[scan-build-py] fix some line separator issues

[rizsotto.mailinglist]

Summary

It's a chunk from D26390

Found interesting bugs on Windows platform. (Functional test will come later for this.)

[dcoughlin]

Looks good to me, although please update the comment to be prose and have proper capitalization.

Also: it really helps to have patches with context -- it makes patches much faster to review. (See http://llvm.org/docs/Phabricator.html#requesting-a-review-via-the-web-interface )

[rizsotto.mailinglist]

Looks good to me, although please update the comment to be prose and have proper capitalization.

Can I ask a temporary exemption on capitalization? Here's why I'm asking... Currently all pydoc comments are capitalized, but inline ones are not. So, it's kind of consistent this way. (I understand that's not what the project guidelines says.) And I would like to catch up with the changes I have on my github project. It will take 20-30 more patches like this. It's much easier to do this if I can diff the two source tree... And when I'm done with it, will make a patch on capitalization only, where we can fix my bad English, or reformulate comments when needed.

Also: it really helps to have patches with context -- it makes patches much faster to review. (See http://llvm.org/docs/Phabricator.html#requesting-a-review-via-the-web-interface )

Sure, will do more context. Have not checked the generated diff, sorry.

[dcoughlin]

In D30600#692629, @rizsotto.mailinglist wrote:

Looks good to me, although please update the comment to be prose and have proper capitalization.

Can I ask a temporary exemption on capitalization? Here's why I'm asking... Currently all pydoc comments are capitalized, but inline ones are not. So, it's kind of consistent this way. (I understand that's not what the project guidelines says.) And I would like to catch up with the changes I have on my github project. It will take 20-30 more patches like this. It's much easier to do this if I can diff the two source tree... And when I'm done with it, will make a patch on capitalization only, where we can fix my bad English, or reformulate comments when needed.

I think skipping the capitalization for now is perfectly fine. But we do need comments in prose, otherwise we can't tell what the intent of the upstreamed patch is and so can't review it properly.

[rizsotto.mailinglist]

Thanks Devin!

cfe/trunk/tools/scan-build-py/libear/__init__.py

class Configure(object):
     def write_by_template(self, template, output):
         def transform(line, definitions):
 
             pattern = re.compile(r'^#cmakedefine\s+(\S+)')
             m = pattern.match(line)
             if m:
                 key = m.group(1)
                 if key not in definitions or not definitions[key]:
-                    return '/* #undef {} */\n'.format(key)
+                    return '/* #undef {0} */{1}'.format(key, os.linesep)
                 else:
-                    return '#define {}\n'.format(key)
+                    return '#define {0}{1}'.format(key, os.linesep)
             return line
 
         with open(template, 'r') as src_handle:
             logging.debug('Writing config to %s', output)
             with open(output, 'w') as dst_handle:
                 for line in src_handle:
                     dst_handle.write(transform(line, self.results))
 

cfe/trunk/tools/scan-build-py/libscanbuild/report.py

def parse_bug_html(filename):
     yield bug
 
 
 def parse_crash(filename):
     """ Parse out the crash information from the report file. """
 
     match = re.match(r'(.*)\.info\.txt', filename)
     name = match.group(1) if match else None
-    with open(filename) as handler:
-        lines = handler.readlines()
+    with open(filename, mode='rb') as handler:
+        # this is a workaround to fix windows read '\r\n' as new lines.
+        lines = [line.decode().rstrip() for line in handler.readlines()]
         return {
-            'source': lines[0].rstrip(),
-            'problem': lines[1].rstrip(),
+            'source': lines[0],
+            'problem': lines[1],
             'file': name,
             'info': name + '.info.txt',
             'stderr': name + '.stderr.txt'
         }
 
 
 def category_type_name(bug):
     """ Create a new bug attribute from bug by category and type.