feat(server.py): update match positions

predictor/__init__.py

@@ -1,4 +1,4 @@
-from typing import Any, Dict,List
+from typing import Any, Dict,List,Union
 from transformers import Pipeline
 import requests
 import re
@@ -474,3 +474,56 @@ class Predictor():
     def __call__(self, *args: Any, **kwds: Any) -> Any:
         return self.process(*args, **kwds)
 
+class PositionPredictor():
+
+    def __init__(self, pipeline: Pipeline) -> None:
+        self.pipeline = pipeline
+        self.__init_split_data()
+        self.logger = logging.getLogger(__name__)
+
+    def __split_blocks(self, text: str) -> List[str]:
+        start,end = 0,0
+        blocks = []
+        while end<len(text):
+            if text[end] in self.splits:
+                if end>start:
+                    blocks.append(text[start:end])
+                start = end+1
+            end += 1
+        if end>start:
+            blocks.append(text[start:end])
+        return blocks
+    
+    def __init_split_data(
+        self
+    ):
+        self.splits = {'\\', '_', '"', '%', '{', '《', ')', '$', '(', '\n', '~', '*', ':', '!', '；', '”', '’', '\t', '？', '-', ';', '》', '】', '`', '、', '+', '“', '[', '—', '·', '）', '=', '‘', '}', '?', '，', '&', '@', '#', ']', '——', ' ', '.', '【', "'", '>', ',', '/', '：', '。', '...', '^', '（', '<', '|', '……', '！'}
+
+    def predict(self, 
+                positions: List[Dict[str,Union[str,List[str]]]], 
+                resume: str
+        ) -> List[Dict[str, Union[str, float]]]:
+            ans = []
+            resume_blocks = self.__split_blocks(resume)
+            print(resume_blocks,positions)
+            for position in positions:
+                requireds = position['required']
+                score = 0.0
+                for required in requireds:
+                    blocks = self.__split_blocks(required)
+                    print(blocks)
+                    for block in blocks:
+                        for block_resume in resume_blocks:
+                            score = max(score, self.pipeline((block_resume, block))[0])
+                            self.logger.info(f"position: {position['name']}, required: {block}, resume: {block_resume}, score: {score}")
+                ans.append({
+                    'position': position['name'],
+                    'score': score
+                })
+            ans.sort(key=lambda x:x['score'], reverse=True)
+            return ans
+    
+    def __call__(self, *args: Any, **kwds: Any) -> Any:
+        return self.predict(*args, **kwds)
+
+            

protos/resume.proto

@@ -4,12 +4,32 @@ service Resume{
     // Get Resume Info
     rpc GetInfo(TextRequest) returns (ResumeEntitiesResponse);
     rpc Health(PingRequest) returns (PongResponse);
+    rpc MatchPosition(PositionMatchRequest) returns (PositionMatchResponse);
 }
 
 message PingRequest{
     string ping = 1;
 }
 
+message Position{
+    string name = 1;
+    repeated string required = 2;
+}
+
+message PositionMatchRequest{
+    repeated Position positions = 1;
+    string resume = 2;
+}
+
+message PositionMatch{
+    string position = 1;
+    float score = 2;
+}
+
+message PositionMatchResponse{
+    repeated PositionMatch matches = 1;
+}
+
 message PongResponse{
     string done = 1;
 }

