Add MIME sniffer overloads that take base::StringPieces, plumbing

StringPieces all the way through the MIME sniffing code.

This CL only updates net/ consumers to use the overloads, which are all
test-only code. I'll update consumers and remove the old methods in
another CL.

Bug: 1123179
Change-Id: I04328913fa1c7bfd9049fa7aaae6a8a538cbe8e3
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2382896
Commit-Queue: Matt Menke <mmenke@chromium.org>
Reviewed-by: Eric Roman <eroman@chromium.org>
Cr-Commit-Position: refs/heads/master@{#803339}

net/base/mime_sniffer.cc

@@ -283,65 +283,53 @@ static const MagicNumber kSniffableTags[] = {
 
 // Compare content header to a magic number where magic_entry can contain '.'
 // for single character of anything, allowing some bytes to be skipped.
-static bool MagicCmp(const char* magic_entry, const char* content, size_t len) {
-  while (len) {
-    if ((*magic_entry != '.') && (*magic_entry != *content))
+static bool MagicCmp(base::StringPiece content, base::StringPiece magic_entry) {
+  DCHECK_GE(content.length(), magic_entry.length());
+
+  for (size_t i = 0; i < magic_entry.length(); ++i) {
+    if (magic_entry[i] != '.' && magic_entry[i] != content[i])
       return false;
-    ++magic_entry;
-    ++content;
-    --len;
   }
   return true;
 }
 
 // Like MagicCmp() except that it ANDs each byte with a mask before
 // the comparison, because there are some bits we don't care about.
-static bool MagicMaskCmp(const char* magic_entry,
-                         const char* content,
-                         size_t len,
-                         const char* mask) {
-  while (len) {
-    if ((*magic_entry != '.') && (*magic_entry != (*mask & *content)))
+static bool MagicMaskCmp(base::StringPiece content,
+                         base::StringPiece magic_entry,
+                         base::StringPiece magic_mask) {
+  DCHECK_GE(content.length(), magic_entry.length());
+
+  for (size_t i = 0; i < magic_entry.length(); ++i) {
+    if (magic_entry[i] != '.' && magic_entry[i] != (magic_mask[i] & content[i]))
       return false;
-    ++magic_entry;
-    ++content;
-    ++mask;
-    --len;
   }
   return true;
 }
 
-static bool MatchMagicNumber(const char* content,
-                             size_t size,
+static bool MatchMagicNumber(base::StringPiece content,
                              const MagicNumber& magic_entry,
                              std::string* result) {
-  const size_t len = magic_entry.magic_len;
+  const size_t magic_len = magic_entry.magic_len;
 
   // Keep kBytesRequiredForMagic honest.
-  DCHECK_LE(len, kBytesRequiredForMagic);
-
-  // To compare with magic strings, we need to compute strlen(content), but
-  // content might not actually have a null terminator.  In that case, we
-  // pretend the length is content_size.
-  const char* end = static_cast<const char*>(memchr(content, '\0', size));
-  const size_t content_strlen =
-      (end != nullptr) ? static_cast<size_t>(end - content) : size;
+  DCHECK_LE(magic_len, kBytesRequiredForMagic);
 
   bool match = false;
-  if (magic_entry.is_string) {
-    if (content_strlen >= len) {
+  base::StringPiece magic_string(magic_entry.magic, magic_len);
+  if (content.length() >= magic_len) {
+    if (magic_entry.is_string) {
+      // Consistency check - string entries should have no embedded nulls.
+      DCHECK_EQ(strlen(magic_entry.magic), magic_len);
+
       // Do a case-insensitive prefix comparison.
-      DCHECK_EQ(strlen(magic_entry.magic), len);
-      match = base::EqualsCaseInsensitiveASCII(magic_entry.magic,
-                                               base::StringPiece(content, len));
-    }
-  } else {
-    if (size >= len) {
-      if (!magic_entry.mask) {
-        match = MagicCmp(magic_entry.magic, content, len);
-      } else {
-        match = MagicMaskCmp(magic_entry.magic, content, len, magic_entry.mask);
-      }
+      match = base::StartsWith(content, magic_string,
+                               base::CompareCase::INSENSITIVE_ASCII);
+    } else if (!magic_entry.mask) {
+      match = MagicCmp(content, magic_string);
+    } else {
+      base::StringPiece magic_mask(magic_entry.mask, magic_len);
+      match = MagicMaskCmp(content, magic_string, magic_mask);
     }
   }
 
@@ -352,78 +340,69 @@ static bool MatchMagicNumber(const char* content,
   return false;
 }
 
