op_desc.cc 28.5 KB
Newer Older
1
/* Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved.
F
fengjiayi 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License. */

Y
Yi Wang 已提交
15
#include "paddle/fluid/framework/op_desc.h"
16
#include <algorithm>
Y
Yu Yang 已提交
17
#include <functional>
18 19
#include <mutex>  // NOLINT
#include <string>
Y
Yu Yang 已提交
20
#include <unordered_map>
21
#include <utility>
22
#include "glog/logging.h"
Y
Yi Wang 已提交
23
#include "paddle/fluid/framework/block_desc.h"
24
#include "paddle/fluid/framework/op_call_stack.h"
Y
yuyang18 已提交
25
#include "paddle/fluid/framework/op_proto_maker.h"
Y
Yi Wang 已提交
26 27 28
#include "paddle/fluid/framework/operator.h"
#include "paddle/fluid/framework/program_desc.h"
#include "paddle/fluid/framework/shape_inference.h"
M
minqiyang 已提交
29
#include "paddle/fluid/framework/var_type_inference.h"
Y
Yu Yang 已提交
30

F
fengjiayi 已提交
31 32 33
namespace paddle {
namespace framework {

Y
Yu Yang 已提交
34 35
class OpDesc;
class BlockDesc;
36 37
class CompileTimeInferShapeContext : public InferShapeContext {
 public:
Y
Yu Yang 已提交
38
  CompileTimeInferShapeContext(const OpDesc &op, const BlockDesc &block);
39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55

  bool HasInput(const std::string &name) const override;

  bool HasOutput(const std::string &name) const override;

  bool HasInputs(const std::string &name) const override;

  bool HasOutputs(const std::string &name) const override;

  AttrReader Attrs() const override;

  const std::vector<std::string> &Inputs(
      const std::string &name) const override;

  const std::vector<std::string> &Outputs(
      const std::string &name) const override;

56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
  void ShareDim(const std::string &in, const std::string &out, size_t i = 0,
                size_t j = 0) override {
    PADDLE_ENFORCE_LT(i, Inputs(in).size());
    PADDLE_ENFORCE_LT(j, Outputs(out).size());
    const std::string &input_n = Inputs(in)[i];
    const std::string &output_n = Outputs(out)[j];

    PADDLE_ENFORCE(input_n != framework::kEmptyVarName, "The %s[%d] is @EMPTY@",
                   in, i);
    PADDLE_ENFORCE(output_n != framework::kEmptyVarName,
                   "The %s[%d] is @EMPTY@", out, j);

    auto *in_var = block_.FindVarRecursive(input_n);
    auto *out_var = block_.FindVarRecursive(output_n);

    PADDLE_ENFORCE(in_var->GetType() == out_var->GetType(),
                   "The type of %s and %s is not the same.", input_n, output_n);

    SetDim(output_n, GetDim(input_n));
  }

Q
Qiao Longfei 已提交
77 78 79 80
  void ShareLoD(const std::string &in, const std::string &out, size_t i = 0,
                size_t j = 0) const override {
    PADDLE_ENFORCE_LT(i, Inputs(in).size());
    PADDLE_ENFORCE_LT(j, Outputs(out).size());
C
chengduo 已提交
81 82 83 84
    PADDLE_ENFORCE(Inputs(in)[i] != framework::kEmptyVarName,
                   "The %s[%d] is @EMPTY@", in, i);
    PADDLE_ENFORCE(Outputs(out)[j] != framework::kEmptyVarName,
                   "The %s[%d] is @EMPTY@", out, j);
Q
Qiao Longfei 已提交
85 86
    auto *in_var = block_.FindVarRecursive(Inputs(in)[i]);
    auto *out_var = block_.FindVarRecursive(Outputs(out)[j]);
C
chengduo 已提交
87 88
    if (in_var->GetType() != proto::VarType::LOD_TENSOR &&
        in_var->GetType() != proto::VarType::LOD_TENSOR_ARRAY) {
89
      VLOG(3) << "input " << in << " is not LoDTensor or LoDTensorArray.";
X
fix  
Xin Pan 已提交
90 91
      return;
    }
92
    out_var->SetLoDLevel(in_var->GetLoDLevel());
Q
Qiao Longfei 已提交
93
  }
D
dzhwinter 已提交
94