protos/resume_pb2.py

@@ -13,7 +13,7 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\x13protos/resume.proto\"\x1b\n\x0bPingRequest\x12\x0c\n\x04ping\x18\x01 \x01(\t\"\x1c\n\x0cPongResponse\x12\x0c\n\x04\x64one\x18\x01 \x01(\t\"\x1b\n\x0bTextRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\"~\n\x06\x45ntity\x12\x0e\n\x06\x65ntity\x18\x01 \x01(\t\x12\r\n\x05start\x18\x02 \x01(\x05\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x05\x12\x11\n\x04text\x18\x04 \x01(\tH\x00\x88\x01\x01\x12\x0e\n\x06origin\x18\x05 \x01(\t\x12\x12\n\x05level\x18\x06 \x01(\tH\x01\x88\x01\x01\x42\x07\n\x05_textB\x08\n\x06_level\"S\n\tJobEntity\x12\x16\n\x05start\x18\x01 \x01(\x0b\x32\x07.Entity\x12\x14\n\x03\x65nd\x18\x02 \x01(\x0b\x32\x07.Entity\x12\x18\n\x07\x63ompany\x18\x03 \x01(\x0b\x32\x07.Entity\"\x8b\x02\n\x16ResumeEntitiesResponse\x12\x16\n\x05names\x18\x01 \x03(\x0b\x32\x07.Entity\x12\x17\n\x06\x65mails\x18\x02 \x03(\x0b\x32\x07.Entity\x12\x17\n\x06phones\x18\x03 \x03(\x0b\x32\x07.Entity\x12\x18\n\x07schools\x18\x04 \x03(\x0b\x32\x07.Entity\x12\x15\n\x04\x65\x64us\x18\x05 \x03(\x0b\x32\x07.Entity\x12\x15\n\x04\x61ges\x18\x06 \x03(\x0b\x32\x07.Entity\x12\x18\n\x07genders\x18\x07 \x03(\x0b\x32\x07.Entity\x12\x12\n\nwork_years\x18\x08 \x01(\x05\x12\x18\n\x04jobs\x18\t \x03(\x0b\x32\n.JobEntity\x12\x17\n\x06titles\x18\n \x03(\x0b\x32\x07.Entity2a\n\x06Resume\x12\x30\n\x07GetInfo\x12\x0c.TextRequest\x1a\x17.ResumeEntitiesResponse\x12%\n\x06Health\x12\x0c.PingRequest\x1a\r.PongResponseb\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\x13protos/resume.proto\"\x1b\n\x0bPingRequest\x12\x0c\n\x04ping\x18\x01 \x01(\t\"*\n\x08Position\x12\x0c\n\x04name\x18\x01 \x01(\t\x12\x10\n\x08required\x18\x02 \x03(\t\"D\n\x14PositionMatchRequest\x12\x1c\n\tpositions\x18\x01 \x03(\x0b\x32\t.Position\x12\x0e\n\x06resume\x18\x02 \x01(\t\"0\n\rPositionMatch\x12\x10\n\x08position\x18\x01 \x01(\t\x12\r\n\x05score\x18\x02 \x01(\x02\"8\n\x15PositionMatchResponse\x12\x1f\n\x07matches\x18\x01 \x03(\x0b\x32\x0e.PositionMatch\"\x1c\n\x0cPongResponse\x12\x0c\n\x04\x64one\x18\x01 \x01(\t\"\x1b\n\x0bTextRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\"~\n\x06\x45ntity\x12\x0e\n\x06\x65ntity\x18\x01 \x01(\t\x12\r\n\x05start\x18\x02 \x01(\x05\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x05\x12\x11\n\x04text\x18\x04 \x01(\tH\x00\x88\x01\x01\x12\x0e\n\x06origin\x18\x05 \x01(\t\x12\x12\n\x05level\x18\x06 \x01(\tH\x01\x88\x01\x01\x42\x07\n\x05_textB\x08\n\x06_level\"S\n\tJobEntity\x12\x16\n\x05start\x18\x01 \x01(\x0b\x32\x07.Entity\x12\x14\n\x03\x65nd\x18\x02 \x01(\x0b\x32\x07.Entity\x12\x18\n\x07\x63ompany\x18\x03 \x01(\x0b\x32\x07.Entity\"\x8b\x02\n\x16ResumeEntitiesResponse\x12\x16\n\x05names\x18\x01 \x03(\x0b\x32\x07.Entity\x12\x17\n\x06\x65mails\x18\x02 \x03(\x0b\x32\x07.Entity\x12\x17\n\x06phones\x18\x03 \x03(\x0b\x32\x07.Entity\x12\x18\n\x07schools\x18\x04 \x03(\x0b\x32\x07.Entity\x12\x15\n\x04\x65\x64us\x18\x05 \x03(\x0b\x32\x07.Entity\x12\x15\n\x04\x61ges\x18\x06 \x03(\x0b\x32\x07.Entity\x12\x18\n\x07genders\x18\x07 \x03(\x0b\x32\x07.Entity\x12\x12\n\nwork_years\x18\x08 \x01(\x05\x12\x18\n\x04jobs\x18\t \x03(\x0b\x32\n.JobEntity\x12\x17\n\x06titles\x18\n \x03(\x0b\x32\x07.Entity2\xa1\x01\n\x06Resume\x12\x30\n\x07GetInfo\x12\x0c.TextRequest\x1a\x17.ResumeEntitiesResponse\x12%\n\x06Health\x12\x0c.PingRequest\x1a\r.PongResponse\x12>\n\rMatchPosition\x12\x15.PositionMatchRequest\x1a\x16.PositionMatchResponseb\x06proto3')
 
