SeqAn3 3.4.0-rc.1
The Modern C++ library for sequence analysis.
Loading...
Searching...
No Matches
misc_input.hpp
Go to the documentation of this file.
1// SPDX-FileCopyrightText: 2006-2024 Knut Reinert & Freie Universität Berlin
2// SPDX-FileCopyrightText: 2016-2024 Knut Reinert & MPI für molekulare Genetik
3// SPDX-License-Identifier: BSD-3-Clause
4
10#pragma once
11
12#include <algorithm>
13#include <concepts>
14#include <filesystem>
15#include <iostream>
16#include <ranges>
17#include <span>
18#include <string>
19#include <tuple>
20
21#if defined(SEQAN3_HAS_BZIP2)
22# include <seqan3/contrib/stream/bz2_istream.hpp>
23#endif
24#if defined(SEQAN3_HAS_ZLIB)
25# include <seqan3/contrib/stream/bgzf_istream.hpp>
27# include <seqan3/contrib/stream/gz_istream.hpp>
28#endif
33
34namespace seqan3::detail
35{
36
42template <std::ranges::forward_range ref_t, std::ranges::forward_range query_t>
43inline bool starts_with(ref_t && reference, query_t && query)
44 requires std::equality_comparable_with<std::ranges::range_reference_t<ref_t>,
45 std::ranges::range_reference_t<query_t>>
46{
47 auto rit = std::ranges::begin(reference);
48 auto rend = std::ranges::end(reference);
49
50 auto qit = std::ranges::begin(query);
51 auto qend = std::ranges::end(query);
52
53 while (true)
54 {
55 if (qit == qend)
56 return true;
57
58 if (rit == rend)
59 return false;
60
61 if (*qit != *rit)
62 return false;
63
64 ++qit;
65 ++rit;
66 }
67}
68
76template <builtin_character char_t>
77inline auto make_secondary_istream(std::basic_istream<char_t> & primary_stream, std::filesystem::path & filename)
79{
80 assert(primary_stream.good());
81
82 // don't assume ownership
83 constexpr auto stream_deleter_noop = [](std::basic_istream<char_t> *) {};
84 // assume ownership
85 [[maybe_unused]] constexpr auto stream_deleter_default = [](std::basic_istream<char_t> * ptr)
86 {
87 delete ptr;
88 };
89
90 // extract "magic header"
91 std::istreambuf_iterator<char_t> it{primary_stream};
92 std::array<char, bgzf_compression::magic_header.size()> magic_number{}; // Largest magic header from bgzf
93 size_t read_chars = 0;
94 for (; read_chars < magic_number.size(); ++read_chars)
95 {
97 break;
98
99 magic_number[read_chars] = *it;
100 ++it;
101 }
102
103 // unget all read chars.
104 for (size_t i = 0; i < read_chars; ++i)
105 primary_stream.unget(); // If you unget() more characters than are present in the get area, badbit is set.
106
107 assert(primary_stream.good() && "`unget()` was called too many times on primary_stream.");
108
109 std::string extension{};
110 if (filename.has_extension())
111 extension = filename.extension().string().substr(1);
112
113 // tests whether the given extension matches with one of the given compression tags.
114 [[maybe_unused]] auto contains_extension = [](auto compression_tag, auto const & extension) constexpr
115 {
116 return std::ranges::find(decltype(compression_tag)::file_extensions, extension)
117 != std::ranges::end(decltype(compression_tag)::file_extensions);
118 };
119
120 // set return value appropriately
121 if (read_chars == magic_number.size() && bgzf_compression::validate_header(std::span{magic_number})) // BGZF
122 {
123#if defined(SEQAN3_HAS_ZLIB)
124 if (contains_extension(gz_compression{}, extension) || contains_extension(bgzf_compression{}, extension))
125 filename.replace_extension();
126
127 return {new contrib::basic_bgzf_istream<char_t>{primary_stream}, stream_deleter_default};
128#else
129 throw file_open_error{"Trying to read from a bgzf file, but no ZLIB available."};
130#endif
131 }
132 else if (starts_with(magic_number, gz_compression::magic_header)) // GZIP
133 {
134#if defined(SEQAN3_HAS_ZLIB)
135 if (contains_extension(gz_compression{}, extension) || contains_extension(bgzf_compression{}, extension))
136 filename.replace_extension();
137
138 return {new contrib::basic_gz_istream<char_t>{primary_stream}, stream_deleter_default};
139#else
140 throw file_open_error{"Trying to read from a gzipped file, but no ZLIB available."};
141#endif
142 }
143 else if (starts_with(magic_number, bz2_compression::magic_header)) // BZip2
144 {
145#if defined(SEQAN3_HAS_BZIP2)
146 if (contains_extension(bz2_compression{}, extension))
147 filename.replace_extension();
148
149 return {new contrib::basic_bz2_istream<char_t>{primary_stream}, stream_deleter_default};
150#else
151 throw file_open_error{"Trying to read from a bzipped file, but no libbz2 available."};
152#endif
153 }
154 else if (starts_with(magic_number, zstd_compression::magic_header)) // ZStd
155 {
156 throw file_open_error{"Trying to read from a zst'ed file, but SeqAn does not yet support this."};
157 }
158
159 return {&primary_stream, stream_deleter_noop};
160}
161
163template <builtin_character char_t>
164inline auto make_secondary_istream(std::basic_istream<char_t> & primary_stream)
165{
167 return make_secondary_istream(primary_stream, p);
168}
169
170} // namespace seqan3::detail
T begin(T... args)
Provides seqan3::contrib::bgzf_thread_count.
Provides stream compression utilities.
T find(T... args)
Provides exceptions used in the I/O module.
Provides seqan3::detail::magic_header.
T rend(T... args)
T size(T... args)
Provides concepts that do not have equivalents in C++20.
Hide me