在YAML格式的文件test.yaml中，某些MAC地址将被判断为数字

Question

the MAC address of [80:41:26:53:24:11] has been judged as number 62745168251 but [84:19:14:15:86:58] has been judged as string

Answer 1

You should always use quotes around string literals that start with a digit and consist of a mix of digits and non-digit characters to avoid confusion from the YAML parser:

MAC1: "80:41:26:53:24:11"

Otherwise, if your string literal consists of a series of one or two-digit numbers delimited by colons and all numbers but the first are between 0 and 59 , it would be interpreted as a sexagesimal number and would be converted to the equivalent number of seconds. For example, your MAC-address 80:41:26:53:24:11 would be treated as a sexagesimal number and converted to:

(((((80*60)+41)*60+26)*60+53)*60+24)*60+11

which is equal to 62745168251 .

Since your other MAC-address, 84:19:14:15:86:58 , has a number after the first number that is over 59 ( 86 that is), it is not considered as a sexagesimal number and is therefore treated literally as a string.

Relevant source code can be found in the int resolver and the int constructor .

Answer 2

You don't explicitly state it, but from the use of yaml.load() and getting that result, my guess is that you are using PyYAML. You should not be using PyYAML yaml.load() as this is potentially unsafe and completely unnecessary . If you need to stick with PyYAML, use yaml.safe_load() instead.

In the latest version of YAML]( (1.2 released in 2009) the core schema of what counts as an int has changed from the previous version (1.1 from 2005). The most notable differences are that 077 is no longer an octal in YAML 1.2 (use 0o77 ) and that sexagesimals , in the YAML 1.1 documentation indicated by this validation rule:

|[-+]?[1-9][0-9_]*(:[0-5]?[0-9])+ # (base 60)

are no longer supported in YAML 1.2 (not explicitly, and the YAML 1.2 specification points to the 1.1 language independent type, but the example has been removed from the document).

In the YAML 1.1 documentation the explicit examples for sexagesimal ints are 190:20:30 , resp. 3:25:45 .¹ This seems to be an integer followed by two sets of (two digit) number and all seperated by two colons ( : ), that can be used to indicate eg hours, minutes and seconds. The rule (regex) however has no limit on only two of these sets as the + at the end of (:[0-5]?[0-9])+ means at least one time and has no restriction. In a sexagesimal, between the colons, you can either have a one digit number from 0-9 or a two digit number where the first is restricted to 0-5 .

PyYAML is documented to feature a YAML 1.1 parser (so it has not been updated although YAML 1.2 came out in 2009). It will read your MAC addresses as sexagesimal, if they match the rule, which roughly said happens when the number between the colons or after the last colon is less than 60 , as happens for both the values for key MAC1 and MAC2 (the values are the same) in the value for NETWORK2 . The other MAC addresses have :86: in them and are not sexagesimals.

If you would dump these MAC addresses from a string using PyYAML, it will quote them for you when necessary:

import sys
import yaml

d = dict(mac1='84:19:14:15:86:58', mac2='80:41:26:53:24:11')
yaml.safe_dump(d, sys.stdout)

which dumps:

{mac1: '84:19:14:15:86:58', mac2: '80:41:26:53:24:11'}

Please note that I did not do print(safe_dump(d)) on purpose, it is slower (as it has to first create and fill an internal buffer, retrieve and stream out the contents). It also adds an extra empty newline at the end of the output.

You can of course quote all (or all necessary) MAC addresses, but the easier way is upgrade to a YAML 1.2 parser like the ruamel.yaml package (disclaimer: I am the author of that package). Which, assuming your input is in network.yaml , can be done using:

import sys
import ruamel.yaml as yaml

with open('network.yaml') as fp:
    data = yaml.safe_load(fp)
print(data)

which gives:

{'NETWORK1': {'BOND_MODE': 4, 'BOND_NAME': 'bond0', 'MAC1': '84:19:14:15:86:58', 'MAC2': '84:19:14:15:86:59'}, 'NETWORK2': {'BOND_MODE': 1, 'BOND_NAME': 'bond0', 'BLAN_ID': 820, 'MAC1': '80:41:26:53:24:11', 'MAC2': '80:41:26:53:24:11'}}

In round-trip mode, using the newer API, ruamel.yaml will not insert quotes when dumping, as they are unnecessary for YAML 1.2:

import sys
import ruamel.yaml
from pathlib import Path

network_file = Path('network.yaml')

yaml = ruamel.yaml.YAML()
# yaml.indent(mapping=4, sequence=4, offset=2)
# yaml.preserve_quotes = True
data = yaml.load(network_file)
yaml.dump(data, sys.stdout)

for result:

NETWORK1:
  BOND_MODE: 4
  BOND_NAME: bond0
  MAC1: 84:19:14:15:86:58
  MAC2: 84:19:14:15:86:59

NETWORK2:
  BOND_MODE: 1
  BOND_NAME: bond0
  BLAN_ID: 820
  MAC1: 80:41:26:53:24:11
  MAC2: 80:41:26:53:24:11

¹ _{You can also have a sexagesimal float: 20:30.15}

Answer 3

Not sure what's going on with your syntax highlighting, but, maybe there's a clue there. Does removing some of the whitepaces before the MAC addresses make any difference?

If you want any of your values in yaml to be explicitly evaluated as strings, wrap the values in double or single quotes.