cannam@147
|
1 // Copyright (c) 2013-2014 Sandstorm Development Group, Inc. and contributors
|
cannam@147
|
2 // Licensed under the MIT License:
|
cannam@147
|
3 //
|
cannam@147
|
4 // Permission is hereby granted, free of charge, to any person obtaining a copy
|
cannam@147
|
5 // of this software and associated documentation files (the "Software"), to deal
|
cannam@147
|
6 // in the Software without restriction, including without limitation the rights
|
cannam@147
|
7 // to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
cannam@147
|
8 // copies of the Software, and to permit persons to whom the Software is
|
cannam@147
|
9 // furnished to do so, subject to the following conditions:
|
cannam@147
|
10 //
|
cannam@147
|
11 // The above copyright notice and this permission notice shall be included in
|
cannam@147
|
12 // all copies or substantial portions of the Software.
|
cannam@147
|
13 //
|
cannam@147
|
14 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
cannam@147
|
15 // IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
cannam@147
|
16 // FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
cannam@147
|
17 // AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
cannam@147
|
18 // LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
cannam@147
|
19 // OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
|
cannam@147
|
20 // THE SOFTWARE.
|
cannam@147
|
21
|
cannam@147
|
22 #ifndef KJ_STRING_TREE_H_
|
cannam@147
|
23 #define KJ_STRING_TREE_H_
|
cannam@147
|
24
|
cannam@147
|
25 #if defined(__GNUC__) && !KJ_HEADER_WARNINGS
|
cannam@147
|
26 #pragma GCC system_header
|
cannam@147
|
27 #endif
|
cannam@147
|
28
|
cannam@147
|
29 #include "string.h"
|
cannam@147
|
30
|
cannam@147
|
31 namespace kj {
|
cannam@147
|
32
|
cannam@147
|
33 class StringTree {
|
cannam@147
|
34 // A long string, represented internally as a tree of strings. This data structure is like a
|
cannam@147
|
35 // String, but optimized for concatenation and iteration at the expense of seek time. The
|
cannam@147
|
36 // structure is intended to be used for building large text blobs from many small pieces, where
|
cannam@147
|
37 // repeatedly concatenating smaller strings into larger ones would waste copies. This structure
|
cannam@147
|
38 // is NOT intended for use cases requiring random access or computing substrings. For those,
|
cannam@147
|
39 // you should use a Rope, which is a much more complicated data structure.
|
cannam@147
|
40 //
|
cannam@147
|
41 // The proper way to construct a StringTree is via kj::strTree(...), which works just like
|
cannam@147
|
42 // kj::str(...) but returns a StringTree rather than a String.
|
cannam@147
|
43 //
|
cannam@147
|
44 // KJ_STRINGIFY() functions that construct large strings from many smaller strings are encouraged
|
cannam@147
|
45 // to return StringTree rather than a flat char container.
|
cannam@147
|
46
|
cannam@147
|
47 public:
|
cannam@147
|
48 inline StringTree(): size_(0) {}
|
cannam@147
|
49 inline StringTree(String&& text): size_(text.size()), text(kj::mv(text)) {}
|
cannam@147
|
50
|
cannam@147
|
51 StringTree(Array<StringTree>&& pieces, StringPtr delim);
|
cannam@147
|
52 // Build a StringTree by concatenating the given pieces, delimited by the given delimiter
|
cannam@147
|
53 // (e.g. ", ").
|
cannam@147
|
54
|
cannam@147
|
55 inline size_t size() const { return size_; }
|
cannam@147
|
56
|
cannam@147
|
57 template <typename Func>
|
cannam@147
|
58 void visit(Func&& func) const;
|
cannam@147
|
59
|
cannam@147
|
60 String flatten() const;
|
cannam@147
|
61 // Return the contents as a string.
|
cannam@147
|
62
|
cannam@147
|
63 // TODO(someday): flatten() when *this is an rvalue and when branches.size() == 0 could simply
|
cannam@147
|
64 // return `kj::mv(text)`. Requires reference qualifiers (Clang 3.3 / GCC 4.8).
|
cannam@147
|
65
|
cannam@147
|
66 void flattenTo(char* __restrict__ target) const;
|
cannam@147
|
67 // Copy the contents to the given character array. Does not add a NUL terminator.
