Add read-only access to POST data for webRequest's onBeforeRequest

chrome/browser/extensions/api/web_request/post_data_parser.cc

+// Copyright (c) 2012 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/extensions/api/web_request/post_data_parser.h"
+
+#include "base/values.h"
+#include "net/base/escape.h"
+#include "net/base/upload_data.h"
+#include "net/url_request/url_request.h"
+
+namespace {
+const char kContentDisposition[] = "Content-Disposition:";
+
+// Takes |dictionary| of <string, list of strings> pairs, and gets the list
+// for |key|, creating it if necessary.
+ListValue* GetOrCreateList(DictionaryValue* dictionary,
+                           const std::string& key) {
+  ListValue* list = NULL;
+  if (!dictionary->GetList(key, &list)) {
+    list = new ListValue();
+    dictionary->Set(key, list);
+  }
+  return list;
+}
+}
+
+namespace extensions {
+
+// Implementation of PostDataParser and PostDataParser::Result .
+
+PostDataParser::Result::Result() {}
+PostDataParser::Result::~Result() {}
+
+void PostDataParser::Result::Reset() {
+  key_.erase();
+  val_.erase();
+}
+
+void PostDataParser::Result::SetKey(const base::StringPiece& str) {
+  key_.replace(0, std::string::npos, str.data(), str.size());

[Matt Perry, 2012/07/31 09:32:14]

can also do: key_ = str.as_string();

[vabr (Chromium), 2012/07/31 12:17:58]

Yes, but then the bytes would be copied twice, first during making a temporary
std::string in as_string(), and then in the assignment to key_.
The way I do it now it is only copied once.
It is no big deal, but unless you don't really hate the key_.replace, I would
leave it with just one pass at copying.

[Matt Perry, 2012/07/31 12:24:28]

Compilers will likely optimize it as a single copy:
http://en.wikipedia.org/wiki/Return_value_optimization

[vabr (Chromium), 2012/07/31 12:55:27]

Thanks for the interesting link about the copy-con optimization!

However, this would be assignment, and I don't see the compiler optimizing that:
1. StringPiece only has a char* pointerand length, and needs to create a
separate string, this is certainly one copy.
2. The string returned from StringPiece needs to be assigned to key_ (or
value_). This means the target string overwrites (and re-allocates, if
necessary) its buffer using the source string data, i.e., another copy.

I doubt there is an option to squash constructing a new string with assigning to
an old one into a single copy operation. At least this is not the cited
optimizing of multiple copy-cons (no copy-cons involved).