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, globals())
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'protos.resume_pb2', globals())
@@ -22,16 +22,24 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   DESCRIPTOR._options = None
   _PINGREQUEST._serialized_start=23
   _PINGREQUEST._serialized_end=50
-  _PONGRESPONSE._serialized_start=52
-  _PONGRESPONSE._serialized_end=80
-  _TEXTREQUEST._serialized_start=82
-  _TEXTREQUEST._serialized_end=109
-  _ENTITY._serialized_start=111
-  _ENTITY._serialized_end=237
-  _JOBENTITY._serialized_start=239
-  _JOBENTITY._serialized_end=322
-  _RESUMEENTITIESRESPONSE._serialized_start=325
-  _RESUMEENTITIESRESPONSE._serialized_end=592
-  _RESUME._serialized_start=594
-  _RESUME._serialized_end=691
+  _POSITION._serialized_start=52
+  _POSITION._serialized_end=94
+  _POSITIONMATCHREQUEST._serialized_start=96
+  _POSITIONMATCHREQUEST._serialized_end=164
+  _POSITIONMATCH._serialized_start=166
+  _POSITIONMATCH._serialized_end=214
+  _POSITIONMATCHRESPONSE._serialized_start=216
+  _POSITIONMATCHRESPONSE._serialized_end=272
+  _PONGRESPONSE._serialized_start=274
+  _PONGRESPONSE._serialized_end=302
+  _TEXTREQUEST._serialized_start=304
+  _TEXTREQUEST._serialized_end=331
+  _ENTITY._serialized_start=333
+  _ENTITY._serialized_end=459
+  _JOBENTITY._serialized_start=461
+  _JOBENTITY._serialized_end=544
+  _RESUMEENTITIESRESPONSE._serialized_start=547
+  _RESUMEENTITIESRESPONSE._serialized_end=814
+  _RESUME._serialized_start=817
+  _RESUME._serialized_end=978
 # @@protoc_insertion_point(module_scope)

protos/resume_pb2.pyi

@@ -43,6 +43,36 @@ class PongResponse(_message.Message):
     done: str
     def __init__(self, done: _Optional[str] = ...) -> None: ...
 
+class Position(_message.Message):
+    __slots__ = ["name", "required"]
+    NAME_FIELD_NUMBER: _ClassVar[int]
+    REQUIRED_FIELD_NUMBER: _ClassVar[int]
+    name: str
+    required: _containers.RepeatedScalarFieldContainer[str]
+    def __init__(self, name: _Optional[str] = ..., required: _Optional[_Iterable[str]] = ...) -> None: ...
+
+class PositionMatch(_message.Message):
+    __slots__ = ["position", "score"]
+    POSITION_FIELD_NUMBER: _ClassVar[int]
+    SCORE_FIELD_NUMBER: _ClassVar[int]
+    position: str
+    score: float
+    def __init__(self, position: _Optional[str] = ..., score: _Optional[float] = ...) -> None: ...
+
+class PositionMatchRequest(_message.Message):
+    __slots__ = ["positions", "resume"]
+    POSITIONS_FIELD_NUMBER: _ClassVar[int]
+    RESUME_FIELD_NUMBER: _ClassVar[int]
+    positions: _containers.RepeatedCompositeFieldContainer[Position]
+    resume: str
+    def __init__(self, positions: _Optional[_Iterable[_Union[Position, _Mapping]]] = ..., resume: _Optional[str] = ...) -> None: ...
+
+class PositionMatchResponse(_message.Message):
+    __slots__ = ["matches"]
+    MATCHES_FIELD_NUMBER: _ClassVar[int]
+    matches: _containers.RepeatedCompositeFieldContainer[PositionMatch]
+    def __init__(self, matches: _Optional[_Iterable[_Union[PositionMatch, _Mapping]]] = ...) -> None: ...
+
 class ResumeEntitiesResponse(_message.Message):
     __slots__ = ["ages", "edus", "emails", "genders", "jobs", "names", "phones", "schools", "titles", "work_years"]
     AGES_FIELD_NUMBER: _ClassVar[int]

protos/resume_pb2_grpc.py

@@ -24,6 +24,11 @@ class ResumeStub(object):
                 request_serializer=protos_dot_resume__pb2.PingRequest.SerializeToString,
                 response_deserializer=protos_dot_resume__pb2.PongResponse.FromString,
                 )
+        self.MatchPosition = channel.unary_unary(
+                '/Resume/MatchPosition',
+                request_serializer=protos_dot_resume__pb2.PositionMatchRequest.SerializeToString,
+                response_deserializer=protos_dot_resume__pb2.PositionMatchResponse.FromString,
+                )
 
 
 class ResumeServicer(object):
@@ -42,6 +47,12 @@ class ResumeServicer(object):
         context.set_details('Method not implemented!')
         raise NotImplementedError('Method not implemented!')
 
