Chrome URL bugfix + added tests for chrome_app.apis.

src/chrome_app/apis.py

Index: src/chrome_app/apis.py
diff --git a/src/chrome_app/apis.py b/src/chrome_app/apis.py
index fa09a1a32f86857e7c6d720440eceed80179ac2e..ad3933b05659ade4e6d7ba4fff3827fad24b4fcb 100755
--- a/src/chrome_app/apis.py
+++ b/src/chrome_app/apis.py
@@ -42,19 +42,21 @@ import walk
 # Regular expression matching Chrome API namespaces, e.g. chrome.tts and
 # chrome.app.window.
 CHROME_API_REGEX = re.compile(
-    r'(?<![\w.])chrome\.((?:(?:app|sockets|system)\.)?\w+)')
+    r'(?<![\w.])chrome\.(?!google)((?:(?:app|sockets|system)\.)?\w+)')
 
 # Regular expression matching Chrome apps API usage, e.g.
 # chrome.tts.speak and chrome.app.runtime.onLaunched.addListener.
-CHROME_API_USAGE_REGEX = re.compile(r'(?<![\w.])chrome\.((?:\w+\.)+\w+)')
+CHROME_API_USAGE_REGEX = re.compile(
+    r'(?<![\w.])chrome\.(?!google)((?:\w+\.)+\w+)')
 
 # Regular expression matching anything in the chrome namespace, e.g. chrome.tts
 # or chrome.app.runtime.onLaunched.addListener.
-CHROME_NAMESPACE_REGEX = re.compile(r'chrome((?:\.\w+)+)')
+CHROME_NAMESPACE_REGEX = re.compile(r'chrome((?:\.(?!google)\w+)+)')
 
 # Regular expression matching both the API and the member, assuming both exist.
 CHROME_API_AND_MEMBER_REGEX = re.compile(r"""
   (?<![\w.])chrome\. # In the chrome namespace
+  (?!google) # chrome.google is not an API

[Matt Giuca, 2016/02/11 02:17:12]

Mmmmmm... I don't like this. As you said in the CL description, this is only
going to catch chrome.google.com, and not any other web domain with chrome in
it.

You also say "hardcoding a list of APIs is a bad idea". Why? The set of APIs is
fixed whereas the set of domains with "chrome" as a subdomain is limitless. So I
think we should hardcode the APIs. (But not in the regex; that would be a
nightmare. Instead, after you apply the regex, check to see if the API name
you've captured is in the list.)

Also: I don't think adding "(?!google)" to this regex is correct. It means
you're going to have potentially weird parses; if you encounter a domain like
"chrome.googlex.com" then you will return "x" as the API name.

   ( # Capture the full API name
     (?:(?:app|sockets|system)\.)? # API may be part of a super-API namespace
     \w+ # Actual API name