C
chengduo 已提交
95 96
  void DecreaseLoDLevel(const std::string &in, const std::string &out,
                        size_t i = 0, size_t j = 0) const override {
97 98
    // When in is a LoDTensor and out is a LoDTensorArray, there may need to
    // decrease the lod_level.
C
chengduo 已提交
99 100 101 102 103 104 105 106
    PADDLE_ENFORCE_LT(i, Inputs(in).size());
    PADDLE_ENFORCE_LT(j, Outputs(out).size());
    PADDLE_ENFORCE(Inputs(in)[i] != framework::kEmptyVarName,
                   "The %s[%d] is @EMPTY@", in, i);
    PADDLE_ENFORCE(Outputs(out)[j] != framework::kEmptyVarName,
                   "The %s[%d] is @EMPTY@", out, j);
    auto *in_var = block_.FindVarRecursive(Inputs(in)[i]);
    auto *out_var = block_.FindVarRecursive(Outputs(out)[j]);
107 108 109 110 111
    PADDLE_ENFORCE_EQ(in_var->GetType(), proto::VarType::LOD_TENSOR,
                      "The input %s should be LoDTensor.", in_var->Name());
    PADDLE_ENFORCE_EQ(out_var->GetType(), proto::VarType::LOD_TENSOR_ARRAY,
                      "The output %s should be LoDTensorArray.",
                      out_var->Name());
C
chengduo 已提交
112 113 114 115 116
    if (in_var->GetLoDLevel() > 0) {
      out_var->SetLoDLevel(in_var->GetLoDLevel() - 1);
    }
  }

117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135
  void IncreaseLoDLevel(const std::string &in, const std::string &out,
                        size_t i = 0, size_t j = 0) const override {
    // When in is a LoDTensorArray and out is a LoDTensor, there may need to
    // increase the lod_level.
    PADDLE_ENFORCE_LT(i, Inputs(in).size());
    PADDLE_ENFORCE_LT(j, Outputs(out).size());
    PADDLE_ENFORCE_NE(Inputs(in)[i], framework::kEmptyVarName,
                      "The %s[%d] is @EMPTY@", in, i);
    PADDLE_ENFORCE_NE(Outputs(out)[j], framework::kEmptyVarName,
                      "The %s[%d] is @EMPTY@", out, j);
    auto *in_var = block_.FindVarRecursive(Inputs(in)[i]);
    auto *out_var = block_.FindVarRecursive(Outputs(out)[j]);
    PADDLE_ENFORCE_EQ(in_var->GetType(), proto::VarType::LOD_TENSOR_ARRAY,
                      "The input %s should be LoDTensorArray.", in_var->Name());
    PADDLE_ENFORCE_EQ(out_var->GetType(), proto::VarType::LOD_TENSOR,
                      "The output %s should be LoDTensor.", out_var->Name());
    out_var->SetLoDLevel(in_var->GetLoDLevel() + 1);
  }

136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
  std::vector<InferShapeVarPtr> GetInputVarPtrs(
      const std::string &name) override {
    const std::vector<std::string> arg_names = Inputs(name);
    std::vector<InferShapeVarPtr> res;
    res.reserve(arg_names.size());
    std::transform(arg_names.begin(), arg_names.end(), std::back_inserter(res),
                   [this](const std::string &name) {
                     return block_.FindVarRecursive(name);
                   });
    return res;
  }

  std::vector<InferShapeVarPtr> GetOutputVarPtrs(
      const std::string &name) override {
    const std::vector<std::string> arg_names = Outputs(name);
    std::vector<InferShapeVarPtr> res;
    res.reserve(arg_names.size());
    std::transform(arg_names.begin(), arg_names.end(), std::back_inserter(res),
                   [this](const std::string &name) {
                     return block_.FindVarRecursive(name);
                   });
    return res;
  }

X
Xin Pan 已提交
160 161 162 163 164 165 166 167 168 169 170 171 172
  DDim GetInputDim(const std::string &name) const override {
    const std::vector<std::string> &arg_names = Inputs(name);
    PADDLE_ENFORCE_EQ(arg_names.size(), 1UL,
                      "Input(%s) should hold one element, but now it holds %d",
                      name, arg_names.size());
    return this->GetDim(arg_names[0]);
  }

  std::vector<DDim> GetInputsDim(const std::string &name) const override {
    const std::vector<std::string> &arg_names = Inputs(name);
    return GetDims(arg_names);
  }

173 174
  bool IsRuntime() const override;

X
Xin Pan 已提交
175 176 177 178 179 180 181 182 183 184
  std::vector<proto::VarType::Type> GetInputsVarType(
      const std::string &name) const override {
    return GetVarTypes(Inputs(name));
  }

  std::vector<proto::VarType::Type> GetOutputsVarType(
      const std::string &name) const override {
    return GetVarTypes(Outputs(name));
  }

X
Xin Pan 已提交
185 186 187 188 189 190 191 192 193 194 195 196 197 198
  void SetOutputDim(const std::string &name, const DDim &dim) override {
    auto &arg_names = Outputs(name);
    PADDLE_ENFORCE_EQ(arg_names.size(), 1UL,
                      "Output(%s) should hold one element, but now it holds %d",
                      name, arg_names.size());
    SetDim(arg_names[0], dim);
  }

  void SetOutputsDim(const std::string &name,
                     const std::vector<DDim> &dims) override {
    auto &names = Outputs(name);
    SetDims(names, dims);
  }

199
 protected:
X
Xin Pan 已提交
200 201 202 203 204 205 206 207 208 209 210 211
  std::vector<proto::VarType::Type> GetVarTypes(
      const std::vector<std::string> &names) const {
    std::vector<proto::VarType::Type> retv;
    retv.resize(names.size());
    std::transform(
        names.begin(), names.end(), retv.begin(),
        std::bind(std::mem_fn(&CompileTimeInferShapeContext::GetVarType), this,
                  std::placeholders::_1));
    return retv;
  }

  proto::VarType::Type GetVarType(const std::string &name) const;
Q
Qiao Longfei 已提交
212

