1*ec63e07aSXin Li // Copyright 2022 Google LLC
2*ec63e07aSXin Li //
3*ec63e07aSXin Li // Licensed under the Apache License, Version 2.0 (the "License");
4*ec63e07aSXin Li // you may not use this file except in compliance with the License.
5*ec63e07aSXin Li // You may obtain a copy of the License at
6*ec63e07aSXin Li //
7*ec63e07aSXin Li // https://www.apache.org/licenses/LICENSE-2.0
8*ec63e07aSXin Li //
9*ec63e07aSXin Li // Unless required by applicable law or agreed to in writing, software
10*ec63e07aSXin Li // distributed under the License is distributed on an "AS IS" BASIS,
11*ec63e07aSXin Li // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12*ec63e07aSXin Li // See the License for the specific language governing permissions and
13*ec63e07aSXin Li // limitations under the License.
14*ec63e07aSXin Li
15*ec63e07aSXin Li #include "contrib/c-blosc/utils/utils_blosc.h"
16*ec63e07aSXin Li
17*ec63e07aSXin Li #include <fstream>
18*ec63e07aSXin Li #include <iostream>
19*ec63e07aSXin Li #include <string>
20*ec63e07aSXin Li
21*ec63e07aSXin Li #include "contrib/c-blosc/sandboxed.h"
22*ec63e07aSXin Li
23*ec63e07aSXin Li constexpr size_t kFileMaxSize = 1024 * 1024 * 1024; // 1GB
24*ec63e07aSXin Li
GetStreamSize(std::ifstream & stream)25*ec63e07aSXin Li std::streamsize GetStreamSize(std::ifstream& stream) {
26*ec63e07aSXin Li stream.seekg(0, std::ios_base::end);
27*ec63e07aSXin Li std::streamsize ssize = stream.tellg();
28*ec63e07aSXin Li stream.seekg(0, std::ios_base::beg);
29*ec63e07aSXin Li
30*ec63e07aSXin Li return ssize;
31*ec63e07aSXin Li }
32*ec63e07aSXin Li
Compress(CbloscApi & api,std::ifstream & in_stream,std::ofstream & out_stream,int clevel,std::string & compressor,int nthreads)33*ec63e07aSXin Li absl::Status Compress(CbloscApi& api, std::ifstream& in_stream,
34*ec63e07aSXin Li std::ofstream& out_stream, int clevel,
35*ec63e07aSXin Li std::string& compressor, int nthreads) {
36*ec63e07aSXin Li std::streamsize ssize = GetStreamSize(in_stream);
37*ec63e07aSXin Li sapi::v::Array<uint8_t> inbuf(ssize);
38*ec63e07aSXin Li sapi::v::Array<uint8_t> outbuf(ssize);
39*ec63e07aSXin Li
40*ec63e07aSXin Li in_stream.read(reinterpret_cast<char*>(inbuf.GetData()), ssize);
41*ec63e07aSXin Li if (in_stream.gcount() != ssize) {
42*ec63e07aSXin Li return absl::UnavailableError("Unable to read file");
43*ec63e07aSXin Li }
44*ec63e07aSXin Li
45*ec63e07aSXin Li int ret;
46*ec63e07aSXin Li SAPI_ASSIGN_OR_RETURN(
47*ec63e07aSXin Li ret, api.blosc_set_compressor(
48*ec63e07aSXin Li sapi::v::ConstCStr(compressor.c_str()).PtrBefore()));
49*ec63e07aSXin Li if (ret < 0) {
50*ec63e07aSXin Li return absl::UnavailableError("Unable to set compressor");
51*ec63e07aSXin Li }
52*ec63e07aSXin Li
53*ec63e07aSXin Li SAPI_ASSIGN_OR_RETURN(ret, api.blosc_set_nthreads(nthreads));
54*ec63e07aSXin Li if (ret < 0) {
55*ec63e07aSXin Li return absl::UnavailableError("Unable to set nthreads");
56*ec63e07aSXin Li }
57*ec63e07aSXin Li
58*ec63e07aSXin Li SAPI_ASSIGN_OR_RETURN(
59*ec63e07aSXin Li ssize_t outsize, api.blosc_compress(clevel, 1, sizeof(uint8_t),
60*ec63e07aSXin Li inbuf.GetSize(), inbuf.PtrBefore(),
