Suppose a string contains a last name, first name, title, and occupation, separated by commas. A regular expression can establish that the fields match the expected pattern; parentheses let you keep just the fields you need. In `/([A-Za-z]+),[A-Za-z]+,(Mrs?\.)/`, the first group captures the last name and the second captures the title. section 11.4.1
Extract two fields
For `m = /([A-Za-z]+),[A-Za-z]+,(Mrs?\.)/.match("Peel,Emma,Mrs.,talented amateur")`, `m[0]` is the entire matched portion, `"Peel,Emma,Mrs."`; `m[1]` is `"Peel"`, and `m[2]` is `"Mrs."`. Ruby also makes those captures available as `$1` and `$2`. A failed `match` returns `nil`, so check for a match before querying it. section 11.4.1 section 11.4.2
Capture numbers follow opening parentheses from left to right, even when groups are nested. With `/((a)((b)c))/` matched against `"abc"`, captures 1 through 4 are `"abc"`, `"a"`, `"bc"`, and `"b"`. `m.captures` returns them as an array starting at index 0, whereas `m[1]` is the first capture; `m[-1]` retrieves the last. section 11.4.3
Choose a grouping form according to what you need afterward:
- Use `(?<first>\w+)` to name a capture; retrieve it with `m[:first]` or inspect `m.named_captures`. An optional group that does not match can have a `nil` value. section 11.4.3
- Use `(?:...)` when parentheses should group and match characters without saving another numbered capture. Unlike a lookaround assertion, this group consumes characters. section 11.5.3
Captures remain useful beyond a single match. `scan` collects repeated matches with captures into arrays of captured fields, or passes each result to a block. In `sub` and `gsub`, a replacement string can refer to captures as `\1` and `\2`; a block can instead use `$1` and `$2`. Do not interpolate `$1` into a replacement string expecting each substitution to refresh it: that string may use a capture from an earlier match. Ruby’s numbered capture variables are thread-local. section 11.7.1 section 11.7.3 section 14.5.4