-static bool CheckForMagicNumbers(const char* content,
-                                 size_t size,
+static bool CheckForMagicNumbers(base::StringPiece content,
                                  base::span<const MagicNumber> magic_numbers,
                                  std::string* result) {
   for (const MagicNumber& magic : magic_numbers) {
-    if (MatchMagicNumber(content, size, magic, result))
+    if (MatchMagicNumber(content, magic, result))
       return true;
   }
   return false;
 }
 
-// Truncates |size| to |max_size| and returns true if |size| is at least
-// |max_size|.
-static bool TruncateSize(const size_t max_size, size_t* size) {
+// Truncates |string_piece| to length |max_size| and returns true if
+// |string_piece| is now exactly |max_size|.
+static bool TruncateStringPiece(const size_t max_size,
+                                base::StringPiece* string_piece) {
   // Keep kMaxBytesToSniff honest.
   DCHECK_LE(static_cast<int>(max_size), kMaxBytesToSniff);
 
-  if (*size >= max_size) {
-    *size = max_size;
-    return true;
-  }
-  return false;
+  *string_piece = string_piece->substr(0, max_size);
+  return string_piece->length() == max_size;
 }
 
 // Returns true and sets result if the content appears to be HTML.
 // Clears have_enough_content if more data could possibly change the result.
-static bool SniffForHTML(const char* content,
-                         size_t size,
+static bool SniffForHTML(base::StringPiece content,
                          bool* have_enough_content,
                          std::string* result) {
   // For HTML, we are willing to consider up to 512 bytes. This may be overly
   // conservative as IE only considers 256.
-  *have_enough_content &= TruncateSize(512, &size);
+  *have_enough_content &= TruncateStringPiece(512, &content);
 
   // We adopt a strategy similar to that used by Mozilla to sniff HTML tags,
   // but with some modifications to better match the HTML5 spec.
-  const char* const end = content + size;
-  const char* pos;
-  for (pos = content; pos < end; ++pos) {
-    if (!base::IsAsciiWhitespace(*pos))
-      break;
-  }
-  // |pos| now points to first non-whitespace character (or at end).
-  return CheckForMagicNumbers(pos, end - pos, kSniffableTags, result);
+  base::StringPiece trimmed =
+      base::TrimWhitespaceASCII(content, base::TRIM_LEADING);
+
+  // |trimmed| now starts at first non-whitespace character (or is empty).
+  return CheckForMagicNumbers(trimmed, kSniffableTags, result);
 }
 
 // Returns true and sets result if the content matches any of kMagicNumbers.
 // Clears have_enough_content if more data could possibly change the result.
-static bool SniffForMagicNumbers(const char* content,
-                                 size_t size,
+static bool SniffForMagicNumbers(base::StringPiece content,
                                  bool* have_enough_content,
                                  std::string* result) {
-  *have_enough_content &= TruncateSize(kBytesRequiredForMagic, &size);
+  *have_enough_content &= TruncateStringPiece(kBytesRequiredForMagic, &content);
 
   // Check our big table of Magic Numbers
-  return CheckForMagicNumbers(content, size, kMagicNumbers, result);
+  return CheckForMagicNumbers(content, kMagicNumbers, result);
 }
 
 // Returns true and sets result if the content matches any of
 // kOfficeMagicNumbers, and the URL has the proper extension.
 // Clears |have_enough_content| if more data could possibly change the result.
-static bool SniffForOfficeDocs(const char* content,
-                               size_t size,
+static bool SniffForOfficeDocs(base::StringPiece content,
                                const GURL& url,
                                bool* have_enough_content,
                                std::string* result) {
-  *have_enough_content &= TruncateSize(kBytesRequiredForOfficeMagic, &size);
+  *have_enough_content &=
+      TruncateStringPiece(kBytesRequiredForOfficeMagic, &content);
 
   // Check our table of magic numbers for Office file types.
   std::string office_version;
-  if (!CheckForMagicNumbers(content, size, kOfficeMagicNumbers,
-                            &office_version))
+  if (!CheckForMagicNumbers(content, kOfficeMagicNumbers, &office_version))
     return false;
 
   OfficeDocType type = DOC_TYPE_NONE;
@@ -513,18 +492,16 @@ static bool IsOfficeType(const std::string& type_hint) {
 //
 // Returns false if additional data is required to determine the file type, or
 // true if there is enough data to make a decision.
-static bool SniffForInvalidOfficeDocs(const char* content,
-                                      size_t size,
+static bool SniffForInvalidOfficeDocs(base::StringPiece content,
                                       const GURL& url,
                                       std::string* result) {
-  if (!TruncateSize(kBytesRequiredForOfficeMagic, &size))
+  if (!TruncateStringPiece(kBytesRequiredForOfficeMagic, &content))
     return false;
 
