Posted by chaley

@capink: I noticed that when I run a chain with several quick-acting actions I get a "screen flash" because of the action chains progress bar. Looking at the code I saw that you already didn't show the progress bar for Chain Caller.

To avoid that flash, using the Chain Caller code as an example I added a checkbox to the chain dialog to explicitly enable/disable the progress bar. This eliminated the flash. The changes are compatible with existing chains and the import/export process.

In case you are interested in seeing or incorporating these changes I am attaching the plugin zip and a patch file. I didn't change the version number.

Thanks for the patch. Changes are now in the new version of the plugin.

[chaley]

Originally Posted by capink

Thanks for the patch. Changes are now in the new version of the plugin.

[ownedbycats]

@chaley: Small issue, resizing the dialog now only resizes the top half.

[chaley]

Originally Posted by ownedbycats

@chaley: Small issue, resizing the dialog now only resizes the top half.

@capink: personally I don't care because the odd stretching only happens once you get 9 or more actions in the chain, but if you want to "fix" this then change line 71 in action_dialogs.py from

        icon_groupbox_layout.addStretch(1)

to

        icon_groupbox_layout.addWidget(QLabel(''))

[capink]

Originally Posted by chaley

if you want to "fix" this then change line 71 in action_dialogs.py from

Fixed.