Merge pull request #604 from MRXLT/0.2.2-web-service

fix app && web service && update version

doc/DESIGN_DOC_CN.md

@@ -26,7 +26,7 @@ serving_io.save_model("serving_model", "client_conf",
                       {"words": data}, {"prediction": prediction},
                       fluid.default_main_program())
 ```
-代码示例中，`{"words": data}`和`{"prediction": prediction}`分别指定了模型的输入和输出，`"words"`和`"prediction"`是输出和输出变量的别名，设计别名的目的是为了使开发者能够记忆自己训练模型的输入输出对应的字段。`data`和`prediction`则是Paddle训练过程中的`[Variable](https://www.paddlepaddle.org.cn/documentation/docs/zh/api_cn/fluid_cn/Variable_cn.html#variable)`，通常代表张量([Tensor](https://www.paddlepaddle.org.cn/documentation/docs/zh/api_cn/fluid_cn/Tensor_cn.html#tensor))或变长张量([LodTensor](https://www.paddlepaddle.org.cn/documentation/docs/zh/beginners_guide/basic_concept/lod_tensor.html#lodtensor))。调用保存命令后，会按照用户指定的`"serving_model"`和`"client_conf"`生成两个目录，内容如下：
+代码示例中，`{"words": data}`和`{"prediction": prediction}`分别指定了模型的输入和输出，`"words"`和`"prediction"`是输入和输出变量的别名，设计别名的目的是为了使开发者能够记忆自己训练模型的输入输出对应的字段。`data`和`prediction`则是Paddle训练过程中的`[Variable](https://www.paddlepaddle.org.cn/documentation/docs/zh/api_cn/fluid_cn/Variable_cn.html#variable)`，通常代表张量([Tensor](https://www.paddlepaddle.org.cn/documentation/docs/zh/api_cn/fluid_cn/Tensor_cn.html#tensor))或变长张量([LodTensor](https://www.paddlepaddle.org.cn/documentation/docs/zh/beginners_guide/basic_concept/lod_tensor.html#lodtensor))。调用保存命令后，会按照用户指定的`"serving_model"`和`"client_conf"`生成两个目录，内容如下：
 ``` shell
 .
 ├── client_conf

python/examples/bert/benchmark.py

@@ -26,7 +26,7 @@ from batching import pad_batch_data
 import tokenization
 import requests
 import json
-from bert_reader import BertReader
+from paddle_serving_app.reader import ChineseBertReader
 
 args = benchmark_args()
 
@@ -37,7 +37,7 @@ def single_func(idx, resource):
     for line in fin:
         dataset.append(line.strip())
     if args.request == "rpc":
-        reader = BertReader(vocab_file="vocab.txt", max_seq_len=20)
+        reader = ChineseBertReader(vocab_file="vocab.txt", max_seq_len=20)
         fetch = ["pooled_output"]
         client = Client()
         client.load_client_config(args.model)

python/examples/bert/bert_client.py

@@ -25,7 +25,7 @@ from paddlehub.common.logger import logger
 import socket
 from paddle_serving_client import Client
 from paddle_serving_client.utils import benchmark_args
-from paddle_serving_app import ChineseBertReader
+from paddle_serving_app.reader import ChineseBertReader
 
 args = benchmark_args()
 

python/examples/bert/bert_web_service.py

@@ -14,14 +14,14 @@
 # limitations under the License.
 # pylint: disable=doc-string-missing
 from paddle_serving_server_gpu.web_service import WebService
-from bert_reader import BertReader
+from paddle_serving_app.reader import ChineseBertReader
 import sys
 import os
 
 
 class BertService(WebService):
     def load(self):
-        self.reader = BertReader(vocab_file="vocab.txt", max_seq_len=128)
+        self.reader = ChineseBertReader(vocab_file="vocab.txt", max_seq_len=128)
 
