implemented article_cache endpoint

src/__init__.py

@@ -8,6 +8,7 @@
 https://github.com/pallets/flask/blob/1.1.2/examples/tutorial/flaskr/__init__.py
 """
 import logging
+import os
 
 from flask import Flask  # type: ignore
 from flask_restful import Api, Resource  # type: ignore
@@ -28,6 +29,8 @@
 from src.views.statistics.references import References
 from src.views.statistics.xhtml import Xhtml
 
+# new stuff apr 2024
+from src.views.v2.article_cache_view_v2 import ArticleCacheV2
 # new stuff jan 2024
 from src.views.v2.article_view_v2 import ArticleV2
 from src.views.version import Version
@@ -37,7 +40,6 @@
 
 app = Flask(__name__)
 
-
 def add_cors_headers(response):
     # Replace "*" with the specific origin(s) you want to allow
     response.headers["Access-Control-Allow-Origin"] = "*"
@@ -46,6 +48,9 @@ def add_cors_headers(response):
     return response
 
 
+# let's see if we can distinguish which server we are on
+server_name = os.getenv('FLASK_SERVER_NAME', 'Unknown Server')
+
 # Register the function as a after_request handler
 app.after_request(add_cors_headers)
 
@@ -56,6 +61,7 @@ def add_cors_headers(response):
 # Here we link together the API views and endpoint urls
 # api.add_resource(LookupByWikidataQid, "/wikidata-qid/<string:qid>")
 api.add_resource(ArticleV2, "/article")
+api.add_resource(ArticleCacheV2, "/article_cache")
 
 api.add_resource(Version, "/version")
 api.add_resource(CheckUrls, "/check-urls")

src/models/v2/file_io/article_cache_file_io_v2.py

@@ -0,0 +1,24 @@
+from typing import Any, Dict, Optional
+
+from src.models.exceptions import MissingInformationError
+from src.models.file_io import FileIo
+from src.models.v2.job.article_cache_job_v2 import ArticleCacheJobV2
+
+
+class ArticleCacheFileIoV2(FileIo):
+    data: Optional[Dict[str, Any]] = None
+
+    job: Optional[ArticleCacheJobV2]
+
+    subfolder = "articles/"
+
+    # we override FileIo::filename property to provide custom one for cached article
+    @property
+    def filename(self) -> str:
+        # raise NotFoundError()
+        if not self.job:
+            raise MissingInformationError("self.job undefined")
+
+        # we got a job, file name should be iari_id +".json"
+        filename = f"{self.job.iari_id}.json"
+        return filename

src/models/v2/job/article_cache_job_v2.py

@@ -0,0 +1,8 @@
+from src.models.v2.job import JobV2
+
+
+class ArticleCacheJobV2(JobV2):
+
+    # the iari id of the article as previously retrieved
+    iari_id: str = 0
+    article_version: int = 0  # should be 1 or 2

src/models/v2/schema/article_cache_schema_v2.py

@@ -0,0 +1,27 @@
+import logging
+
+from marshmallow import fields, post_load
+
+from src.models.v2.job.article_cache_job_v2 import ArticleCacheJobV2
+from src.models.v2.schema import BaseSchemaV2
+
+logger = logging.getLogger(__name__)
+
+
+class ArticleCacheSchemaV2(BaseSchemaV2):
+    # marshmallow style declarations
+    iari_id = fields.Str(required=True)
+    article_version = fields.Int(default=1)  # version 1 is original, 2 is V2
+
+    # noinspection PyUnusedLocal
+    @post_load
+    # runs after request args are loaded
+    # **kwargs is needed here despite what the validator claims
+    def return_object(self, data, **kwargs) -> ArticleCacheJobV2:  # type: ignore # dead: disable
+        """Return job object"""
+        from src import app
+
+        app.logger.debug("ArticleCacheSchemaV2::@post_load:return_object")
+
+        job = ArticleCacheJobV2(**data)
+        return job

