scaffolder-tools 0.1.2 → 0.1.3

data/.rspec

@@ -1 +1 @@
---color
+--color --format=progress

data/.travis.yml

@@ -0,0 +1,4 @@
+rvm:
+  - 1.8.7
+  - 1.9.2
+  - jruby

data/Gemfile

@@ -10,14 +10,12 @@ end
 group :development do
   gem "bundler",  "~> 1.0"
   gem "jeweler",  "~> 1.5"
-  gem "gherkin",  "~> 2.3.3"
   gem "rspec",    "~> 2.4"
   gem "cucumber", "~> 0.10"
   gem "fakefs",   "~> 0.2"
   gem "aruba",    "~> 0.2"
   gem "mocha",    "~> 0.9"
-  gem "hashie",   "~> 0.4"
   gem "yard",     "~> 0.6"
 
-  gem "scaffolder-test-helpers", "~> 0.1"
+  gem "scaffolder-test-helpers", "~> 0.4"
 end

data/README.rdoc

@@ -27,10 +27,12 @@ A unix man page is available for each scaffolder command by typing:
 
 == Contact
 
-Scaffolder was developed by Michael Barton (http://www.michaelbarton.me.uk).
-Pull requests, patches and bug reports are welcome. The source code is
-available on github[http://github.com/michaelbarton/scaffolder]. Bug reports
-and feature requests should be made here.
+Scaffolder was developed by Michael Barton (www.michaelbarton.me.uk). The
+website for scaffolder is at http://next.gs and a {twitter feed for
+updates}[http://twitter.com/nxtgs] is available. There is a {Google
+group}[http://groups.google.com/group/scaffolder] for questions and comments.
+The source code is available on github[http://github.com/scaffolder]. Bug
+reports and feature requests may also be made there.
 
 == Copyright
 

data/Rakefile

@@ -14,10 +14,11 @@ Jeweler::Tasks.new do |gem|
   gem.name = "scaffolder-tools"
   gem.summary = "Tools for manipulating genome scaffolds"
   gem.description = "Binary to use with scaffolder genome scaffolds"
-  gem.email = "mail@michaelbarton.me.uk"
-  gem.homepage = "http://www.michaelbarton.me.uk/projects/scaffolder/"
+  gem.email = "mail@next.gs"
+  gem.homepage = "http://next.gs"
   gem.authors = ["Michael Barton"]
   gem.license = "MIT"
+  gem.test_files = Dir['spec/**/*.rb']
 end
 Jeweler::RubygemsDotOrgTasks.new
 

data/VERSION

@@ -1 +1 @@
-0.1.2
+0.1.3

data/features/error_checking.feature

@@ -0,0 +1,137 @@
+Feature: Error checking command line arguments
+  In order to recoginise incorrect input
+  A user should be able to use the binaries with incorrect inputs
+  and be given clear error messages
+
+
+  Scenario: Using sequence where the sequence file specified does not exist
+    Given a file named "scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: "seq"
+      """
+    When I call "scaffolder" with arguments "sequence scaffold.yml missing_file"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Sequence file not found:"
+
+  Scenario: Using sequence where the sequence file doesn't contain any thing
+    Given an empty file named "sequence.fna"
+    Given a file named "scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: "seq1"
+      """
+    When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Sequence file is empty"
+
+  Scenario: Using sequence where the scaffold file specified does not exist
+    Given a file named "sequence.fna" with:
+      """
+      >seq
+      ATGGC
+      """
+    When I call "scaffolder" with arguments "sequence missing_file sequence.fna"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Scaffold file not found:"
+
+  Scenario: Using sequence where the scaffold file doesn't contain anything
+    Given an empty file named "scaffold.yml"
+    Given a file named "sequence.fna" with:
+      """
+      >seq
+      ATGGC
+      """
+    When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Scaffold file is empty"
+
+  Scenario: Using sequence where a scaffold sequence is missing
+    Given a file named "sequence.fna" with:
+      """
+      >seq1
+      ATGGC
+      """
+    Given a file named "scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: "seq1"
+        -
+          sequence:
+            source: "seq2"
+      """
+    When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Unknown sequence: seq2"
+
+  Scenario: Using validate where the sequence file specified does not exist
+    Given a file named "scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: "seq"
+      """
+    When I call "scaffolder" with arguments "validate scaffold.yml missing_file"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Sequence file not found:"
+
+  Scenario: Using validate where the sequence file doesn't contain any thing
+    Given an empty file named "sequence.fna"
+    Given a file named "scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: "seq1"
+      """
+    When I call "scaffolder" with arguments "validate scaffold.yml sequence.fna"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Sequence file is empty"
+
+  Scenario: Using validate where the scaffold file specified does not exist
+    Given a file named "sequence.fna" with:
+      """
+      >seq
+      ATGGC
+      """
+    When I call "scaffolder" with arguments "validate missing_file sequence.fna"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Scaffold file not found:"
+
+  Scenario: Using validate where the scaffold file doesn't contain anything
+    Given an empty file named "scaffold.yml"
+    Given a file named "sequence.fna" with:
+      """
+      >seq
+      ATGGC
+      """
+    When I call "scaffolder" with arguments "validate scaffold.yml sequence.fna"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Scaffold file is empty"
+
+  Scenario: Using validate where a scaffold sequence is missing
+    Given a file named "sequence.fna" with:
+      """
+      >seq1
+      ATGGC
+      """
+    Given a file named "scaffold.yml" with:
+      """
+      ---
+        -
+          sequence:
+            source: "seq1"
+        -
+          sequence:
+            source: "seq2"
+      """
+    When I call "scaffolder" with arguments "validate scaffold.yml sequence.fna"
+    Then the exit status should be 1
+    And the stderr should contain "Error. Unknown sequence: seq2"

