Home Explore Help
Register Sign In
J1X
/
huginn
Watch 2
Star 0
Fork 0

Browse Source
Merge pull request #909 from cantino/fix_regex_replace

Fix regex_replace

Akinori MUSHA 9 years ago
parent
a76c68bffb b707b683e4
commit
9c9cd011f4
3 changed files with 102 additions and 23 deletions
Show Diff Stats
  1. 64 7
      app/concerns/liquid_interpolatable.rb
  2. 8 0
      config/initializers/liquid.rb
  3. 30 16
      spec/concerns/liquid_interpolatable_spec.rb

+ 64 - 7
app/concerns/liquid_interpolatable.rb
View File 

@@ -201,15 +201,15 @@ module LiquidInterpolatable
201 201 
         'concat(' << subs.join(', ') << ')'
202 202 
       end
203 203 
     end
204 
-
205 
-    def regex_replace(input, regex, replacement = ''.freeze)
206 
-      input.to_s.gsub(Regexp.new(regex), replacement.to_s)
204 
+
205 
+    def regex_replace(input, regex, replacement = nil)
206 
+      input.to_s.gsub(Regexp.new(regex), unescape_replacement(replacement.to_s))
207 207 
     end
208 
-
209 
-    def regex_replace_first(input, regex, replacement = ''.freeze)
210 
-      input.to_s.sub(Regexp.new(regex), replacement.to_s)
208 
+
209 
+    def regex_replace_first(input, regex, replacement = nil)
210 
+      input.to_s.sub(Regexp.new(regex), unescape_replacement(replacement.to_s))
211 211 
     end
212 
-
212 
+
213 213 
     private
214 214
215 215 
     def logger
 
@@ -221,6 +221,63 @@ module LiquidInterpolatable
221 221 
           Logger.new(STDERR)
222 222 
         end
223 223 
     end
224 
+
225 
+    BACKSLASH = "\\".freeze
226 
+
227 
+    UNESCAPE = {
228 
+      "a" => "\a",
229 
+      "b" => "\b",
230 
+      "e" => "\e",
231 
+      "f" => "\f",
232 
+      "n" => "\n",
233 
+      "r" => "\r",
234 
+      "s" => " ",
235 
+      "t" => "\t",
236 
+      "v" => "\v",
237 
+    }
238 
+
239 
+    # Unescape a replacement text for use in the second argument of
240 
+    # gsub/sub.  The following escape sequences are recognized:
241 
+    #
242 
+    # - "\\" (backslash itself)
243 
+    # - "\a" (alert)
244 
+    # - "\b" (backspace)
245 
+    # - "\e" (escape)
246 
+    # - "\f" (form feed)
247 
+    # - "\n" (new line)
248 
+    # - "\r" (carriage return)
249 
+    # - "\s" (space)
250 
+    # - "\t" (horizontal tab)
251 
+    # - "\u{XXXX}" (unicode codepoint)
252 
+    # - "\v" (vertical tab)
253 
+    # - "\xXX" (hexadecimal character)
254 
+    # - "\1".."\9" (numbered capture groups)
255 
+    # - "\+" (last capture group)
256 
+    # - "\k<name>" (named capture group)
257 
+    # - "\&" or "\0" (complete matched text)
258 
+    # - "\`" (string before match)
259 
+    # - "\'" (string after match)
260 
+    #
261 
+    # Octal escape sequences are deliberately unsupported to avoid
262 
+    # conflict with numbered capture groups.  Rather obscure Emacs
263 
+    # style character codes ("\C-x", "\M-\C-x" etc.) are also omitted
264 
+    # from this implementation.
265 
+    def unescape_replacement(s)
266 
+      s.gsub(/\\(?:([\d+&`'\\]|k<\w+>)|u\{([[:xdigit:]]+)\}|x([[:xdigit:]]{2})|(.))/) {
267 
+        if c = $1
268 
+          BACKSLASH + c
269 
+        elsif c = ($2 && [$2.to_i(16)].pack('U')) ||
270 
+                  ($3 && [$3.to_i(16)].pack('C'))
271 
+          if c == BACKSLASH
272 
+            BACKSLASH + c
273 
+          else
274 
+            c
275 
+          end
276 
+        else
277 
+          UNESCAPE[$4] || $4
278 
+        end
279 
+      }
280 
+    end
224 281 
   end
225 282 
   Liquid::Template.register_filter(LiquidInterpolatable::Filters)
226 283

