Improve Zip File Scanning on Mac

third_party/zlib/google/zip_reader.cc

Index: third_party/zlib/google/zip_reader.cc
diff --git a/third_party/zlib/google/zip_reader.cc b/third_party/zlib/google/zip_reader.cc
index ff5a39514af113d28f5266d2c119d51a4d386328..860a6eb20863a60c47157330ea84eac701b7cdc2 100644
--- a/third_party/zlib/google/zip_reader.cc
+++ b/third_party/zlib/google/zip_reader.cc
@@ -289,7 +289,8 @@ bool ZipReader::LocateAndOpenEntry(const base::FilePath& path_in_zip) {
   return OpenCurrentEntryInZip();
 }
 
-bool ZipReader::ExtractCurrentEntry(WriterDelegate* delegate) const {
+bool ZipReader::ExtractCurrentEntry(WriterDelegate* delegate,
+                                    int64_t num_bytes_to_extract) const {
   DCHECK(zip_file_);
 
   const int open_result = unzOpenCurrentFile(zip_file_);
@@ -298,26 +299,61 @@ bool ZipReader::ExtractCurrentEntry(WriterDelegate* delegate) const {
 
   if (!delegate->PrepareOutput())
     return false;
-
+  LOG(ERROR) << "done with init stuff";
   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);
+    LOG(ERROR) << "total_num_bytes_read: " << total_num_bytes_read;
+    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;
+        LOG(ERROR) << "tried to read past end of file.";
+        break;
+      }
     } else if (num_bytes_read < 0) {
       // If num_bytes_read < 0, then it's a specific UNZ_* error code.
+      LOG(ERROR) << "read failed";
       success = false;
       break;
     } else if (num_bytes_read > 0) {
       // Some data is read.
       if (!delegate->WriteBytes(buf.get(), num_bytes_read)) {
+        LOG(ERROR) << "write failed";
         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;
+    }
+    LOG(ERROR) << "total_num_bytes_read: " << total_num_bytes_read;
   }
 
   unzCloseCurrentFile(zip_file_);
@@ -336,7 +372,7 @@ bool ZipReader::ExtractCurrentEntryToFilePath(
   bool success = false;
   {
     FilePathWriterDelegate writer(output_file_path);
-    success = ExtractCurrentEntry(&writer);
+    success = ExtractCurrentEntry(&writer, -1);
   }
 
   if (success &&
@@ -416,15 +452,20 @@ bool ZipReader::ExtractCurrentEntryToFile(base::File* file) const {
     return false;
 
   FileWriterDelegate writer(file);
-  return ExtractCurrentEntry(&writer);
+  return ExtractCurrentEntry(&writer, -1);
 }
 
-bool ZipReader::ExtractCurrentEntryToString(size_t max_read_bytes,
+bool ZipReader::ExtractCurrentEntryToString(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;
+  }
+  LOG(ERROR) << "here1";
   if (current_entry_info()->is_directory()) {
     output->clear();
     return true;
@@ -435,13 +476,23 @@ bool ZipReader::ExtractCurrentEntryToString(size_t max_read_bytes,
   // 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))
+  StringWriterDelegate writer(num_bytes, &contents);
+  int64_t size_arg = read_entire_file ? ((int64_t)-1) : num_bytes;
+  LOG(ERROR) << "size_arg: " << size_arg;
+  LOG(ERROR) << "num_bytes: " << num_bytes;
+  if (!ExtractCurrentEntry(&writer, size_arg)) {
+    output->clear();
+    LOG(ERROR) << "ExtractCurrentEntry failed";
     return false;
+  }
   output->swap(contents);
   return true;
 }