Allow url::SchemeHostPort to hold non-file scheme without port

url/url_util.cc

 // Copyright 2013 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 "url/url_util.h"
 
 #include <string.h>
 #include <vector>
 
 #include "base/debug/leak_annotations.h"
 #include "base/logging.h"
 #include "base/strings/string_util.h"
 #include "url/url_canon_internal.h"
 #include "url/url_file.h"
 #include "url/url_util_internal.h"
 
 namespace url {
 
 namespace {
 
 const int kNumStandardURLSchemes = 8;
-const char* kStandardURLSchemes[kNumStandardURLSchemes] = {
-  kHttpScheme,
-  kHttpsScheme,
-  kFileScheme,  // Yes, file urls can have a hostname!
-  kFtpScheme,
-  kGopherScheme,
-  kWsScheme,    // WebSocket.
-  kWssScheme,   // WebSocket secure.
-  kFileSystemScheme,
+const SchemeWithType kStandardURLSchemes[kNumStandardURLSchemes] = {
+  {kHttpScheme, SCHEME_WITH_PORT},
+  {kHttpsScheme, SCHEME_WITH_PORT},
+  {kFileScheme, SCHEME_WITHOUT_PORT},  // Yes, file urls can have a hostname!

[Ryan Sleevi, 2015/08/13 22:32:12]

nit: It took me a while to find out why this is 'true'

RFC 3986 defines the generic syntax, but the IANA registry for file (
www.iana.org/assignments/uri-schemes/uri-schemes.xhtml ) refers to
http://tools.ietf.org/html/rfc1738#section-3.10 , but that's then updated by
http://tools.ietf.org/html/rfc3986 

I'm not 100% sure this is universally true, but it's consistent enough with the
previous implementation, so yolo.

[tyoshino (SeeGerritForStatus), 2015/08/14 06:29:01]

Ya. At least we won't break anything new, I think.

+  {kFtpScheme, SCHEME_WITH_PORT},
+  {kGopherScheme, SCHEME_WITH_PORT},
+  {kWsScheme, SCHEME_WITH_PORT},    // WebSocket.
+  {kWssScheme, SCHEME_WITH_PORT},   // WebSocket secure.
+  {kFileSystemScheme, SCHEME_WITHOUT_AUTHORITY},
 };
 
 // List of the currently installed standard schemes. This list is lazily
 // initialized by InitStandardSchemes and is leaked on shutdown to prevent
 // any destructors from being called that will slow us down or cause problems.
-std::vector<const char*>* standard_schemes = NULL;
+std::vector<SchemeWithType>* standard_schemes = NULL;
 
 // See the LockStandardSchemes declaration in the header.
 bool standard_schemes_locked = false;
 
 // This template converts a given character type to the corresponding
 // StringPiece type.
 template<typename CHAR> struct CharToStringPiece {
 };
 template<> struct CharToStringPiece<char> {
   typedef base::StringPiece Piece;
 };
 template<> struct CharToStringPiece<base::char16> {
   typedef base::StringPiece16 Piece;
 };
 
 // Ensures that the standard_schemes list is initialized, does nothing if it
 // already has values.
 void InitStandardSchemes() {
   if (standard_schemes)
     return;
-  standard_schemes = new std::vector<const char*>;
+  standard_schemes = new std::vector<SchemeWithType>;
   for (int i = 0; i < kNumStandardURLSchemes; i++)
     standard_schemes->push_back(kStandardURLSchemes[i]);
 }
 
 // Given a string and a range inside the string, compares it to the given
 // lower-case |compare_to| buffer.
 template<typename CHAR>
 inline bool DoCompareSchemeComponent(const CHAR* spec,
                                      const Component& component,
                                      const char* compare_to) {
   if (!component.is_nonempty())
     return compare_to[0] == 0;  // When component is empty, match empty scheme.
   return base::LowerCaseEqualsASCII(
       typename CharToStringPiece<CHAR>::Piece(
           &spec[component.begin], component.len),
       compare_to);
 }
 
-// Returns true if the given scheme identified by |scheme| within |spec| is one
-// of the registered "standard" schemes.
+// Returns true and sets |type| to the SchemeType of the given scheme
+// identified by |scheme| within |spec| if the scheme is one of the registered
+// "standard" schemes.
 template<typename CHAR>
-bool DoIsStandard(const CHAR* spec, const Component& scheme) {
+bool DoIsStandard(const CHAR* spec,
+                  const Component& scheme,
+                  SchemeType* type) {
   if (!scheme.is_nonempty())
     return false;  // Empty or invalid schemes are non-standard.
 
