pyaml 16.11.3

What this module does and why

YAML is generally nice and easy format to read if it was written by humans.

PyYAML can a do fairly decent job of making stuff readable, and the best combination of parameters for such output that I’ve seen so far is probably this one:

>>> m = [123, 45.67, {1: None, 2: False}, u'some text']
>>> data = dict(a=u'asldnsa\nasldpáknsa\n', b=u'whatever text', ma=m, mb=m)
>>> yaml.safe_dump(data, sys.stdout, allow_unicode=True, default_flow_style=False)
a: 'asldnsa

  asldpáknsa

  '
b: whatever text
ma: &id001
- 123
- 45.67
- 1: null
  2: false
- some text
mb: *id001

pyaml tries to improve on that a bit, with the following tweaks:

• Most human-friendly representation options in PyYAML (that I know of) get picked as defaults.

• Does not dump “null” values, if possible, replacing these with just empty strings, which have the same meaning but reduce visual clutter and are easier to edit.

• Dicts, sets, OrderedDicts, defaultdicts, namedtuples, etc are representable and get sorted on output (OrderedDicts and namedtuples keep their ordering), so that output would be as diff-friendly as possible, and not arbitrarily depend on python internals.

  It appears that at least recent PyYAML versions also do such sorting for python dicts.

• List items get indented, as they should be.

• bytestrings that can’t be auto-converted to unicode raise error, as yaml has no “binary bytes” (i.e. unix strings) type.

• Attempt is made to pick more readable string representation styles, depending on the value, e.g.:

  >>> yaml.safe_dump(cert, sys.stdout)
  cert: '-----BEGIN CERTIFICATE-----
  
    MIIH3jCCBcagAwIBAgIJAJi7AjQ4Z87OMA0GCSqGSIb3DQEBCwUAMIHBMRcwFQYD
  
    VQQKFA52YWxlcm9uLm5vX2lzcDEeMBwGA1UECxMVQ2VydGlmaWNhdGUgQXV0aG9y
  ...
  
  >>> pyaml.p(cert):
  cert: |
    -----BEGIN CERTIFICATE-----
    MIIH3jCCBcagAwIBAgIJAJi7AjQ4Z87OMA0GCSqGSIb3DQEBCwUAMIHBMRcwFQYD
    VQQKFA52YWxlcm9uLm5vX2lzcDEeMBwGA1UECxMVQ2VydGlmaWNhdGUgQXV0aG9y
  ...

• “force_embed” option to avoid having &id stuff scattered all over the output (which might be beneficial in some cases, hence the option).

• “&id” anchors, if used, get labels from the keys they get attached to, not just use meaningless enumerators.

• “string_val_style” option to only apply to strings that are values, not keys, i.e:

  >>> pyaml.p(data, string_val_style='"')
  key: "value\nasldpáknsa\n"
  >>> yaml.safe_dump(data, sys.stdout, allow_unicode=True, default_style='"')
  "key": "value\nasldpáknsa\n"

• Has an option to add vertical spacing (empty lines) between keys on different depths, to make output much more seekable.

Result for the (rather meaningless) example above (without any additional tweaks):

>>> pyaml.p(data)
a: |
  asldnsa
  asldpáknsa
b: 'whatever text'
ma: &ma
  - 123
  - 45.67
  - 1:
    2: false
  - 'some text'
mb: *ma

Extended example:

>>> pyaml.dump(conf, sys.stdout, vspacing=[2, 1]):
destination:

  encoding:
    xz:
      enabled: true
      min_size: 5120
      options:
      path_filter:
        - \.(gz|bz2|t[gb]z2?|xz|lzma|7z|zip|rar)$
        - \.(rpm|deb|iso)$
        - \.(jpe?g|gif|png|mov|avi|ogg|mkv|webm|mp[34g]|flv|flac|ape|pdf|djvu)$
        - \.(sqlite3?|fossil|fsl)$
        - \.git/objects/[0-9a-f]+/[0-9a-f]+$