X
Xin Pan 已提交
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234
  DDim GetDim(const std::string &name) const {
    auto var = block_.FindVarRecursive(name);
    PADDLE_ENFORCE(var != nullptr, "Cannot find variable %s", name);
    DDim res;
    try {
      auto shape = var->GetShape();
      res = shape.empty() ? make_ddim({0UL}) : make_ddim(shape);
    } catch (...) {
      VLOG(5) << "GetDim of variable " << name << " error";
      std::rethrow_exception(std::current_exception());
    }
    return res;
  }

  std::vector<DDim> GetDims(const std::vector<std::string> &names) const {
    std::vector<DDim> ret;
    ret.reserve(names.size());
    std::transform(
        names.begin(), names.end(), std::back_inserter(ret),
        [this](const std::string &name) { return this->GetDim(name); });
    return ret;
  }
235

X
Xin Pan 已提交
236 237 238 239 240 241 242 243 244 245 246 247 248
  void SetDim(const std::string &name, const DDim &dim);

  void SetDims(const std::vector<std::string> &names,
               const std::vector<DDim> &dims) {
    size_t length = names.size();
    PADDLE_ENFORCE_EQ(length, dims.size());
    for (size_t i = 0; i < length; ++i) {
      if (names[i] == framework::kEmptyVarName) {
        continue;
      }
      SetDim(names[i], dims[i]);
    }
  }
249

F
fengjiayi 已提交
250 251 252 253
  std::vector<DDim> GetRepeatedDims(const std::string &name) const override;

  void SetRepeatedDims(const std::string &name,
                       const std::vector<DDim> &dims) override;
F
fengjiayi 已提交
254

Y
Yu Yang 已提交
255 256
  const OpDesc &op_;
  const BlockDesc &block_;
257 258
};

Y
Yu Yang 已提交
259 260
OpDesc::OpDesc(const std::string &type, const VariableNameMap &inputs,
               const VariableNameMap &outputs, const AttributeMap &attrs) {
261
  desc_.set_type(type);
F
fengjiayi 已提交
262 263 264
  inputs_ = inputs;
  outputs_ = outputs;
  attrs_ = attrs;
F
Fix bug  
fengjiayi 已提交
265
  need_update_ = true;
L
liuwei1031 已提交
266
  block_ = nullptr;
F
fengjiayi 已提交
267 268
}

X
Xin Pan 已提交
269 270 271 272 273 274
OpDesc::OpDesc(const OpDesc &other, BlockDesc *block) {
  CopyFrom(other);
  block_ = block;
  need_update_ = true;
}

275
void OpDesc::CopyFrom(const OpDesc &op_desc) {
F
fengjiayi 已提交
276 277 278 279 280 281 282
  desc_.set_type(op_desc.Type());
  inputs_ = op_desc.inputs_;
  outputs_ = op_desc.outputs_;
  attrs_ = op_desc.attrs_;
  need_update_ = true;
}

F
fengjiayi 已提交
283
OpDesc::OpDesc(const proto::OpDesc &desc, BlockDesc *block)
284 285 286 287
    : desc_(desc), need_update_(false) {
  // restore inputs_
  int input_size = desc_.inputs_size();
  for (int i = 0; i < input_size; ++i) {
288
    const proto::OpDesc::Var &var = desc_.inputs(i);
289 290 291 292 293 294 295 296 297 298
    std::vector<std::string> &args = inputs_[var.parameter()];
    int argu_size = var.arguments_size();
    args.reserve(argu_size);
    for (int j = 0; j < argu_size; ++j) {
      args.push_back(var.arguments(j));
    }
  }
  // restore outputs_
  int output_size = desc_.outputs_size();
  for (int i = 0; i < output_size; ++i) {
299
    const proto::OpDesc::Var &var = desc_.outputs(i);
300 301 302 303 304 305 306 307
    std::vector<std::string> &args = outputs_[var.parameter()];
    int argu_size = var.arguments_size();
    args.reserve(argu_size);
    for (int j = 0; j < argu_size; ++j) {
      args.push_back(var.arguments(j));
    }
  }
  // restore attrs_
308
  for (const proto::OpDesc::Attr &attr : desc_.attrs()) {
309
    std::string attr_name = attr.name();
310
    // The sub_block referred to by the BLOCK attr hasn't been added
X
Xin Pan 已提交
311 312 313
    // to ProgramDesc class yet, we skip setting BLOCK/BLOCKS attr here.
    if (attr.type() != proto::AttrType::BLOCK &&
        attr.type() != proto::AttrType::BLOCKS) {
314 315
      attrs_[attr_name] = GetAttrValue(attr);
    }
316
  }
317
  this->block_ = block;
318 319
}

