multiline regex pattern that stops at empty line-CodePudding

I am looking for a regex pattern that starts with a specific string on the first line, contains one of a set of strings on the following line, and ends with an empty line. For example, it must start with - hello: world; it must contain a line with fruit: apple or fruit: banana, and it must end with an empty lines. So the pattern would match the first two blocks here, but not the third:

- hello: world
  fruit: apple
  foo: bar
  key: value

- hello: world
  fruit: banana
  message: hi

- hello: world
  fruit: orange
  message: hi

This is what I have so far:

/- hello: world\s*fruit: (apple|banana)/g

What I'm looking for is the rest that will stop at the empty line.

CodePudding user response：

Instead of using a Regex, use a parsed like yq that is build for parsing YAML.

So if the input file looks something like:

myData:
  - hello: world
    fruit: apple
    foo: bar
    key: value
  - hello: world
    fruit: banana
    message: hi
  - hello: world
    fruit: orange
    message: hi

Use an yq filter to filter the output where the following conditions are true

.hello == "world"
.fruit == "apple" or .fruit == "banana":

yq e '.myData | map(select(.hello == "world" and (.fruit == "apple" or .fruit == "banana")))' /path/to/input/file

Output:

- hello: world
  fruit: apple
  foo: bar
  key: value
- hello: world
  fruit: banana
  message: hi

CodePudding user response：

Using \s* matches optional whitespace characters which might also match newline.

The pattern world\s*fruit that you are using could also match worldfruit or world fruit

If there should be a newline in between, and not matching empty lines:

- hello: world\n[^\S\n]*fruit: (?:apple|banana)\b(?:\n[^\S\n]*\S.*)*

Explanation

- hello: world\n Match literally followed by a newline
[^\S\n]*fruit: Match optional spaces followed by fruit:
(?:apple|banana)\b Match either apple or banana followed by a word boundary
(?: Non capture group to repeat as a whole part
- \n[^\S\n]*\S.* Match a newline, optional spaces and a non white space character followed by the rest of the line
)* Close the non capture group and optionally repeat it to match all lines

See a regex101 demo.