   InitStandardSchemes();
   for (size_t i = 0; i < standard_schemes->size(); i++) {
     if (base::LowerCaseEqualsASCII(
             typename CharToStringPiece<CHAR>::Piece(
                 &spec[scheme.begin], scheme.len),
-            standard_schemes->at(i)))
+            standard_schemes->at(i).scheme)) {
+      *type = standard_schemes->at(i).type;
       return true;
+    }
   }
   return false;
 }
 
 template<typename CHAR>
 bool DoFindAndCompareScheme(const CHAR* str,
                             int str_len,
                             const char* compare,
                             Component* found_scheme) {
   // Before extracting scheme, canonicalize the URL to remove any whitespace.
@@ skipping 49 matching lines @@
   }
 #endif
 
   Component scheme;
   if (!ExtractScheme(spec, spec_len, &scheme))
     return false;
 
   // This is the parsed version of the input URL, we have to canonicalize it
   // before storing it in our object.
   bool success;
+  SchemeType unused_scheme_type = SCHEME_WITH_PORT;
   if (DoCompareSchemeComponent(spec, scheme, url::kFileScheme)) {
     // File URLs are special.
     ParseFileURL(spec, spec_len, &parsed_input);
     success = CanonicalizeFileURL(spec, spec_len, parsed_input,
                                   charset_converter, output, output_parsed);
   } else if (DoCompareSchemeComponent(spec, scheme, url::kFileSystemScheme)) {
     // Filesystem URLs are special.
     ParseFileSystemURL(spec, spec_len, &parsed_input);
     success = CanonicalizeFileSystemURL(spec, spec_len, parsed_input,
                                         charset_converter, output,
                                         output_parsed);
 
-  } else if (DoIsStandard(spec, scheme)) {
+  } else if (DoIsStandard(spec, scheme, &unused_scheme_type)) {
     // All "normal" URLs.
     ParseStandardURL(spec, spec_len, &parsed_input);
     success = CanonicalizeStandardURL(spec, spec_len, parsed_input,
                                       charset_converter, output, output_parsed);
 
   } else if (DoCompareSchemeComponent(spec, scheme, url::kMailToScheme)) {
     // Mailto are treated like a standard url with only a scheme, path, query
     ParseMailtoURL(spec, spec_len, &parsed_input);
     success = CanonicalizeMailtoURL(spec, spec_len, parsed_input, output,
                                     output_parsed);
@@ skipping 27 matching lines @@
   bool base_is_hierarchical = false;
   if (base_spec &&
       base_parsed.scheme.is_nonempty()) {
     int after_scheme = base_parsed.scheme.end() + 1;  // Skip past the colon.
     int num_slashes = CountConsecutiveSlashes(base_spec, after_scheme,
                                               base_spec_len);
     base_is_authority_based = num_slashes > 1;
     base_is_hierarchical = num_slashes > 0;
   }
 
+  SchemeType unused_scheme_type = SCHEME_WITH_PORT;
   bool standard_base_scheme =
       base_parsed.scheme.is_nonempty() &&
-      DoIsStandard(base_spec, base_parsed.scheme);
+      DoIsStandard(base_spec, base_parsed.scheme, &unused_scheme_type);
 
   bool is_relative;
   Component relative_component;
   if (!IsRelativeURL(base_spec, base_parsed, relative, relative_length,
                      (base_is_hierarchical || standard_base_scheme),
                      &is_relative, &relative_component)) {
     // Error resolving.
     return false;
   }
 