   // Check our table of magic numbers for Office file types.  If it does not
   // match one, the MIME type was invalid.  Set it instead to a safe value.
   std::string office_version;
-  if (!CheckForMagicNumbers(content, size, kOfficeMagicNumbers,
-                            &office_version)) {
+  if (!CheckForMagicNumbers(content, kOfficeMagicNumbers, &office_version)) {
     *result = "application/octet-stream";
   }
 
@@ -547,29 +524,27 @@ static const MagicNumber kMagicXML[] = {
 // while HTML5 has a different recommendation -- what should we do?
 // TODO(evanm): this is incorrect for documents whose encoding isn't a superset
 // of ASCII -- do we care?
-static bool SniffXML(const char* content,
-                     size_t size,
+static bool SniffXML(base::StringPiece content,
                      bool* have_enough_content,
                      std::string* result) {
   // We allow at most 300 bytes of content before we expect the opening tag.
-  *have_enough_content &= TruncateSize(300, &size);
-  const char* pos = content;
-  const char* const end = content + size;
+  *have_enough_content &= TruncateStringPiece(300, &content);
 
   // This loop iterates through tag-looking offsets in the file.
   // We want to skip XML processing instructions (of the form "<?xml ...")
   // and stop at the first "plain" tag, then make a decision on the mime-type
   // based on the name (or possibly attributes) of that tag.
   const int kMaxTagIterations = 5;
-  for (int i = 0; i < kMaxTagIterations && pos < end; ++i) {
-    pos = reinterpret_cast<const char*>(memchr(pos, '<', end - pos));
-    if (!pos)
+  size_t pos = 0;
+  for (size_t i = 0; i < kMaxTagIterations && pos < content.length(); ++i) {
+    pos = content.find('<', pos);
+    if (pos == base::StringPiece::npos)
       return false;
 
     static constexpr base::StringPiece kXmlPrefix("<?xml");
     static constexpr base::StringPiece kDocTypePrefix("<!DOCTYPE");
 
-    base::StringPiece current(pos, end - pos);
+    base::StringPiece current = content.substr(pos);
     if (base::EqualsCaseInsensitiveASCII(current.substr(0, kXmlPrefix.size()),
                                          kXmlPrefix)) {
       // Skip XML declarations.
@@ -584,7 +559,7 @@ static bool SniffXML(const char* content,
       continue;
     }
 
-    if (CheckForMagicNumbers(pos, end - pos, kMagicXML, result))
+    if (CheckForMagicNumbers(current, kMagicXML, result))
       return true;
 
     // TODO(evanm): handle RSS 1.0, which is an RDF format and more difficult
@@ -598,7 +573,7 @@ static bool SniffXML(const char* content,
   // We iterated too far without finding a start tag.
   // If we have more content to look at, we aren't going to change our mind by
   // seeing more bytes from the network.
-  return pos < end;
+  return pos < content.length();
 }
 
 // Byte order marks
@@ -611,8 +586,7 @@ static const MagicNumber kByteOrderMark[] = {
 // Returns true and sets result to "application/octet-stream" if the content
 // appears to be binary data. Otherwise, returns false and sets "text/plain".
 // Clears have_enough_content if more data could possibly change the result.
-static bool SniffBinary(const char* content,
-                        size_t size,
+static bool SniffBinary(base::StringPiece content,
                         bool* have_enough_content,
                         std::string* result) {
   // There is no concensus about exactly how to sniff for binary content.
@@ -622,18 +596,18 @@ static bool SniffBinary(const char* content,
   // because it is small enough to comfortably fit into a single packet (after
   // allowing for headers) and yet large enough to account for binary formats
   // that have a significant amount of ASCII at the beginning (crbug.com/15314).
-  const bool is_truncated = TruncateSize(kMaxBytesToSniff, &size);
+  const bool is_truncated = TruncateStringPiece(kMaxBytesToSniff, &content);
 
   // First, we look for a BOM.
   std::string unused;
-  if (CheckForMagicNumbers(content, size, kByteOrderMark, &unused)) {
+  if (CheckForMagicNumbers(content, kByteOrderMark, &unused)) {
     // If there is BOM, we think the buffer is not binary.
     result->assign("text/plain");
     return false;
   }
 
   // Next we look to see if any of the bytes "look binary."
-  if (LooksLikeBinary(content, size)) {
+  if (LooksLikeBinary(content)) {
     result->assign("application/octet-stream");
     return true;
   }
@@ -646,7 +620,7 @@ static bool SniffBinary(const char* content,
   return false;
 }
 