     def preprocess(self, feed=[], fetch=[]):
         feed_res = [
@@ -37,5 +37,5 @@ gpu_ids = os.environ["CUDA_VISIBLE_DEVICES"]
 bert_service.set_gpus(gpu_ids)
 bert_service.prepare_server(
     workdir="workdir", port=int(sys.argv[2]), device="gpu")
-bert_service.run_server()
-bert_service.run_flask()
+bert_service.run_rpc_service()
+bert_service.run_web_service()

python/examples/imagenet/resnet50_web_service.py

@@ -68,5 +68,5 @@ if device == "gpu":
     image_service.set_gpus("0,1")
 image_service.prepare_server(
     workdir="workdir", port=int(sys.argv[3]), device=device)
-image_service.run_server()
-image_service.run_flask()
+image_service.run_rpc_service()
+image_service.run_web_service()

python/examples/imdb/benchmark.py

@@ -16,7 +16,7 @@
 import sys
 import time
 import requests
-from paddle_serving_app import IMDBDataset
+from paddle_serving_app.reader import IMDBDataset
 from paddle_serving_client import Client
 from paddle_serving_client.utils import MultiThreadRunner
 from paddle_serving_client.utils import benchmark_args

python/examples/imdb/test_client.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 # pylint: disable=doc-string-missing
 from paddle_serving_client import Client
-from paddle_serving_app import IMDBDataset
+from paddle_serving_app.reader import IMDBDataset
 import sys
 
 client = Client()

python/examples/imdb/text_classify_service.py

@@ -14,7 +14,7 @@
 # pylint: disable=doc-string-missing
 
 from paddle_serving_server.web_service import WebService
-from paddle_serving_app import IMDBDataset
+from paddle_serving_app.reader import IMDBDataset
 import sys
 
 
@@ -37,5 +37,5 @@ imdb_service.load_model_config(sys.argv[1])
 imdb_service.prepare_server(
     workdir=sys.argv[2], port=int(sys.argv[3]), device="cpu")
 imdb_service.prepare_dict({"dict_file_path": sys.argv[4]})
-imdb_service.run_server()
-imdb_service.run_flask()
+imdb_service.run_rpc_service()
+imdb_service.run_web_service()

python/examples/lac/lac_web_service.py

@@ -47,5 +47,5 @@ lac_service.load_model_config(sys.argv[1])
 lac_service.load_reader()
 lac_service.prepare_server(
     workdir=sys.argv[2], port=int(sys.argv[3]), device="cpu")
-lac_service.run_server()
-lac_service.run_flask()
+lac_service.run_rpc_service()
+lac_service.run_web_service()

python/examples/resnet_v2_50/resnet50_debug.py

@@ -14,7 +14,7 @@
 
 from paddle_serving_app.reader import Sequential, File2Image, Resize, CenterCrop
 from paddle_serving_app.reader import RGB2BGR, Transpose, Div, Normalize
-from paddle_serving_app import Debugger
+from paddle_serving_app.local_predict import Debugger
 import sys
 
 debugger = Debugger()

python/examples/senta/get_data.sh

 wget https://paddle-serving.bj.bcebos.com/paddle_hub_models/text/SentimentAnalysis/senta_bilstm.tar.gz --no-check-certificate
 tar -xzvf senta_bilstm.tar.gz
-wget https://paddle-serving.bj.bcebos.com/paddle_hub_models/text/LexicalAnalysis/lac_model.tar.gz --no-check-certificate
+wget https://paddle-serving.bj.bcebos.com/paddle_hub_models/text/LexicalAnalysis/lac.tar.gz --no-check-certificate
 tar -xzvf lac_model.tar.gz
 wget https://paddle-serving.bj.bcebos.com/reader/lac/lac_dict.tar.gz  --no-check-certificate
 tar -xzvf lac_dict.tar.gz

