From 0f3ece775d455fadb301c8d8609a424b4a4f508c Mon Sep 17 00:00:00 2001
From: Qiao Longfei <qiaolongfei@baidu.com>
Date: Thu, 18 Oct 2018 18:33:05 +0800
Subject: [PATCH] use gzstream

---
 paddle/fluid/operators/reader/CMakeLists.txt |  2 +-
 paddle/fluid/operators/reader/ctr_reader.cc  | 46 +++++++-------------
 paddle/fluid/operators/reader/ctr_reader.h   |  4 --
 3 files changed, 17 insertions(+), 35 deletions(-)

diff --git a/paddle/fluid/operators/reader/CMakeLists.txt b/paddle/fluid/operators/reader/CMakeLists.txt
index 341aeda4e41..4ad376c6170 100644
--- a/paddle/fluid/operators/reader/CMakeLists.txt
+++ b/paddle/fluid/operators/reader/CMakeLists.txt
@@ -16,7 +16,7 @@ function(reader_library TARGET_NAME)
 endfunction()
 
 cc_library(buffered_reader SRCS buffered_reader.cc DEPS reader simple_threadpool)
-cc_library(ctr_reader SRCS ctr_reader.cc DEPS reader simple_threadpool boost)
+cc_library(ctr_reader SRCS ctr_reader.cc DEPS reader simple_threadpool boost gzstream)
 reader_library(open_files_op SRCS open_files_op.cc DEPS buffered_reader)
 reader_library(create_ctr_reader_op SRCS create_ctr_reader_op.cc DEPS ctr_reader)
 reader_library(create_random_data_generator_op SRCS create_random_data_generator_op.cc)
diff --git a/paddle/fluid/operators/reader/ctr_reader.cc b/paddle/fluid/operators/reader/ctr_reader.cc
index a4197a54349..8be9f68c941 100644
--- a/paddle/fluid/operators/reader/ctr_reader.cc
+++ b/paddle/fluid/operators/reader/ctr_reader.cc
@@ -14,6 +14,8 @@
 
 #include "paddle/fluid/operators/reader/ctr_reader.h"
 
+#include <gzstream.h>
+
 #include <cstdlib>
 #include <fstream>
 #include <iostream>
@@ -24,10 +26,6 @@
 #include <algorithm>
 #include <random>
 
-#include <boost/iostreams/copy.hpp>
-#include <boost/iostreams/filter/gzip.hpp>
-#include <boost/iostreams/filtering_streambuf.hpp>
-
 namespace paddle {
 namespace operators {
 namespace reader {
@@ -75,23 +73,19 @@ static inline void parse_line(
 
 class GzipReader {
  public:
-  explicit GzipReader(const std::string& file_name) : instream_(&inbuf_) {
-    file_ = std::ifstream(file_name, std::ios_base::in | std::ios_base::binary);
-    inbuf_.push(boost::iostreams::gzip_decompressor());
-    inbuf_.push(file_);
-    // Convert streambuf to istream
-  }
+  explicit GzipReader(const std::string& file_name)
+      : gzstream_(file_name.c_str()) {}
 
-  ~GzipReader() { file_.close(); }
+  ~GzipReader() {}
 
-  bool HasNext() { return instream_.peek() != EOF; }
+  bool HasNext() { return gzstream_.peek() != EOF; }
 
-  void NextLine(std::string& line) { std::getline(instream_, line); }  // NOLINT
+  void NextLine(std::string* line) {  // NOLINT
+    std::getline(gzstream_, line);
+  }
 
  private:
-  boost::iostreams::filtering_streambuf<boost::iostreams::input> inbuf_;
-  std::ifstream file_;
-  std::istream instream_;
+  igzstream gzstream_;
 };
 
 class MultiGzipReader {
@@ -113,8 +107,8 @@ class MultiGzipReader {
     return true;
   }
 
-  void NextLine(std::string& line) {  // NOLINT
-    readers_[current_reader_index_]->NextLine(line);
+  void NextLine(std::string* line) {
+    readers_[current_reader_index_]->NextLine(*line);
   }
 
  private:
@@ -122,12 +116,6 @@ class MultiGzipReader {
   size_t current_reader_index_ = 0;
 };
 
-// void CTRReader::ReadThread(
-//        const std::vector<std::string> &file_list,
-//        const std::vector<std::string>& slots,
-//        int batch_size,
-//        std::shared_ptr<LoDTensorBlockingQueue>& queue) {}
-
 void CTRReader::ReadThread(const std::vector<std::string>& file_list,
                            const std::vector<std::string>& slots,
                            int batch_size,
@@ -135,14 +123,12 @@ void CTRReader::ReadThread(const std::vector<std::string>& file_list,
   std::string line;
 
   // read all files
-  std::vector<std::string> all_lines;
   MultiGzipReader reader(file_list);
+  reader.NextLine(&line);
 
-  for (int j = 0; j < all_lines.size(); ++j) {
-    std::unordered_map<std::string, std::vector<int64_t>> slots_to_data;
-    int64_t label;
-    parse_line(all_lines[j], slots, &label, &slots_to_data);
-  }
+  std::unordered_map<std::string, std::vector<int64_t>> slots_to_data;
+  int64_t label;
+  parse_line(line, slots, &label, &slots_to_data);
 }
 
 }  // namespace reader
diff --git a/paddle/fluid/operators/reader/ctr_reader.h b/paddle/fluid/operators/reader/ctr_reader.h
index 1ef6e6d551f..11eb4f97864 100644
--- a/paddle/fluid/operators/reader/ctr_reader.h
+++ b/paddle/fluid/operators/reader/ctr_reader.h
@@ -22,10 +22,6 @@
 #include <unordered_map>
 #include <vector>
 
-#include <boost/iostreams/copy.hpp>
-#include <boost/iostreams/filter/gzip.hpp>
-#include <boost/iostreams/filtering_streambuf.hpp>
-
 #include "paddle/fluid/framework/reader.h"
 #include "paddle/fluid/framework/threadpool.h"
 #include "paddle/fluid/operators/reader/lod_tensor_blocking_queue.h"
-- 
GitLab