Add Platform Independent Eddystone Encoder

components/physical_web/eddystone/eddystone_encoder.cc

+// Copyright 2017 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 "components/physical_web/eddystone/eddystone_encoder.h"
+
+#include <string>
+#include <vector>
+
+#include "url/gurl.h"
+
+namespace {
+const char* kPrefixes[] = {"http://www.", "https://www.", "http://",
+                           "https://"};
+const size_t kNumPrefixes = sizeof(kPrefixes) / sizeof(char*);
+
+const char* kSuffixes[] = {".com/", ".org/", ".edu/", ".net/", ".info/",
+                           ".biz/", ".gov/", ".com",  ".org",  ".edu",
+                           ".net",  ".info", ".biz",  ".gov"};
+const size_t kNumSuffixes = sizeof(kSuffixes) / sizeof(char*);
+}
+
+namespace physical_web {
+int EncodeUrl(const std::string& url, std::vector<uint8_t>* encoded_url) {
+  /*
+   * Sanitize input.
+   */
+
+  if (encoded_url == NULL)

[agrieve, 2017/03/14 00:30:15]

nit: Should be using nullptr rather than NULL for new code.

As an aside though, is this check useful? Would an assert be more appropriate
here (e.g. would encoded_url == NULL imply a programming error?)

[iankc, 2017/03/14 17:42:15]

Not necessarily. We're writing this for a particular use case now, where it
would imply a program error. However, we're trying to make this extensible and
we can envision use cases in the future where this would be legal.

+    return kNullEncodedUrl;
+  if (url.empty())
+    return kEmptyUrl;
+
+  GURL gurl(url);
+  if (!gurl.is_valid())
+    return kInvalidUrl;
+  if (gurl.HostIsIPAddress() || !gurl.SchemeIsHTTPOrHTTPS())
+    return kInvalidFormat;
+
+  /*
+   * Necessary Declarations.
+   */
+  const char* raw_url_position = url.c_str();
+
+  static bool lengths_loaded = false;
+  static size_t prefix_lengths[kNumPrefixes];
+  static size_t suffix_lengths[kNumSuffixes];
+
+  if (!lengths_loaded) {

[agrieve, 2017/03/14 00:30:15]

Did you measure that this was necessary? The strings here are extremely small,
so I'd guess this caching is just using up memory unnecessarily.

[iankc, 2017/03/14 17:42:15]

Done.

+    for (size_t i = 0; i < kNumPrefixes; i++) {
+      prefix_lengths[i] = strlen(kPrefixes[i]);
+    }
+    for (size_t i = 0; i < kNumSuffixes; i++) {
+      suffix_lengths[i] = strlen(kSuffixes[i]);
+    }
+    lengths_loaded = true;
+  }
+
+  /*
+   * Handle prefix.
+   */
+  for (size_t i = 0; i < kNumPrefixes; i++) {
+    size_t prefix_len = prefix_lengths[i];
+    if (strncmp(raw_url_position, kPrefixes[i], prefix_len) == 0) {
+      encoded_url->push_back(i);
+      raw_url_position += prefix_len;
+      break;
+    }
+  }
+
+  /*
+   * Handle suffixes.
+   */
+  while (*raw_url_position) {
+    /* check for suffix match */
+    size_t i;
+    for (i = 0; i < kNumSuffixes; i++) {
+      size_t suffix_len = suffix_lengths[i];
+      if (strncmp(raw_url_position, kSuffixes[i], suffix_len) == 0) {
+        encoded_url->push_back(i);
+        raw_url_position += suffix_len;
+        break; /* from the for loop for checking against suffixes */
+      }
+    }
+    /* This is the default case where we've got an ordinary character which
+     * doesn't match a suffix. */
+    if (i == kNumSuffixes) {
+      encoded_url->push_back(*raw_url_position++);

[agrieve, 2017/03/14 00:30:15]

nit: this would be more readable to put on two lines (one for the dereference,
one for the increment)

[iankc, 2017/03/14 17:42:16]

Done.

+    }
+  }
+
+  return encoded_url->size();
+}
+}