[telemetry] Refactoring script.

tools/telemetry/telemetry/util/path.py

Index: tools/telemetry/telemetry/util/path.py
diff --git a/tools/telemetry/telemetry/util/path.py b/tools/telemetry/telemetry/util/path.py
index 7feb861bdfba2c627cae5c062d998c5da32f7d29..fce896e1666cc87fe18f0465dcf39caed2b18092 100644
--- a/tools/telemetry/telemetry/util/path.py
+++ b/tools/telemetry/telemetry/util/path.py
@@ -41,3 +41,17 @@ def FindInstalledWindowsApplication(application_path):
       return path
 
   return None
+
+
+def InDirectory(subdirectory, directory):
+  """Returns True iff subdirectory is or is in directory.
+
+  A little bit of a misnomer, since it works even if they're files."""

[eakuefner, 2015/03/27 17:55:06]

Do we actually need this? Why not just get the paths as strings and use
string.startswith?

If you're going to keep this I'd say remove this disclaimer.

Either SubpathOf(subpath, superpath) and then it will obvious that it can be
used for either, or keep the signature as-is; it's awkward if I call InDirectory
with subdirectory == something that isn't actually a directory.

[dtu, 2015/03/27 19:17:15]

This is surprisingly tricky, and people get it wrong all the time. Using
startswith() or os.path.commonprefix() will fail if one directory is a substring
of another. Here's an example:
'foo/bar_test/nom.py'.startswith('foo/bar') is True, but we want False.

(My original version of this function in find_dependencies was incorrect.)

Did the rename, though.

+  subdirectory = os.path.realpath(subdirectory)
+  directory = os.path.realpath(directory)
+
+  while len(subdirectory) >= len(directory):
+    if subdirectory == directory:
+      return True
+    subdirectory = os.path.split(subdirectory)[0]
+  return False