  result:
    append_to_file:
    append_to_lafs_dir:
    print_to_stdout: true

  url: http://localhost:3456/uri


filter:
  - /(CVS|RCS|SCCS|_darcs|\{arch\})/$
  - /\.(git|hg|bzr|svn|cvs)(/|ignore|attributes|tags)?$
  - /=(RELEASE-ID|meta-update|update)$


http:

  ca_certs_files: /etc/ssl/certs/ca-certificates.crt

  debug_requests: false

  request_pool_options:
    cachedConnectionTimeout: 600
    maxPersistentPerHost: 10
    retryAutomatically: true


logging:

  formatters:
    basic:
      datefmt: '%Y-%m-%d %H:%M:%S'
      format: '%(asctime)s :: %(name)s :: %(levelname)s: %(message)s'

  handlers:
    console:
      class: logging.StreamHandler
      formatter: basic
      level: custom
      stream: ext://sys.stderr

  loggers:
    twisted:
      handlers:
        - console
      level: 0

  root:
    handlers:
      - console
    level: custom

Note that unless there are many moderately wide and deep trees of data, which are expected to be read and edited by people, it might be preferrable to directly use PyYAML regardless, as it won’t introduce another (rather pointless in that case) dependency and a point of failure.

Obligatory warning

Prime concern for this module is to chew simple types/values gracefully, and internally there are some nasty hacks (that I’m not too proud of) used to do that, which may not work with more complex serialization cases, possibly even producing non-deserializable (but still easily fixable) output.

Again, prime goal is not to serialize, say, gigabytes of complex document-storage db contents, but rather individual simple human-parseable documents, please keep that in mind (and of course, patches for hacks are welcome!).

Some Tricks

• Pretty-print any yaml or json (yaml subset) file from the shell:

  python -m pyaml /path/to/some/file.yaml
  curl -s https://status.github.com/api.json | python -m pyaml

• Easier “debug printf” for more complex data (all funcs below are aliases to same thing):

  pyaml.p(stuff)
  pyaml.pprint(my_data)
  pyaml.pprint('----- HOW DOES THAT BREAKS!?!?', input_data, some_var, more_stuff)
  pyaml.print(data, file=sys.stderr) # needs "from __future__ import print_function"

• Force all string values to a certain style (see info on these in PyYAML docs):

  pyaml.dump(many_weird_strings, string_val_style='|')
  pyaml.dump(multiline_words, string_val_style='>')
  pyaml.dump(no_want_quotes, string_val_style='plain')

  Using pyaml.add_representer() (note *p*yaml) as suggested in this SO thread (or github-issue-7) should also work.

• Control indent and width of the results:

  pyaml.dump(wide_and_deep, indent=4, width=120)

  These are actually keywords for PyYAML Emitter (passed to it from Dumper), see more info on these in PyYAML docs.

Installation

It’s a regular package for Python (3.x or 2.x).

Module uses PyYAML for processing of the actual YAML files and should pull it in as a dependency.

Dependency on unidecode module is optional and should only be necessary if same-id objects or recursion is used within serialized data.

Be sure to use python3/python2, pip3/pip2, easy_install-… binaries below, based on which python version you want to install the module for, if you have several on the system (as is norm these days for py2-py3 transition).

Using pip is the best way:

% pip install pyaml

(add –user option to install into $HOME for current user only)

Or, if you don’t have “pip” command:

% python -m ensurepip
% python -m pip install --upgrade pip
% python -m pip install pyaml

(same suggestion wrt “install –user” as above)

On a very old systems, one of these might work:

% curl https://raw.github.com/pypa/pip/master/contrib/get-pip.py | python
% pip install pyaml

% easy_install pyaml

% git clone --depth=1 https://github.com/mk-fg/pretty-yaml
% cd pretty-yaml
% python setup.py install

(all of install-commands here also have –user option, see also pip docs “installing” section)

Current-git version can be installed like this:

% pip install 'git+https://github.com/mk-fg/pretty-yaml#egg=pyaml'

Note that to install stuff to system-wide PATH and site-packages (without –user), elevated privileges (i.e. root and su/sudo) are often required.

Use “…install –user”, ~/.pydistutils.cfg or virtualenv to do unprivileged installs into custom paths.

More info on python packaging can be found at packaging.python.org.