python/examples/senta/senta_web_service.py

@@ -14,13 +14,10 @@
 
 from paddle_serving_server_gpu.web_service import WebService
 from paddle_serving_client import Client
-from paddle_serving_app import LACReader, SentaReader
-import numpy as np
+from paddle_serving_app.reader import LACReader, SentaReader
 import os
-import io
 import sys
-import subprocess
-from multiprocessing import Process, Queue
+from multiprocessing import Process
 
 
 class SentaService(WebService):
@@ -33,10 +30,6 @@ class SentaService(WebService):
         self.lac_client_config_path = lac_model_path + "/serving_server_conf.prototxt"
         self.lac_dict_path = lac_dict_path
         self.senta_dict_path = senta_dict_path
-        self.show = False
-
-    def show_detail(self, show=False):
-        self.show = show
 
     def start_lac_service(self):
         if not os.path.exists('./lac_serving'):
@@ -64,34 +57,29 @@ class SentaService(WebService):
         self.lac_client.connect(["127.0.0.1:{}".format(self.lac_port)])
 
     def init_lac_reader(self):
-        self.lac_reader = LACReader(self.lac_dict_path)
+        self.lac_reader = LACReader()
 
     def init_senta_reader(self):
-        self.senta_reader = SentaReader(vocab_path=self.senta_dict_path)
+        self.senta_reader = SentaReader()
 
     def preprocess(self, feed=[], fetch=[]):
-        feed_data = self.lac_reader.process(feed[0]["words"])
-        if self.show:
-            print("---- lac reader ----")
-            print(feed_data)
-        lac_result = self.lac_predict(feed_data)
-        if self.show:
-            print("---- lac out ----")
-            print(lac_result)
-        segs = self.lac_reader.parse_result(feed[0]["words"],
-                                            lac_result["crf_decode"])
-        if self.show:
-            print("---- lac parse ----")
-            print(segs)
+        feed_data = [{
+            "words": self.lac_reader.process(x["words"])
+        } for x in feed]
+        lac_result = self.lac_client.predict(
+            feed=feed_data, fetch=["crf_decode"])
+        feed_batch = []
+        result_lod = lac_result["crf_decode.lod"]
+        for i in range(len(feed)):
+            segs = self.lac_reader.parse_result(
+                feed[i]["words"],
+                lac_result["crf_decode"][result_lod[i]:result_lod[i + 1]])
             feed_data = self.senta_reader.process(segs)
-        if self.show:
-            print("---- senta reader ----")
-            print("feed_data", feed_data)
-        return [{"words": feed_data}], fetch
+            feed_batch.append({"words": feed_data})
+        return feed_batch, fetch
 
 
 senta_service = SentaService(name="senta")
-#senta_service.show_detail(True)
 senta_service.set_config(
     lac_model_path="./lac_model",
     lac_dict_path="./lac_dict",
@@ -102,5 +90,5 @@ senta_service.prepare_server(
 senta_service.init_lac_reader()
 senta_service.init_senta_reader()
 senta_service.init_lac_service()
-senta_service.run_server()
-senta_service.run_flask()
+senta_service.run_rpc_service()
+senta_service.run_web_service()

python/paddle_serving_app/README.md

@@ -158,7 +158,7 @@ Therefore, a local prediction tool is built into the paddle_serving_app, which i
 Taking [fit_a_line prediction service](../examples/fit_a_line) as an example, the following code can be used to run local prediction.
 
 ```python
-from paddle_serving_app import Debugger
+from paddle_serving_app.local_predict import Debugger
 import numpy as np
 
 debugger = Debugger()

python/paddle_serving_app/README_CN.md

@@ -147,7 +147,7 @@ Paddle Serving框架的server预测op使用了Paddle 的预测框架，在部署
 以[fit_a_line预测服务](../examples/fit_a_line)为例，使用以下代码即可执行本地预测。
 