Y
Yu Yang 已提交
320
proto::OpDesc *OpDesc::Proto() {
321
  Flush();
322
  return &desc_;
F
fengjiayi 已提交
323 324
}

Y
Yu Yang 已提交
325
const std::vector<std::string> &OpDesc::Input(const std::string &name) const {
F
fengjiayi 已提交
326 327 328 329 330 331
  auto it = inputs_.find(name);
  PADDLE_ENFORCE(it != inputs_.end(), "Input %s cannot be found in Op %s", name,
                 Type());
  return it->second;
}

Y
Yu Yang 已提交
332
std::vector<std::string> OpDesc::InputArgumentNames() const {
F
Update  
fengjiayi 已提交
333 334 335 336 337 338 339
  std::vector<std::string> retv;
  for (auto &ipt : this->inputs_) {
    retv.insert(retv.end(), ipt.second.begin(), ipt.second.end());
  }
  return retv;
}

Y
Yu Yang 已提交
340 341
void OpDesc::SetInput(const std::string &param_name,
                      const std::vector<std::string> &args) {
F
fengjiayi 已提交
342 343 344 345
  need_update_ = true;
  inputs_[param_name] = args;
}

Y
Yu Yang 已提交
346
const std::vector<std::string> &OpDesc::Output(const std::string &name) const {
F
fengjiayi 已提交
347 348 349 350 351 352
  auto it = outputs_.find(name);
  PADDLE_ENFORCE(it != outputs_.end(), "Output %s cannot be found in Op %s",
                 name, Type());
  return it->second;
}

Y
Yu Yang 已提交
353
std::vector<std::string> OpDesc::OutputArgumentNames() const {
F
Update  
fengjiayi 已提交
354 355 356 357 358 359 360
  std::vector<std::string> retv;
  for (auto &ipt : this->outputs_) {
    retv.insert(retv.end(), ipt.second.begin(), ipt.second.end());
  }
  return retv;
}

Y
Yu Yang 已提交
361 362
void OpDesc::SetOutput(const std::string &param_name,
                       const std::vector<std::string> &args) {
F
fengjiayi 已提交
363 364 365 366
  need_update_ = true;
  this->outputs_[param_name] = args;
}

367 368 369 370 371 372 373 374 375 376
bool OpDesc::HasProtoAttr(const std::string &name) const {
  auto &op_info = OpInfoMap::Instance();
  if (op_info.Has(desc_.type())) {
    auto op_info_ptr = op_info.Get(desc_.type());
    if (op_info_ptr.HasOpProtoAndChecker()) {
      const proto::OpProto &proto = op_info_ptr.Proto();
      for (int i = 0; i != proto.attrs_size(); ++i) {
        const proto::OpProto::Attr &attr = proto.attrs(i);
        if (attr.name() == name) {
          return true;
L
luotao1 已提交
377 378
        }
      }
L
luotao1 已提交
379 380 381 382 383
    }
  }
  return false;
}

Y
Yu Yang 已提交
384
proto::AttrType OpDesc::GetAttrType(const std::string &name) const {
F
fengjiayi 已提交
385 386
  auto it = attrs_.find(name);
  PADDLE_ENFORCE(it != attrs_.end(), "Attribute %s is not found", name);
387
  return static_cast<proto::AttrType>(it->second.which() - 1);
F
fengjiayi 已提交
388 389
}

Y
Yu Yang 已提交
390
std::vector<std::string> OpDesc::AttrNames() const {
F
fengjiayi 已提交
391 392 393 394 395 396 397 398
  std::vector<std::string> retv;
  retv.reserve(attrs_.size());
  for (auto &attr : attrs_) {
    retv.push_back(attr.first);
  }
  return retv;
}

399 400 401 402 403
void OpDesc::RemoveAttr(const std::string &name) {
  attrs_.erase(name);
  need_update_ = true;
}