-static bool IsUnknownMimeType(const std::string& mime_type) {
+static bool IsUnknownMimeType(base::StringPiece mime_type) {
   // TODO(tc): Maybe reuse some code in net/http/http_response_headers.* here.
   // If we do, please be careful not to alter the semantics at all.
   static const char* const kUnknownMimeTypes[] = {
@@ -663,7 +637,7 @@ static bool IsUnknownMimeType(const std::string& mime_type) {
     if (mime_type == unknown_mime_type)
       return true;
   }
-  if (mime_type.find('/') == std::string::npos) {
+  if (mime_type.find('/') == base::StringPiece::npos) {
     // Firefox rejects a mime type if it does not contain a slash
     return true;
   }
@@ -673,8 +647,7 @@ static bool IsUnknownMimeType(const std::string& mime_type) {
 // Returns true and sets result if the content appears to be a crx (Chrome
 // extension) file.
 // Clears have_enough_content if more data could possibly change the result.
-static bool SniffCRX(const char* content,
-                     size_t size,
+static bool SniffCRX(base::StringPiece content,
                      const GURL& url,
                      bool* have_enough_content,
                      std::string* result) {
@@ -690,11 +663,11 @@ static bool SniffCRX(const char* content,
   if (!base::EndsWith(url.path_piece(), ".crx", base::CompareCase::SENSITIVE))
     return false;
 
-  *have_enough_content &= TruncateSize(kBytesRequiredForMagic, &size);
-  return CheckForMagicNumbers(content, size, kCRXMagicNumbers, result);
+  *have_enough_content &= TruncateStringPiece(kBytesRequiredForMagic, &content);
+  return CheckForMagicNumbers(content, kCRXMagicNumbers, result);
 }
 
-bool ShouldSniffMimeType(const GURL& url, const std::string& mime_type) {
+bool ShouldSniffMimeType(const GURL& url, base::StringPiece mime_type) {
   bool sniffable_scheme = url.is_empty() ||
                           url.SchemeIsHTTPOrHTTPS() ||
 #if defined(OS_ANDROID)
@@ -744,14 +717,13 @@ bool ShouldSniffMimeType(const GURL& url, const std::string& mime_type) {
   return false;
 }
 
-bool SniffMimeType(const char* content,
-                   size_t content_size,
+bool SniffMimeType(base::StringPiece content,
                    const GURL& url,
                    const std::string& type_hint,
                    ForceSniffFileUrlsForHtml force_sniff_file_url_for_html,
                    std::string* result) {
-  DCHECK_LT(content_size, 1000000U);  // sanity check
-  DCHECK(content);
+  // Sanity check.
+  DCHECK_LT(content.length(), 1000000U);
   DCHECK(result);
 
   // By default, we assume we have enough content.
@@ -766,7 +738,7 @@ bool SniffMimeType(const char* content,
   // is a valid Office file.  Because this is the only reason we sniff files
   // with a Microsoft Office MIME type, we can return early.
   if (IsOfficeType(type_hint))
-    return SniffForInvalidOfficeDocs(content, content_size, url, result);
+    return SniffForInvalidOfficeDocs(content, url, result);
 
   // Cache information about the type_hint
   bool hint_is_unknown_mime_type = IsUnknownMimeType(type_hint);
@@ -779,7 +751,7 @@ bool SniffMimeType(const char* content,
     // We're only willing to sniff HTML if the server has not supplied a mime
     // type, or if the type it did supply indicates that it doesn't know what
     // the type should be.
-    if (SniffForHTML(content, content_size, &have_enough_content, result))
+    if (SniffForHTML(content, &have_enough_content, result))
       return true;  // We succeeded in sniffing HTML.  No more content needed.
   }
 
@@ -791,7 +763,7 @@ bool SniffMimeType(const char* content,
   //    could be indicative of a mis-configuration that we shield the user from.
   const bool hint_is_text_plain = (type_hint == "text/plain");
   if (hint_is_unknown_mime_type || hint_is_text_plain) {
-    if (!SniffBinary(content, content_size, &have_enough_content, result)) {
+    if (!SniffBinary(content, &have_enough_content, result)) {
       // If the server said the content was text/plain and it doesn't appear
       // to be binary, then we trust it.
       if (hint_is_text_plain) {
@@ -805,22 +777,22 @@ bool SniffMimeType(const char* content,
     // We're not interested in sniffing these types for images and the like.
     // Instead, we're looking explicitly for a feed.  If we don't find one
     // we're done and return early.
-    if (SniffXML(content, content_size, &have_enough_content, result))
+    if (SniffXML(content, &have_enough_content, result))
       return true;
     return have_enough_content;
   }
 
   // CRX files (Chrome extensions) have a special sniffing algorithm. It is
   // tighter than the others because we don't have to match legacy behavior.
-  if (SniffCRX(content, content_size, url, &have_enough_content, result))
+  if (SniffCRX(content, url, &have_enough_content, result))
     return true;
 
