Is there a way to convert juniper "json" or "xml" config to "set" or "show" config?

Question

We use juniper hardware with junos version 15. In this version we can export our config as "json" or "xml" which we want to use to edit it with our automation tooling. Importing however is only possible in "set" or "show" format.

Is there a tool to convert "json" or "xml" format to "set" or "show" format? I can only find converters between "show" and "set".

We can't upgrade to version 16 where the import of "json" would be possible.

Answer 1

Here's a script I made at work, throw it in your bin and you can it via providing a filename or piping output. This assumes linux or mac so the os.isatty function works, but the logic can work anywhere:

usage demo:

person@laptop ~ > head router.cfg
## Last commit: 2021-04-20 21:21:39 UTC by vit
version 15.1X12.2;
groups {
    BACKBONE-PORT {
        interfaces {
            <*> {
                mtu 9216;
                unit <*> {
                    family inet {
                        mtu 9150;
person@laptop ~ > convert.py router.cfg | head
set groups BACKBONE-PORT interfaces <*> mtu 9216
set groups BACKBONE-PORT interfaces <*> unit <*> family inet mtu 9150
set groups BACKBONE-PORT interfaces <*> unit <*> family inet6 mtu 9150
set groups BACKBONE-PORT interfaces <*> unit <*> family mpls maximum-labels 5
<... output removed... >

convert.py:

#!/usr/bin/env python3
# Class that attempts to parse out Juniper JSON into set format
#   I think it works? still testing
#
#   TODO: 
#      accumulate annotations and provide them as commands at the end. Will be weird as annotations have to be done after an edit command
from argparse import ArgumentParser, RawTextHelpFormatter
import sys, os, re

class TokenStack():
    def __init__(self):
        self._tokens = []

    def push(self, token):
        self._tokens.append(token)

    def pop(self):
        if not self._tokens:
            return None
        item = self._tokens[-1]
        self._tokens = self._tokens[:-1]
        return item

    def peek(self):
        if not self._tokens:
            return None
        return self._tokens[-1]

    def __str__(self):
        return " ".join(self._tokens)

    def __repr__(self):
        return " ".join(self._tokens)

def main():
    # get file
    a = ArgumentParser(prog="convert_jpr_json",
            description="This program takes in Juniper style JSON (blah { format) and prints it in a copy pastable display set format",
            epilog=f"Either supply with a filename or pipe config contents into this program and it'll print out the display set view.\nEx:\n{B}convert_jpr_json <FILENAME>\ncat <FILENAME> | convert_jpr_json{WHITE}",
            formatter_class=RawTextHelpFormatter)
    a.add_argument('file', help="juniper config in JSON format", nargs="?")
    args = a.parse_args()
    if not args.file and os.isatty(0):
        a.print_help()
        die("Please supply filename or provide piped input")
    file_contents = None
    if args.file:
        try:
            file_contents = open(args.file, "r").readlines()
        except IOError as e:
            die(f"Issue opening file {args.file}: {e}")
            print(output_text)
    else:
        file_contents = sys.stdin.readlines()

    tokens = TokenStack()
    in_comment = False
    new_config = []

    for line_num, line in enumerate(file_contents):
        if line.startswith("version ") or len(line) == 0:
            continue
        token = re.sub(r"^(.+?)#+[^\"]*$", r"\1", line.strip())
        token = token.strip()
        if (any(token.startswith(_) for _ in ["!", "#"])):
            # annotations currently not supported
            continue
    
        if token.startswith("/*"):
            # we're in a comment now until the next token (this will break if a multiline comment with # style { happens, but hopefully no-one is that dumb
            in_comment = True
            continue
    
        if "inactive: " in token:
            token = token.split("inactive: ")[1]
            new_config.append(f"deactivate {tokens} {token}")
        if token[-1] == "{":
            in_comment = False
            tokens.push(token.strip("{ "))
        elif token[-1] == "}":
            if not tokens.pop():
                die("Invalid json supplied: unmatched closing } encountered on line " + f"{line_num}")
        elif token[-1] == ";":
            new_config.append(f"set {tokens} {token[:-1]}")
    if tokens.peek():
        print(tokens)
        die("Unbalanced JSON: expected closing }, but encountered EOF")
    print("\n".join(new_config))

def die(msg): print(f"\n{B}{RED}FATAL ERROR{WHITE}: {msg}"); exit(1)
RED = "\033[31m"; GREEN = "\033[32m"; YELLOW = "\033[33m"; B = "\033[1m"; WHITE = "\033[0m"
if __name__ == "__main__": main()

Answer 2

You can load XML configuration using edit-config RPC or load-configuration RPC. For more details:

https://www.juniper.net/documentation/en_US/junos/topics/reference/tag-summary/netconf-edit-config.html

https://www.juniper.net/documentation/en_US/junos/topics/reference/tag-summary/junos-xml-protocol-load-configuration.html

Answer 3

XML content can be loaded via an "op" script by placing the content inside a call to junos:load-configuration() template defined in "junos.xsl". Something like the following:

version 1.1;

ns jcs = "http://xml.juniper.net/junos/commit-scripts/1.0";

import "../import/junos.xsl";

var $arguments = {
    <argument> {
        <name> "file";
        <description> "Filename of XML content to load";
    }
    <argument> {
        <name> "action";
        <description> "Mode for the load (override, replace, merge)";
    }
}

param $file;
param $action = "replace";

match / {
    <op-script-results> {
        var $configuration = slax:document($file);
        var $connection = jcs:open();
        call jcs:load-configuration($connection, $configuration, $action);
    }
}

Thanks, Phil