Y
Yu Yang 已提交
404
void OpDesc::SetAttr(const std::string &name, const Attribute &v) {
M
minqiyang 已提交
405 406 407 408 409 410 411
  // NOTICE(minqiyang): pybind11 will take the empty list in python as
  // the std::vector<int> type in C++; so we have to change the attr's type
  // here if we meet this issue
  proto::AttrType attr_type = static_cast<proto::AttrType>(v.which() - 1);
  if (attr_type == proto::AttrType::INTS &&
      boost::get<std::vector<int>>(v).size() == 0u) {
    // Find current attr via attr name and set the correct attribute value
M
minqiyang 已提交
412
    const proto::OpProto::Attr &attr = GetProtoAttr(name);
M
minqiyang 已提交
413 414
    switch (attr.type()) {
      case proto::AttrType::BOOLEANS: {
M
minqiyang 已提交
415 416
        VLOG(11) << "SetAttr: " << Type() << ", " << name
                 << " from INTS to BOOLEANS";
M
minqiyang 已提交
417 418 419 420
        this->attrs_[name] = std::vector<bool>();
        break;
      }
      case proto::AttrType::INTS: {
M
minqiyang 已提交
421 422
        VLOG(11) << "SetAttr: " << Type() << ", " << name
                 << " from INTS to INTS";
M
minqiyang 已提交
423 424 425
        this->attrs_[name] = std::vector<int>();
        break;
      }
426
      case proto::AttrType::LONGS: {
M
minqiyang 已提交
427 428
        VLOG(11) << "SetAttr: " << Type() << ", " << name
                 << " from LONGS to LONGS";
429 430 431
        this->attrs_[name] = std::vector<int64_t>();
        break;
      }
M
minqiyang 已提交
432
      case proto::AttrType::FLOATS: {
M
minqiyang 已提交
433 434
        VLOG(11) << "SetAttr: " << Type() << ", " << name
                 << " from INTS to FLOATS";
M
minqiyang 已提交
435 436 437 438
        this->attrs_[name] = std::vector<float>();
        break;
      }
      case proto::AttrType::STRINGS: {
M
minqiyang 已提交
439 440
        VLOG(11) << "SetAttr: " << Type() << ", " << name
                 << " from INTS to STRINGS";
M
minqiyang 已提交
441 442 443 444
        this->attrs_[name] = std::vector<std::string>();
        break;
      }
      case proto::AttrType::BLOCKS: {
M
minqiyang 已提交
445 446
        VLOG(11) << "SetAttr: " << Type() << ", " << name
                 << " from INTS to BLOCKS";
M
minqiyang 已提交
447
        this->SetBlocksAttr(name, std::vector<BlockDesc *>());
M
minqiyang 已提交
448 449
        return;
      }
M
minqiyang 已提交
450 451
      default:
        PADDLE_THROW("Wrong attr type %d", attr.type());
M
minqiyang 已提交
452
    }
M
minqiyang 已提交
453 454
    need_update_ = true;
    return;
M
minqiyang 已提交
455 456
  }

F
fengjiayi 已提交
457 458 459 460
  this->attrs_[name] = v;
  need_update_ = true;
}

A
Abhinav Arora 已提交
461 462
void OpDesc::SetBlockAttr(const std::string &name, BlockDesc *block) {
  this->attrs_[name] = block;
F
fengjiayi 已提交
463
  need_update_ = true;
F
fengjiayi 已提交
464 465
}

466 467 468 469 470 471
void OpDesc::SetBlocksAttr(const std::string &name,
                           std::vector<BlockDesc *> blocks) {
  this->attrs_[name] = blocks;
  need_update_ = true;
}

Y
Yu Yang 已提交
472
void OpDesc::SetAttrMap(
F
fengjiayi 已提交
473 474 475 476 477
    const std::unordered_map<std::string, Attribute> &attr_map) {
  attrs_ = attr_map;
  need_update_ = true;
}

Y
Yu Yang 已提交
478
Attribute OpDesc::GetAttr(const std::string &name) const {
F
fengjiayi 已提交
479 480 481 482 483
  auto it = attrs_.find(name);
  PADDLE_ENFORCE(it != attrs_.end(), "Attribute %s is not found", name);
  return it->second;
}

M
minqiyang 已提交
484 485 486
const proto::OpProto::Attr &OpDesc::GetProtoAttr(
    const std::string &name) const {
  const proto::OpProto &proto = OpInfoMap::Instance().Get(Type()).Proto();
M
minqiyang 已提交
487 488 489 490 491 492 493 494 495 496
  for (int i = 0; i != proto.attrs_size(); ++i) {
    const proto::OpProto::Attr &attr = proto.attrs(i);
    if (attr.name() == name) {
      return attr;
    }
  }

  PADDLE_THROW("Attribute %s is not found in proto %s", name, proto.type());
}

Y
yuyang18 已提交
497
Attribute OpDesc::GetNullableAttr(const std::string &name) const {
Y
Fix bug  
yuyang18 已提交
498 499 500 501
  auto it = attrs_.find(name);
  if (it != attrs_.end()) {
    return it->second;
  } else {
Y
yuyang18 已提交
502
    return Attribute();
Y
Fix bug  
yuyang18 已提交
503 504 505
  }
}

G
gongweibao 已提交
506 507 508 509 510 511 512 513 514 515 516 517 518 519
std::vector<int> OpDesc::GetBlocksAttrIds(const std::string &name) const {
  auto it = attrs_.find(name);
  PADDLE_ENFORCE(it != attrs_.end(), "Attribute %s is not found", name);
  auto blocks = boost::get<std::vector<BlockDesc *>>(it->second);

  std::vector<int> ids;
  for (auto n : blocks) {
    ids.push_back(n->ID());
  }

  return ids;
}

int OpDesc::GetBlockAttrId(const std::string &name) const {
F
fengjiayi 已提交
520 521
  auto it = attrs_.find(name);
  PADDLE_ENFORCE(it != attrs_.end(), "Attribute %s is not found", name);
Y
Yu Yang 已提交
522
  return boost::get<BlockDesc *>(it->second)->ID();
F
fengjiayi 已提交
523 524
}

Y
Yu Yang 已提交
525
const std::unordered_map<std::string, Attribute> &OpDesc::GetAttrMap() const {
F
fengjiayi 已提交
526 527 528
  return attrs_;
}

