EES-4727: Tidying up the data_ingestion code by adding types and spli…

data_ingestion/routers/maintenance.py

@@ -12,26 +12,24 @@
 router = APIRouter(prefix="/api/maintenance")
 
 
-@router.post("/publications/build")
-async def build_publications():
-    slugs = fetch_publication_slugs()
+@router.post(path="/publications/build")
+async def build_publications() -> JSONResponse:
     try:
-        data_upsertion(slugs, extract_releases)
+        data_upsertion(records=extract_releases(slugs=fetch_publication_slugs()))
     except Exception as e:
         return JSONResponse(status_code=500, content={"Content": e})
     return JSONResponse(status_code=200, content={"Content": "Successful"})
 
 
-@router.post("/methodologies/build")
-async def build_methodologies():
-    slugs = fetch_methodology_slugs()
+@router.post(path="/methodologies/build")
+async def build_methodologies() -> JSONResponse:
     try:
-        data_upsertion(slugs, extract_methodologies)
+        data_upsertion(records=extract_methodologies(slugs=fetch_methodology_slugs()))
     except Exception as e:
         return JSONResponse(status_code=500, content={"Content": e})
     return JSONResponse(status_code=200, content={"Content": "Successful"})
 
 
-@router.delete("/clear", status_code=status.HTTP_204_NO_CONTENT)
-async def clear():
+@router.delete(path="/clear", status_code=status.HTTP_204_NO_CONTENT)
+async def clear() -> None:
     recreate_collection()

data_ingestion/routers/methodologies.py

@@ -7,11 +7,11 @@
 router = APIRouter(prefix="/api/methodologies")
 
 
-@router.post("/{slug}/update")
-def update(slug: str):
+@router.post(path="/{slug}/update")
+def update(slug: str) -> JSONResponse:
     try:
         delete_methodology(slug=slug)
-        data_upsertion([slug], extract_methodologies)
+        data_upsertion(records=extract_methodologies(slugs=[slug]))
     except Exception as e:
         return JSONResponse(status_code=500, content={"Content": e})
     return JSONResponse(status_code=200, content={"Content": "Succesful"})

data_ingestion/routers/publications.py

@@ -8,11 +8,11 @@
 router = APIRouter(prefix="/api/publications")
 
 
-@router.post("/{slug}/update")
-async def update(slug: str):
+@router.post(path="/{slug}/update")
+async def update(slug: str) -> JSONResponse:
     try:
         delete_publication(slug=slug)
-        data_upsertion([slug], extract_releases)
+        data_upsertion(records=extract_releases(slugs=[slug]))
     except Exception as e:
         return JSONResponse(status_code=500, content={"Content": e})
     return JSONResponse(status_code=200, content={"Content": "Successful"})

data_ingestion/services/methodology_service.py

@@ -1,4 +1,3 @@
-import json
 import logging
 
 import requests
@@ -10,50 +9,54 @@
 logger = logging.getLogger(__name__)
 
 
-def delete_methodology(slug: str):
+def delete_methodology(slug: str) -> None:
     delete_url(url=f"{settings.ees_url_api_content}/methodology{slug}")
 
 
-def extract_methodologies(slugs):
-    texts = []
-    for slug in slugs:
-        methodology_info = {}
-        content = fetch_methodology(slug)
-        methodology_info["text"] = content["data"]
-        methodology_info["link"] = content["link"]
-        texts.append(methodology_info)
-    return texts
+def extract_methodologies(slugs: list[str]) -> list[dict[str, str]]:
+    return list(map(fetch_methodology, slugs))
 