   // Check the file extension and magic numbers to see if this is an Office
   // document.  This needs to be checked before the general magic numbers
   // because zip files and Office documents (OOXML) have the same magic number.
-  if (SniffForOfficeDocs(content, content_size, url,
-                         &have_enough_content, result))
+  if (SniffForOfficeDocs(content, url, &have_enough_content, result)) {
     return true;  // We've matched a magic number.  No more content needed.
+  }
 
   // We're not interested in sniffing for magic numbers when the type_hint
   // is application/octet-stream.  Time to bail out.
@@ -829,24 +801,38 @@ bool SniffMimeType(const char* content,
 
   // Now we look in our large table of magic numbers to see if we can find
   // anything that matches the content.
-  if (SniffForMagicNumbers(content, content_size,
-                           &have_enough_content, result))
+  if (SniffForMagicNumbers(content, &have_enough_content, result))
     return true;  // We've matched a magic number.  No more content needed.
 
   return have_enough_content;
 }
 
-bool SniffMimeTypeFromLocalData(const char* content,
-                                size_t size,
-                                std::string* result) {
+bool SniffMimeType(const char* content,
+                   size_t content_size,
+                   const GURL& url,
+                   const std::string& type_hint,
+                   ForceSniffFileUrlsForHtml force_sniff_file_url_for_html,
+                   std::string* result) {
+  return SniffMimeType(base::StringPiece(content, content_size), url, type_hint,
+                       force_sniff_file_url_for_html, result);
+}
+
+NET_EXPORT bool SniffMimeTypeFromLocalData(base::StringPiece content,
+                                           std::string* result) {
   // First check the extra table.
-  if (CheckForMagicNumbers(content, size, kExtraMagicNumbers, result))
+  if (CheckForMagicNumbers(content, kExtraMagicNumbers, result))
     return true;
   // Finally check the original table.
-  return CheckForMagicNumbers(content, size, kMagicNumbers, result);
+  return CheckForMagicNumbers(content, kMagicNumbers, result);
+}
+
+bool SniffMimeTypeFromLocalData(const char* content,
+                                size_t size,
+                                std::string* result) {
+  return SniffMimeTypeFromLocalData(base::StringPiece(content, size), result);
 }
 
-bool LooksLikeBinary(const char* content, size_t size) {
+bool LooksLikeBinary(base::StringPiece content) {
   // The definition of "binary bytes" is from the spec at
   // https://mimesniff.spec.whatwg.org/#binary-data-byte
   //
@@ -856,7 +842,7 @@ bool LooksLikeBinary(const char* content, size_t size) {
   // represents byte 0x1F.
   const uint32_t kBinaryBits =
       ~(1u << '\t' | 1u << '\n' | 1u << '\r' | 1u << '\f' | 1u << '\x1b');
-  for (size_t i = 0; i < size; ++i) {
+  for (size_t i = 0; i < content.length(); ++i) {
     uint8_t byte = static_cast<uint8_t>(content[i]);
     if (byte < 0x20 && (kBinaryBits & (1u << byte)))
       return true;

net/base/mime_sniffer.h

@@ -9,6 +9,7 @@
 
 #include <string>
 
+#include "base/strings/string_piece.h"
 #include "net/base/net_export.h"
 
 class GURL;
@@ -35,13 +36,12 @@ enum class ForceSniffFileUrlsForHtml {
 // |mime_type| is the current mime type, e.g. from the Content-Type header.
 // Returns true if the mime type should be sniffed.
 NET_EXPORT bool ShouldSniffMimeType(const GURL& url,
-                                    const std::string& mime_type);
+                                    base::StringPiece mime_type);
 
 // Guess a mime type from the first few bytes of content an its URL.  Always
 // assigns |result| with its best guess of a mime type.
 //
-// |content| is the buffer containing the bytes to sniff.
-// |content_size| is the number of bytes in the |content| buffer.
+// |content| contains the bytes to sniff.
 // |url| is the URL from which the content was obtained.
 // |type_hint| is the current mime type, e.g. from the Content-Type header.
 // |result| is the address at which to place the sniffed mime type.
@@ -52,6 +52,16 @@ NET_EXPORT bool ShouldSniffMimeType(const GURL& url,
 // Returns true if |content| had enough data to guess the mime type. Otherwise,
 // |result| will be populated with a putative MIME type, but the method should
 // be called again with more of the content.
+NET_EXPORT bool SniffMimeType(
+    base::StringPiece content,
+    const GURL& url,
+    const std::string& type_hint,
+    ForceSniffFileUrlsForHtml force_sniff_file_url_for_html,
+    std::string* result);
+
+// Older deprecated version of the above function.
+//
+// TODO(https://crbug.com/1123179): Update callers and remove this method.
 NET_EXPORT bool SniffMimeType(
     const char* content,
     size_t content_size,
@@ -68,20 +78,22 @@ NET_EXPORT bool SniffMimeType(
 // The caller should understand the security ramifications of trusting
 // uncontrolled data before accepting the results of this function.
 //
-// @param content A buffer containing the bytes to sniff.
-// @param content_size The number of bytes in the |content| buffer.
-// @param result Address at which to place the sniffed mime type.
-// @return Returns true if a MIME type match was found.
+// |content| contains the bytes to sniff.
+// |result| is address at which to place the sniffed mime type.
+// Returns true if a MIME type match was found.
+NET_EXPORT bool SniffMimeTypeFromLocalData(base::StringPiece content,
+                                           std::string* result);
+
+// Older deprecated version of the above function.
+//
+// TODO(https://crbug.com/1123179): Update callers and remove this method.
 NET_EXPORT bool SniffMimeTypeFromLocalData(const char* content,
                                            size_t content_size,
                                            std::string* result);
 