|
cannam@147
|
68
|
cannam@147
|
69 private:
|
cannam@147
|
70 size_t size_;
|
cannam@147
|
71 String text;
|
cannam@147
|
72
|
cannam@147
|
73 struct Branch;
|
cannam@147
|
74 Array<Branch> branches; // In order.
|
cannam@147
|
75
|
cannam@147
|
76 inline void fill(char* pos, size_t branchIndex);
|
cannam@147
|
77 template <typename First, typename... Rest>
|
cannam@147
|
78 void fill(char* pos, size_t branchIndex, First&& first, Rest&&... rest);
|
cannam@147
|
79 template <typename... Rest>
|
cannam@147
|
80 void fill(char* pos, size_t branchIndex, StringTree&& first, Rest&&... rest);
|
cannam@147
|
81 template <typename... Rest>
|
cannam@147
|
82 void fill(char* pos, size_t branchIndex, Array<char>&& first, Rest&&... rest);
|
cannam@147
|
83 template <typename... Rest>
|
cannam@147
|
84 void fill(char* pos, size_t branchIndex, String&& first, Rest&&... rest);
|
cannam@147
|
85
|
cannam@147
|
86 template <typename... Params>
|
cannam@147
|
87 static StringTree concat(Params&&... params);
|
cannam@147
|
88 static StringTree&& concat(StringTree&& param) { return kj::mv(param); }
|
cannam@147
|
89
|
cannam@147
|
90 template <typename T>
|
cannam@147
|
91 static inline size_t flatSize(const T& t) { return t.size(); }
|
cannam@147
|
92 static inline size_t flatSize(String&& s) { return 0; }
|
cannam@147
|
93 static inline size_t flatSize(StringTree&& s) { return 0; }
|
cannam@147
|
94
|
cannam@147
|
95 template <typename T>
|
cannam@147
|
96 static inline size_t branchCount(const T& t) { return 0; }
|
cannam@147
|
97 static inline size_t branchCount(String&& s) { return 1; }
|
cannam@147
|
98 static inline size_t branchCount(StringTree&& s) { return 1; }
|
cannam@147
|
99
|
cannam@147
|
100 template <typename... Params>
|
cannam@147
|
101 friend StringTree strTree(Params&&... params);
|
cannam@147
|
102 };
|
cannam@147
|
103
|
cannam@147
|
104 inline StringTree&& KJ_STRINGIFY(StringTree&& tree) { return kj::mv(tree); }
|
cannam@147
|
105 inline const StringTree& KJ_STRINGIFY(const StringTree& tree) { return tree; }
|
cannam@147
|
106
|
cannam@147
|
107 inline StringTree KJ_STRINGIFY(Array<StringTree>&& trees) { return StringTree(kj::mv(trees), ""); }
|
cannam@147
|
108
|
cannam@147
|
109 template <typename... Params>
|
cannam@147
|
110 StringTree strTree(Params&&... params);
|
cannam@147
|
111 // Build a StringTree by stringifying the given parameters and concatenating the results.
|
cannam@147
|
112 // If any of the parameters stringify to StringTree rvalues, they will be incorporated as
|
cannam@147
|
113 // branches to avoid a copy.
|
cannam@147
|
114
|
cannam@147
|
115 // =======================================================================================
|
cannam@147
|
116 // Inline implementation details
|
cannam@147
|
117
|
cannam@147
|
118 namespace _ { // private
|
cannam@147
|
119
|
cannam@147
|
120 template <typename... Rest>
|
cannam@147
|
121 char* fill(char* __restrict__ target, const StringTree& first, Rest&&... rest) {
|
cannam@147
|
122 // Make str() work with stringifiers that return StringTree by patching fill().
|
cannam@147
|
123
|
cannam@147
|
124 first.flattenTo(target);
|
cannam@147
|
125 return fill(target + first.size(), kj::fwd<Rest>(rest)...);
|
cannam@147
|
126 }
|
cannam@147
|
127
|
cannam@147
|
128 template <typename T> constexpr bool isStringTree() { return false; }
|
cannam@147
|
129 template <> constexpr bool isStringTree<StringTree>() { return true; }
|
cannam@147
|
130
|
cannam@147
|
131 inline StringTree&& toStringTreeOrCharSequence(StringTree&& tree) { return kj::mv(tree); }
|
cannam@147
|
132 inline StringTree toStringTreeOrCharSequence(String&& str) { return StringTree(kj::mv(str)); }
|
cannam@147
|
133
|
cannam@147
|
134 template <typename T>
|
cannam@147
|
135 inline auto toStringTreeOrCharSequence(T&& value)
|
cannam@147
|
136 -> decltype(toCharSequence(kj::fwd<T>(value))) {
|
cannam@147
|
137 static_assert(!isStringTree<Decay<T>>(),
|
cannam@147
|
138 "When passing a StringTree into kj::strTree(), either pass it by rvalue "
|
cannam@147
|
139 "(use kj::mv(value)) or explicitly call value.flatten() to make a copy.");
|
cannam@147
|
140
|
cannam@147
|
141 return toCharSequence(kj::fwd<T>(value));
|
cannam@147
|
142 }
|
cannam@147
|
143
|
cannam@147
|
144 } // namespace _ (private)
|
cannam@147
|
145
|
cannam@147
|
146 struct StringTree::Branch {
|
cannam@147
|
147 size_t index;
|
cannam@147
|
148 // Index in `text` where this branch should be inserted.