 
-def fetch_methodology(slug: str):
-    methodology_content = {}
-    methodology_content["link"] = f"{settings.ees_url_public_ui}/methodology/{slug}"
-    res = requests.get(f"{settings.ees_url_api_content}/methodologies/{slug}")
-    text = json.loads(res.text)
+def fetch_methodology(slug: str) -> dict[str, str]:
     try:
-        methodology_content["data"] = "Headlines Section: "
-        methodology_content["data"] += BeautifulSoup(
-            text["headlinesSection"]["content"][0]["body"], "html.parser"
-        ).get_text()
-    except KeyError as e:
-        logger.error(f" Error: Key '{e.args[0]}' not found whilst reading content for methodology with slug: '{slug}'")
-
-    methodology_content["data"] += "Content Section"
-    for i in range(len(text["content"])):
-        for j in range(len(text["content"][i]["content"])):
-            try:
-                methodology_content["data"] += BeautifulSoup(
-                    text["content"][i]["content"][j]["body"], "html.parser"
-                ).get_text()
-            except KeyError:
-                logger.debug(f"Key does not exist for {slug} at {i}")
-    return methodology_content
-
-
-def fetch_methodology_slugs():
-    data = requests.get(f"{settings.ees_url_api_content}/methodology-themes").json()
-    slugs = []
-    for item in data:
+        response = requests.get(url=f"{settings.ees_url_api_content}/methodologies/{slug}")
+        response.raise_for_status()
+        response_json = response.json()
+        methodology_version_id = response_json["id"]
+
+        logger.debug(f"Processing content for methodology version: {methodology_version_id}")
+
+        return {
+            "link": f"{settings.ees_url_public_ui}/methodology/{slug}",
+            "text": get_general_content(res=response_json),
+        }
+    except requests.exceptions.HTTPError as err:
+        # TODO Why are some methodologies not found?
+        if err.response.status_code == 404:
+            logger.error(f"Methodology version for slug {slug} was not found")
+            return {}
+        else:
+            raise
+
+
+def get_general_content(res: dict) -> str:
+    content_sections = res["content"]
+    result = "Content: "
+    for section_index in range(len(content_sections)):
+        content_blocks = content_sections[section_index]["content"]
+        for block_index in range(len(content_blocks)):
+            content_block = content_blocks[block_index]
+            if content_block["type"] == "HtmlBlock":
+                result += BeautifulSoup(markup=content_block["body"], features="html.parser").get_text()
+    return result
+
+
+def fetch_methodology_slugs() -> list[str]:
+    response = requests.get(url=f"{settings.ees_url_api_content}/methodology-themes")
+    response.raise_for_status()
+    response_json = response.json()
+    slugs: list[str] = []
+    for item in response_json:
         for topic in item["topics"]:
             for publication in topic["publications"]:
                 for methodology in publication["methodologies"]:

data_ingestion/services/publication_service.py

@@ -1,4 +1,3 @@
-import json
 import logging
 
 import requests
@@ -9,20 +8,15 @@
 logger = logging.getLogger(__name__)
 
 
-def delete_publication(slug: str):
+def delete_publication(slug: str) -> None:
     delete_url(url=f"{settings.ees_url_public_ui}/find-statistics/{slug}")
 
 
-def fetch_publication_slugs():
-    try:
-        response = requests.get(
-            f"{settings.ees_url_api_content}/publications?page=1&pageSize=9999&sort=published&order=asc"
-        )
-        response.raise_for_status()
-        publications = json.loads(response.text)["results"]
-        slugs = [publications[i]["slug"] for i in range(len(publications))]
-        return slugs
-    except requests.HTTPError as http_err:
-        logger.error(f"HTTP error occurred: {http_err}")
-    except Exception as err:
-        logger.error(f"Other error occurred: {err}")
+def fetch_publication_slugs() -> list[str]:
+    response = requests.get(
+        url=f"{settings.ees_url_api_content}/publications?page=1&pageSize=9999&sort=published&order=asc"
+    )
+    response.raise_for_status()
+    response_json = response.json()
+    publications = response_json["results"]
+    return list(map(lambda publication: publication["slug"], publications))

data_ingestion/services/release_service.py

@@ -1,60 +1,71 @@
 import logging
-from typing import Dict, List
 
 import requests
 from bs4 import BeautifulSoup
 
 from ..config import settings
-from .tablebuilder_service import fetch_key_stat
+from .tablebuilder_service import fetch_data_block
 
 logger = logging.getLogger(__name__)
 