data/features/sequence.feature

@@ -18,73 +18,12 @@ Feature: The scaffolder-sequence binary
       """
     When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna"
     Then the exit status should be 0
-    And the stdout should contain "ATGGC"
-
-  Scenario: The sequence file specified does not exist
-    Given a file named "scaffold.yml" with:
-      """
-      ---
-        -
-          sequence:
-            source: "seq"
-      """
-    When I call "scaffolder" with arguments "sequence scaffold.yml missing_file"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Sequence file not found:"
-
-  Scenario: The sequence file doesn't contain any thing
-    Given an empty file named "sequence.fna"
-    Given a file named "scaffold.yml" with:
-      """
-      ---
-        -
-          sequence:
-            source: "seq1"
+    And the stdout should contain exactly:
       """
-    When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Sequence file is empty"
-
-  Scenario: The scaffold file specified does not exist
-    Given a file named "sequence.fna" with:
-      """
-      >seq
+      >
       ATGGC
-      """
-    When I call "scaffolder" with arguments "sequence missing_file sequence.fna"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Scaffold file not found:"
 
-  Scenario: The scaffold file doesn't contain anything
-    Given an empty file named "scaffold.yml"
-    Given a file named "sequence.fna" with:
       """
-      >seq
-      ATGGC
-      """
-    When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Scaffold file is empty"
-
-  Scenario: One of the sequences specified in the scaffold is missing
-    Given a file named "sequence.fna" with:
-      """
-      >seq1
-      ATGGC
-      """
-    Given a file named "scaffold.yml" with:
-      """
-      ---
-        -
-          sequence:
-            source: "seq1"
-        -
-          sequence:
-            source: "seq2"
-      """
-    When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Unknown sequence: seq2"
 
   Scenario: Using the definition argument before the file arguments
     Given a file named "sequence.fna" with:
@@ -101,8 +40,12 @@ Feature: The scaffolder-sequence binary
       """
     When I call "scaffolder" with arguments "sequence --definition='name' scaffold.yml sequence.fna"
     Then the exit status should be 0
-    And the stdout should contain "ATGGC"
-    And the stdout should contain ">name"
+    And the stdout should contain exactly:
+      """
+      >name
+      ATGGC
+
+      """
 
   Scenario: Using the definition argument after the file arguments
     Given a file named "sequence.fna" with:
@@ -119,10 +62,14 @@ Feature: The scaffolder-sequence binary
       """
     When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna --definition='name'"
     Then the exit status should be 0
-    And the stdout should contain "ATGGC"
-    And the stdout should contain ">name"
+    And the stdout should contain exactly:
+      """
+      >name
+      ATGGC
 
-  Scenario: Using the argument --no-sequence-hash
+      """
+
+  Scenario: Using the argument --with-sequence-digest
     Given a file named "sequence.fna" with:
       """
       >seq
@@ -135,11 +82,16 @@ Feature: The scaffolder-sequence binary
           sequence:
             source: "seq"
       """
-    When I call "scaffolder" with arguments "sequence --no-sequence-hash scaffold.yml sequence.fna"
+    When I call "scaffolder" with arguments "sequence --with-sequence-digest scaffold.yml sequence.fna"
     Then the exit status should be 0
-    And the stdout should contain ">\nATGGC"
+    And the stdout should contain exactly:
+      """
+      >[sha1=32848c64b5bac47e23002c989a9d1bf3d21b8f92]
+      ATGGC
 
