First commit

Dockerfile

@@ -0,0 +1,5 @@
+FROM python:2.7
+RUN pip install pattern grpcio-tools==1.16.1
+COPY . /app
+WORKDIR /app
+CMD ["./start.sh"]

README.md

@@ -0,0 +1,29 @@
+# grpc-pattern as microservice
+
+Pattern is a great framework for Natural Language Processing for Python.
+
+gRPC is the new efficient and effective transfer protocol powered by Google to create microservice.
+
+grpc-pattern brings Pattern as microservice through Docker container.
+
+# Run it
+
+To use it, just execute:
+```
+docker run -p 50051:50051 digitalilusion/grpc-pattern
+```
+
+## Try out
+
+You can use `grpcc` tool to try the microservice
+
+```
+grpcc -i -p api.proto -a localhost:50051
+```
+A REPL interface with node is launched. Let's try the parser. `language = 1` for English, `language = 2` for Spanish, ... (see `api.proto` for full language options)
+
+```
+client.parse({'language': 1, 'text': 'The cat is over the roof. And the dog?'}, pr)
+```
+
+And this is the result:

api.proto

@@ -0,0 +1,45 @@
+syntax = "proto3";
+
+package api;
+
+service API {
+  rpc parse (ParseRequest) returns (ParseResponse);
+}
+
+enum Language {
+  NONE = 0;
+  EN = 1;
+  ES = 2;
+  DE = 3;
+  FR = 4;
+  IT = 5;
+  NL = 6;
+
+}
+
+message ParseRequest {
+  Language language = 1;
+  string text = 2;
+}
+
+message ParseResponse {
+  bool isOk = 1;
+  string reason = 2;
+  repeated Sentence sentences = 3;
+}
+
+message Sentence {
+  repeated Word words = 1;
+  repeated Chunk chunks = 2;
+
+}
+
+message Word {
+  string text = 1;
+  string type = 2;
+}
+
+message Chunk {
+  string type = 1;
+  repeated Word words = 2;
+}

server.py

@@ -0,0 +1,87 @@
+# -*- coding: utf-8 -*-
+import os
+import time
+import logging
+from concurrent import futures
+
+from google.protobuf import json_format
+import grpc
+
+from api import api_pb2_grpc, api_pb2
+
+
+DEFAULT_GRPC_PORT = '[::]:50051'
+
+GRPC_PORT = os.getenv('GRPC_PORT', DEFAULT_GRPC_PORT)
+
+parsers = {
+    api_pb2.EN: None,
+    api_pb2.ES: None,
+    api_pb2.DE: None,
+    api_pb2.FR: None,
+    api_pb2.IT: None,
+    api_pb2.NL: None
+}
+
+langCode = {
+    api_pb2.EN: 'en',
+    api_pb2.ES: 'es',
+    api_pb2.DE: 'de',
+    api_pb2.FR: 'fr',
+    api_pb2.IT: 'it',
+    api_pb2.NL: 'nl'
+}
+
+
+def getParser(lang):
+    """
+    Layz-load parsetree if available
+    """
+    if parsers[lang] is None:
+        # Dynamic load
+        logging.debug('Loading parsetree for %s...' % langCode[lang])
+        parsers[lang] = __import__('pattern.' + langCode[lang], None, None,
+                                   langCode[lang])
+        logging.debug('loaded!')
+    return parsers[lang].parsetree
+
+
+class APIServicer(api_pb2_grpc.APIServicer):
+
+    def parse(self, request, context):
+        """
+        Get the parsed tree of sentences
+        """
+        rl = request.language
+        if rl != api_pb2.EN and rl != api_pb2.ES and rl != api_pb2.DE and \
+                rl != api_pb2.FR and rl != api_pb2.IT and rl != api_pb2.NL:
+            return api_pb2.ParseResponse(
+                isOk=False,
+                reason='Requested language is not implemented')
+        parsetree = getParser(request.language)
+        tree = parsetree(request.text, relations=True, lemmata=True)
+        data = {'isOk': True, 'reason': None, 'sentences': []}
+        for sentence in tree:
+            words = []
+            for word in sentence.words:
+                words.append({'type': word.type, 'text': word.string})
+            chunks = []
+            for chunk in sentence.chunks:
+                words = [{'text': w.string, 'type': w.type}
+                         for w in chunk.words]
+                chunk = {'type': chunk.type, 'words': words}
+                chunks.append(chunk)
+            data['sentences'].append({'words': words, 'chunks': chunks})
+        ret = json_format.ParseDict(data, api_pb2.ParseResponse())
+        return ret
+
+
+if __name__ == "__main__":
+    logging.basicConfig(level=logging.DEBUG)
+    server = grpc.server(futures.ThreadPoolExecutor(max_workers=10))
+    api_pb2_grpc.add_APIServicer_to_server(APIServicer(), server)
+    server.add_insecure_port(GRPC_PORT)
+    logging.debug('Starting gRPC-Pattern service in %s...' % GRPC_PORT)
+    server.start()
+    while True:
+        time.sleep(100)

start.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+
+python -m grpc_tools.protoc -I . --python_out=api --grpc_python_out=api api.proto
+python server.py