Y
Yu Yang 已提交
529
void OpDesc::Rename(const std::string &old_name, const std::string &new_name) {
Y
Yancey1989 已提交
530 531
  RenameInput(old_name, new_name);
  RenameOutput(old_name, new_name);
F
fengjiayi 已提交
532 533 534
  need_update_ = true;
}

Y
Yu Yang 已提交
535 536
void OpDesc::RenameOutput(const std::string &old_name,
                          const std::string &new_name) {
Y
Yang Yang(Tony) 已提交
537 538 539 540
  for (auto &output : outputs_) {
    std::replace(output.second.begin(), output.second.end(), old_name,
                 new_name);
  }
Y
yuyang18 已提交
541 542 543 544 545 546 547

  auto it = attrs_.find(framework::OpProtoAndCheckerMaker::OpRoleVarAttrName());
  if (it != attrs_.end()) {
    auto &op_vars = boost::get<std::vector<std::string>>(it->second);
    std::replace(op_vars.begin(), op_vars.end(), old_name, new_name);
  }

Y
Yang Yang(Tony) 已提交
548 549 550
  need_update_ = true;
}

Y
Yu Yang 已提交
551 552
void OpDesc::RenameInput(const std::string &old_name,
                         const std::string &new_name) {
Y
Yang Yang(Tony) 已提交
553 554 555
  for (auto &input : inputs_) {
    std::replace(input.second.begin(), input.second.end(), old_name, new_name);
  }
Y
Yancey1989 已提交
556 557 558 559 560 561 562

  auto it = attrs_.find(framework::OpProtoAndCheckerMaker::OpRoleVarAttrName());
  if (it != attrs_.end()) {
    auto &op_vars = boost::get<std::vector<std::string>>(it->second);
    std::replace(op_vars.begin(), op_vars.end(), old_name, new_name);
  }

Y
Yang Yang(Tony) 已提交
563 564 565
  need_update_ = true;
}

Y
Yu Yang 已提交
566
struct SetAttrDescVisitor : public boost::static_visitor<void> {
567 568
  explicit SetAttrDescVisitor(proto::OpDesc::Attr *attr) : attr_(attr) {}
  mutable proto::OpDesc::Attr *attr_;
Y
Yu Yang 已提交
569 570 571
  void operator()(int v) const { attr_->set_i(v); }
  void operator()(float v) const { attr_->set_f(v); }
  void operator()(const std::string &v) const { attr_->set_s(v); }
Q
QI JUN 已提交
572 573 574 575 576 577 578

  // Please refer to https://github.com/PaddlePaddle/Paddle/issues/7162
  template <class T,
            class = typename std::enable_if<std::is_same<bool, T>::value>::type>
  void operator()(T b) const {
    attr_->set_b(b);
  }
Y
Yu Yang 已提交
579 580 581 582 583 584 585 586 587 588 589 590 591

  void operator()(const std::vector<int> &v) const {
    VectorToRepeated(v, attr_->mutable_ints());
  }
  void operator()(const std::vector<float> &v) const {
    VectorToRepeated(v, attr_->mutable_floats());
  }
  void operator()(const std::vector<std::string> &v) const {
    VectorToRepeated(v, attr_->mutable_strings());
  }
  void operator()(const std::vector<bool> &v) const {
    VectorToRepeated(v, attr_->mutable_bools());
  }
592 593 594
  void operator()(const std::vector<BlockDesc *> &v) const {
    std::vector<int> blocks_idx;
    for (auto blk : v) {
T
tangwei12 已提交
595
      blocks_idx.push_back(blk->ID());
596 597 598
    }
    VectorToRepeated(blocks_idx, attr_->mutable_blocks_idx());
  }
T
tangwei12 已提交
599 600 601

  void operator()(BlockDesc *desc) const { attr_->set_block_idx(desc->ID()); }

602
  void operator()(int64_t v) const { attr_->set_l(v); }
T
tangwei12 已提交
603 604 605 606 607

  void operator()(const std::vector<int64_t> &v) const {
    VectorToRepeated(v, attr_->mutable_longs());
  }

Y
Yu Yang 已提交
608 609 610
  void operator()(boost::blank) const { PADDLE_THROW("Unexpected branch"); }
};

Y
Yu Yang 已提交
611
void OpDesc::Flush() {
F
fengjiayi 已提交
612
  if (need_update_) {
613
    this->desc_.mutable_inputs()->Clear();
F
fengjiayi 已提交
614
    for (auto &ipt : inputs_) {
615
      auto *input = desc_.add_inputs();
F
fengjiayi 已提交
616 617 618 619
      input->set_parameter(ipt.first);
      VectorToRepeated(ipt.second, input->mutable_arguments());
    }

620
    this->desc_.mutable_outputs()->Clear();
F
fengjiayi 已提交
621
    for (auto &opt : outputs_) {
622
      auto *output = desc_.add_outputs();
F
fengjiayi 已提交
623 624 625 626
      output->set_parameter(opt.first);
      VectorToRepeated(opt.second, output->mutable_arguments());
    }