+ 8 - 0
config/initializers/liquid.rb
View File 

@@ -0,0 +1,8 @@
1 
+module Liquid
2 
+  # https://github.com/Shopify/liquid/pull/623
3 
+  remove_const :PartialTemplateParser
4 
+  remove_const :TemplateParser
5 
+
6 
+  PartialTemplateParser       = /#{TagStart}.*?#{TagEnd}|#{VariableStart}(?:(?:[^'"{}]+|#{QuotedString})*?|.*?)#{VariableIncompleteEnd}/m
7 
+  TemplateParser              = /(#{PartialTemplateParser}|#{AnyStartingTag})/m
8 
+end

+ 30 - 16
spec/concerns/liquid_interpolatable_spec.rb
View File 

@@ -177,23 +177,37 @@ describe LiquidInterpolatable::Filters do
177 177 
         expect(@agent.interpolated['long_url']).to eq('http://2many.x/6')
178 178 
       end
179 179 
     end
180 
-
181 
-    describe 'regex replace' do
182 
-      let(:agent) { Agents::InterpolatableAgent.new(name: "test") }
183 
-
184 
-      it 'should replace the first occurrence of a string using regex' do
185 
-        agent.interpolation_context['something'] = 'foobar foobar'
186 
-        agent.options['cleaned'] = '{{ something | regex_replace_first: "\S+bar", "foobaz"  }}'
187 
-        expect(agent.interpolated['cleaned']).to eq('foobaz foobar')
188 
-      end
180 
+  end
189 181
190 
-      it 'should replace the all occurrences of a string using regex' do
191 
-        agent.interpolation_context['something'] = 'foobar foobar'
192 
-        agent.options['cleaned'] = '{{ something | regex_replace: "\S+bar", "foobaz"  }}'
193 
-        expect(agent.interpolated['cleaned']).to eq('foobaz foobaz')
194 
-      end
195 
-
182 
+  describe 'regex_replace_first' do
183 
+    let(:agent) { Agents::InterpolatableAgent.new(name: "test") }
184 
+
185 
+    it 'should replace the first occurrence of a string using regex' do
186 
+      agent.interpolation_context['something'] = 'foobar foobar'
187 
+      agent.options['cleaned'] = '{{ something | regex_replace_first: "\S+bar", "foobaz"  }}'
188 
+      expect(agent.interpolated['cleaned']).to eq('foobaz foobar')
189 
+    end
190 
+
191 
+    it 'should support escaped characters' do
192 
+      agent.interpolation_context['something'] = "foo\\1\n\nfoo\\bar\n\nfoo\\baz"
193 
+      agent.options['test'] = "{{ something | regex_replace_first: '\\\\(\\w{2,})', '\\1\\\\' | regex_replace_first: '\\n+', '\\n'  }}"
194 
+      expect(agent.interpolated['test']).to eq("foo\\1\nfoobar\\\n\nfoo\\baz")
195 
+    end
196 
+  end
197 
+
198 
+  describe 'regex_replace' do
199 
+    let(:agent) { Agents::InterpolatableAgent.new(name: "test") }
200 
+
201 
+    it 'should replace the all occurrences of a string using regex' do
202 
+      agent.interpolation_context['something'] = 'foobar foobar'
203 
+      agent.options['cleaned'] = '{{ something | regex_replace: "\S+bar", "foobaz"  }}'
204 
+      expect(agent.interpolated['cleaned']).to eq('foobaz foobaz')
205 
+    end
206 
+
207 
+    it 'should support escaped characters' do
208 
+      agent.interpolation_context['something'] = "foo\\1\n\nfoo\\bar\n\nfoo\\baz"
209 
+      agent.options['test'] = "{{ something | regex_replace: '\\\\(\\w{2,})', '\\1\\\\' | regex_replace: '\\n+', '\\n'  }}"
210 
+      expect(agent.interpolated['test']).to eq("foo\\1\nfoobar\\\nfoobaz\\")
196 211 
     end
197 
-
198 212 
   end
199 213 
 end