+}
+
+void PostDataParser::Result::SetVal(const base::StringPiece& str) {
+  val_.replace(0, std::string::npos, str.data(), str.size());
+}
+
+void PostDataParser::Result::SetKey(const std::string& str) {
+  key_ = str;
+}
+
+void PostDataParser::Result::SetVal(const std::string& str) {
+  val_ = str;
+}
+
+PostDataParser::~PostDataParser() {}
+
+// static
+scoped_ptr<PostDataParser> PostDataParser::CreatePostDataParser(
+    const net::URLRequest* request) {
+  std::string value;
+  const bool found = request->extra_request_headers().GetHeader(
+      net::HttpRequestHeaders::kContentType, &value);
+  return CreatePostDataParser(found ? &value : NULL);
+}
+
+// static
+scoped_ptr<PostDataParser> PostDataParser::CreatePostDataParser(
+    const std::string* content_type_header) {
+  enum ParserChoice {kUrlEncoded, kMultipart, kError};
+  ParserChoice choice = kError;
+  std::string boundary;
+
+  if (content_type_header == NULL) {
+    choice = kUrlEncoded;
+  } else {
+    const std::string content_type(
+        content_type_header->substr(0, content_type_header->find(';')));
+    if (content_type == "application/x-www-form-urlencoded") {
+      choice = kUrlEncoded;
+    } else if (content_type == "multipart/form-data") {
+      const char kBoundaryString[] = "boundary=";
+      size_t offset = content_type_header->find(kBoundaryString);
+      if (offset == std::string::npos) {
+        // Malformed header.
+        return scoped_ptr<PostDataParser>();
+      }
+      offset += strlen(kBoundaryString);
+      boundary = content_type_header->substr(
+          offset, content_type_header->find(';', offset));
+      choice = kMultipart;
+    }
+  }
+  // Other cases are unparseable, including when |content_type| is "text/plain".
+
+  switch (choice) {
+    case kUrlEncoded:
+      return scoped_ptr<PostDataParser>(new PostDataParserUrlEncoded());
+    case kMultipart:
+      return scoped_ptr<PostDataParser>(new PostDataParserMultipart(boundary));
+    default:  // In other words, case kError:
+      return scoped_ptr<PostDataParser>();
+  }
+}
+
+// static
+scoped_ptr<base::DictionaryValue> PostDataParser::ParseURLRequestData(
+    const net::URLRequest* request) {
+  if (request->method() != "POST")
+    return scoped_ptr<base::DictionaryValue>();
+  const std::vector<net::UploadData::Element>* elements =
+      request->get_upload()->elements();
+  scoped_ptr<PostDataParser> parser = CreatePostDataParser(request);
+  if (parser.get() == NULL) {
+    // No parser means most probably unsupported form encoding.
+    return scoped_ptr<base::DictionaryValue>();
+  }
+  scoped_ptr<base::DictionaryValue> form_data(new base::DictionaryValue);
+  std::vector<net::UploadData::Element>::const_iterator element;
+  bool data_valid = true;
+  for (element = elements->begin();
+       data_valid && element != elements->end(); ++element) {
+    if (element->type() != net::UploadData::TYPE_BYTES) {
+      // We do not handle data including blobs or chunks.
+      if (element->type() != net::UploadData::TYPE_FILE)
+        data_valid = false;
+      continue;
+    }
+    if (!parser->SetSource(&(element->bytes())))
+      continue;
+    Result result;
+    while (parser->GetNextPair(&result)) {
+      GetOrCreateList(form_data.get(), result.key())->Append(
+          new StringValue(result.val()));
+    }
+  }
+  if (data_valid && parser->AllDataReadOK())
+    return form_data.Pass();
+  else
+    return scoped_ptr<base::DictionaryValue>();
+}
+
+// Implementation of PostDataParserUrlEncoded.
+
+PostDataParserUrlEncoded::PostDataParserUrlEncoded() : source_(NULL) {}
+
+PostDataParserUrlEncoded::~PostDataParserUrlEncoded() {}
+
+bool PostDataParserUrlEncoded::AllDataReadOK() {
+  return source_ != NULL && offset_ == source_->end();
+}
+
+bool PostDataParserUrlEncoded::GetNextPair(Result* result) {
+  result->Reset();
+  if (source_ == NULL)
+    return false;
+  if (offset_ == source_->end())
+    return false;
+  std::vector<char>::const_iterator seek = offset_;
+  // (*) Now we have |seek| >= |offset_| until the end of this function:
+  while (seek != source_->end() && *seek != '=')
+    ++seek;
+  if (seek == source_->end()) {
+    // This means the data is malformed.
+    offset_ = seek;
+    return false;
+  }
+  std::string encoded_key(&(*offset_), seek - offset_);  // Safe, see (*).
+  const net::UnescapeRule::Type unescape_rules =
+      net::UnescapeRule::URL_SPECIAL_CHARS | net::UnescapeRule::CONTROL_CHARS |
+      net::UnescapeRule::SPACES | net::UnescapeRule::REPLACE_PLUS_WITH_SPACE;
+  result->SetKey(net::UnescapeURLComponent(encoded_key, unescape_rules));
+  offset_ = ++seek;
+  while (seek != source_->end() && *seek != '&')
+    ++seek;
+  std::string encoded_val(&(*offset_), seek - offset_);  // Safe, see (*).
+  result->SetVal(net::UnescapeURLComponent(encoded_val, unescape_rules));
+  offset_ = (seek == source_->end()) ? seek : seek + 1;
+  return true;
+}
+
+bool PostDataParserUrlEncoded::SetSource(const std::vector<char>* source) {
+  if (source_ != NULL)
+    return false;
+  source_ = source;
+  offset_ = source_->begin();
+  return true;