627
    this->desc_.mutable_attrs()->Clear();
F
fengjiayi 已提交
628
    for (auto &attr : attrs_) {
629
      auto *attr_desc = desc_.add_attrs();
F
fengjiayi 已提交
630 631
      attr_desc->set_name(attr.first);
      attr_desc->set_type(
632
          static_cast<proto::AttrType>(attr.second.which() - 1));
Y
Yu Yang 已提交
633 634
      SetAttrDescVisitor visitor(attr_desc);
      boost::apply_visitor(visitor, attr.second);
F
fengjiayi 已提交
635 636 637 638 639
    }

    need_update_ = false;
  }
}
Y
Yu Yang 已提交
640

641 642
static std::once_flag init_infer_shape_funcs;

643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661
/**
 * NOTE(paddle-dev): Very tricky code here. Maybe we should find a
 * better way to register compile-time infershape method gentlely.
 *
 * Normally, we can register a class derived from InferShapeBase, so that
 * we can set the field of `infer_shape_` inside OpInfo when registering op.
 *
 * However, there is another way we can set the field of `infer_shape_` inside
 * OpInfo. Usually, we overload InferShape method of OperatorWithKernel. After
 * running the following method InitInferShapeFuncs, `infer_shape_` would be set
 * to be the InferShape method of OperatorWithKernel. That is to say, we borrow
 * the run-time InferShape method of OperatorWithKernel to be the compile-time
 * InferShape method.
 *
 * However, during compiling time, we may not know inputs, outputs and attrs of
 * run-time OperatorWithKernel. So the following code creates a fake
 * OperatorWithKernel object. That is why the field info_ of OperatorBase
 * would be null.
 */
662 663 664 665 666 667 668
static void InitInferShapeFuncs() {
  std::call_once(init_infer_shape_funcs, [] {
    auto &map = OpInfoMap::Instance();
    auto &info_map = *map.mutable_map();

    for (auto &kern_pair : OperatorWithKernel::AllOpKernels()) {
      auto op_type = kern_pair.first;
C
chengduoZH 已提交
669 670 671 672
      auto it = info_map.find(op_type);
      PADDLE_ENFORCE(it != info_map.end(), "%s has not been registered",
                     op_type);
      auto &op_info = it->second;
673 674 675
      if (op_info.infer_shape_) {  // infer_shape has been registered.
        continue;
      }
676 677 678 679 680 681 682

      auto op = dynamic_cast<OperatorWithKernel *>(op_info.Creator()(
          "", VariableNameMap{}, VariableNameMap{}, AttributeMap{}));

      PADDLE_ENFORCE_NOT_NULL(
          op, "InferShapeBase is not registered to Operator %s", op_type);

683 684 685
      op_info.infer_shape_ = [op](InferShapeContext *ctx) {
        op->InferShape(ctx);
      };
Y
Yu Yang 已提交
686
    }
687
  });
Y
Yu Yang 已提交
688 689
}

Y
Yu Yang 已提交
690
void OpDesc::CheckAttrs() {
F
fengjiayi 已提交
691 692
  PADDLE_ENFORCE(!Type().empty(),
                 "CheckAttr() can not be called before type is setted.");
Y
Yu Yang 已提交
693 694 695 696 697 698
  auto *checker = OpInfoMap::Instance().Get(Type()).Checker();
  if (checker == nullptr) {
    // checker is not configured. That operator could be generated by Paddle,
    // not by users.
    return;
  }
699
  VLOG(10) << "begin to check attribute of " << Type();
T
tangwei12 已提交
700
  checker->Check(&attrs_);
F
fengjiayi 已提交
701 702
}

Y
Yu Yang 已提交
703
void OpDesc::InferShape(const BlockDesc &block) const {
704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724
  try {
    VLOG(3) << "CompileTime infer shape on " << Type();
    InitInferShapeFuncs();
    auto &infer_shape = OpInfoMap::Instance().Get(this->Type()).infer_shape_;
    PADDLE_ENFORCE(static_cast<bool>(infer_shape),
                   "%s's infer_shape has not been registered", this->Type());
    CompileTimeInferShapeContext ctx(*this, block);
    if (VLOG_IS_ON(10)) {
      std::ostringstream sout;
      auto inames = this->InputArgumentNames();
      sout << " From [";
      std::copy(inames.begin(), inames.end(),
                std::ostream_iterator<std::string>(sout, ", "));
      sout << "] to [";
      auto onames = this->OutputArgumentNames();
      std::copy(onames.begin(), onames.end(),
                std::ostream_iterator<std::string>(sout, ", "));
      sout << "]";
      VLOG(10) << sout.str();
    }
    infer_shape(&ctx);
725
  } catch (platform::EnforceNotMet &exception) {
726 727 728 729 730
    framework::InsertCallStackInfo(Type(), attrs_, &exception);
    throw std::move(exception);
  } catch (...) {
    std::rethrow_exception(std::current_exception());
  }
Y
Yu Yang 已提交
731 732
}

