URL Formatter: Add StripSubdomain method that preserves eTLD + 1.

components/url_formatter/url_formatter.h

Index: components/url_formatter/url_formatter.h
diff --git a/components/url_formatter/url_formatter.h b/components/url_formatter/url_formatter.h
index 90f2bbf9b73128e35a080382b35fe91d08db3d09..72dd55d43dc4357bc3b644bcf59b0655afd71df1 100644
--- a/components/url_formatter/url_formatter.h
+++ b/components/url_formatter/url_formatter.h
@@ -137,11 +137,20 @@ void AppendFormattedHost(const GURL& url, base::string16* output);
 base::string16 IDNToUnicode(base::StringPiece host);
 
 // If |text| starts with "www." it is removed, otherwise |text| is returned
-// unmodified.
+// unmodified. |text| is an arbitrary string and not necessarily a valid host.
+// Even if |text| is a valid host, this method may return an invalid host.

[Peter Kasting, 2017/06/17 02:46:55]

I think these two sentences use "valid" in a way that differs from the GURL
sense.  For example, in GURL, "www." is a valid host, as is "com".

I think what you mean is something like "Both |text| and the returned value are
arbitrary strings and are not guaranteed to be navigable hostnames."

[tommycli, 2017/06/19 23:03:08]

Done.

+// This is useful for finding search engine keywords inside user input.
 base::string16 StripWWW(const base::string16& text);
 
-// Runs |url|'s host through StripWWW().  |url| must be valid.
-base::string16 StripWWWFromHost(const GURL& url);
+// Returns the host of |url| with subdomains specified by |subdomains|
+// removed. |url| must be valid. If the |url| host is already an eTLD+1,
+// the host is returned unmodified.
+typedef uint32_t StripSubdomainType;
+typedef uint32_t StripSubdomainTypes;
+constexpr StripSubdomainType kStripWWW = 1 << 0;
+constexpr StripSubdomainType kStripM = 1 << 1;
+base::StringPiece StripSubdomains(const GURL& url,
+                                  StripSubdomainTypes subdomains);
 
 }  // namespace url_formatter