+    def MatchPosition(self, request, context):
+        """Missing associated documentation comment in .proto file."""
+        context.set_code(grpc.StatusCode.UNIMPLEMENTED)
+        context.set_details('Method not implemented!')
+        raise NotImplementedError('Method not implemented!')
+
 
 def add_ResumeServicer_to_server(servicer, server):
     rpc_method_handlers = {
@@ -55,6 +66,11 @@ def add_ResumeServicer_to_server(servicer, server):
                     request_deserializer=protos_dot_resume__pb2.PingRequest.FromString,
                     response_serializer=protos_dot_resume__pb2.PongResponse.SerializeToString,
             ),
+            'MatchPosition': grpc.unary_unary_rpc_method_handler(
+                    servicer.MatchPosition,
+                    request_deserializer=protos_dot_resume__pb2.PositionMatchRequest.FromString,
+                    response_serializer=protos_dot_resume__pb2.PositionMatchResponse.SerializeToString,
+            ),
     }
     generic_handler = grpc.method_handlers_generic_handler(
             'Resume', rpc_method_handlers)
@@ -98,3 +114,20 @@ class Resume(object):
             protos_dot_resume__pb2.PongResponse.FromString,
             options, channel_credentials,
             insecure, call_credentials, compression, wait_for_ready, timeout, metadata)
+
+    @staticmethod
+    def MatchPosition(request,
+            target,
+            options=(),
+            channel_credentials=None,
+            call_credentials=None,
+            insecure=False,
+            compression=None,
+            wait_for_ready=None,
+            timeout=None,
+            metadata=None):
+        return grpc.experimental.unary_unary(request, target, '/Resume/MatchPosition',
+            protos_dot_resume__pb2.PositionMatchRequest.SerializeToString,
+            protos_dot_resume__pb2.PositionMatchResponse.FromString,
+            options, channel_credentials,
+            insecure, call_credentials, compression, wait_for_ready, timeout, metadata)

server.py

@@ -5,7 +5,7 @@ import protos.resume_pb2
 import protos.resume_pb2_grpc
 from huggingface_hub import login
 import os
-from predictor import Predictor
+from predictor import Predictor, PositionPredictor
 from transformers import pipeline
 from datetime import date
 
@@ -17,7 +17,16 @@ class Resume(protos.resume_pb2_grpc.ResumeServicer):
 
     def __init__(self):
         self.done = False
-        self.logger = logging.getLogger(__name__)   
+        self.logger = logging.getLogger(__name__)  
+        self.position_predictor =  PositionPredictor(
+            pipeline=pipeline(
+                "sentences_sim",
+                model="minskiter/simbert-chinese-bert-wwm-ext",
+                device="cpu",
+                trust_remote_code=True,
+                use_auth_token=True
+            )
+        )
         self.predictor = Predictor(
             pipelines={
                 "name": pipeline("nerpipe", model="minskiter/resume-token-classification-name-0708",trust_remote_code=True,use_auth_token=True),
@@ -39,6 +48,25 @@ class Resume(protos.resume_pb2_grpc.ResumeServicer):
             else:
                 return protos.resume_pb2.PongResponse(done="Pending")
         return protos.resume_pb2.PongResponse(done="PING request is not valid")
+    
+    def MatchPosition(self, request, context):
+        positions = [] 
+        for position in request.positions:
+            required = list(text for text in position.required)
+            name = position.name
+            positions.append({
+                "name": name,
+                "required": required
+            })
+        resume = request.resume
+        scores = self.position_predictor(positions,resume)
+        res = protos.resume_pb2.PositionMatchResponse()
+        for score in scores:
+            res.matches.append(protos.resume_pb2.PositionMatch(
+                position=score["position"],
+                score=score["score"]
+            ))
+        return res
 
     def GetInfo(self, request, context):
         entities = self.predictor(request.text)
@@ -137,7 +165,6 @@ class Resume(protos.resume_pb2_grpc.ResumeServicer):
         res.work_years = entities['work_time']
         return res
         
-
 def serve(port = "50051"):
     logger = logging.getLogger(__name__)
     server = grpc.server(futures.ThreadPoolExecutor(max_workers=10))