Basically, I am trying to use regex to match repeating patterns containing special characters in Ruby. I've been able to do this if I am given the number of times a pattern repeats but not dynamically. An example string I am looking to match is:
Draw a square that is {{coords.width}} pixels wide by {{coords.height}} pixels tall.
This can be easily done by using
arr = value.scan(/\{\{(\w+?\.\w+?)\}\}/).flatten
arr looks like this after I run this
["coords.width", "coords.height"]
But how do I write a regex which can match in case this pattern follows arbitrarily, for example
Draw a square that is {{shape.rectangle.coords.width}} pixels wide by {{shape.rectangle.coords.height}} pixels tall.
while also matching in case of the following(no ".")
Draw a square that is {{width}} pixels wide by {{height}} pixels tall.
You can match the regular expression
r = /(?<=\{\{)[a-z]+(?:\.[a-z]+)*(?=\}\})/
Rubular demo / PCRE demo at regex 101.com
I've included the PCRE demo because regex101.com provides a detailed explanation of each element of the regex (hover the cursor).
For example,
str = "Draw a square {{coords.width}} wide by {{coords.height}} " +
"tall by {{coords deep}} deep"
str.scan(r)
#=> ["coords.width", "coords.height"]
Notice that "coords deep" does not match because it does not have (what I have assumed is) a valid form. Notice also that I did not have to flatten the return value from scan because the regex has no capture groups.
We can write the regular expression in free-spacing mode to make it self-documenting.
/
(?<= # begin a positive lookbehind
\{\{ # match 1 or more lower case letters
) # end the positive lookbehind
[a-z]+ # match 1 or more lower case letters
(?: # begin a non-capture group
\. # match a period
[a-z]+ # match 1 or more lower case letters
) # end the non-capture group
* # execute the non-capture group zero or more times
(?= # begin a positive lookahead
\}\} # match '}}'
) # end positive lookahead
/x # free-spacing regex definition mode