<add> klang controller and service

app/klang/__init__.py

@@ -1,7 +1,7 @@
 BASE_ROUTE = "klang"
 
 
-# def register_routes(api, app, root="api"):
-#     from .controller import api as project_api
+def register_routes(api, app, root="api"):
+    from .controller import api as project_api
 
-#     api.add_namespace(project_api, path=f"/{root}/{BASE_ROUTE}")
+    api.add_namespace(project_api, path=f"/{root}/{BASE_ROUTE}")

app/klang/controller.py

@@ -0,0 +1,31 @@
+from datetime import datetime
+from typing import List
+
+from flask import session
+from flask_accepts.decorators.decorators import responds
+from flask_restx import Namespace, Resource
+
+from .service import ConllService
+
+api = Namespace("Klang", description="Single namespace, single entity")  # noqa
+
+@api.route("/conlls")
+class ConllServiceResource(Resource):
+    "ConllService"
+
+    def get(self) :
+        return ConllService.get_all()
+
+
+@api.route("/conll/<string:conll_name>")
+class ConllNameServiceResource(Resource):
+    "ConllService"
+
+    def get(self, conll_name) :
+        conll_string = ConllService.get_by_name(conll_name)
+        sentences_string = ConllService.seperate_conll_sentences(conll_string)
+        sentences_audio_token = []
+        for sentence_string in sentences_string:
+          audio_tokens = ConllService.sentence_to_audio_tokens(sentence_string)
+          sentences_audio_token.append(audio_tokens)
+        return sentences_audio_token

app/klang/service.py

@@ -1,33 +1,55 @@
 import os
+import re
+from typing import List
+
 from app import klang_config
 
+align_begin_and_end_regex = re.compile(
+    r"^\d+\t(.+?)\t.*AlignBegin=(\d+).*AlignEnd=(\d+)"
+)
+
 class ConllService:
     @staticmethod
     def get_path_data():
         path_data = klang_config.path
         return path_data
-    
+
     @staticmethod
     def get_path_conll(file_name_suffix):
         file_name = file_name_suffix + ".intervals.conll"
         path_data = ConllService.get_path_data()
-        path_conll = os.path.join(path_data,file_name_suffix, file_name)
+        path_conll = os.path.join(path_data, file_name_suffix, file_name)
         return path_conll
 
     @staticmethod
     def read_conll(path_conll):
         with open(path_conll, "r", encoding="utf-8") as infile:
             conll = infile.read()
-        return conll 
-
+        return conll
 
     @staticmethod
     def get_all():
         path_data = ConllService.get_path_data()
         conlls = os.listdir(path_data)
         return conlls
 
-    # @staticmethod
-    # def get_by_name(conll_name):
-    #     with
-    #     return True
+    @staticmethod
+    def get_by_name(conll_name):
+        path_conll = ConllService.get_path_conll(conll_name)
+        conll_string = ConllService.read_conll(path_conll)
+        return conll_string
+
+    @staticmethod
+    def seperate_conll_sentences(conll_string: str) -> List[str]:
+        return list(filter(lambda x: x != "", conll_string.split("\n\n")))
+
+    @staticmethod
+    def sentence_to_audio_tokens(sentence_string: str):
+        audio_tokens = []
+        for line in sentence_string.split("\n"):
+            if line:
+                if not line.startswith("#"):
+                    m = align_begin_and_end_regex.search(line)
+                    audio_tokens += [(m.group(1), int(m.group(2)), int(m.group(3)))]
+
+        return audio_tokens

app/klang/service_test.py

@@ -1,10 +1,13 @@
-from .service import ConllService
-from app.test.fixtures import app, db  # noqa
 import os
 
+from app.test.fixtures import app, db  # noqa
+
+from .service import ConllService
+
 folder_name = "data_test"
 file_name = "John_Doe"
 
+
 def test_get_path_data():
     path_data = ConllService.get_path_data()
     print("KK path_data", path_data)
@@ -23,9 +26,19 @@ def test_get_all():
     conlls = ConllService.get_all()
     assert conlls == ["John_Doe"]
 
-# # def test_get_by_name():
-# #     # conll = ConllService.get_by_name("John_Doe")
-# #     conll_string = ConllService.get_by_name(file_name)
-# #     path_conll = ConllService.get_path_conll(folder_name, file_name)
-# #     conll_string = ConllService.read_conll(path_conll)
-# #     assert type(conll_string) == str
+def test_get_by_name():
+    path_conll = ConllService.get_path_conll(file_name)
+    conll_string = ConllService.read_conll(path_conll)
+    assert conll_string == ConllService.get_by_name(file_name)
+
+def test_seperate_conll_sentences():
+    conll_string = "# sent_id = test_sentence_1\n1\ttest_token\ntest_lemma\ntest_upos\n\n# sent_id = test_sentence_2\n1\ttest_token\ntest_lemma\ntest_upos\n\n"
+    sentences = ConllService.seperate_conll_sentences(conll_string)
+    assert len(sentences) == 2
+
+def test_sentence_to_audio_tokens():
+    sentence = "# sent_id = test_sentence_1\n1\tdonc\tdonc\t_\t_\t_\t_\t_\t_\tAlignBegin=0|AlignEnd=454"
+    audio_tokens = ConllService.sentence_to_audio_tokens(sentence)
+    assert audio_tokens[0][0] == "donc"
+    assert audio_tokens[0][1] == 0
+    assert audio_tokens[0][2] == 454

app/routes.py

@@ -5,6 +5,7 @@ def register_routes(api, app, root="api"):
     from app.trees import register_routes as attach_trees
     from app.lexicon import register_routes as attach_lexicon
     from app.grew import register_routes as attach_grew
+    from app.klang import register_routes as attach_klang
 
     # Add routes
     attach_user(api, app, root)
@@ -13,3 +14,4 @@ def register_routes(api, app, root="api"):
     attach_trees(api, app, root)
     attach_lexicon(api, app, root)
     attach_grew(api, app, root)
+    attach_klang(api, app, root)

wsgi.py

@@ -8,6 +8,5 @@
 env = os.getenv("FLASK_ENV") or "test"
 
 app = create_app(env)
-print("KK app", app)
 if __name__ == "__main__":
     app.run("0.0.0.0", 8080, debug=True)