-  Scenario: Using the arguments --no-sequence-hash and --definition
+      """
+
+  Scenario: Using the arguments --with-sequence-digest and --definition
     Given a file named "sequence.fna" with:
       """
       >seq
@@ -152,6 +104,11 @@ Feature: The scaffolder-sequence binary
           sequence:
             source: "seq"
       """
-      When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna --no-sequence-hash --definition='name'"
+      When I call "scaffolder" with arguments "sequence scaffold.yml sequence.fna --with-sequence-digest --definition='name'"
     Then the exit status should be 0
-    And the stdout should contain ">name \nATGGC"
+    And the stdout should contain exactly:
+      """
+      >name [sha1=32848c64b5bac47e23002c989a9d1bf3d21b8f92]
+      ATGGC
+
+      """

data/features/step_definitions/scaffolder-tools.rb

@@ -1,4 +1,8 @@
 When /^I call "([^"]*)" with arguments "([^"]*)"$/ do |command,args|
   bin = File.join(File.dirname(__FILE__),'..','..','bin',command)
-  When "I run \"#{bin} #{args}\""
+  When "I run `#{bin} #{args}`"
+end
+
+Then /^the stdout yaml should contain exactly:$/ do |string|
+  YAML.load(string).should == YAML.load(all_stdout)
 end

data/features/validate.feature

@@ -3,72 +3,6 @@ Feature: The scaffolder-validate binary
   A user can use the scaffolder binary with the argument validate
   to test that inserts are correctly inserted
 
-  Scenario: The sequence file specified does not exist
-    Given a file named "scaffold.yml" with:
-      """
-      ---
-        -
-          sequence:
-            source: "seq"
-      """
-    When I call "scaffolder" with arguments "validate scaffold.yml missing_file"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Sequence file not found:"
-
-  Scenario: The sequence file doesn't contain any thing
-    Given an empty file named "sequence.fna"
-    Given a file named "scaffold.yml" with:
-      """
-      ---
-        -
-          sequence:
-            source: "seq1"
-      """
-    When I call "scaffolder" with arguments "validate scaffold.yml sequence.fna"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Sequence file is empty"
-
-  Scenario: The scaffold file specified does not exist
-    Given a file named "sequence.fna" with:
-      """
-      >seq
-      ATGGC
-      """
-    When I call "scaffolder" with arguments "validate missing_file sequence.fna"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Scaffold file not found:"
-
-  Scenario: The scaffold file doesn't contain anything
-    Given an empty file named "scaffold.yml"
-    Given a file named "sequence.fna" with:
-      """
-      >seq
-      ATGGC
-      """
-    When I call "scaffolder" with arguments "validate scaffold.yml sequence.fna"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Scaffold file is empty"
-
-  Scenario: One of the sequences specified in the scaffold is missing
-    Given a file named "sequence.fna" with:
-      """
-      >seq1
-      ATGGC
-      """
-    Given a file named "scaffold.yml" with:
-      """
-      ---
-        -
-          sequence:
-            source: "seq1"
-        -
-          sequence:
-            source: "seq2"
-      """
-    When I call "scaffolder" with arguments "validate scaffold.yml sequence.fna"
-    Then the exit status should be 1
-    And the stderr should contain "Error. Unknown sequence: seq2"
-
   Scenario: Validating a scaffold with no overlapping inserts
     Given a file named "sequence.fna" with:
       """
@@ -169,18 +103,18 @@ Feature: The scaffolder-validate binary
       """
     When I call "scaffolder" with arguments "validate scaffold.yml sequence.fna"
     Then the exit status should be 0
-    And the stdout should contain exactly:
+    And the stdout yaml should contain exactly:
     """
-    --- 
-    - sequence-insert-overlap: 
-        inserts: 
+    ---
+    - sequence-insert-overlap:
+        source: seq
+        inserts:
         - open: 2
           close: 4
           source: ins1
         - open: 3
           close: 5
           source: ins2
-        source: seq
 
     """
 
@@ -220,26 +154,26 @@ Feature: The scaffolder-validate binary
       """
     When I call "scaffolder" with arguments "validate scaffold.yml sequence.fna"
     Then the exit status should be 0
-    And the stdout should contain exactly:
+    And the stdout yaml should contain exactly:
     """
-    --- 
-    - sequence-insert-overlap: 
-        inserts: 
+    ---
+    - sequence-insert-overlap:
+        source: seq
+        inserts:
         - open: 2
           close: 4
           source: ins1
         - open: 3
           close: 5
           source: ins2
+    - sequence-insert-overlap:
         source: seq
-    - sequence-insert-overlap: 
-        inserts: 
+        inserts:
         - open: 6
           close: 8
           source: ins1
         - open: 7
           close: 9
           source: ins2
-        source: seq
 
     """