src/views/v2/article_cache_view_v2.py

@@ -0,0 +1,86 @@
+# from flask_restful import Resource, abort  # type: ignore
+# from marshmallow import Schema
+from datetime import datetime
+from typing import Any, Optional, Tuple
+import traceback
+
+from src.models.exceptions import MissingInformationError, WikipediaApiFetchError
+
+from src.models.v2.file_io.article_cache_file_io_v2 import ArticleCacheFileIoV2
+from src.models.v2.job.article_cache_job_v2 import ArticleCacheJobV2
+from src.models.v2.schema.article_cache_schema_v2 import ArticleCacheSchemaV2
+
+from src.models.v2.wikimedia.wikipedia.analyzer_v2 import WikipediaAnalyzerV2
+from src.models.wikimedia.enums import AnalyzerReturnValues, WikimediaDomain
+from src.views.v2.statistics import StatisticsViewV2
+
+from src.helpers.get_version import get_poetry_version
+
+
+class ArticleCacheV2(StatisticsViewV2):
+    """
+    returns data associated with article specified by the schema
+    """
+
+    schema = ArticleCacheSchemaV2()  # overrides StatisticsViewV2's schema property
+    job: ArticleCacheJobV2  # overrides StatisticsViewV2's job property
+
+    def __setup_io__(self):
+        """
+        implementation for StatisticsWriteView.__setup_io__
+        """
+        self.io = ArticleCacheFileIoV2(job=self.job)
+
+    def __return_article_data__(self):
+        from src import app
+
+        app.logger.debug("ArticleCacheV2::__return_article_data__")
+
+        self.__setup_io__()  # defined right above!
+
+        self.__read_from_cache__()  # inherited from StatisticsWriteView; fills io.data if successful
+
+        # if self.io.data and not self.job.refresh:
+        if self.io.data:
+            # cached data has been successfully retrieved - return it
+            app.logger.info(
+                f"Returning cached articleV2 json data, date: {self.time_of_analysis}"
+            )
+            return self.io.data, 200
+
+        # else no cache, so return"no cached data" error
+
+
+    def get(self):
+        """
+        main entrypoint for flask
+        must return a tuple (Any,response_code)
+        """
+        from src import app
+        app.logger.debug("ArticleCacheV2::get")
+
+        try:
+            self.__validate_and_get_job__()
+            # inherited from StatisticsWriteView -> StatisticsViewV2
+            # sets up job parameters, possibly with some massaging from the @post_load function
+            # (@postload is courtesy of the marshmallow module addition)
+
+            if self.job.iari_id:
+                return self.__return_article_data__()
+
+            else:
+                return self.__return_article_error__()
+
+        except WikipediaApiFetchError as e:
+            return {"error": f"API Error: {str(e)}"}, 500
+
+        except Exception as e:
+            traceback.print_exc()
+            return {"error": f"General Error: {str(e)}"}, 500
+
+    def __return_article_error__(self):
+        from src import app
+
+        if self.job.iari_id == "":
+            app.logger.error("ArticleCacheV2: ERROR: iari_id is missing")
+            return "iari_id is missing", 400

src/views/v2/article_view_v2.py

@@ -62,7 +62,8 @@ def get(self):
         app.logger.debug("ArticleV2::get")
 
         try:
-            self.__validate_and_get_job__()  # inherited from StatisticsWriteView -> StatisticsView
+            self.__validate_and_get_job__()
+            # inherited from StatisticsWriteView -> StatisticsViewV2
             # sets up job parameters, possibly with some massaging from the @post_load function
             # (@postload is courtesy of the marshmallow module addition)
 

src/views/v2/statistics/__init__.py

@@ -1,5 +1,6 @@
 from datetime import datetime
 from typing import Optional
+import json
 
 from flask import request
 from flask_restful import Resource, abort  # type: ignore
@@ -39,7 +40,7 @@ class StatisticsViewV2(Resource):
 
     # derived ("child") class must implement __setup_io__ from this base ("parent") class
     def __setup_io__(self):
-        raise NotImplementedError()
+        raise NotImplementedError()  # must be defined in parent class
 
     def __setup_and_read_from_cache__(self):
         self.__setup_io__()  # sets up "io" property as FileIo instance