 // Returns true if |content| contains bytes that are control codes that do
 // not usually appear in plain text.
-// @param content A buffer contains bytes that may be binary.
-// @param size    The number of bytes in the |content| buffer.
-// @return Returns true if |content| looks like binary.
-NET_EXPORT_PRIVATE bool LooksLikeBinary(const char* content, size_t size);
+NET_EXPORT_PRIVATE bool LooksLikeBinary(base::StringPiece content);
 
 }  // namespace net
 

net/base/mime_sniffer_fuzzer.cc

@@ -44,10 +44,10 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   std::string input = data_provider.ConsumeRemainingBytesAsString();
 
   std::string result;
-  net::SniffMimeType(input.data(), input.length(), GURL(url_string),
-                     mime_type_hint, force_sniff_file_urls_for_html, &result);
+  net::SniffMimeType(input, GURL(url_string), mime_type_hint,
+                     force_sniff_file_urls_for_html, &result);
 
-  net::SniffMimeTypeFromLocalData(input.data(), input.length(), &result);
+  net::SniffMimeTypeFromLocalData(input, &result);
 
   return 0;
 }

net/base/mime_sniffer_perftest.cc

@@ -72,7 +72,7 @@ const char kRepresentativePlainText[] =
 void RunLooksLikeBinary(const std::string& plaintext, size_t iterations) {
   bool looks_like_binary = false;
   for (size_t i = 0; i < iterations; ++i) {
-    if (LooksLikeBinary(plaintext.data(), plaintext.size()))
+    if (LooksLikeBinary(plaintext))
       looks_like_binary = true;
   }
   CHECK(!looks_like_binary);

net/base/mime_sniffer_unittest.cc

@@ -24,11 +24,11 @@ std::string MakeConstantString(const char (&str)[N]) {
   return std::string(str, N - 1);
 }
 
-static std::string SniffMimeType(const std::string& content,
+static std::string SniffMimeType(base::StringPiece content,
                                  const std::string& url,
                                  const std::string& mime_type_hint) {
   std::string mime_type;
-  SniffMimeType(content.data(), content.size(), GURL(url), mime_type_hint,
+  SniffMimeType(content, GURL(url), mime_type_hint,
                 ForceSniffFileUrlsForHtml::kDisabled, &mime_type);
   return mime_type;
 }
@@ -74,14 +74,14 @@ TEST(MimeSnifferTest, BoundaryConditionsTest) {
 
   GURL url;
 
-  SniffMimeType(buf, 0, url, type_hint, ForceSniffFileUrlsForHtml::kDisabled,
-                &mime_type);
+  SniffMimeType(base::StringPiece(), url, type_hint,
+                ForceSniffFileUrlsForHtml::kDisabled, &mime_type);
   EXPECT_EQ("text/plain", mime_type);
-  SniffMimeType(buf, 1, url, type_hint, ForceSniffFileUrlsForHtml::kDisabled,
-                &mime_type);
+  SniffMimeType(base::StringPiece(buf, 1), url, type_hint,
+                ForceSniffFileUrlsForHtml::kDisabled, &mime_type);
   EXPECT_EQ("text/plain", mime_type);
-  SniffMimeType(buf, 2, url, type_hint, ForceSniffFileUrlsForHtml::kDisabled,
-                &mime_type);
+  SniffMimeType(base::StringPiece(buf, 2), url, type_hint,
+                ForceSniffFileUrlsForHtml::kDisabled, &mime_type);
   EXPECT_EQ("application/octet-stream", mime_type);
 }
 
@@ -256,11 +256,11 @@ TEST(MimeSnifferTest, SniffFilesAsHtml) {
   const GURL kUrl("file:///C/test.unusualextension");
 
   std::string mime_type;
-  SniffMimeType(kContent.c_str(), kContent.length(), kUrl, "" /* type_hint */,
+  SniffMimeType(kContent, kUrl, "" /* type_hint */,
                 ForceSniffFileUrlsForHtml::kDisabled, &mime_type);
   EXPECT_EQ("text/plain", mime_type);
 
