mekkyz / ymlio Goto Github PK

I suggest to add some documentation to the Readme.md file to make uses know what this is about.

It might be a good to include version as a feature in the tool i.e. implement ymlio -v command.

One can then also always include a comment: # written using ymlio vX.x in all the files that are written using the tool.

At present, using the split command meaningfully requires that keys in the parent YAML file have an extension i.e.

my_yaml_file.yml:
  key1: dict
  key2: str

produces a YAML file called my_yaml_file.yml. This then fails if the file is supplied from an external resource, e.g. is a valid docker-compose file.

This issue suggests a specification that changes this requirement and makes the tool more flexible. The specification is as follows:

1. Extension .yml is automatically appended to the key that is being exported as a file. So, the following will result in files my_yaml_file1.yml and my_yaml_file2.yml.

my_yaml_file1:
  key1: dict
my_yaml_file2:
  key2: str

2. If it is required that a given extension is used, the following format for key can be used. Following would create four files, namely my_text_file1.txt, my_file_with_no_extension, .dotfile and my_yaml_file3.yml.

./my_text_file1.txt:
./my_file_with_no_extension:
my_yaml_file3:
  key3: float
./.dotfile:

3. The same can be extended to encode folders by using a trailing slash. The following should create two folders my_folder and my_second_folder, each containing a subfolder called subfolder.

./my_folder/:
./my_folder/subfolder/:
./my_second_folder/subfolder/:

4. A combination of all of the above is also possible using the following notation:

./my_third_folder/my_yaml_file4:
  key4: int
./my_fourth_folder/./my_file_without_extension

A suggestion would be to implement 1, 2, and 3 first with the possibility of implementing 4 if required.

test2.yml:
  int_key2: 2
  string_key2: value2
test3.yaml:
  int_key: 1
  string_key: value

Running ymlio split <file.something> fails. If test3.yaml is replaced with test3.yml, then it works.

When using ymlio split and no filename is passed or the passed filename equals -, then the tool should read the input file from stdin.

At present, the tool expects that, when combining YML files, the name of the output file will be the last argument in the input string i.e. the command will be as follows:
ymlio combine A.yml B.yml AB.yml or
ymlio combine A.yml B.yml C.yml ABC.ymal
when combining files A.yml, B.yml, ... to produce AB.yml or AB....yml.

This is contrary to standard practice for CLI tools that expect the output file to be specified by using the -o/--output flag or by simply redirecting the output from stdout to a file using > or >>. This issue suggests that the tool be modified to follow the standard practice of redirecting the stdout in the short term. The implementation of a flag for output can then be done at a later stage.

The compiled main binary should be excluded from the repo, as it can easily be generated and just adds clutter.

Cobra has a feature to automatically generate documentation for you. Please have a look if you can use it.

Consider the two files:

production:
  int_key: 1
  string_key: value

development:
  int_key2: 2
  string_key2: value2

running ymlio combine file1.yml file2.yml output.yml results in this:

/tmp/tmp.RJycLxVELk.yml:
  production:
  int_key: 1
  string_key: value

/tmp/tmp.ixXOqTjG5U.yml:
  development:
  int_key2: 2
  string_key2: value2

There should be an indentation.

Use os.CreateTemp and make the operation on it then delete it to avoid editing the original file. A to-be-split-file should not be touched

A lot of unnecessary empty lines are present in the source files, i.e.:

Please add a license to the code. We can discuss this if you like. A good resource is https://choosealicense.com/.

The yq binary is necessary for this one feature only:

Can we somehow do the conversion on our own without having to introduce this dependency?

see header, goal would be to have a better separation of source files and other files.

If the output file of a combine operation exists, the output file should be overwritten instead of the data being appended.