11#pragma once
22
3- #include < stdexcept>
4- #include < iterator>
53#include < cstdint>
4+ #include < iterator>
5+ #include < stdexcept>
6+ #include < type_traits>
7+
68#include " mio/mmap.hpp"
9+
710#include " util/util.hpp"
811
912#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
1215
1316namespace ds2i {
1417
15- class binary_collection {
16- public:
17- typedef uint32_t posting_type;
18+ template <typename Source = mio::mmap_source>
19+ class base_binary_collection {
20+ public:
21+ using posting_type = uint32_t ;
22+ using pointer = typename std::conditional<std::is_same<Source, mio::mmap_source>::value,
23+ posting_type const ,
24+ posting_type>::type *;
1825
19- binary_collection (const char * filename)
20- {
26+ base_binary_collection (const char *filename) {
2127 std::error_code error;
2228 m_file.map (filename, error);
2329 if ( error ) {
2430 std::cerr << " error mapping file: " message () << " , exiting..."
2531 throw std::runtime_error (" Error opening file"
2632 }
27- m_data = (posting_type const *) m_file.data ();
33+ m_data = reinterpret_cast <pointer>( m_file.data () );
2834 m_data_size = m_file.size () / sizeof (m_data[0 ]);
2935
3036#if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__))
@@ -34,100 +40,70 @@ namespace ds2i {
3440#endif
3541 }
3642
37- class iterator ;
38-
39- iterator begin () const
40- {
41- return iterator (this , 0 );
42- }
43-
44- iterator end () const
45- {
46- return iterator (this , m_data_size);
47- }
48-
4943 class sequence {
50- public:
51- sequence ()
52- : m_begin(nullptr )
53- , m_end(nullptr )
54- {}
55-
56- posting_type const * begin () const
57- {
58- return m_begin;
59- }
44+ public:
45+ sequence (pointer begin, pointer end) : m_begin(begin), m_end(end) {}
46+ sequence () : m_begin(nullptr ), m_end(nullptr ) {}
6047
61- posting_type const * end () const
62- {
63- return m_end;
64- }
48+ pointer begin () const { return m_begin; }
49+ pointer end () const { return m_end; }
50+ size_t size () const { return m_end - m_begin; }
6551
6652 posting_type back () const
6753 {
6854 assert (size ());
6955 return *(m_end - 1 );
7056 }
7157
72- size_t size () const
73- {
74- return m_end - m_begin ;
75- }
58+ private:
59+ pointer m_begin;
60+ pointer m_end;
61+ };
7662
77- private:
78- friend class binary_collection ::iterator;
63+ using const_sequence = sequence;
7964
80- sequence (posting_type const * begin, posting_type const * end)
81- : m_begin(begin)
82- , m_end(end)
83- {}
65+ template <typename S>
66+ class base_iterator ;
8467
85- posting_type const * m_begin;
86- posting_type const * m_end;
87- };
68+ using const_iterator = base_iterator<const_sequence>;
69+ using iterator = typename std::conditional<std::is_same<Source, mio::mmap_source>::value,
70+ const_iterator,
71+ base_iterator<sequence>>::type;
8872
89- class iterator : public std :: iterator<std::forward_iterator_tag,
90- sequence> {
91- public:
92- iterator ()
93- : m_collection( nullptr )
94- { }
73+ iterator begin () { return iterator ( this , 0 ); }
74+ iterator end () { return iterator ( this , m_data_size); }
75+ const_iterator begin () const { return const_iterator ( this , 0 ); }
76+ const_iterator end () const { return const_iterator ( this , m_data_size); }
77+ const_iterator cbegin () const { return const_iterator ( this , 0 ); }
78+ const_iterator cend () const { return const_iterator ( this , m_data_size); }
9579
96- value_type const & operator *() const
97- {
98- return m_cur_seq;
99- }
80+ template < typename S>
81+ class base_iterator : public std ::iterator<std::forward_iterator_tag, S> {
82+ public:
83+ base_iterator () : m_collection( nullptr ) { }
10084
101- value_type const * operator ->() const
102- {
103- return &m_cur_seq;
104- }
85+ auto const &operator *() const { return m_cur_seq; }
10586
106- iterator& operator ++()
107- {
87+ auto const *operator -> () const { return &m_cur_seq; }
88+
89+ base_iterator &operator ++() {
10890 m_pos = m_next_pos;
10991 read ();
11092 return *this ;
11193 }
11294
113- bool operator ==(iterator const & other) const
114- {
95+ bool operator ==(base_iterator const &other) const {
11596 assert (m_collection == other.m_collection );
11697 return m_pos == other.m_pos ;
11798 }
11899
119- bool operator !=(iterator const & other) const
120- {
121- return !(*this == other);
122- }
100+ bool operator !=(base_iterator const &other) const { return !(*this == other); }
123101
124- private:
125- friend class binary_collection ;
102+ private:
103+ friend class base_binary_collection ;
126104
127- iterator (binary_collection const * coll, size_t pos)
128- : m_collection(coll)
129- , m_pos(pos)
130- {
105+ base_iterator (base_binary_collection const *coll, size_t pos)
106+ : m_collection(coll), m_pos(pos) {
131107 read ();
132108 }
133109
@@ -138,24 +114,26 @@ namespace ds2i {
138114
139115 size_t n = 0 ;
140116 size_t pos = m_pos;
141- while (!( n = m_collection->m_data [pos++])); // skip empty seqs
117+ n = m_collection->m_data [pos++];
142118 // file might be truncated
143119 n = std::min (n, size_t (m_collection->m_data_size - pos));
144- posting_type const * begin = &m_collection->m_data [pos];
145- posting_type const * end = begin + n;
120+ auto begin = &m_collection->m_data [pos];
146121
147122 m_next_pos = pos + n;
148- m_cur_seq = sequence (begin, end );
123+ m_cur_seq = S (begin, begin + n );
149124 }
150125
151- binary_collection const * m_collection;
152- size_t m_pos, m_next_pos;
153- sequence m_cur_seq;
126+ base_binary_collection const * m_collection;
127+ size_t m_pos, m_next_pos;
128+ S m_cur_seq;
154129 };
155130
156- private:
157- mio::mmap_source m_file;
158- posting_type const * m_data;
159- size_t m_data_size;
131+ private:
132+ Source m_file;
133+ pointer m_data;
134+ size_t m_data_size;
160135 };
136+
137+ using binary_collection = base_binary_collection<>;
138+ using writable_binary_collection = base_binary_collection<mio::mmap_sink>;
161139}
0 commit comments