iwyu: pass source root into fix_includes

fix_includes tries to identify the "main compilation unit header" for
each file using some heuristics. For example, "foo-test.cc" is
associated with "foo.h".

This was previously broken due to the fact that the tool runs from the
build/ directory instead of the source directory, so it saw the CC file
as "../../src/kudu/.../foo.cc" while it saw the header file as
"kudu/.../foo.h".

This adds a --source_root parameter to fix_includes so that the source
files can be relativized to this path prior to fixing.

Change-Id: Iac9f0b47b9a3c8180e178ed59863d33840434433
Reviewed-on: http://gerrit.cloudera.org:8080/10159
Tested-by: Todd Lipcon <[email protected]>
Reviewed-by: Alexey Serbin <[email protected]>

build-support/iwyu.py

@@ -134,7 +134,9 @@ def crash(output):
 
   try:
     output = check_output(cmdline, env=env, stderr=subprocess.STDOUT)
-    if '\nFATAL ERROR: ' in output or _RE_CLANG_ERROR.search(output):
+    if '\nFATAL ERROR: ' in output or \
+       'Assertion failed: ' in output or \
+       _RE_CLANG_ERROR.search(output):
       crash(output)
     return output
   except subprocess.CalledProcessError, e:
@@ -161,7 +163,9 @@ def _get_thirdparty_include_dirs():
 
 
 def _get_fixer_flags(flags):
-  args = ['--quiet', '--nosafe_headers']
+  args = ['--quiet',
+          '--nosafe_headers',
+          '--source_root=%s' % os.path.join(ROOT, 'src')]
   if flags.dry_run:
     args.append("--dry_run")
   for d in _get_thirdparty_include_dirs():

build-support/iwyu/fix_includes.py

@@ -1503,31 +1503,32 @@ def _IsSameProject(line_info, edited_file, project):
   return (included_root and edited_root and included_root == edited_root)
 
 
-def _GetLineKind(file_line, filename, separate_project_includes, thirdparty_include_dirs):
+def _GetLineKind(file_line, filename, flags):
   """Given a file_line + file being edited, return best *_KIND value or None.
 
   Arguments:
     file_line: the LineInfo structure to be analyzed
     filename: the file which contains the line to be analyzed
-    separate_project_includes: see 'project' parameter of '_IsSameProject()'
-    thirdparty_include_dirs: see help output for the corresponding flag
+    flags: the program flags. Uses 'separate_project_includes' and 'thirdparty_include_dirs'
   """
+  if flags.source_root:
+    filename = os.path.relpath(filename, flags.source_root)
   line_without_coments = _COMMENT_RE.sub('', file_line.line)
   if file_line.deleted:
     return None
   elif _IsMainCUInclude(file_line, filename):
     return _MAIN_CU_INCLUDE_KIND
   elif _IsSystemInclude(file_line) and \
-         not _IsThirdpartyInclude(file_line, thirdparty_include_dirs):
+         not _IsThirdpartyInclude(file_line, flags.thirdparty_include_dirs):
     if '.' in line_without_coments:
       # e.g. <string.h>
       return _C_SYSTEM_INCLUDE_KIND
     else:
       # e.g. <string>
       return _CXX_SYSTEM_INCLUDE_KIND
   elif file_line.type == _INCLUDE_RE:
-    if (separate_project_includes and
-        _IsSameProject(file_line, filename, separate_project_includes)):
+    if (flags.separate_project_includes and
+        _IsSameProject(file_line, filename, flags.separate_project_includes)):
       return _PROJECT_INCLUDE_KIND
     return _NONSYSTEM_INCLUDE_KIND
   elif file_line.type == _FORWARD_DECLARE_RE:
@@ -1606,9 +1607,8 @@ class Bang;
   last_reorder_spans = {}
   for reorder_span in good_reorder_spans:
     for line_number in range(*reorder_span):
-      line_kind = _GetLineKind(file_lines[line_number], filename,
-                               flags.separate_project_includes,
-                               flags.thirdparty_include_dirs)
+      line_kind = _GetLineKind(file_lines[line_number], filename, flags)
+
       # Ignore forward-declares that come after 'contentful' code; we
       # never want to insert new forward-declares there.
       if (line_kind == _FORWARD_DECLARE_KIND and
@@ -1787,9 +1787,7 @@ def _DecoratedMoveSpanLines(iwyu_record, file_lines, move_span_lines, flags):
   sort_key = sort_key.replace('-inl.h', '_inl.h')
 
   # Next figure out the kind.
-  kind = _GetLineKind(firstline, iwyu_record.filename,
-                      flags.separate_project_includes,
-                      flags.thirdparty_include_dirs)
+  kind = _GetLineKind(firstline, iwyu_record.filename, flags)
 
   # All we're left to do is the reorder-span we're in.  Hopefully it's easy.
   reorder_span = firstline.reorder_span
@@ -2270,6 +2268,14 @@ def ParseArgs(args):
                     help=('Any includes which are found in <dir> are considered '
                           '"other-project" includes rather than "system" includes.'))
 
+  parser.add_option('--source_root', type='str',
+                    metavar="dir",
+                    help=('The path to the root of the source tree. When processing '
+                          'changes to a file /project/src/bar/baz.cc, setting '
+                          '--source_root to /project/src will ensure that an include '
+                          'of "bar/baz.h" is properly identified as the corresponding '
+                          'main-compilation-unit include.'))
+
   parser.add_option('--invoking_command_line', default=None,
                     help=('Internal flag used by iwyu.py, It should be the'
                           ' command line used to invoke iwyu.py'))