Load YAML file into Python preserving aliases

Question

I am making a program that needs to check that certain fields are defined with the correct alias string. For example:

networks:
  base: 
    name: build
    address: &dummyname
     url: 192.168.1.1
     port: 8080 
  first: 
    name: masterA
    address: *dummyname
  second: 
    name: masterB
    address: *dummyname

I need to check whether field address is defined with the alias "*dummyname" in first and second, no matter what the content of the alias is.

When executing load with PyYaml, aliases are always rendered, so I am not able to check that:

data = yaml.safe_load(file_data)

data rendered as python dict:

networks:
  base: 
    name: build
    address: 
      url: 192.168.1.1
      port: 8080 
  first: 
    name: masterA
    address: 
      url: 192.168.1.1
      port: 8080 
  second: 
    name: masterB
    address: 
      url: 192.168.1.1
      port: 8080 

I have seen similar posts for the other way around, dumping python object to YAML without creating aliases/anchors, but I haven't found a solution for this.

How can I access the alias used in the YAML document?

Answer 1

As you indicated, PyYAML does not give you access to the anchor/alias name, it uses it internally to resolve this. When you dump data again, you'll notice that you get a generic anchor (e.g. &id0001).

If you would use ruamel.yaml to round-trip your data in this way, you can see that your actual anchor/alias is preserved:

import sys
import ruamel.yaml

file_in = Path('input.yaml')
    
yaml = ruamel.yaml.YAML()
data = yaml.load(file_in)
yaml.dump(data, sys.stdout)

as this gives:

networks:
  base:
    name: build
    address: &dummyname
      url: 192.168.1.1
      port: 8080
  first:
    name: masterA
    address: *dummyname
  second:
    name: masterB
    address: *dummyname

You can inspect the loaded data structure:

# you get the same object, whether using `first`, `second` or `base`
address = data['networks']['first']['address']
print(address, type(address))
print('\n'.join([k for k in dir(address) if k[0] != '_']))  # skip the build-in attributes

which gives:

ordereddict([('url', '192.168.1.1'), ('port', 8080)]) <class 'ruamel.yaml.comments.CommentedMap'>
add_referent
add_yaml_merge
anchor
ca
clear
copy
copy_attributes
fa
fromkeys
get
insert
items
keys
lc
merge
mlget
move_to_end
non_merged_items
pop
popitem
rya
setdefault
tag
update
update_key_value
values
yaml_add_eol_comment
yaml_anchor
yaml_end_comment_extend
yaml_key_comment_extend
yaml_set_anchor
yaml_set_comment_before_after_key
yaml_set_start_comment
yaml_set_tag
yaml_value_comment_extend

The likely candidate is attribute anchor, this actually an Anchor instance on which the original string can be retrieved:

print(f'anchor: {address.anchor.value}')

giving:

anchor: dummyname

Please note that these kind of internals might change, so pin the version of ruamel.yaml you are using and test before upgrading.