1 | /* Copyright 2018 The TensorFlow Authors. All Rights Reserved. |
2 | |
3 | Licensed under the Apache License, Version 2.0 (the "License"); |
4 | you may not use this file except in compliance with the License. |
5 | You may obtain a copy of the License at |
6 | |
7 | http://www.apache.org/licenses/LICENSE-2.0 |
8 | |
9 | Unless required by applicable law or agreed to in writing, software |
10 | distributed under the License is distributed on an "AS IS" BASIS, |
11 | WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
12 | See the License for the specific language governing permissions and |
13 | limitations under the License. |
14 | ==============================================================================*/ |
15 | |
16 | #include <string> |
17 | |
18 | #include "re2/re2.h" |
19 | #include "tensorflow/core/framework/op_kernel.h" |
20 | #include "tensorflow/core/framework/tensor.h" |
21 | #include "tensorflow/core/lib/core/errors.h" |
22 | #include "tensorflow/core/lib/core/status.h" |
23 | #include "tensorflow/core/platform/mutex.h" |
24 | #include "tensorflow/core/platform/thread_annotations.h" |
25 | #include "tensorflow/core/util/ptr_util.h" |
26 | |
27 | namespace tensorflow { |
28 | |
29 | class RegexFullMatchOp : public OpKernel { |
30 | public: |
31 | explicit RegexFullMatchOp(OpKernelConstruction* ctx) : OpKernel(ctx) {} |
32 | |
33 | ~RegexFullMatchOp() override {} |
34 | |
35 | void Compute(OpKernelContext* ctx) override { |
36 | const Tensor* input_tensor; |
37 | OP_REQUIRES_OK(ctx, ctx->input("input" , &input_tensor)); |
38 | const auto& input_flat = input_tensor->flat<tstring>(); |
39 | |
40 | const Tensor* pattern_tensor; |
41 | OP_REQUIRES_OK(ctx, ctx->input("pattern" , &pattern_tensor)); |
42 | OP_REQUIRES(ctx, TensorShapeUtils::IsScalar(pattern_tensor->shape()), |
43 | errors::InvalidArgument("Pattern must be scalar, but received " , |
44 | pattern_tensor->shape().DebugString())); |
45 | const string pattern = pattern_tensor->flat<tstring>()(0); |
46 | std::shared_ptr<RE2> regex = CachedRE2(pattern); |
47 | OP_REQUIRES(ctx, regex->ok(), |
48 | errors::InvalidArgument("Invalid pattern: " , pattern, |
49 | ", error: " , regex->error())); |
50 | |
51 | Tensor* output_tensor = nullptr; |
52 | OP_REQUIRES_OK(ctx, ctx->allocate_output("output" , input_tensor->shape(), |
53 | &output_tensor)); |
54 | auto output_flat = output_tensor->flat<bool>(); |
55 | for (size_t i = 0; i < input_flat.size(); ++i) { |
56 | output_flat(i) = RE2::FullMatch(input_flat(i), *regex); |
57 | } |
58 | } |
59 | |
60 | private: |
61 | std::shared_ptr<RE2> CachedRE2(const string& pattern) { |
62 | { |
63 | tf_shared_lock l(mu_); |
64 | if (regex_ != nullptr && regex_->pattern() == pattern) { |
65 | return regex_; |
66 | } |
67 | } |
68 | // Construct the new RE2 object before acquiring the lock. |
69 | auto regex = std::make_shared<RE2>(pattern); |
70 | { |
71 | mutex_lock l(mu_); |
72 | // Swap instead of assigning so that we destruct the old |
73 | // RE2 object (when necessary) after releasing the lock. |
74 | regex_.swap(regex); |
75 | return regex_; |
76 | } |
77 | } |
78 | |
79 | mutex mu_; |
80 | std::shared_ptr<RE2> regex_ TF_GUARDED_BY(mu_); |
81 | |
82 | TF_DISALLOW_COPY_AND_ASSIGN(RegexFullMatchOp); |
83 | }; |
84 | |
85 | REGISTER_KERNEL_BUILDER(Name("RegexFullMatch" ).Device(DEVICE_CPU), |
86 | RegexFullMatchOp); |
87 | |
88 | class StaticRegexFullMatchOp : public OpKernel { |
89 | public: |
90 | explicit StaticRegexFullMatchOp(OpKernelConstruction* ctx) : OpKernel(ctx) { |
91 | string pattern; |
92 | OP_REQUIRES_OK(ctx, ctx->GetAttr("pattern" , &pattern)); |
93 | re_ = MakeUnique<RE2>(pattern); |
94 | OP_REQUIRES(ctx, re_->ok(), |
95 | errors::InvalidArgument("Invalid pattern: " , pattern, |
96 | ", error: " , re_->error())); |
97 | } |
98 | |
99 | void Compute(OpKernelContext* ctx) override { |
100 | const Tensor* input_tensor; |
101 | OP_REQUIRES_OK(ctx, ctx->input("input" , &input_tensor)); |
102 | const auto& input_flat = input_tensor->flat<tstring>(); |
103 | |
104 | Tensor* output_tensor = nullptr; |
105 | OP_REQUIRES_OK(ctx, ctx->allocate_output("output" , input_tensor->shape(), |
106 | &output_tensor)); |
107 | auto output_flat = output_tensor->flat<bool>(); |
108 | for (size_t i = 0; i < input_flat.size(); ++i) { |
109 | output_flat(i) = RE2::FullMatch(input_flat(i), *re_); |
110 | } |
111 | } |
112 | |
113 | private: |
114 | std::unique_ptr<RE2> re_; |
115 | }; |
116 | |
117 | REGISTER_KERNEL_BUILDER(Name("StaticRegexFullMatch" ).Device(DEVICE_CPU), |
118 | StaticRegexFullMatchOp); |
119 | |
120 | } // namespace tensorflow |
121 | |