2 files changed, 103 insertions, 0 deletions
diff --git a/lib/ref_extractor.rb b/lib/ref_extractor.rb
new file mode 100644
index 00000000000..d3f02d6ec3e
--- /dev/null
+++ b/lib/ref_extractor.rb
@@ -0,0 +1,64 @@
+# Module providing an extract_ref method for controllers working with Git
+# tree-ish + path params
+#
+# Given a string containing both a Git ref - such as a branch or tag - and a
+# filesystem path joined by forward slashes, attempts to separate the two.
+#
+# Expects a @project instance variable to contain the active project. Used to
+# check the input against a list of valid repository refs.
+#
+# Examples
+#
+#   # No @project available
+#   extract_ref('master')
+#   # => ['', '']
+#
+#   extract_ref('master')
+#   # => ['master', '/']
+#
+#   extract_ref("f4b14494ef6abf3d144c28e4af0c20143383e062/CHANGELOG")
+#   # => ['f4b14494ef6abf3d144c28e4af0c20143383e062', '/CHANGELOG']
+#
+#   extract_ref("v2.0.0/README.md")
+#   # => ['v2.0.0', '/README.md']
+#
+#   extract_ref('issues/1234/app/models/project.rb')
+#   # => ['issues/1234', '/app/models/project.rb']
+#
+#   # Given an invalid branch, we fall back to just splitting on the first slash
+#   extract_ref('non/existent/branch/README.md')
+#   # => ['non', '/existent/branch/README.md']
+#
+# Returns an Array where the first value is the tree-ish and the second is the
+# path
+module RefExtractor
+  def extract_ref(input)
+    pair = ['', '']
+
+    return pair unless @project
+
+    if input.match(/^([[:alnum:]]{40})(.+)/)
+      # If the ref appears to be a SHA, we're done, just split the string
+      pair = $~.captures
+    else
+      # Append a trailing slash if we only get a ref and no file path
+      id = input
+      id += '/' unless id.include?('/')
+
+      # Otherwise, attempt to detect the ref using a list of the project's
+      # branches and tags
+      valid_refs = @project.branches + @project.tags
+      valid_refs.select! { |v| id.start_with?("#{v}/") }
+
+      if valid_refs.length != 1
+        # No exact ref match, so just try our best
+        pair = id.match(/([^\/]+)(.+)/).captures
+      else
+        # Partition the string into the ref and the path, ignoring the empty first value
+        pair = id.partition(valid_refs.first)[1..-1]
+      end
+    end
+
+    pair
+  end
+end
diff --git a/spec/lib/ref_extractor_spec.rb b/spec/lib/ref_extractor_spec.rb
new file mode 100644
index 00000000000..84347977bc4
--- /dev/null
+++ b/spec/lib/ref_extractor_spec.rb
@@ -0,0 +1,39 @@
+require 'spec_helper'
+
+describe RefExtractor do
+  include RefExtractor
+
+  let(:project) { double('project') }
+
+  before do
+    @project = project
+    project.stub(:branches).and_return(['master', 'foo/bar/baz'])
+    project.stub(:tags).and_return(['v1.0.0', 'v2.0.0'])
+  end
+
+  it "extracts a ref without a path" do
+    extract_ref('master').should == ['master', '/']
+  end
+
+  it "extracts a valid branch ref" do
+    extract_ref('foo/bar/baz/CHANGELOG').should == ['foo/bar/baz', '/CHANGELOG']
+  end
+
+  it "extracts a valid tag ref" do
+    extract_ref('v2.0.0/CHANGELOG').should == ['v2.0.0', '/CHANGELOG']
+  end
+
+  it "extracts a valid commit ref" do
+    extract_ref('f4b14494ef6abf3d144c28e4af0c20143383e062/CHANGELOG').should ==
+      ['f4b14494ef6abf3d144c28e4af0c20143383e062', '/CHANGELOG']
+  end
+
+  it "falls back to a primitive split for an invalid ref" do
+    extract_ref('stable/CHANGELOG').should == ['stable', '/CHANGELOG']
+  end
+
+  it "returns an empty pair when no @project is set" do
+    @project = nil
+    extract_ref('master/CHANGELOG').should == ['', '']
+  end
+end