@@ skipping 100 matching lines @@
   // If we get here, then we know the scheme doesn't need to be replaced, so can
   // just key off the scheme in the spec to know how to do the replacements.
   if (DoCompareSchemeComponent(spec, parsed.scheme, url::kFileScheme)) {
     return ReplaceFileURL(spec, parsed, replacements, charset_converter, output,
                           out_parsed);
   }
   if (DoCompareSchemeComponent(spec, parsed.scheme, url::kFileSystemScheme)) {
     return ReplaceFileSystemURL(spec, parsed, replacements, charset_converter,
                                 output, out_parsed);
   }
-  if (DoIsStandard(spec, parsed.scheme)) {
+  SchemeType unused_scheme_type = SCHEME_WITH_PORT;
+  if (DoIsStandard(spec, parsed.scheme, &unused_scheme_type)) {
     return ReplaceStandardURL(spec, parsed, replacements, charset_converter,
                               output, out_parsed);
   }
   if (DoCompareSchemeComponent(spec, parsed.scheme, url::kMailToScheme)) {
     return ReplaceMailtoURL(spec, parsed, replacements, output, out_parsed);
   }
 
   // Default is a path URL.
   return ReplacePathURL(spec, parsed, replacements, output, out_parsed);
 }
 
 }  // namespace
 
 void Initialize() {
   InitStandardSchemes();
 }
 
 void Shutdown() {
   if (standard_schemes) {
     delete standard_schemes;
     standard_schemes = NULL;
   }
 }
 
-void AddStandardScheme(const char* new_scheme) {
+void AddStandardScheme(const char* new_scheme,
+                       SchemeType type) {
   // If this assert triggers, it means you've called AddStandardScheme after
   // LockStandardSchemes have been called (see the header file for
   // LockStandardSchemes for more).
   //
   // This normally means you're trying to set up a new standard scheme too late
   // in your application's init process. Locate where your app does this
   // initialization and calls LockStandardScheme, and add your new standard
   // scheme there.
   DCHECK(!standard_schemes_locked) <<
       "Trying to add a standard scheme after the list has been locked.";
 
   size_t scheme_len = strlen(new_scheme);
   if (scheme_len == 0)
     return;
 
   // Dulicate the scheme into a new buffer and add it to the list of standard
   // schemes. This pointer will be leaked on shutdown.
   char* dup_scheme = new char[scheme_len + 1];
   ANNOTATE_LEAKING_OBJECT_PTR(dup_scheme);
   memcpy(dup_scheme, new_scheme, scheme_len + 1);
 
   InitStandardSchemes();
-  standard_schemes->push_back(dup_scheme);
+  standard_schemes->push_back(SchemeWithType(dup_scheme, type));
 }
 
 void LockStandardSchemes() {
   standard_schemes_locked = true;
 }
 
 bool IsStandard(const char* spec, const Component& scheme) {
-  return DoIsStandard(spec, scheme);
+  SchemeType unused_scheme_type;
+  return DoIsStandard(spec, scheme, &unused_scheme_type);
+}
+
+bool GetStandardSchemeType(const char* spec,
+                           const Component& scheme,
+                           SchemeType* type) {
+  return DoIsStandard(spec, scheme, type);
 }
 
 bool IsStandard(const base::char16* spec, const Component& scheme) {
-  return DoIsStandard(spec, scheme);
+  SchemeType unused_scheme_type;
+  return DoIsStandard(spec, scheme, &unused_scheme_type);
 }
 
 bool FindAndCompareScheme(const char* str,
                           int str_len,
                           const char* compare,
                           Component* found_scheme) {
   return DoFindAndCompareScheme(str, str_len, compare, found_scheme);
 }
 
 bool FindAndCompareScheme(const base::char16* str,
@@ skipping 137 matching lines @@
   return DoCompareSchemeComponent(spec, component, compare_to);
 }
 
 bool CompareSchemeComponent(const base::char16* spec,
                             const Component& component,
                             const char* compare_to) {
   return DoCompareSchemeComponent(spec, component, compare_to);
 }
 
 }  // namespace url