Add `slurp_fields_from_readme` function.

* gd_diff.py (slurp_fields_from_readme): New function.
* tests/test_gd_diff.py (test_slurp_fields_from_readme): Test method for
`slurp_fields_from_readme`

2 files changed, 74 insertions, 0 deletions

diff --git a/gd_diff.py b/gd_diff.py
index 4709d30..8d332f4 100644
--- a/gd_diff.py
+++ b/gd_diff.py
@@ -7,6 +7,7 @@
 #  <https://creativecommons.org/publicdomain/zero/1.0>
 
 import os
+import re
 import shlex
 import sys
 
@@ -143,3 +144,32 @@ def read_gns_readme(release, pkg, local_dir):
     readme_content = read_file(readme_path)
 
     return readme_content
+
+
+def slurp_fields_from_readme(content):
+    """Returns dict containing fields slurped from `content`
+
+    - If a field is not defined or if its value is empty in the
+    `content`, then its corresponding value in the dict will be None.
+
+    """
+    # list of recognized fields.
+    field_list = [
+        'Change-Type',
+        'Changed-From-Debian',
+    ]
+
+    field_values = {}
+    for field in field_list:
+        pattern = r'{}:[ ]*(.+)'.format(field)
+        field_pattern = re.compile(pattern)
+        field_match = field_pattern.search(content)
+
+        if (field_match and
+            field_match.group(1) and
+            field_match.group(1).strip()):
+            field_values[field] = field_match.group(1).strip()
+        else:
+            field_values[field] = None
+
+    return field_values
diff --git a/tests/test_gd_diff.py b/tests/test_gd_diff.py
index d377439..145461f 100644
--- a/tests/test_gd_diff.py
+++ b/tests/test_gd_diff.py
@@ -154,6 +154,50 @@ class TestGdDiff(object):
         assert_equal(antlr_readme_content, expected_antlr_readme_content)
 
 
+    def test_slurp_fields_from_readme(self):
+        readme_content = ''
+        field_values = slurp_fields_from_readme(readme_content)
+        assert_equal(field_values['Change-Type'], None)
+        assert_equal(field_values['Changed-From-Debian'], None)
+
+        readme_content = 'Changed-From-Debian: \n'
+        field_values = slurp_fields_from_readme(readme_content)
+        assert_equal(field_values['Change-Type'], None)
+        assert_equal(field_values['Changed-From-Debian'], None)
+
+        readme_content = 'Change-Type: Deblob\n'
+        field_values = slurp_fields_from_readme(readme_content)
+        assert_equal(field_values['Change-Type'], 'Deblob')
+        assert_equal(field_values['Changed-From-Debian'], None)
+
+        readme_content = 'Changed-From-Debian: \nChange-Type: \n'
+        field_values = slurp_fields_from_readme(readme_content)
+        assert_equal(field_values['Change-Type'], None)
+        assert_equal(field_values['Changed-From-Debian'], None)
+
+        readme_content = 'Changed-From-Debian: Branding.    \nChange-Type: \n'
+        field_values = slurp_fields_from_readme(readme_content)
+        assert_equal(field_values['Change-Type'], None)
+        assert_equal(field_values['Changed-From-Debian'], 'Branding.')
+
+        readme_content = 'Changed-From-Debian: \nChange-Type: Modified'
+        field_values = slurp_fields_from_readme(readme_content)
+        assert_equal(field_values['Change-Type'], 'Modified')
+        assert_equal(field_values['Changed-From-Debian'], None)
+
+        readme_content = 'Changed-From-Debian:Fixed ambiguous use of the word free.\n\n\nChange-Type:Deblob\n'
+        field_values = slurp_fields_from_readme(readme_content)
+        assert_equal(field_values['Change-Type'], 'Deblob')
+        assert_equal(field_values['Changed-From-Debian'],
+                     'Fixed ambiguous use of the word free.')
+
+        readme_content  = 'Changed-From-Debian: Removed example with non-free files.\nChange-Type: Modified\n\nFor gNewSense, the non-free unicode.IDENTs files are *actually* removed (see\nalso README.source). See gNewSense bug #34218 for details.\n'
+        field_values = slurp_fields_from_readme(readme_content)
+        assert_equal(field_values['Change-Type'], 'Modified')
+        assert_equal(field_values['Changed-From-Debian'],
+                     'Removed example with non-free files.')
+
+
     def teardown(self):
         """Teardown method for this class."""
         if(path.exists(self.gns_pkgs_dir)):