Merge pull request #5 from nexB/develop

Updating my fork

Author: akugarg

src/cluecode/copyrights.py

@@ -17,7 +17,6 @@
 from cluecode import copyrights_hint
 from commoncode.text import toascii
 from commoncode.text import unixlinesep
-from textcode import analysis
 
 # Tracing flags
 TRACE = False or os.environ.get('SCANCODE_DEBUG_COPYRIGHT', False)
@@ -60,6 +59,7 @@ def logger_debug(*args):
 
 def detect_copyrights(location, copyrights=True, holders=True, authors=True,
                       include_years=True, include_allrights=False,
+                      demarkup=True,
                       deadline=sys.maxsize):
     """
     Yield tuples of (detection type, detected string, start line, end line)
@@ -68,14 +68,36 @@ def detect_copyrights(location, copyrights=True, holders=True, authors=True,
     Valid detection types are: copyrights, authors, holders.
     These are included in the yielded tuples based on the values of `copyrights=True`, `holders=True`, `authors=True`,
     """
-    detector = CopyrightDetector()
-    numbered_lines = analysis.numbered_text_lines(location, demarkup=True)
+    from textcode.analysis import numbered_text_lines
+    numbered_lines = numbered_text_lines(location, demarkup=demarkup)
     numbered_lines = list(numbered_lines)
     if TRACE:
         numbered_lines = list(numbered_lines)
         for nl in numbered_lines:
             logger_debug('numbered_line:', repr(nl))
 
+    yield from detect_copyrights_from_lines(
+        numbered_lines,
+        copyrights=copyrights,
+        holders=holders,
+        authors=authors,
+        include_years=include_years,
+        include_allrights=include_allrights,
+        deadline=deadline)
+
+
+def detect_copyrights_from_lines(numbered_lines, copyrights=True, holders=True, authors=True,
+          include_years=True, include_allrights=False,
+          deadline=sys.maxsize):
+    """
+    Yield tuples of (detection type, detected string, start line, end line)
+    detected in numbered lines
+    Include years in copyrights if include_years is True.
+    Valid detection types are: copyrights, authors, holders.
+    These are included in the yielded tuples based on the values of `copyrights=True`, `holders=True`, `authors=True`,
+    """
+    detector = CopyrightDetector()
+
     for candidates in candidate_lines(numbered_lines):
 
         detections = detector.detect(
@@ -93,6 +115,7 @@ def detect_copyrights(location, copyrights=True, holders=True, authors=True,
         if time() > deadline:
             break
 
+
 ################################################################################
 # DETECTION PROPER
 ################################################################################

tests/cluecode/test_plugin_filter_clues.py

@@ -30,7 +30,7 @@ def test_is_empty_():
     assert not is_empty(Detections(urls=['Foo']))
 
 
-def test_scan_plugin_filter_cluesfor_rule():
+def test_scan_plugin_filter_clues_for_rule():
     # this test fies is a copy of apache-1.1_63.RULE that contains
     # several emails, authors, urls and copyrights
     # it has been modified to include more unrelated clues
@@ -39,7 +39,7 @@ def test_scan_plugin_filter_cluesfor_rule():
     args = ['-clieu', '--filter-clues', test_dir, '--json', result_file]
     run_scan_click(args)
     expected = test_env.get_test_loc('plugin_filter_clues/filtered-expected.json')
-    check_json_scan(expected, result_file, regen=False)
+    check_json_scan(expected, result_file, remove_file_date=True, ignore_headers=True, regen=False)
 
 
 def test_scan_plugin_filter_clues_does_not_filter_incorrectly():
@@ -51,7 +51,7 @@ def test_scan_plugin_filter_clues_does_not_filter_incorrectly():
     args = ['-clieu', '--filter-clues', test_dir, '--json', result_file]
     run_scan_click(args)
     expected = test_env.get_test_loc('plugin_filter_clues/filtered-expected2.json')
-    check_json_scan(expected, result_file, regen=False)
+    check_json_scan(expected, result_file, remove_file_date=True, ignore_headers=True, regen=False)
 
 
 def test_scan_plugin_filter_clues_for_license():
@@ -62,4 +62,4 @@ def test_scan_plugin_filter_clues_for_license():
     args = ['-clieu', '--filter-clues', test_dir, '--json', result_file]
     run_scan_click(args)
     expected = test_env.get_test_loc('plugin_filter_clues/filtered-expected3.json')
-    check_json_scan(expected, result_file, regen=False)
+    check_json_scan(expected, result_file, remove_file_date=True, ignore_headers=True, regen=False)