61*ec63e07aSXin Li outbuf.PtrAfter(), outbuf.GetSize()));
62*ec63e07aSXin Li if (outsize <= 0) {
63*ec63e07aSXin Li return absl::UnavailableError("Unable to compress file.");
64*ec63e07aSXin Li }
65*ec63e07aSXin Li
66*ec63e07aSXin Li out_stream.write(reinterpret_cast<char*>(outbuf.GetData()), outsize);
67*ec63e07aSXin Li if (!out_stream.good()) {
68*ec63e07aSXin Li return absl::UnavailableError("Unable to write file");
69*ec63e07aSXin Li }
70*ec63e07aSXin Li
71*ec63e07aSXin Li return absl::OkStatus();
72*ec63e07aSXin Li }
73*ec63e07aSXin Li
Decompress(CbloscApi & api,std::ifstream & in_stream,std::ofstream & out_stream,int nthreads)74*ec63e07aSXin Li absl::Status Decompress(CbloscApi& api, std::ifstream& in_stream,
75*ec63e07aSXin Li std::ofstream& out_stream, int nthreads) {
76*ec63e07aSXin Li std::streamsize ssize = GetStreamSize(in_stream);
77*ec63e07aSXin Li sapi::v::Array<uint8_t> inbuf(ssize);
78*ec63e07aSXin Li
79*ec63e07aSXin Li in_stream.read(reinterpret_cast<char*>(inbuf.GetData()), ssize);
80*ec63e07aSXin Li if (in_stream.gcount() != ssize) {
81*ec63e07aSXin Li return absl::UnavailableError("Unable to read file");
82*ec63e07aSXin Li }
83*ec63e07aSXin Li
84*ec63e07aSXin Li int ret;
85*ec63e07aSXin Li SAPI_ASSIGN_OR_RETURN(ret, api.blosc_set_nthreads(nthreads));
86*ec63e07aSXin Li if (ret < 0) {
87*ec63e07aSXin Li return absl::UnavailableError("Unable to set nthreads");
88*ec63e07aSXin Li }
89*ec63e07aSXin Li
90*ec63e07aSXin Li // To not transfer memory twice (for blosc_cbuffer_sizes and decopmress),
91*ec63e07aSXin Li // tranfer memory before using it.
92*ec63e07aSXin Li SAPI_RETURN_IF_ERROR(api.GetSandbox()->Allocate(&inbuf, true));
93*ec63e07aSXin Li SAPI_RETURN_IF_ERROR(api.GetSandbox()->TransferToSandboxee(&inbuf));
94*ec63e07aSXin Li
95*ec63e07aSXin Li sapi::v::IntBase<size_t> nbytes;
96*ec63e07aSXin Li sapi::v::IntBase<size_t> cbytes;
97*ec63e07aSXin Li sapi::v::IntBase<size_t> blocksize;
98*ec63e07aSXin Li SAPI_RETURN_IF_ERROR(
99*ec63e07aSXin Li api.blosc_cbuffer_sizes(inbuf.PtrNone(), nbytes.PtrAfter(),
100*ec63e07aSXin Li cbytes.PtrAfter(), blocksize.PtrAfter()));
101*ec63e07aSXin Li if (nbytes.GetValue() == 0) {
102*ec63e07aSXin Li return absl::UnavailableError("Unable to get size");
103*ec63e07aSXin Li }
104*ec63e07aSXin Li if (nbytes.GetValue() > kFileMaxSize) {
105*ec63e07aSXin Li return absl::UnavailableError("The file is to large");
106*ec63e07aSXin Li }
107*ec63e07aSXin Li
108*ec63e07aSXin Li sapi::v::Array<uint8_t> outbuf(nbytes.GetValue());
109*ec63e07aSXin Li SAPI_ASSIGN_OR_RETURN(ssize_t outsize,
110*ec63e07aSXin Li api.blosc_decompress(inbuf.PtrNone(), outbuf.PtrAfter(),
111*ec63e07aSXin Li outbuf.GetSize()));
112*ec63e07aSXin Li if (outsize <= 0) {
113*ec63e07aSXin Li return absl::UnavailableError("Unable to decompress file");
114*ec63e07aSXin Li }
115*ec63e07aSXin Li
116*ec63e07aSXin Li out_stream.write(reinterpret_cast<char*>(outbuf.GetData()), outsize);
117*ec63e07aSXin Li if (!out_stream.good()) {
118*ec63e07aSXin Li return absl::UnavailableError("Unable to write file");
119*ec63e07aSXin Li }
120*ec63e07aSXin Li
121*ec63e07aSXin Li return absl::OkStatus();
122*ec63e07aSXin Li }
123