|
cannam@147
|
149
|
cannam@147
|
150 StringTree content;
|
cannam@147
|
151 };
|
cannam@147
|
152
|
cannam@147
|
153 template <typename Func>
|
cannam@147
|
154 void StringTree::visit(Func&& func) const {
|
cannam@147
|
155 size_t pos = 0;
|
cannam@147
|
156 for (auto& branch: branches) {
|
cannam@147
|
157 if (branch.index > pos) {
|
cannam@147
|
158 func(text.slice(pos, branch.index));
|
cannam@147
|
159 pos = branch.index;
|
cannam@147
|
160 }
|
cannam@147
|
161 branch.content.visit(func);
|
cannam@147
|
162 }
|
cannam@147
|
163 if (text.size() > pos) {
|
cannam@147
|
164 func(text.slice(pos, text.size()));
|
cannam@147
|
165 }
|
cannam@147
|
166 }
|
cannam@147
|
167
|
cannam@147
|
168 inline void StringTree::fill(char* pos, size_t branchIndex) {
|
cannam@147
|
169 KJ_IREQUIRE(pos == text.end() && branchIndex == branches.size(),
|
cannam@147
|
170 kj::str(text.end() - pos, ' ', branches.size() - branchIndex).cStr());
|
cannam@147
|
171 }
|
cannam@147
|
172
|
cannam@147
|
173 template <typename First, typename... Rest>
|
cannam@147
|
174 void StringTree::fill(char* pos, size_t branchIndex, First&& first, Rest&&... rest) {
|
cannam@147
|
175 pos = _::fill(pos, kj::fwd<First>(first));
|
cannam@147
|
176 fill(pos, branchIndex, kj::fwd<Rest>(rest)...);
|
cannam@147
|
177 }
|
cannam@147
|
178
|
cannam@147
|
179 template <typename... Rest>
|
cannam@147
|
180 void StringTree::fill(char* pos, size_t branchIndex, StringTree&& first, Rest&&... rest) {
|
cannam@147
|
181 branches[branchIndex].index = pos - text.begin();
|
cannam@147
|
182 branches[branchIndex].content = kj::mv(first);
|
cannam@147
|
183 fill(pos, branchIndex + 1, kj::fwd<Rest>(rest)...);
|
cannam@147
|
184 }
|
cannam@147
|
185
|
cannam@147
|
186 template <typename... Rest>
|
cannam@147
|
187 void StringTree::fill(char* pos, size_t branchIndex, String&& first, Rest&&... rest) {
|
cannam@147
|
188 branches[branchIndex].index = pos - text.begin();
|
cannam@147
|
189 branches[branchIndex].content = StringTree(kj::mv(first));
|
cannam@147
|
190 fill(pos, branchIndex + 1, kj::fwd<Rest>(rest)...);
|
cannam@147
|
191 }
|
cannam@147
|
192
|
cannam@147
|
193 template <typename... Params>
|
cannam@147
|
194 StringTree StringTree::concat(Params&&... params) {
|
cannam@147
|
195 StringTree result;
|
cannam@147
|
196 result.size_ = _::sum({params.size()...});
|
cannam@147
|
197 result.text = heapString(
|
cannam@147
|
198 _::sum({StringTree::flatSize(kj::fwd<Params>(params))...}));
|
cannam@147
|
199 result.branches = heapArray<StringTree::Branch>(
|
cannam@147
|
200 _::sum({StringTree::branchCount(kj::fwd<Params>(params))...}));
|
cannam@147
|
201 result.fill(result.text.begin(), 0, kj::fwd<Params>(params)...);
|
cannam@147
|
202 return result;
|
cannam@147
|
203 }
|
cannam@147
|
204
|
cannam@147
|
205 template <typename... Params>
|
cannam@147
|
206 StringTree strTree(Params&&... params) {
|
cannam@147
|
207 return StringTree::concat(_::toStringTreeOrCharSequence(kj::fwd<Params>(params))...);
|
cannam@147
|
208 }
|
cannam@147
|
209
|
cannam@147
|
210 } // namespace kj
|
cannam@147
|
211
|
cannam@147
|
212 #endif // KJ_STRING_TREE_H_
|