Merge branch 'master' of https://github.com/tom-lord/regexp-examples

Author: xiejiangzhi

.ruby-version

@@ -0,0 +1 @@
+2.6.3

.travis.yml

@@ -1,12 +1,9 @@
 language: ruby
 rvm:
-  - 2.0.0
-  - 2.1.10
-  - 2.2.10
-  - 2.3.8
   - 2.4.5
   - 2.5.5
-  - 2.6.2
+  - 2.6.2 # Uses unicode 12.0.0
+  - 2.6.3 # Uses unicode 12.1.0
   - ruby-head
 matrix:
   allow_failures:

README.md

@@ -52,21 +52,18 @@ Obviously, you will get different (random) results if you try these yourself!
 ```
 
 ## Supported ruby versions
-* MRI 2.0.x
-* MRI 2.1.x
-* MRI 2.2.x
-* MRI 2.3.x
-* MRI 2.4.x
-* MRI 2.5.x
-* MRI 2.6.x
-
-MRI ≤ 1.9.3 are not supported. This is primarily because MRI 2.0.0 introduced a new
-regexp engine (`Oniguruma` was replaced by `Onigmo`). Whilst *most* of this gem could
-be made to work with MRI 1.9.x (or even 1.8.x), I feel the changes are too significant
-to implement backwards compatability (especially since [long-term support for MRI
-1.9.3 has now ended](https://www.ruby-lang.org/en/news/2014/01/10/ruby-1-9-3-will-end-on-2015/)).
-
-For example, named properties (e.g. `/\p{Alpha}/`) are illegal syntax on MRI 1.9.3.
+MRI 2.4.0 (oldest non-[EOL](https://www.ruby-lang.org/en/news/2019/03/31/support-of-ruby-2-3-has-ended/) version) --> 2.6.3 (latest stable version)
+
+MRI 2.0.0 --> 2.3.x were supported until version `1.5.0` of this library. Support was dropped primarily
+because of the need to use `RbConfig::CONFIG['UNICODE_VERSION']`, which was added to ruby version `2.4.0`.
+
+MRI versions ≤ 1.9.3 were never supported by this library. This is primarily because MRI 2.0.0 introduced a new
+regexp engine (`Oniguruma` was replaced by `Onigmo` -- For example, named properties like `/\p{Alpha}/`
+are illegal syntax on MRI 1.9.3.). Whilst *most* of this gem could be made to work with MRI 1.9.x
+(or even 1.8.x), I considered the changes too significant to implement backwards compatability
+(especially since [long-term support for MRI 1.9.3 has long
+ended](https://www.ruby-lang.org/en/news/2014/01/10/ruby-1-9-3-will-end-on-2015/)).
+
 
 Other implementations, such as JRuby, could probably work fine -
 but I haven't fully tried/tested it. Pull requests are welcome.

db/unicode_ranges_2.5.pstore renamed to db/unicode_ranges_10.0.0.pstore

@@ -11,8 +11,7 @@ class UnicodeCharRanges
     # These values were generated by: scripts/unicode_lister.rb
     # Note: Only the first 128 results are listed, for performance.
     # Also, some groups seem to have no matches (weird!)
-    # (Don't care about ruby micro version number)
-    STORE_FILENAME = "unicode_ranges_#{RUBY_VERSION[0..2]}.pstore".freeze
+    STORE_FILENAME = "unicode_ranges_#{RbConfig::CONFIG['UNICODE_VERSION']}.pstore".freeze
 
     attr_reader :range_store
 
@@ -30,10 +29,12 @@ def get(key)
 
     private
 
+    # The method is written like this to future-proof it a little,
+    # i.e. the gem won't completely break for a new ruby version release
     def unicode_ranges_file
       db_path = File.join(__dir__, '../../db')
       Dir["#{db_path}/*.pstore"].sort.select do |file|
-        file <= "#{db_path}/unicode_ranges_#{RUBY_VERSION[0..2]}.pstore"
+        file <= "#{db_path}/#{STORE_FILENAME}"
       end.last
     end