+}
+
+// Implementation of PostDataParserMultipart.
+
+PostDataParserMultipart::PostDataParserMultipart(
+    const std::string& boundary_separator)
+    : source_(NULL),
+      length_(0),  // Dummy value.
+      line_start_(0),  // Dummy value.
+      line_end_(0),  // Dummy value.
+      next_line_(0),  // Dummy value.
+      boundary_("--" + boundary_separator),
+      final_boundary_(boundary_ + "--"),
+      state_(kInit),
+      line_type_(kEmpty)  // Dummy value.
+{}
+
+PostDataParserMultipart::~PostDataParserMultipart() {}
+
+bool PostDataParserMultipart::AllDataReadOK() {
+  return source_ != NULL && next_line_ >= length_ && state_ == kFinal;
+}
+
+// This function reads one block of the data, between two boundaries.
+// First it reads the header to learn the key, and possibly also the
+// value, if this block is for a file input element.
+// Otherwise it then reads the value from the body.
+bool PostDataParserMultipart::GetNextPair(Result* result) {
+  result->Reset();
+  if (state_ == kError)
+    return false;
+  while (state_ != kHeadRead) {
+    if (!DoStep())
+      return false;
+  }
+  bool val_extracted = false;
+  bool name_parsed = ParseHead(result, &val_extracted);
+  while (state_ != kBody) {
+    if (!DoStep())
+      return false;
+  }
+  size_t val_start;
+  size_t val_end = 0;  // Dummy value, replaced below, see (*).
+  // There may not be more to read from |source_| if the current result comes
+  // from a "file" input element. But then |result| is complete already.
+  if (!DoStep())
+    return val_extracted;
+  val_start = line_start_;
+  // (*) Now state_ == kBody, so val_end gets updated below.
+  while (state_ != kHeadStart && state_ != kFinal) {
+    val_end = line_end_;
+    if (!DoStep()) break;
+  }
+  if (name_parsed && !val_extracted) {
+    result->SetVal(base::StringPiece(source_ + val_start, val_end - val_start));
+  }
+  return name_parsed;
+}
+
+bool PostDataParserMultipart::SetSource(const std::vector<char>* source) {
+  if (state_ == kError)
+    return false;
+  if (source_ != NULL && next_line_ < length_)
+    return false;
+  source_ = &(source->front());
+  length_ = source->size();
+  next_line_ = 0;
+  return true;
+}
+
+bool PostDataParserMultipart::DoStep() {
+  if (!SeekNextLine())
+    return false;
+  switch (state_) {
+    case kInit:
+      if (line_type_ == kBoundary)
+        state_ = kHeadStart;
+      else
+        state_ = kError;
+      break;
+    case kHeadStart:
+      if (line_type_ == kDisposition)
+        state_ = kHeadRead;
+      else
+        state_ = kHead;
+      break;
+    case kHead:
+      if (line_type_ == kDisposition)
+        state_ = kHeadRead;
+      break;
+    case kHeadRead:
+      if (line_type_ == kEmpty)
+        state_ = kBody;
+      break;
+    case kBody:
+      if (line_type_ == kBoundary)
+        state_ = kHeadStart;
+      else if (line_type_ == kEndBoundary)
+        state_ = kFinal;
+      break;
+    case kFinal:
+      if (line_type_ != kEmpty)
+        state_ = kError;
+    case kError:
+      break;
+  }
+  return true;
+}
+
+PostDataParserMultipart::LineType PostDataParserMultipart::GetLineType() {
+  const size_t line_length = line_end_ - line_start_;
+  const base::StringPiece line(source_ + line_start_, line_length);
+  if (line == boundary_)
+    return kBoundary;
+  else if (line == final_boundary_)
+    return kEndBoundary;
+  else if (line.starts_with(kContentDisposition))
+    return kDisposition;
+  else if (line_start_ == line_end_)
+    return kEmpty;
+  else
+    return kOther;
+}
+
+// Contract: only to be called from DoStep().
+bool PostDataParserMultipart::SeekNextLine() {
+  if (source_ == NULL || state_ == kError)
+    return false;
+  if (next_line_ >= length_)
+    return false;
+  line_start_ = next_line_;
+  size_t seek = line_start_;
+  while (seek < length_ && *(source_ + seek) != '\r')
+    ++seek;
+  line_end_ = seek;
+  line_type_ = GetLineType();
+  if ((seek+1) < length_ && strncmp(source_ + seek, "\r\n", 2) != 0)
+    return false;
+  next_line_ = seek + 2;
+  return true;
+}
+
+// Contract: line_type_ == kDisposition.
+bool PostDataParserMultipart::ParseHead(Result* result, bool* val_extracted) {
+  DCHECK_EQ(kDisposition, line_type_);
+  base::StringPiece line(source_ + line_start_, line_end_ - line_start_);
+  const char kNameEquals[] = " name=\"";
+  const char kFilenameEquals[] = " filename=\"";
+  size_t key_offset = line.find(kNameEquals);
+  if (key_offset == base::StringPiece::npos)
+    return false;
+  key_offset += strlen(kNameEquals);
+  result->SetKey(base::StringPiece(source_ + line_start_ + key_offset,
+                                   line.find('"', key_offset) - key_offset));
+  size_t val_offset = line.find(kFilenameEquals);
+  if (val_offset == std::string::npos) {
+    *val_extracted = false;
+  } else {
+    *val_extracted = true;
+    val_offset += strlen(kFilenameEquals);
+    result->SetVal(base::StringPiece(source_ + line_start_ + val_offset,
+                                     line.find('"', val_offset) - val_offset));
+  }
+  return true;
+}
+
+}  // namespace extensions