xref: /aosp_15_r20/external/abseil-cpp/absl/strings/str_split.cc (revision 9356374a3709195abf420251b3e825997ff56c0f)
1*9356374aSAndroid Build Coastguard Worker // Copyright 2017 The Abseil Authors.
2*9356374aSAndroid Build Coastguard Worker //
3*9356374aSAndroid Build Coastguard Worker // Licensed under the Apache License, Version 2.0 (the "License");
4*9356374aSAndroid Build Coastguard Worker // you may not use this file except in compliance with the License.
5*9356374aSAndroid Build Coastguard Worker // You may obtain a copy of the License at
6*9356374aSAndroid Build Coastguard Worker //
7*9356374aSAndroid Build Coastguard Worker //      https://www.apache.org/licenses/LICENSE-2.0
8*9356374aSAndroid Build Coastguard Worker //
9*9356374aSAndroid Build Coastguard Worker // Unless required by applicable law or agreed to in writing, software
10*9356374aSAndroid Build Coastguard Worker // distributed under the License is distributed on an "AS IS" BASIS,
11*9356374aSAndroid Build Coastguard Worker // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12*9356374aSAndroid Build Coastguard Worker // See the License for the specific language governing permissions and
13*9356374aSAndroid Build Coastguard Worker // limitations under the License.
14*9356374aSAndroid Build Coastguard Worker 
15*9356374aSAndroid Build Coastguard Worker #include "absl/strings/str_split.h"
16*9356374aSAndroid Build Coastguard Worker 
17*9356374aSAndroid Build Coastguard Worker #include <algorithm>
18*9356374aSAndroid Build Coastguard Worker #include <cstddef>
19*9356374aSAndroid Build Coastguard Worker #include <cstdlib>
20*9356374aSAndroid Build Coastguard Worker #include <cstring>
21*9356374aSAndroid Build Coastguard Worker 
22*9356374aSAndroid Build Coastguard Worker #include "absl/base/config.h"
23*9356374aSAndroid Build Coastguard Worker #include "absl/base/internal/raw_logging.h"
24*9356374aSAndroid Build Coastguard Worker #include "absl/strings/string_view.h"
25*9356374aSAndroid Build Coastguard Worker 
26*9356374aSAndroid Build Coastguard Worker namespace absl {
27*9356374aSAndroid Build Coastguard Worker ABSL_NAMESPACE_BEGIN
28*9356374aSAndroid Build Coastguard Worker 
29*9356374aSAndroid Build Coastguard Worker namespace {
30*9356374aSAndroid Build Coastguard Worker 
31*9356374aSAndroid Build Coastguard Worker // This GenericFind() template function encapsulates the finding algorithm
32*9356374aSAndroid Build Coastguard Worker // shared between the ByString and ByAnyChar delimiters. The FindPolicy
33*9356374aSAndroid Build Coastguard Worker // template parameter allows each delimiter to customize the actual find
34*9356374aSAndroid Build Coastguard Worker // function to use and the length of the found delimiter. For example, the
35*9356374aSAndroid Build Coastguard Worker // Literal delimiter will ultimately use absl::string_view::find(), and the
36*9356374aSAndroid Build Coastguard Worker // AnyOf delimiter will use absl::string_view::find_first_of().
37*9356374aSAndroid Build Coastguard Worker template <typename FindPolicy>
GenericFind(absl::string_view text,absl::string_view delimiter,size_t pos,FindPolicy find_policy)38*9356374aSAndroid Build Coastguard Worker absl::string_view GenericFind(absl::string_view text,
39*9356374aSAndroid Build Coastguard Worker                               absl::string_view delimiter, size_t pos,
40*9356374aSAndroid Build Coastguard Worker                               FindPolicy find_policy) {
41*9356374aSAndroid Build Coastguard Worker   if (delimiter.empty() && text.length() > 0) {
42*9356374aSAndroid Build Coastguard Worker     // Special case for empty string delimiters: always return a zero-length
43*9356374aSAndroid Build Coastguard Worker     // absl::string_view referring to the item at position 1 past pos.
