Skip to content

StackAdapt/ds-pyhocon

 
 

Repository files navigation

pyhocon

pypi Supported Python Versions Build Status Downloads Codacy Badge License Coverage Status Requirements Status

HOCON parser for Python

Specs

https://github.com/typesafehub/config/blob/master/HOCON.md

Installation

It is available on StackAdapt's artifactory so you can install it as follows:

$ pip install stackadapt.ds-pyhocon

Usage

The parsed config can be seen as a nested dictionary (with types automatically inferred) where values can be accessed using normal dictionary getter (e.g., conf['a']['b'] or using paths like conf['a.b']) or via the methods get, get_int (throws an exception if it is not an int), get_string, get_list, get_float, get_bool, get_config.

from ds_pyhocon import ConfigFactory

conf = ConfigFactory.parse_file('samples/database.conf')
host = conf.get_string('databases.mysql.host')
same_host = conf.get('databases.mysql.host')
same_host = conf['databases.mysql.host']
same_host = conf['databases']['mysql.host']
port = conf['databases.mysql.port']
username = conf['databases']['mysql']['username']
password = conf.get_config('databases')['mysql.password']
password = conf.get('databases.mysql.password', 'default_password')  # use default value if key not found

Example of HOCON file

//
// You can use # or // for comments
//
{
  databases {
    # MySQL
    active = true
    enable_logging = false
    resolver = null
    # you can use substitution with unquoted strings. If it it not found in the document, it defaults to environment variables
    home_dir = ${HOME} # you can substitute with environment variables
    "mysql" = {
      host = "abc.com" # change it
      port = 3306 # default
      username: scott // can use : or =
      password = tiger, // can optionally use a comma
      // number of retries
      retries = 3
    }
  }

  // multi line support
  motd = """
            Hello "man"!
            How is it going?
         """
  // this will be appended to the databases dictionary above
  databases.ips = [
    192.168.0.1 // use white space or comma as separator
    "192.168.0.2" // optional quotes
    192.168.0.3, # can have a trailing , which is ok
  ]

  # you can use substitution with unquoted strings
  retries_msg = You have ${databases.mysql.retries} retries

  # retries message will be overriden if environment variable CUSTOM_MSG is set
  retries_msg = ${?CUSTOM_MSG}
}

// dict merge
data-center-generic = { cluster-size = 6 }
data-center-east = ${data-center-generic} { name = "east" }

// list merge
default-jvm-opts = [-XX:+UseParNewGC]
large-jvm-opts = ${default-jvm-opts} [-Xm16g]

Conversion tool

We provide a conversion tool to convert from HOCON to the JSON, .properties and YAML formats.

usage: tool.py [-h] [-i INPUT] [-o OUTPUT] [-f FORMAT] [-n INDENT] [-v]

pyhocon tool

optional arguments:
  -h, --help                 show this help message and exit
  -i INPUT, --input INPUT    input file
  -o OUTPUT, --output OUTPUT output file
  -c, --compact              compact format
  -f FORMAT, --format FORMAT output format: json, properties, yaml or hocon
  -n INDENT, --indent INDENT indentation step (default is 2)
  -v, --verbosity            increase output verbosity

If -i is omitted, the tool will read from the standard input. If -o is omitted, the result will be written to the standard output. If -c is used, HOCON will use a compact representation for nested dictionaries of one element (e.g., a.b.c = 1)

JSON

$ cat samples/database.conf | pyhocon -f json

{
  "databases": {
    "active": true,
    "enable_logging": false,
    "resolver": null,
    "home_dir": "/Users/darthbear",
    "mysql": {
      "host": "abc.com",
      "port": 3306,
      "username": "scott",
      "password": "tiger",
      "retries": 3
    },
    "ips": [
      "192.168.0.1",
      "192.168.0.2",
      "192.168.0.3"
    ]
  },
  "motd": "\n            Hello \"man\"!\n            How is it going?\n         ",
  "retries_msg": "You have 3 retries"
}