Y
Yu Yang 已提交
733
void OpDesc::InferVarType(BlockDesc *block) const {
X
Xin Pan 已提交
734 735 736 737 738
  // There are a few places that var type can be set.
  // When VarDesc is created, default set to LOD_TENSOR.
  // When output variable is created, default is defaut set to LOD_TENSOR.
  // We limit here to be the only place that operator defines its customized
  // var type inference. Hence, we don't do any "default" setting here.
Y
Yu Yang 已提交
739 740
  auto &info = OpInfoMap::Instance().Get(this->Type());
  if (info.infer_var_type_) {
M
minqiyang 已提交
741
    InferVarTypeContext context(this, block);
M
minqiyang 已提交
742
    info.infer_var_type_(&context);
Y
Yu Yang 已提交
743 744 745
  }
}

746
CompileTimeInferShapeContext::CompileTimeInferShapeContext(
Y
Yu Yang 已提交
747
    const OpDesc &op, const BlockDesc &block)
748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811
    : op_(op), block_(block) {}

bool CompileTimeInferShapeContext::HasInput(const std::string &name) const {
  const std::vector<std::string> &input_names = op_.Input(name);
  auto length = input_names.size();
  if (length == 0) {
    return false;
  }
  PADDLE_ENFORCE_EQ(length, 1UL,
                    "Input(%s) should have only one value, "
                    "but it have %d now",
                    name, length);
  return block_.HasVarRecursive(input_names[0]);
}

bool CompileTimeInferShapeContext::HasOutput(const std::string &name) const {
  const std::vector<std::string> &output_names = op_.Output(name);
  auto length = output_names.size();
  if (length == 0) {
    return false;
  }
  PADDLE_ENFORCE_EQ(length, 1UL,
                    "Output(%s) should have only one value, "
                    "but it have %d now",
                    name, length);
  return block_.HasVarRecursive(output_names[0]);
}

bool CompileTimeInferShapeContext::HasInputs(const std::string &name) const {
  const std::vector<std::string> &input_names = op_.Input(name);
  if (input_names.empty()) {
    return false;
  }
  for (auto &input : input_names) {
    if (!block_.HasVarRecursive(input)) return false;
  }
  return true;
}

bool CompileTimeInferShapeContext::HasOutputs(const std::string &name) const {
  const std::vector<std::string> &output_names = op_.Output(name);
  if (output_names.empty()) {
    return false;
  }
  for (auto &output : output_names) {
    if (!block_.HasVarRecursive(output)) return false;
  }
  return true;
}

AttrReader CompileTimeInferShapeContext::Attrs() const {
  return AttrReader(op_.GetAttrMap());
}

const std::vector<std::string> &CompileTimeInferShapeContext::Inputs(
    const std::string &name) const {
  return op_.Input(name);
}

const std::vector<std::string> &CompileTimeInferShapeContext::Outputs(
    const std::string &name) const {
  return op_.Output(name);
}

F
fengjiayi 已提交
812
std::vector<DDim> CompileTimeInferShapeContext::GetRepeatedDims(
F
fengjiayi 已提交
813 814 815 816 817 818 819 820 821 822
    const std::string &name) const {
  auto var = block_.FindVarRecursive(name);
  PADDLE_ENFORCE(var != nullptr, "Cannot find variable %s", name);
  std::vector<DDim> res;
  try {
    auto shapes = var->GetShapes();
    for (const auto &s : shapes) {
      res.push_back(s.empty() ? make_ddim({0UL}) : make_ddim(s));
    }
  } catch (...) {
M
minqiyang 已提交
823
    VLOG(5) << "GetRepeatedDim of variable " << name << " error.";
F
fengjiayi 已提交
824 825 826
    std::rethrow_exception(std::current_exception());
  }
  return res;
827 828 829 830
}

void CompileTimeInferShapeContext::SetDim(const std::string &name,
                                          const DDim &dim) {
F
fengjiayi 已提交
831
  block_.FindVarRecursive(name)->SetShape(vectorize(dim));
832
}
F
fengjiayi 已提交
833 834 835 836 837 838

void CompileTimeInferShapeContext::SetRepeatedDims(
    const std::string &name, const std::vector<DDim> &dims) {
  auto var = block_.FindVarRecursive(name);
  PADDLE_ENFORCE(var != nullptr, "Cannot find variable %s", name);
  std::vector<std::vector<int64_t>> dim_vec(dims.size());
839
  std::transform(dims.begin(), dims.end(), dim_vec.begin(), vectorize<>);
F
fengjiayi 已提交
840
  var->SetShapes(dim_vec);
841
}
F
fengjiayi 已提交
842

843 844
bool CompileTimeInferShapeContext::IsRuntime() const { return false; }

845
proto::VarType::Type CompileTimeInferShapeContext::GetVarType(
846 847 848
    const std::string &name) const {
  return block_.FindVarRecursive(name)->GetType();
}
849

F
fengjiayi 已提交
850 851
}  // namespace framework
}  // namespace paddle