 ```python
-from paddle_serving_app import Debugger
+from paddle_serving_app.local_predict import Debugger
 import numpy as np
 
 debugger = Debugger()

python/paddle_serving_app/__init__.py

@@ -11,10 +11,4 @@
 # 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.
-from .reader.chinese_bert_reader import ChineseBertReader
-from .reader.image_reader import ImageReader, File2Image, URL2Image, Sequential, Normalize, CenterCrop, Resize, PadStride
-from .reader.lac_reader import LACReader
-from .reader.senta_reader import SentaReader
-from .reader.imdb_reader import IMDBDataset
 from .models import ServingModels
-from .local_predict import Debugger

python/paddle_serving_app/models/model_list.py

@@ -37,7 +37,7 @@ class ServingModels(object):
         object_detection_url = "https://paddle-serving.bj.bcebos.com/paddle_hub_models/image/ObjectDetection/"
         senta_url = "https://paddle-serving.bj.bcebos.com/paddle_hub_models/text/SentimentAnalysis/"
         semantic_url = "https://paddle-serving.bj.bcebos.com/paddle_hub_models/text/SemanticRepresentation/"
-        wordseg_url = "https://paddle-serving.bj.bcebos.com/paddle_hub_models/text/ChineseWordSegmentation/"
+        wordseg_url = "https://paddle-serving.bj.bcebos.com/paddle_hub_models/text/LexicalAnalysis/"
 
         self.url_dict = {}
 

python/paddle_serving_app/reader/__init__.py

@@ -11,4 +11,8 @@
 # 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.
+from .chinese_bert_reader import ChineseBertReader
 from .image_reader import ImageReader, File2Image, URL2Image, Sequential, Normalize, CenterCrop, Resize, Transpose, Div, RGB2BGR, BGR2RGB, RCNNPostprocess, SegPostprocess, PadStride
+from .lac_reader import LACReader
+from .senta_reader import SentaReader
+from .imdb_reader import IMDBDataset

python/paddle_serving_app/reader/lac_reader.py

@@ -48,10 +48,16 @@ def load_kv_dict(dict_path,
 class LACReader(object):
     """data reader"""
 
-    def __init__(self, dict_folder):
+    def __init__(self, dict_folder=""):
         # read dict
         #basepath = os.path.abspath(__file__)
         #folder = os.path.dirname(basepath)
+        if dict_folder == "":
+            dict_folder = "lac_dict"
+            if not os.path.exists(dict_folder):
+                r = os.system(
+                    "wget https://paddle-serving.bj.bcebos.com/reader/lac/lac_dict.tar.gz  --no-check-certificate && tar -xzvf lac_dict.tar.gz"
+                )
         word_dict_path = os.path.join(dict_folder, "word.dic")
         label_dict_path = os.path.join(dict_folder, "tag.dic")
         replace_dict_path = os.path.join(dict_folder, "q2b.dic")

python/paddle_serving_app/reader/senta_reader.py

@@ -14,10 +14,11 @@
 
 import sys
 import io
+import os
 
 
 class SentaReader():
-    def __init__(self, vocab_path, max_seq_len=20):
+    def __init__(self, vocab_path="", max_seq_len=20):
         self.max_seq_len = max_seq_len
         self.word_dict = self.load_vocab(vocab_path)
 
@@ -25,6 +26,13 @@ class SentaReader():
         """
         load the given vocabulary
         """
+        if vocab_path == "":
+            vocab_path = "senta_vocab.txt"
+            if not os.path.exists(vocab_path):
+                r = os.system(
+                    " wget https://paddle-serving.bj.bcebos.com/reader/senta/senta_vocab.txt --no-check-certificate"
+                )
+
         vocab = {}
         with io.open(vocab_path, 'r', encoding='utf8') as f:
             for line in f:

python/paddle_serving_app/version.py

@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """ Paddle Serving App version string """
-serving_app_version = "0.0.3"
+serving_app_version = "0.1.0"

