Merge pull request #91 from bothub-it/staging

Update master from staging

Pipfile

@@ -9,7 +9,7 @@ lint = "flake8"
 [packages]
 raven = "==6.10.0"
 requests = "==2.20.1"
-bothub_backend = {ref = "1.0.13-hotfix",git = "https://github.com/Ilhasoft/bothub-backend"}
+bothub_backend = {ref = "1.0.20",git = "https://github.com/Ilhasoft/bothub-backend"}
 kombu = "==4.6.11"
 celery = "==4.4.0"
 gunicorn = "==19.9.0"

bothub_nlp_api/handlers/evaluate.py

@@ -16,8 +16,66 @@
 EVALUATE_STATUS_FAILED = "failed"
 
 
+def crossvalidation_evaluate_handler(
+    authorization, language, repository_version=None
+):
+    if language and (
+        language not in settings.SUPPORTED_LANGUAGES.keys()
+        and language not in DEFAULT_LANGS_PRIORITY.keys()
+    ):
+        raise ValidationError("Language '{}' not supported by now.".format(language))
+
+    repository_authorization = get_repository_authorization(authorization)
+    if not repository_authorization:
+        raise AuthorizationIsRequired()
+
+    try:
+        update = backend().request_backend_start_automatic_evaluate(
+            repository_authorization, repository_version, language
+        )
+    except Exception as err:
+        print(err)
+        update = {}
+
+    if not update.get("can_run_automatic_evaluate"):
+        raise ValidationError("Validation error")
+
+    model = get_language_model(update)
+
+    try:
+        job_id = f'bothub_{settings.ENVIRONMENT}_evaluate_{update.get("repository_version_language_id")}_{language}_{str(int(time.time()))}'
+        send_job_train_ai_platform(
+            jobId=job_id,
+            repository_version=str(update.get("repository_version_language_id")),
+            by_id=str(update.get("user_id")),
+            repository_authorization=str(repository_authorization),
+            language=language,
+            type_model=model,
+            operation="evaluate",
+        )
+        backend().request_backend_save_queue_id(
+            update_id=str(update.get("repository_version_language_id")),
+            repository_authorization=str(repository_authorization),
+            task_id=job_id,
+            from_queue=0,
+            type_processing=2,
+        )
+        evaluate_report = {
+            "language": language,
+            "status": EVALUATE_STATUS_PROCESSING,
+            "repository_version": update.get("repository_version_language_id"),
+            "evaluate_id": None,
+            "evaluate_version": None,
+            "cross_validation": True,
+        }
+    except Exception as e:
+        evaluate_report = {"status": EVALUATE_STATUS_FAILED, "error": str(e)}
+
+    return evaluate_report
+
+
 def evaluate_handler(
-    authorization, language, repository_version=None, cross_validation=False
+    authorization, language, repository_version=None
 ):
     if language and (
         language not in settings.SUPPORTED_LANGUAGES.keys()
@@ -43,36 +101,20 @@ def evaluate_handler(
 
     try:
         evaluate = None
-        if cross_validation is False:
-            evaluate_task = celery_app.send_task(
-                TASK_NLU_EVALUATE_UPDATE,
-                args=[
-                    update.get("repository_version"),
-                    update.get("user_id"),
-                    repository_authorization,
-                    cross_validation,
-                ],
-                queue=queue_name(update.get("language"), ACTION_EVALUATE, model),
-            )
-            evaluate_task.wait()
-            evaluate = evaluate_task.result
-        else:
-            job_id = f'bothub_{settings.ENVIRONMENT}_evaluate_{str(update.get("repository_version"))}_{language}_{str(int(time.time()))}'
-            send_job_train_ai_platform(
-                jobId=job_id,
-                repository_version=str(update.get("repository_version")),
-                by_id=str(update.get("user_id")),
-                repository_authorization=str(repository_authorization),
-                language=language,
-                type_model=model,
-                operation="evaluate",
-            )
-            backend().request_backend_save_queue_id(
-                update_id=str(update.get("repository_version")),
-                repository_authorization=str(repository_authorization),
-                task_id=job_id,
-                from_queue=0,
-            )
+        cross_validation = False
+        evaluate_task = celery_app.send_task(
+            TASK_NLU_EVALUATE_UPDATE,
+            args=[
+                repository_version,
+                update.get("repository_version"),  # repository_version_language_id
+                repository_authorization,
+                cross_validation,
+                update.get("language")
+            ],
+            queue=queue_name(update.get("language"), ACTION_EVALUATE, model),
+        )
+        evaluate_task.wait()
+        evaluate = evaluate_task.result
 
         evaluate_report = {
             "language": language,

bothub_nlp_api/handlers/train.py

@@ -57,6 +57,7 @@ def train_handler(authorization, repository_version=None, language=None):
                 repository_authorization=str(repository_authorization),
                 task_id=job_id,
                 from_queue=0,
+                type_processing=0,
             )
         languages_report[update.get("language")] = {"status": TRAIN_STATUS_PROCESSING}
 

bothub_nlp_api/routers/v2.py

@@ -1,7 +1,8 @@
 from fastapi import Depends, APIRouter, Header, HTTPException
 from starlette.requests import Request
 