44*9356374aSAndroid Build Coastguard Worker     return absl::string_view(text.data() + pos + 1, 0);
45*9356374aSAndroid Build Coastguard Worker   }
46*9356374aSAndroid Build Coastguard Worker   size_t found_pos = absl::string_view::npos;
47*9356374aSAndroid Build Coastguard Worker   absl::string_view found(text.data() + text.size(),
48*9356374aSAndroid Build Coastguard Worker                           0);  // By default, not found
49*9356374aSAndroid Build Coastguard Worker   found_pos = find_policy.Find(text, delimiter, pos);
50*9356374aSAndroid Build Coastguard Worker   if (found_pos != absl::string_view::npos) {
51*9356374aSAndroid Build Coastguard Worker     found = absl::string_view(text.data() + found_pos,
52*9356374aSAndroid Build Coastguard Worker                               find_policy.Length(delimiter));
53*9356374aSAndroid Build Coastguard Worker   }
54*9356374aSAndroid Build Coastguard Worker   return found;
55*9356374aSAndroid Build Coastguard Worker }
56*9356374aSAndroid Build Coastguard Worker 
57*9356374aSAndroid Build Coastguard Worker // Finds using absl::string_view::find(), therefore the length of the found
58*9356374aSAndroid Build Coastguard Worker // delimiter is delimiter.length().
59*9356374aSAndroid Build Coastguard Worker struct LiteralPolicy {
Findabsl::__anone98186780111::LiteralPolicy60*9356374aSAndroid Build Coastguard Worker   static size_t Find(absl::string_view text, absl::string_view delimiter,
61*9356374aSAndroid Build Coastguard Worker                      size_t pos) {
62*9356374aSAndroid Build Coastguard Worker     return text.find(delimiter, pos);
63*9356374aSAndroid Build Coastguard Worker   }
Lengthabsl::__anone98186780111::LiteralPolicy64*9356374aSAndroid Build Coastguard Worker   static size_t Length(absl::string_view delimiter) {
65*9356374aSAndroid Build Coastguard Worker     return delimiter.length();
66*9356374aSAndroid Build Coastguard Worker   }
67*9356374aSAndroid Build Coastguard Worker };
68*9356374aSAndroid Build Coastguard Worker 
69*9356374aSAndroid Build Coastguard Worker // Finds using absl::string_view::find_first_of(), therefore the length of the
70*9356374aSAndroid Build Coastguard Worker // found delimiter is 1.
71*9356374aSAndroid Build Coastguard Worker struct AnyOfPolicy {
Findabsl::__anone98186780111::AnyOfPolicy72*9356374aSAndroid Build Coastguard Worker   static size_t Find(absl::string_view text, absl::string_view delimiter,
73*9356374aSAndroid Build Coastguard Worker                      size_t pos) {
74*9356374aSAndroid Build Coastguard Worker     return text.find_first_of(delimiter, pos);
75*9356374aSAndroid Build Coastguard Worker   }
Lengthabsl::__anone98186780111::AnyOfPolicy76*9356374aSAndroid Build Coastguard Worker   static size_t Length(absl::string_view /* delimiter */) { return 1; }
77*9356374aSAndroid Build Coastguard Worker };
78*9356374aSAndroid Build Coastguard Worker 
79*9356374aSAndroid Build Coastguard Worker }  // namespace
80*9356374aSAndroid Build Coastguard Worker 
81*9356374aSAndroid Build Coastguard Worker //
82*9356374aSAndroid Build Coastguard Worker // ByString
83*9356374aSAndroid Build Coastguard Worker //
84*9356374aSAndroid Build Coastguard Worker 
ByString(absl::string_view sp)85*9356374aSAndroid Build Coastguard Worker ByString::ByString(absl::string_view sp) : delimiter_(sp) {}
86*9356374aSAndroid Build Coastguard Worker 
Find(absl::string_view text,size_t pos) const87*9356374aSAndroid Build Coastguard Worker absl::string_view ByString::Find(absl::string_view text, size_t pos) const {
88*9356374aSAndroid Build Coastguard Worker   if (delimiter_.length() == 1) {
89*9356374aSAndroid Build Coastguard Worker     // Much faster to call find on a single character than on an
90*9356374aSAndroid Build Coastguard Worker     // absl::string_view.