-  SniffMimeType(kContent.c_str(), kContent.length(), kUrl, "" /* type_hint */,
+  SniffMimeType(kContent, kUrl, "" /* type_hint */,
                 ForceSniffFileUrlsForHtml::kEnabled, &mime_type);
   EXPECT_EQ("text/html", mime_type);
 }
@@ -371,7 +371,7 @@ TEST(MimeSnifferTest, XMLTestLargeNoAngledBracket) {
 
   // content.size() >= 1024 so the sniff is unambiguous.
   std::string mime_type;
-  EXPECT_TRUE(SniffMimeType(content.data(), content.size(), GURL(), "text/xml",
+  EXPECT_TRUE(SniffMimeType(content, GURL(), "text/xml",
                             ForceSniffFileUrlsForHtml::kDisabled, &mime_type));
   EXPECT_EQ("text/xml", mime_type);
 }
@@ -387,9 +387,8 @@ TEST(MimeSnifferTest, LooksBinary) {
 
   // content.size() >= 1024 so the sniff is unambiguous.
   std::string mime_type;
-  EXPECT_TRUE(SniffMimeType(content.data(), content.size(), GURL(),
-                            "text/plain", ForceSniffFileUrlsForHtml::kDisabled,
-                            &mime_type));
+  EXPECT_TRUE(SniffMimeType(content, GURL(), "text/plain",
+                            ForceSniffFileUrlsForHtml::kDisabled, &mime_type));
   EXPECT_EQ("application/octet-stream", mime_type);
 }
 
@@ -450,27 +449,27 @@ TEST(MimeSnifferTest, OfficeTest) {
 TEST(MimeSnifferTest, AudioVideoTest) {
   std::string mime_type;
   const char kOggTestData[] = "OggS\x00";
-  EXPECT_TRUE(SniffMimeTypeFromLocalData(kOggTestData, sizeof(kOggTestData) - 1,
-                                         &mime_type));
+  EXPECT_TRUE(SniffMimeTypeFromLocalData(
+      base::StringPiece(kOggTestData, sizeof(kOggTestData) - 1), &mime_type));
   EXPECT_EQ("audio/ogg", mime_type);
   mime_type.clear();
   // Check ogg header requires the terminal '\0' to be sniffed.
   EXPECT_FALSE(SniffMimeTypeFromLocalData(
-      kOggTestData, sizeof(kOggTestData) - 2, &mime_type));
+      base::StringPiece(kOggTestData, sizeof(kOggTestData) - 2), &mime_type));
   EXPECT_EQ("", mime_type);
   mime_type.clear();
 
   const char kFlacTestData[] =
       "fLaC\x00\x00\x00\x22\x12\x00\x12\x00\x00\x00\x00\x00";
   EXPECT_TRUE(SniffMimeTypeFromLocalData(
-      kFlacTestData, sizeof(kFlacTestData) - 1, &mime_type));
+      base::StringPiece(kFlacTestData, sizeof(kFlacTestData) - 1), &mime_type));
   EXPECT_EQ("audio/x-flac", mime_type);
   mime_type.clear();
 
   const char kWMATestData[] =
       "\x30\x26\xb2\x75\x8e\x66\xcf\x11\xa6\xd9\x00\xaa\x00\x62\xce\x6c";
-  EXPECT_TRUE(SniffMimeTypeFromLocalData(kWMATestData, sizeof(kWMATestData) - 1,
-                                         &mime_type));
+  EXPECT_TRUE(SniffMimeTypeFromLocalData(
+      base::StringPiece(kWMATestData, sizeof(kWMATestData) - 1), &mime_type));
   EXPECT_EQ("video/x-ms-asf", mime_type);
   mime_type.clear();
 
@@ -478,22 +477,22 @@ TEST(MimeSnifferTest, AudioVideoTest) {
   // format.
   const char kMP4TestData[] =
       "\x00\x00\x00\x20\x66\x74\x79\x70\x4d\x34\x41\x20\x00\x00\x00\x00";
-  EXPECT_TRUE(SniffMimeTypeFromLocalData(kMP4TestData, sizeof(kMP4TestData) - 1,
-                                         &mime_type));
+  EXPECT_TRUE(SniffMimeTypeFromLocalData(
+      base::StringPiece(kMP4TestData, sizeof(kMP4TestData) - 1), &mime_type));
   EXPECT_EQ("video/mp4", mime_type);
   mime_type.clear();
 