-from bothub_nlp_api.handlers import evaluate, task_queue
+from bothub_nlp_api.handlers import evaluate
+from bothub_nlp_api.handlers import task_queue
 from bothub_nlp_api.handlers import parse
 from bothub_nlp_api.handlers import debug_parse
 from bothub_nlp_api.handlers import sentence_suggestion
@@ -83,7 +84,6 @@ async def sentence_suggestion_post_handler(item: SentenceSuggestionRequest,):
         item.language,
         item.n_sentences_to_generate,
         item.percentage_to_replace,
-        item.intent,
     )
 
 
@@ -188,9 +188,14 @@ async def evaluate_handler(
     request: Request = Depends(AuthorizationRequired()),
     Authorization: str = Header(..., description="Bearer your_key"),
 ):
-    result = evaluate.evaluate_handler(
-        Authorization, item.language, item.repository_version, item.cross_validation
-    )
+    if item.cross_validation:
+        result = evaluate.crossvalidation_evaluate_handler(
+            Authorization, item.language, item.repository_version
+        )
+    else:
+        result = evaluate.evaluate_handler(
+            Authorization, item.language, item.repository_version
+        )
     if result.get("status") and result.get("error"):
         raise HTTPException(status_code=400, detail=result)
     return result

bothub_nlp_api/settings.py

@@ -34,7 +34,11 @@ def cast_supported_languages(i):
             "gs://poc-training-ai-platform/bothub-nlp-ai-platform/bothub-nlp-ai-platform-0.1.tar.gz"
         ],
     ),
-    BOTHUB_GOOGLE_AI_PLATFORM_JOB_TIMEOUT=(int, None)
+    BOTHUB_GOOGLE_AI_PLATFORM_JOB_TIMEOUT=(int, None),
+    BOTHUB_NLP_AWS_S3_BUCKET_NAME=(str, None),
+    BOTHUB_NLP_AWS_ACCESS_KEY_ID=(str, None),
+    BOTHUB_NLP_AWS_SECRET_ACCESS_KEY=(str, None),
+    BOTHUB_NLP_AWS_REGION_NAME=(str, None)
 )
 
 ENVIRONMENT = env.str("ENVIRONMENT")
@@ -78,3 +82,8 @@ def cast_supported_languages(i):
 )
 
 BOTHUB_GOOGLE_AI_PLATFORM_JOB_TIMEOUT = env.int("BOTHUB_GOOGLE_AI_PLATFORM_JOB_TIMEOUT")
+
+BOTHUB_NLP_AWS_S3_BUCKET_NAME = env.str("BOTHUB_NLP_AWS_S3_BUCKET_NAME")
+BOTHUB_NLP_AWS_ACCESS_KEY_ID = env.str("BOTHUB_NLP_AWS_ACCESS_KEY_ID")
+BOTHUB_NLP_AWS_SECRET_ACCESS_KEY = env.str("BOTHUB_NLP_AWS_SECRET_ACCESS_KEY")
+BOTHUB_NLP_AWS_REGION_NAME = env.str("BOTHUB_NLP_AWS_REGION_NAME")

bothub_nlp_api/utils.py

@@ -149,6 +149,18 @@ def send_job_train_ai_platform(
 
     if type_model is not None:
         args.extend(["--AIPLATFORM_LANGUAGE_MODEL", type_model])
+
+    if operation == "evaluate":
+        args.extend(
+            [
+                "--BOTHUB_NLP_AWS_S3_BUCKET_NAME", settings.BOTHUB_NLP_AWS_S3_BUCKET_NAME,
+                "--BOTHUB_NLP_AWS_ACCESS_KEY_ID", settings.BOTHUB_NLP_AWS_ACCESS_KEY_ID,
+                "--BOTHUB_NLP_AWS_SECRET_ACCESS_KEY", settings.BOTHUB_NLP_AWS_SECRET_ACCESS_KEY,
+                "--BOTHUB_NLP_AWS_REGION_NAME", settings.BOTHUB_NLP_AWS_REGION_NAME,
+            ]
+        )
+
+
     training_inputs = {
         "scaleTier": "CUSTOM",
         "masterType": "standard_gpu",