Improve Zip File Scanning on Mac

third_party/zlib/google/zip_reader.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 "third_party/zlib/google/zip_reader.h"
 
 #include <utility>
 
 #include "base/bind.h"
 #include "base/files/file.h"
@@ skipping 272 matching lines @@
                                    path_in_zip.AsUTF8Unsafe().c_str(),
                                    kDefaultCaseSensivityOfOS);
   if (result != UNZ_OK)
     return false;
 
   // Then Open the entry.
   return OpenCurrentEntryInZip();
 }
 
 bool ZipReader::ExtractCurrentEntry(WriterDelegate* delegate) const {
+  return ExtractFromBeginningOfCurrentEntry(delegate, -1);
+}
+
+bool ZipReader::ExtractFromBeginningOfCurrentEntry(
+    WriterDelegate* delegate,
+    int64_t num_bytes_to_extract) const {
   DCHECK(zip_file_);
 
   const int open_result = unzOpenCurrentFile(zip_file_);
   if (open_result != UNZ_OK)
     return false;
 
   if (!delegate->PrepareOutput())
     return false;
 
   bool success = true;  // This becomes false when something bad happens.
   std::unique_ptr<char[]> buf(new char[internal::kZipBufSize]);
+  uint64_t total_num_bytes_read = 0;
   while (true) {
-    const int num_bytes_read = unzReadCurrentFile(zip_file_, buf.get(),
-                                                  internal::kZipBufSize);
+    uint64_t bytes_to_read = internal::kZipBufSize;
+    // This should never result in a negative value, as |num_bytes_to_extract|
+    // will always be greater than |total_num_bytes_read|.
+    if (num_bytes_to_extract >= 0) {
+      auto unread_bytes =
+          base::CheckedNumeric<uint64_t>(
+              base::checked_cast<uint64_t>(num_bytes_to_extract)) -
+          total_num_bytes_read;
+      if (unread_bytes.ValueOrDie() < internal::kZipBufSize) {
+        bytes_to_read = unread_bytes.ValueOrDie();
+      }
+    }
+
+    const int num_bytes_read =
+        unzReadCurrentFile(zip_file_, buf.get(), bytes_to_read);
+
     if (num_bytes_read == 0) {
-      // Reached the end of the file.
-      break;
+      // Reached the end of the file. Return true if entire file was supposed to
+      // be read, false if not.
+      if (num_bytes_to_extract == -1) {
+        success = true;
+        break;
+      } else {
+        success = false;
+        break;
+      }
     } else if (num_bytes_read < 0) {
       // If num_bytes_read < 0, then it's a specific UNZ_* error code.
       success = false;
       break;
     } else if (num_bytes_read > 0) {
       // Some data is read.
       if (!delegate->WriteBytes(buf.get(), num_bytes_read)) {
         success = false;
         break;
       }
     }
+    // Previous check ensures that num_bytes_read is not negative here.
+    total_num_bytes_read += base::checked_cast<uint64_t>(num_bytes_read);
+    if (num_bytes_to_extract >= 0 &&
+        total_num_bytes_read ==
+            base::checked_cast<uint64_t>(num_bytes_to_extract)) {
+      // Read all the data.
+      break;
+    }
   }
 
   unzCloseCurrentFile(zip_file_);
 
   return success;
 }
 
 bool ZipReader::ExtractCurrentEntryToFilePath(
     const base::FilePath& output_file_path) const {
   DCHECK(zip_file_);
@@ skipping 83 matching lines @@
   // false.
   if (current_entry_info()->is_directory())
     return false;
 
   FileWriterDelegate writer(file);
   return ExtractCurrentEntry(&writer);
 }
 
 bool ZipReader::ExtractCurrentEntryToString(size_t max_read_bytes,
                                             std::string* output) const {
+  return ExtractFromBeginningOfCurrentEntryToString(max_read_bytes, true,
+                                                    output);
+}
+
+bool ZipReader::ExtractFromBeginningOfCurrentEntryToString(
+    size_t num_bytes,
+    bool read_entire_file,
+    std::string* output) const {
   DCHECK(output);
   DCHECK(zip_file_);
-  DCHECK_NE(0U, max_read_bytes);
+
+  if (num_bytes == 0) {
+    output->clear();
+    return true;
+  }
 
   if (current_entry_info()->is_directory()) {
     output->clear();
     return true;
   }
 
   // The original_size() is the best hint for the real size, so it saves
   // doing reallocations for the common case when the uncompressed size is
   // correct. However, we need to assume that the uncompressed size could be
   // incorrect therefore this function needs to read as much data as possible.
   std::string contents;
-  contents.reserve(
-      static_cast<size_t>(std::min(static_cast<int64_t>(max_read_bytes),
-                                   current_entry_info()->original_size())));
+  if (read_entire_file) {
+    contents.reserve(
+        static_cast<size_t>(std::min(static_cast<int64_t>(num_bytes),
+                                     current_entry_info()->original_size())));
+  } else {
+    contents.reserve(num_bytes);
+  }
 
-  StringWriterDelegate writer(max_read_bytes, &contents);
-  if (!ExtractCurrentEntry(&writer))
-    return false;
-  output->swap(contents);
+  StringWriterDelegate writer(num_bytes, &contents);
+  if (read_entire_file) {
+    if (!ExtractCurrentEntry(&writer)) {
+      output->clear();
+      return false;
+    }
+    output->swap(contents);
+  } else {
+    if (!ExtractFromBeginningOfCurrentEntry(&writer, num_bytes)) {
+      output->clear();
+      return false;
+    }
+    output->swap(contents);
+  }
   return true;
 }
 
 bool ZipReader::OpenInternal() {
   DCHECK(zip_file_);
 
   unz_global_info zip_info = {};  // Zero-clear.
   if (unzGetGlobalInfo(zip_file_, &zip_info) != UNZ_OK) {
     return false;
   }
@@ skipping 68 matching lines @@
 }
 
 bool FileWriterDelegate::WriteBytes(const char* data, int num_bytes) {
   int bytes_written = file_->WriteAtCurrentPos(data, num_bytes);
   if (bytes_written > 0)
     file_length_ += bytes_written;
   return bytes_written == num_bytes;
 }
 
 }  // namespace zip