   const char kAACTestData[] =
       "\xff\xf1\x50\x80\x02\x20\xb0\x23\x0a\x83\x20\x7d\x61\x90\x3e\xb1";
-  EXPECT_TRUE(SniffMimeTypeFromLocalData(kAACTestData, sizeof(kAACTestData) - 1,
-                                         &mime_type));
+  EXPECT_TRUE(SniffMimeTypeFromLocalData(
+      base::StringPiece(kAACTestData, sizeof(kAACTestData) - 1), &mime_type));
   EXPECT_EQ("audio/mpeg", mime_type);
   mime_type.clear();
 
   const char kAMRTestData[] =
       "\x23\x21\x41\x4d\x52\x0a\x3c\x53\x0a\x7c\xe8\xb8\x41\xa5\x80\xca";
-  EXPECT_TRUE(SniffMimeTypeFromLocalData(kAMRTestData, sizeof(kAMRTestData) - 1,
-                                         &mime_type));
+  EXPECT_TRUE(SniffMimeTypeFromLocalData(
+      base::StringPiece(kAMRTestData, sizeof(kAMRTestData) - 1), &mime_type));
   EXPECT_EQ("audio/amr", mime_type);
   mime_type.clear();
 }
@@ -502,19 +501,22 @@ TEST(MimeSnifferTest, ImageTest) {
   std::string mime_type;
   const char kWebPSimpleFormat[] = "RIFF\xee\x81\x00\x00WEBPVP8 ";
   EXPECT_TRUE(SniffMimeTypeFromLocalData(
-      kWebPSimpleFormat, sizeof(kWebPSimpleFormat) - 1, &mime_type));
+      base::StringPiece(kWebPSimpleFormat, sizeof(kWebPSimpleFormat) - 1),
+      &mime_type));
   EXPECT_EQ("image/webp", mime_type);
   mime_type.clear();
 
   const char kWebPLosslessFormat[] = "RIFF\xee\x81\x00\x00WEBPVP8L";
   EXPECT_TRUE(SniffMimeTypeFromLocalData(
-      kWebPLosslessFormat, sizeof(kWebPLosslessFormat) - 1, &mime_type));
+      base::StringPiece(kWebPLosslessFormat, sizeof(kWebPLosslessFormat) - 1),
+      &mime_type));
   EXPECT_EQ("image/webp", mime_type);
   mime_type.clear();
 
   const char kWebPExtendedFormat[] = "RIFF\xee\x81\x00\x00WEBPVP8X";
   EXPECT_TRUE(SniffMimeTypeFromLocalData(
-      kWebPExtendedFormat, sizeof(kWebPExtendedFormat) - 1, &mime_type));
+      base::StringPiece(kWebPExtendedFormat, sizeof(kWebPExtendedFormat) - 1),
+      &mime_type));
   EXPECT_EQ("image/webp", mime_type);
   mime_type.clear();
 }
@@ -529,8 +531,8 @@ class MimeSnifferBinaryTest : public ::testing::TestWithParam<int> {};
 // 0x0B (VT), a byte in the range 0x0E to 0x1A (SO to SUB), or a byte in the
 // range 0x1C to 0x1F (FS to US).
 TEST_P(MimeSnifferBinaryTest, IsBinaryControlCode) {
-  char param = static_cast<char>(GetParam());
-  EXPECT_TRUE(LooksLikeBinary(&param, 1));
+  std::string param(1, static_cast<char>(GetParam()));
+  EXPECT_TRUE(LooksLikeBinary(param));
 }
 
 // ::testing::Range(a, b) tests an open-ended range, ie. "b" is not included.
@@ -553,8 +555,8 @@ INSTANTIATE_TEST_SUITE_P(MimeSnifferBinaryTestRange3,
 class MimeSnifferPlainTextTest : public ::testing::TestWithParam<int> {};
 
 TEST_P(MimeSnifferPlainTextTest, NotBinaryControlCode) {
-  char param = static_cast<char>(GetParam());
-  EXPECT_FALSE(LooksLikeBinary(&param, 1));
+  std::string param(1, static_cast<char>(GetParam()));
+  EXPECT_FALSE(LooksLikeBinary(param));
 }
 
 INSTANTIATE_TEST_SUITE_P(MimeSnifferPlainTextTestPlainTextControlCodes,
@@ -569,8 +571,7 @@ class MimeSnifferControlCodesEdgeCaseTest
     : public ::testing::TestWithParam<const char*> {};
 
 TEST_P(MimeSnifferControlCodesEdgeCaseTest, EdgeCase) {
-  const char* param = GetParam();
-  EXPECT_TRUE(LooksLikeBinary(param, strlen(param)));
+  EXPECT_TRUE(LooksLikeBinary(GetParam()));
 }
 
 INSTANTIATE_TEST_SUITE_P(MimeSnifferControlCodesEdgeCaseTest,