91*9356374aSAndroid Build Coastguard Worker     size_t found_pos = text.find(delimiter_[0], pos);
92*9356374aSAndroid Build Coastguard Worker     if (found_pos == absl::string_view::npos)
93*9356374aSAndroid Build Coastguard Worker       return absl::string_view(text.data() + text.size(), 0);
94*9356374aSAndroid Build Coastguard Worker     return text.substr(found_pos, 1);
95*9356374aSAndroid Build Coastguard Worker   }
96*9356374aSAndroid Build Coastguard Worker   return GenericFind(text, delimiter_, pos, LiteralPolicy());
97*9356374aSAndroid Build Coastguard Worker }
98*9356374aSAndroid Build Coastguard Worker 
Find(absl::string_view text,size_t pos) const99*9356374aSAndroid Build Coastguard Worker absl::string_view ByAsciiWhitespace::Find(absl::string_view text,
100*9356374aSAndroid Build Coastguard Worker                                           size_t pos) const {
101*9356374aSAndroid Build Coastguard Worker   return GenericFind(text, " \t\v\f\r\n", pos, AnyOfPolicy());
102*9356374aSAndroid Build Coastguard Worker }
103*9356374aSAndroid Build Coastguard Worker 
104*9356374aSAndroid Build Coastguard Worker //
105*9356374aSAndroid Build Coastguard Worker // ByChar
106*9356374aSAndroid Build Coastguard Worker //
107*9356374aSAndroid Build Coastguard Worker 
Find(absl::string_view text,size_t pos) const108*9356374aSAndroid Build Coastguard Worker absl::string_view ByChar::Find(absl::string_view text, size_t pos) const {
109*9356374aSAndroid Build Coastguard Worker   size_t found_pos = text.find(c_, pos);
110*9356374aSAndroid Build Coastguard Worker   if (found_pos == absl::string_view::npos)
111*9356374aSAndroid Build Coastguard Worker     return absl::string_view(text.data() + text.size(), 0);
112*9356374aSAndroid Build Coastguard Worker   return text.substr(found_pos, 1);
113*9356374aSAndroid Build Coastguard Worker }
114*9356374aSAndroid Build Coastguard Worker 
115*9356374aSAndroid Build Coastguard Worker //
116*9356374aSAndroid Build Coastguard Worker // ByAnyChar
117*9356374aSAndroid Build Coastguard Worker //
118*9356374aSAndroid Build Coastguard Worker 
ByAnyChar(absl::string_view sp)119*9356374aSAndroid Build Coastguard Worker ByAnyChar::ByAnyChar(absl::string_view sp) : delimiters_(sp) {}
120*9356374aSAndroid Build Coastguard Worker 
Find(absl::string_view text,size_t pos) const121*9356374aSAndroid Build Coastguard Worker absl::string_view ByAnyChar::Find(absl::string_view text, size_t pos) const {
122*9356374aSAndroid Build Coastguard Worker   return GenericFind(text, delimiters_, pos, AnyOfPolicy());
123*9356374aSAndroid Build Coastguard Worker }
124*9356374aSAndroid Build Coastguard Worker 
125*9356374aSAndroid Build Coastguard Worker //
126*9356374aSAndroid Build Coastguard Worker // ByLength
127*9356374aSAndroid Build Coastguard Worker //
ByLength(ptrdiff_t length)128*9356374aSAndroid Build Coastguard Worker ByLength::ByLength(ptrdiff_t length) : length_(length) {
129*9356374aSAndroid Build Coastguard Worker   ABSL_RAW_CHECK(length > 0, "");
130*9356374aSAndroid Build Coastguard Worker }
131*9356374aSAndroid Build Coastguard Worker 
Find(absl::string_view text,size_t pos) const132*9356374aSAndroid Build Coastguard Worker absl::string_view ByLength::Find(absl::string_view text, size_t pos) const {
133*9356374aSAndroid Build Coastguard Worker   pos = std::min(pos, text.size());  // truncate `pos`
134*9356374aSAndroid Build Coastguard Worker   absl::string_view substr = text.substr(pos);
135*9356374aSAndroid Build Coastguard Worker   // If the string is shorter than the chunk size we say we
136*9356374aSAndroid Build Coastguard Worker   // "can't find the delimiter" so this will be the last chunk.
137*9356374aSAndroid Build Coastguard Worker   if (substr.length() <= static_cast<size_t>(length_))
138*9356374aSAndroid Build Coastguard Worker     return absl::string_view(text.data() + text.size(), 0);
139*9356374aSAndroid Build Coastguard Worker 
140*9356374aSAndroid Build Coastguard Worker   return absl::string_view(substr.data() + length_, 0);
141*9356374aSAndroid Build Coastguard Worker }
142*9356374aSAndroid Build Coastguard Worker 
143*9356374aSAndroid Build Coastguard Worker ABSL_NAMESPACE_END
144*9356374aSAndroid Build Coastguard Worker }  // namespace absl
145