CODEWITHSUNDEEP
regexregex-group
Alex
Alex

Reputation: 2402

How to adapt this regex to be agnostic of the order of the key-value pairs

I have constructed this regex to pull the version number of a package from an automatically generated lockfile:

\[\[package\]\]\s(?:[a-z-]+ = \"?.*\"?\s)*name = \"NAME\"\s(?:[a-z-]+ = \"?.*\"?\s)*version = \"([ab0-9.]+)\"

The subject file looks like this (shortened, there are many such blocks):

[[package]]
category = "main"
description = "Some. , - description"
name = "django"
optional = false
python-versions = ">=3.5"
version = "2.2.17"

[package.dependencies]
django = ">=1.8.0"
redis = ">=3"
rq = ">=0.13,<1.0"

[package.extras]
Sentry = ["raven (>=6.1.0)"]
testing = ["mock (>=2.0.0)"]

This seems to work well. The problem is that sometimes, the ordering of the two important keys might be different, e.g.:

[[package]]
category = "main"
description = "Some. , - description"
version = "2.2.17"
optional = false
name = "django"
python-versions = ">=3.5"

Which will cause this regex to fail.

I would like to find a block (starting with [[package]] and ending with a newline, that contains the string ^name = \"NAME\", and within that block, do find the value of the version key, regardless of what order they are in.

I have done some reading around lookaheads/lookbehinds, but I could not apply it to this.

Upvotes: 1

Views: 67

Answers (1)

anubhava
anubhava

Reputation: 785156

You can use a lookahead assertion to match the name of the package and capture version in main regex:

\[\[package]]\s(?=(?:[a-z-]+ = "?[^"]*"?\s)*?name = "django"\s)(?:[a-z-]+ = "?[^"]*"?\s)*?version = "([ab0-9.]+)"

RegEx Demo

RegEx Details:

  • \[\[package]]\s: Match [[package]] followed by a whitespace
  • (?=(?:[a-z-]+ = "?[^"]*"?\s)*?name = "django"\s): Positive lookahead to assert that we have a property name = "django" somewhere within this package
  • (?:[a-z-]+ = "?[^"]*"?\s)*?: Match 0 or more property lines
  • version = "([ab0-9.]+)": Match version property and capture version number in capture group #1

Upvotes: 2

Related Questions