python/paddle_serving_client/__init__.py

@@ -207,8 +207,9 @@ class Client(object):
                 key))
         if type(feed[key]).__module__ == np.__name__ and np.size(feed[
                 key]) != self.feed_tensor_len[key]:
-            raise SystemExit("The shape of feed tensor {} not match.".format(
-                key))
+            #raise SystemExit("The shape of feed tensor {} not match.".format(
+            #    key))
+            pass
 
     def predict(self, feed=None, fetch=None, need_variant_tag=False):
         self.profile_.record('py_prepro_0')

python/paddle_serving_client/io/__init__.py

@@ -33,7 +33,6 @@ def save_model(server_model_folder,
     executor = Executor(place=CPUPlace())
 
     feed_var_names = [feed_var_dict[x].name for x in feed_var_dict]
-    #target_vars = list(fetch_var_dict.values())
     target_vars = []
     target_var_names = []
     for key in sorted(fetch_var_dict.keys()):

python/paddle_serving_client/version.py

@@ -12,6 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """ Paddle Serving Client version string """
-serving_client_version = "0.2.2"
-serving_server_version = "0.2.2"
-module_proto_version = "0.2.2"
+serving_client_version = "0.3.0"
+serving_server_version = "0.3.0"
+module_proto_version = "0.3.0"

python/paddle_serving_server/serve.py

@@ -103,7 +103,7 @@ if __name__ == "__main__":
         service.load_model_config(args.model)
         service.prepare_server(
             workdir=args.workdir, port=args.port, device=args.device)
-        service.run_server()
+        service.run_rpc_service()
 
         app_instance = Flask(__name__)
 

python/paddle_serving_server/version.py

@@ -12,6 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """ Paddle Serving Client version string """
-serving_client_version = "0.2.2"
-serving_server_version = "0.2.2"
-module_proto_version = "0.2.2"
+serving_client_version = "0.3.0"
+serving_server_version = "0.3.0"
+module_proto_version = "0.3.0"

python/paddle_serving_server/web_service.py

@@ -92,7 +92,7 @@ class WebService(object):
             result = {"result": "Request Value Error"}
         return result
 
-    def run_server(self):
+    def run_rpc_service(self):
         import socket
         localIP = socket.gethostbyname(socket.gethostname())
         print("web service address:")
@@ -115,7 +115,7 @@ class WebService(object):
 
         self.app_instance = app_instance
 
-    def run_flask(self):
+    def run_web_service(self):
         self.app_instance.run(host="0.0.0.0",
                               port=self.port,
                               threaded=False,

python/paddle_serving_server_gpu/serve.py

@@ -118,7 +118,7 @@ if __name__ == "__main__":
             web_service.set_gpus(gpu_ids)
         web_service.prepare_server(
             workdir=args.workdir, port=args.port, device=args.device)
-        web_service.run_server()
+        web_service.run_rpc_service()
 
         app_instance = Flask(__name__)
 

python/paddle_serving_server_gpu/version.py

@@ -12,6 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """ Paddle Serving Client version string """
-serving_client_version = "0.2.2"
-serving_server_version = "0.2.2"
-module_proto_version = "0.2.2"
+serving_client_version = "0.3.0"
+serving_server_version = "0.3.0"
+module_proto_version = "0.3.0"

python/paddle_serving_server_gpu/web_service.py

@@ -133,12 +133,11 @@ class WebService(object):
             result = self.postprocess(
                 feed=feed, fetch=fetch, fetch_map=fetch_map)
             result = {"result": result}
-            result = {"result": fetch_map}
         except ValueError:
             result = {"result": "Request Value Error"}
         return result
 
-    def run_server(self):
+    def run_rpc_service(self):
         import socket
         localIP = socket.gethostbyname(socket.gethostname())
         print("web service address:")
@@ -165,7 +164,7 @@ class WebService(object):
 
         self.app_instance = app_instance
 
-    def run_flask(self):
+    def run_web_service(self):
         self.app_instance.run(host="0.0.0.0",
                               port=self.port,
                               threaded=False,