Final phase of the URL History Quick autocomplete provider integration. This phase hooks up the InMemoryURLIndex to the HistoryQuickProvider which is itself made active during the autocomplete process. A small amount of refactoring was done to liberate some common functionality from the history_url_provider.

BUG=None
TEST=None

Review URL: http://codereview.chromium.org/3375002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@62193 0039d316-1c4b-4281-b951-d872f2087c98

1 files changed, 103 insertions, 0 deletions

diff --git a/chrome/browser/autocomplete/history_provider.cc b/chrome/browser/autocomplete/history_provider.cc
new file mode 100644
index 0000000..bbc2fe5
--- /dev/null
+++ b/chrome/browser/autocomplete/history_provider.cc
@@ -0,0 +1,103 @@
+// Copyright (c) 2010 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "chrome/browser/autocomplete/history_url_provider.h"
+
+#include <string>
+
+#include "base/string_util.h"
+#include "base/utf_string_conversions.h"
+#include "chrome/browser/net/url_fixer_upper.h"
+#include "chrome/common/url_constants.h"
+#include "googleurl/src/url_util.h"
+
+HistoryProvider::HistoryProvider(ACProviderListener* listener,
+                                 Profile* profile,
+                                 const char* name)
+    : AutocompleteProvider(listener, profile, name) {
+}
+
+// static
+std::wstring HistoryProvider::FixupUserInput(const AutocompleteInput& input) {
+  const std::wstring& input_text = input.text();
+  // Fixup and canonicalize user input.
+  const GURL canonical_gurl(URLFixerUpper::FixupURL(WideToUTF8(input_text),
+                                                    std::string()));
+  std::string canonical_gurl_str(canonical_gurl.possibly_invalid_spec());
+  if (canonical_gurl_str.empty()) {
+    // This probably won't happen, but there are no guarantees.
+    return input_text;
+  }
+
+  // If the user types a number, GURL will convert it to a dotted quad.
+  // However, if the parser did not mark this as a URL, then the user probably
+  // didn't intend this interpretation.  Since this can break history matching
+  // for hostname beginning with numbers (e.g. input of "17173" will be matched
+  // against "0.0.67.21" instead of the original "17173", failing to find
+  // "17173.com"), swap the original hostname in for the fixed-up one.
+  if ((input.type() != AutocompleteInput::URL) &&
+      canonical_gurl.HostIsIPAddress()) {
+    std::string original_hostname =
+        WideToUTF8(input_text.substr(input.parts().host.begin,
+                                     input.parts().host.len));
+    const url_parse::Parsed& parts =
+        canonical_gurl.parsed_for_possibly_invalid_spec();
+    // parts.host must not be empty when HostIsIPAddress() is true.
+    DCHECK(parts.host.is_nonempty());
+    canonical_gurl_str.replace(parts.host.begin, parts.host.len,
+                               original_hostname);
+  }
+  std::wstring output = UTF8ToWide(canonical_gurl_str);
+  // Don't prepend a scheme when the user didn't have one.  Since the fixer
+  // upper only prepends the "http" scheme, that's all we need to check for.
+  if (canonical_gurl.SchemeIs(chrome::kHttpScheme) &&
+      !url_util::FindAndCompareScheme(WideToUTF8(input_text),
+                                      chrome::kHttpScheme, NULL))
+    TrimHttpPrefix(&output);
+
+  // Make the number of trailing slashes on the output exactly match the input.
+  // Examples of why not doing this would matter:
+  // * The user types "a" and has this fixed up to "a/".  Now no other sites
+  //   beginning with "a" will match.
+  // * The user types "file:" and has this fixed up to "file://".  Now inline
+  //   autocomplete will append too few slashes, resulting in e.g. "file:/b..."
+  //   instead of "file:///b..."
+  // * The user types "http:/" and has this fixed up to "http:".  Now inline
+  //   autocomplete will append too many slashes, resulting in e.g.
+  //   "http:///c..." instead of "http://c...".
+  // NOTE: We do this after calling TrimHttpPrefix() since that can strip
+  // trailing slashes (if the scheme is the only thing in the input).  It's not
+  // clear that the result of fixup really matters in this case, but there's no
+  // harm in making sure.
+  const size_t last_input_nonslash = input_text.find_last_not_of(L"/\\");
+  const size_t num_input_slashes = (last_input_nonslash == std::wstring::npos) ?
+      input_text.length() : (input_text.length() - 1 - last_input_nonslash);
+  const size_t last_output_nonslash = output.find_last_not_of(L"/\\");
+  const size_t num_output_slashes =
+      (last_output_nonslash == std::wstring::npos) ?
+      output.length() : (output.length() - 1 - last_output_nonslash);
+  if (num_output_slashes < num_input_slashes)
+    output.append(num_input_slashes - num_output_slashes, '/');
+  else if (num_output_slashes > num_input_slashes)
+    output.erase(output.length() - num_output_slashes + num_input_slashes);
+
+  return output;
+}
+
+// static
+size_t HistoryProvider::TrimHttpPrefix(std::wstring* url) {
+  // Find any "http:".
+  if (!HasHTTPScheme(*url))
+    return 0;
+  size_t scheme_pos = url->find(ASCIIToWide(chrome::kHttpScheme) + L":");
+  DCHECK(scheme_pos != std::wstring::npos);
+
+  // Erase scheme plus up to two slashes.
+  size_t prefix_end = scheme_pos + strlen(chrome::kHttpScheme) + 1;
+  const size_t after_slashes = std::min(url->length(), prefix_end + 2);
+  while ((prefix_end < after_slashes) && ((*url)[prefix_end] == L'/'))
+    ++prefix_end;
+  url->erase(scheme_pos, prefix_end - scheme_pos);
+  return (scheme_pos == 0) ? prefix_end : 0;
+}