protocol.cc

/* Copyright (c) 2015 Brian R. Bondy. Distributed under the MPL2 license.
 * This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

#include <ctype.h>
#include "./protocol.h"

enum ProtocolParseState {
    ProtocolParseStateStart,
    ProtocolParseStateReadingBlob,
    ProtocolParseStatePostBlob,
    ProtocolParseStateReadingProtoWebSocket,
    ProtocolParseStateReadingProtoHTTP,
    ProtocolParseStatePostProto,
    ProtocolParseStateReadingSeperator,
};

/**
 * Checks to see if a URL is "blockable".
 *
 * Blockable URLs are ones that use one of the following protocols (any of
 * which can be prefixed by "blob:")
 *  - http
 *  - https
 *  - ws
 *  - wss
 */
bool isBlockableProtocol(const char *url, int urlLen) {
  // First check to see if this is a blob URL.  If the URL is very short,
  // then trivially it isn't of the above protocols.
  if (urlLen <= 5) {
    return false;
  }

  const char *curChar = url;
  int totalCharsRead = 0;
  int numCharsReadInState;
  char lowerChar;
  ProtocolParseState parseState = ProtocolParseStateStart;

  // The below loop encodes a state machine.  Free transitions between states
  // are continues.  States that consume input "break" so that the can
  // share the common incrementing statements at the bottom of the loop.
  //
  // Its not quite as optimized as possible (some state transitions could
  // be collapsed) but its written in this _slightly_ more verbose way
  // to make it easier to grok.
  while (true) {
    switch (parseState) {
      case ProtocolParseStateStart:
        if (tolower(*curChar) == 'b') {
          parseState = ProtocolParseStateReadingBlob;
          continue;
        }
        // Intentional fall through
        [[fallthrough]];
      case ProtocolParseStatePostBlob:
        lowerChar = tolower(*curChar);
        if (lowerChar == 'w') {
          parseState = ProtocolParseStateReadingProtoWebSocket;
          continue;
        }
        if (lowerChar == 'h') {
          parseState = ProtocolParseStateReadingProtoHTTP;
          continue;
        }
        // If we're in ProtocolParseStateStart and didn't see "blob:",
        // "ws" or "http", or in ProtocolParseStatePostBlob
        // and don't see "ws" or "http" starting, then the URL doesn't match
        // any protocol we're interested in.
        return false;

      case ProtocolParseStateReadingBlob:
        if (tolower(*curChar) == 'b' &&
            tolower(*(curChar + 1)) == 'l' &&
            tolower(*(curChar + 2)) == 'o' &&
            tolower(*(curChar + 3)) == 'b' &&
            tolower(*(curChar + 4)) == ':') {
          parseState = ProtocolParseStatePostBlob;
          numCharsReadInState = 5;
          break;
        }
        // Unexpected character read when consuming "blob:"
        return false;

      case ProtocolParseStateReadingProtoHTTP:
        if (tolower(*curChar) == 'h' &&
            tolower(*(curChar + 1)) == 't' &&
            tolower(*(curChar + 2)) == 't' &&
            tolower(*(curChar + 3)) == 'p') {
          parseState = ProtocolParseStatePostProto;
          numCharsReadInState = 4;
          break;
        }
        // Unexpected character read when consuming "http"
        return false;

      case ProtocolParseStateReadingProtoWebSocket:
        if (tolower(*curChar) == 'w' &&
            tolower(*(curChar + 1)) == 's') {
          parseState = ProtocolParseStatePostProto;
          numCharsReadInState = 2;
          break;
        }
        // Unexpected character read when consuming "ws"
        return false;

      case ProtocolParseStatePostProto:
        if (tolower(*curChar) == 's') {
          parseState = ProtocolParseStateReadingSeperator;
          numCharsReadInState = 1;
          break;
        }
        [[fallthrough]];
      // Intentional fall through
      case ProtocolParseStateReadingSeperator:
        if (*curChar == ':' &&
            (*(curChar + 1)) == '/' &&
            (*(curChar + 2)) == '/') {
          return true;
        }
        // Unexpected character read when consuming "://"
        return false;
    }

    // If we've read the entire URL and we haven't been able to determine
    // the protocol, then its trivially not a blockable protocol.
    totalCharsRead += numCharsReadInState;
    if (totalCharsRead >= urlLen) {
      return false;
    }
    curChar += numCharsReadInState;
  }
}