如何使用ruamel.yaml自动添加参考？

Question

I asked a very similar question some time ago, but I am still confused on how to add references on YAML dump.

My goal is to add anchors on default values to minimize redundancy on my dump. So I wrote this:

import collections
import ruamel.yaml as yaml

default = {'a': 1, 'b': 2, 'c': 3}

data = {
    (1,2,3,4): {1: {'a': 10}, 2: {'b': 20}},
    (5,6,7,8): {1: {}, 2: {'a': 100, 'b': 200, 'c': 300}},
}

d = yaml.comments.CommentedMap()
d.update(default)
d.yaml_set_anchor('default')
default = d

for m, a in data.items():
    for k in a.keys():
        u = yaml.comments.CommentedMap()
        u.update(a[k])
        u.add_yaml_merge([(0, default)])
        a[k] = u

data[None] = default

def my_key_repr(self, data):
    if isinstance(data, tuple):
        return self.represent_sequence(u'tag:yaml.org,2002:seq', data, flow_style=True)
    return yaml.representer.SafeRepresenter.represent_key(self, data)

yaml.representer.RoundTripRepresenter.represent_key = my_key_repr

print yaml.dump(data, Dumper=yaml.RoundTripDumper, width=100, allow_unicode=True,
                explicit_start=True)

The expected output is:

---
~: &default
  a: 1
  c: 3
  b: 2
[1, 2, 3, 4]:
  1:
    <<: *default
    a: 10
  2:
    <<: *default
    b: 20
[5, 6, 7, 8]:
  1:
    <<: *default
  2:
    <<: *default
    a: 100
    c: 300
    b: 200

And what I get is:

---
?
: &default
  a: 1
  c: 3
  b: 2
[1, 2, 3, 4]:
  1:
    <<: *default
    a: 10
  2:
    <<: *default
    b: 20
&id001 [5, 6, 7, 8]:
  1:
    <<: *id001
  2:
    <<: *id001
    a: 100
    c: 300
    b: 200

The *id001 comes from nowhere...

Answer 1

There are several problems here:

• If you expect the keys foo and bar to show up in your output, you'll have to specify them in your source somewhere.

• If you expect your YAML document to be implicit (ie not starting with --- ), then you should not specify explicit_start=True

• Keys that are None are dumped as ? not as ~

• If you use update() to fill a CommentedMap() (ie a ordered dictionary) from a dict , you cannot expect the keys to be added in a specific order. If you want the keys in order a , c , b (as you indicate), you'll have to make sure that that is the order in which they are presented to the CommentedMap()

• Since your toplevel data is not ordered, there is no guarantee that in your YAML document the toplevel mapping is the null key ( ? : or ~: ).

The following:

from ruamel import yaml

d = {'a': 1, 'b': 2, 'c': 3}

abc = yaml.comments.CommentedMap()
abc['a'] = 100
abc['c'] = 300
abc['b'] = 200

base = [
    ('foo', {1: {'a': 10}, 2: {'b': 20}}),
    ('bar', {1: {}, 2: abc}),
]

data = yaml.comments.CommentedMap()
default = yaml.comments.CommentedMap()
for m, a in base:
    data[m] = a
for k in sorted(d):
    default[k] = d[k]
default.yaml_set_anchor('default')


for m, a in data.items():
    for k in sorted(a.keys()):
        u = yaml.comments.CommentedMap()
        u.update(a[k])
        u.add_yaml_merge([(0, default)])
        a[k] = u

data.insert(0, None, default)

x = yaml.round_trip_dump(data, width=100).replace('?\n:', '~:')
print(x)

gives exactly the output that you expected.