From 2707d1db67ed782a0e523679fa960d65a20c6a78 Mon Sep 17 00:00:00 2001
From: Andreas Bolka <a@bolka.at>
Date: Wed, 29 Jun 2011 01:37:30 +0200
Subject: [PATCH] Guess if .r is R or REBOL

.r is the default extension used widely for both, R and REBOL. This
patch attempts to guess if .r is a REBOL file by looking for the
following very common REBOL code fragments:

- `REBOL`
- `: func [`
- `make object! [`
- `context [`

If any of those is found in a .r file, it is identified as a REBOL file.
Otherwise the language for a .r file is R.

Signed-off-by: Andreas Bolka <a@bolka.at>
---
 lib/linguist/blob_helper.rb | 16 ++++++++++++++++
 test/fixtures/hello-r.R     |  4 ++++
 test/fixtures/hello-rebol.r |  5 +++++
 test/test_blob.rb           |  4 ++++
 4 files changed, 29 insertions(+)
 create mode 100644 test/fixtures/hello-r.R
 create mode 100644 test/fixtures/hello-rebol.r

diff --git a/lib/linguist/blob_helper.rb b/lib/linguist/blob_helper.rb
index 844ac03a..97b6fa06 100644
--- a/lib/linguist/blob_helper.rb
+++ b/lib/linguist/blob_helper.rb
@@ -259,6 +259,9 @@ module Linguist
       # If its a header file (.h) try to guess the language
       header_language ||
 
+        # If it's a .r file, try to guess the language
+        r_language ||
+
         # See if there is a Language for the extension
         pathname.language ||
 
@@ -288,6 +291,19 @@ module Linguist
       end
     end
 
+    # Internal: Guess language of .r files.
+    #
+    # Returns a Language.
+    def r_language
+      return unless extname == '.r'
+
+      if lines.grep(/(rebol|(:\s+func|make\s+object!|^\s*context)\s*\[)/i).any?
+        Language['Rebol']
+      else
+        Language['R']
+      end
+    end
+
     # Internal: Extract the script name from the shebang line
     #
     # Requires Blob#data
diff --git a/test/fixtures/hello-r.R b/test/fixtures/hello-r.R
new file mode 100644
index 00000000..30d715c3
--- /dev/null
+++ b/test/fixtures/hello-r.R
@@ -0,0 +1,4 @@
+hello <- function() {
+    print("hello, world!")
+}
+hello()
diff --git a/test/fixtures/hello-rebol.r b/test/fixtures/hello-rebol.r
new file mode 100644
index 00000000..4c1b7983
--- /dev/null
+++ b/test/fixtures/hello-rebol.r
@@ -0,0 +1,5 @@
+REBOL []
+hello: func [] [
+    print "hello, world!"
+]
+hello
diff --git a/test/test_blob.rb b/test/test_blob.rb
index 62844590..bcc63a98 100644
--- a/test/test_blob.rb
+++ b/test/test_blob.rb
@@ -224,6 +224,10 @@ class TestBlob < Test::Unit::TestCase
     assert_equal Language['Ruby'],        blob("wrong_shebang.rb").language
     assert_nil blob("octocat.png").language
 
+    # .r disambiguation
+    assert_equal Language['R'],           blob("hello-r.R").language
+    assert_equal Language['Rebol'],       blob("hello-rebol.r").language
+
     # ML
     assert_equal Language['OCaml'],       blob("Foo.ml").language
     assert_equal Language['Standard ML'], blob("Foo.sig").language