.properties

$ cat samples/database.conf | pyhocon -f properties

databases.active = true
databases.enable_logging = false
databases.home_dir = /Users/darthbear
databases.mysql.host = abc.com
databases.mysql.port = 3306
databases.mysql.username = scott
databases.mysql.password = tiger
databases.mysql.retries = 3
databases.ips.0 = 192.168.0.1
databases.ips.1 = 192.168.0.2
databases.ips.2 = 192.168.0.3
motd = \
            Hello "man"\!\
            How is it going?\

retries_msg = You have 3 retries

YAML

$ cat samples/database.conf | pyhocon -f yaml

databases:
  active: true
  enable_logging: false
  resolver: None
  home_dir: /Users/darthbear
  mysql:
    host: abc.com
    port: 3306
    username: scott
    password: tiger
    retries: 3
  ips:
    - 192.168.0.1
    - 192.168.0.2
    - 192.168.0.3
motd: |

            Hello "man"!
            How is it going?

retries_msg: You have 3 retries

Includes

We support the include semantics using one of the followings:

include "test.conf"
include "http://abc.com/test.conf"
include "https://abc.com/test.conf"
include "file://abc.com/test.conf"
include file("test.conf")
include required(file("test.conf"))
include url("http://abc.com/test.conf")
include url("https://abc.com/test.conf")
include url("file://abc.com/test.conf")
include package("package:assets/test.conf")

When one uses a relative path (e.g., test.conf), we use the same directory as the file that includes the new file as a base directory. If the standard input is used, we use the current directory as a base directory.

For example if we have the following files:

cat.conf:

{
  garfield: {
    say: meow
  }
}

dog.conf:

{
  mutt: {
    say: woof
    hates: {
      garfield: {
        notes: I don't like him
        say: meeeeeeeooooowww
      }
      include "cat.conf"
    }
  }
}

animals.conf:

{
  cat : {
    include "cat.conf"
  }

  dog: {
    include "dog.conf"
  }
}

Then evaluating animals.conf will result in the followings:

$ pyhocon -i samples/animals.conf
{
  "cat": {
    "garfield": {
      "say": "meow"
    }
  },
  "dog": {
    "mutt": {
      "say": "woof",
      "hates": {
        "garfield": {
          "notes": "I don't like him",
          "say": "meow"
        }
      }
    }
  }
}

As you can see, the attributes in cat.conf were merged to the ones in dog.conf. Note that the attribute "say" in dog.conf got overwritten by the one in cat.conf.

Duration/Period support

Difference from HOCON spec

  • nanoseconds supported only in the sense that it is converted to microseconds with lowered accuracy (divided by 1000 and rounded to int).
  • m suffix only applies to minutes. Spec specifies that m can also be applied to months, but that would cause a conflict in syntax.
  • months and years only available if dateutils is installed (relativedelta is used instead of timedelta).

Misc

with_fallback

  • with_fallback: Usage: config3 = config1.with_fallback(config2) or config3 = config1.with_fallback('samples/aws.conf')

from_dict

d = OrderedDict()
d['banana'] = 3
d['apple'] = 4
d['pear'] = 1
d['orange'] = 2
config = ConfigFactory.from_dict(d)
assert config == d

TODO

Items Status
Comments
Omit root braces
Key-value separator
Commas
Whitespace
Duplicate keys and object merging
Unquoted strings
Multi-line strings
String value concatenation
Array concatenation
Object concatenation
Arrays without commas
Path expressions
Paths as keys
Substitutions
Self-referential substitutions
The += separator
Includes
Include semantics: merging
Include semantics: substitution
Include semantics: missing files
Include semantics: file formats and extensions
Include semantics: locating resources
Include semantics: preventing cycles
Conversion of numerically-index objects to arrays
API Recommendations Status
Conversion of numerically-index objects to arrays
Automatic type conversions
Units format
Duration format
Size in bytes format
Config object merging and file merging
Java properties mapping

Contributors

Thanks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%