[zorg] Lit parser should ignore leading whitespace in log file.

[rfoos]

Summary

Use case: MSBuild indents Lit test output, and test results are not displayed.

Everything except the Stop RE strips the log line, and this one line change allows indented logs to be displayed.

[rfoos]

Drop file not part of change.

[gkistanova]

LGTM with a minor note for line 51.

[gkistanova]

It seems my comment to the line 51 was lost in the original message. Sorry for this.
Here it is.

zorg/buildbot/commands/LitTestCommand.py

51

lstrip() if you really want to remove the leading whitespace.

[rfoos]

zorg/buildbot/commands/LitTestCommand.py

51

I agree, lstrip is more correct.

strip should to lstrip + rstrip.

I used strip because the StartRE, and TestLineRE also use strip.

m = self.kTestVerboseLogStartRE.match(line.strip())
m = self.kTestLineRE.match(line.strip())

Should I

• leave the strip on stopRE
• Change the three RE's to lstrip
• change only the stopRE to lstrip.

zorg/buildbot/builders/LLDBBuilder.py

 def cleanSVNSourceTree(f, srcdir='llvm'):
     f.addStep(ShellCommand(name='clean svn source %s' % srcdir,
                            command="svn status --no-ignore | grep '^[I?]' | cut -c 9- | while IFS= read -r f; do echo \"$f\"; rm -rf \"$f\"; done",
                            description="clean SVN source tree",
                            workdir='%s' % srcdir))
     return f
 
 # CMake Windows builds
     def getLLDBWindowsCMakeBuildFactory(
             clean=False,
             cmake='cmake',
                 jobs="%(jobs)s",
 
             # Source directory containing a built python
             python_source_dir=r'C:/src/python',
 
             # Default values for VS devenv and build configuration
             vs=r"""%VS120COMNTOOLS%""",
             config='Release',
             target_arch='x86',

zorg/buildbot/commands/LitTestCommand.py

for code in self.failingCodes:
       if self.resultCounts.get(code):
         return True
 
   def handleVerboseLogLine(self, line):
     # Append to the log.
     self.activeVerboseLog.append(line)
 
     # If this is a stop marker, process the test info.
-    if self.kTestVerboseLogStopRE.match(line):
+    if self.kTestVerboseLogStopRE.match(line.strip()):
       self.testInfoFinished()
 
   def testInfoFinished(self):
     # We have finished getting information for one test, handle it.
     code, name = self.lastTestResult
 
     # If the test failed, add a log entry for it (unless we have reached the
     # max).
""")
     self.assertEqual(obs.step.logs, [
         ('FAIL: test-one', 'FAIL: test-one'),
         ('FAIL: test-two', """\
 **** TEST 'test-two' FAILED ****
 bla bla bla
 **********"""),
         ('FAIL: test-three', 'FAIL: test-three')])
 
+  def test_indented_log(self):
+    obs = self.parse_log("""
+    FAIL: test-one (1 of 3)
+    FAIL: test-two (2 of 3)
+    **** TEST 'test-two' FAILED ****
+    bla bla bla
+    **********
+    FAIL: test-three (3 of 3)
+""")
+
+    self.assertEqual(obs.resultCounts, { 'FAIL' : 3 })
+    self.assertEqual(obs.step.logs, [
+        ('FAIL: test-one', 'FAIL: test-one'),
+        ('FAIL: test-two', """\
+    **** TEST 'test-two' FAILED ****
+    bla bla bla
+    **********"""),
+        ('FAIL: test-three', 'FAIL: test-three')])
+
 class TestCommand(unittest.TestCase):
   def parse_log(self, text, **kwargs):
     cmd = LitTestCommand(**kwargs)
     cmd.logObserver.step = StepProxy()
     for ln in text.split('\n'):
       cmd.logObserver.outLineReceived(ln)
     return cmd