 
-def extract_releases(slugs: str) -> List[Dict]:
-    texts = []
-    for slug in slugs:
-        slug_info = {}
-        res = requests.get(f"{settings.ees_url_api_content}/publications/{slug}/releases/latest")
-        key_stats = {}
-        response_json = res.json()
-        release_id = response_json["publication"]["releases"][0]["id"]
-        try:
-            key_statistics = response_json["keyStatistics"]
-            if key_statistics != []:
-                data_strings = []
-                for i, statistic in enumerate(key_statistics):
-                    data_strings.append(fetch_key_stat(statistic, release_id, i))
-                key_stats["data"] = "Key Statistics section: ".join(data_strings)
-        except KeyError:
-            logger.warn(f"{slug} doesnt contain key stats")
-        try:
-            slug_info["text"] = key_stats["data"]
-            content = fetch_release(slug, response_json)
-            slug_info["text"] += content["data"]
-            slug_info["link"] = content["link"]
-        except Exception:
-            logger.warn(f"{slug} doesnt contain key stats")
-            content = fetch_release(slug, response_json)
-            slug_info["text"] = content["data"]
-            slug_info["link"] = content["link"]
-        texts.append(slug_info)
-    return texts
-
-
-def fetch_release(slug: str, res: dict) -> dict:
-    slug_content = {}
-    slug_content["link"] = f"{settings.ees_url_public_ui}/find-statistics/{slug}"
-    try:
-        slug_content["data"] = "Headlines Section: "
-        slug_content["data"] += BeautifulSoup(res["headlinesSection"]["content"][0]["body"], "html.parser").get_text()
-    except Exception as e:
-        logger.info(f" Error: {e}. For {slug} the headlines section doesnt exist")
-
-    slug_content["data"] += "Content Section"
-    for i in range(len(res["content"])):
-        for j in range(len(res["content"][i]["content"])):
-            try:
-                slug_content["data"] += BeautifulSoup(res["content"][i]["content"][j]["body"], "html.parser").get_text()
-            except KeyError:
-                logger.debug(f"Key does not exist for {slug} at {i}")
-    return slug_content
+def extract_releases(slugs: list[str]) -> list[dict[str, str]]:
+    return list(map(fetch_release, slugs))
+
+
+def fetch_release(slug: str) -> dict[str, str]:
+    response = requests.get(url=f"{settings.ees_url_api_content}/publications/{slug}/releases/latest")
+    response.raise_for_status()
+    response_json = response.json()
+    release_id = response_json["id"]
+
+    logger.debug(f"Processing content for release id: {release_id}")
+
+    headlines_content = str(get_headlines_content(res=response_json))
+    key_stats_content = get_key_statistics_content(release_id=release_id, res=response_json)
+    general_content = get_general_content(res=response_json)
+
+    return {
+        "link": f"{settings.ees_url_public_ui}/find-statistics/{slug}",
+        "text": f"{headlines_content}{key_stats_content}{general_content}",
+    }
+
+
+def get_headlines_content(res: dict) -> str | None:
+    headlines_section = res["headlinesSection"]["content"]
+    if headlines_section:
+        headlines_content_block = headlines_section[0]
+        headlines = BeautifulSoup(markup=headlines_content_block["body"], features="html.parser").get_text()
+        return f"Headline: {headlines}"
+
+
+def get_key_statistics_content(release_id: str, res: dict) -> str | None:
+    key_statistics = res["keyStatistics"]
+    if key_statistics:
+        key_statistics_content = list(
+            map(
+                lambda item: get_key_statistic_content(release_id=release_id, index_and_key_statistic=item),
+                enumerate(key_statistics),
+            )
+        )
+        return "Key statistic ".join(key_statistics_content)
+
+
+def get_key_statistic_content(release_id: str, index_and_key_statistic: tuple[int, dict[str, str]]) -> str:
+    index, key_statistic = index_and_key_statistic
+    data_block_id = key_statistic["dataBlockId"]
+    return fetch_data_block(
+        release_id=release_id, data_block_id=data_block_id, key_statistic=key_statistic, index=index
+    )
+
+
+def get_general_content(res: dict) -> str:
+    content_sections = res["content"]
+    result = "Content: "
+    for section_index in range(len(content_sections)):
+        content_blocks = content_sections[section_index]["content"]
+        for block_index in range(len(content_blocks)):
+            content_block = content_blocks[block_index]
+            if content_block["type"] == "HtmlBlock":
+                result += BeautifulSoup(markup=content_block["body"], features="html.parser").get_text()
+    return result

data_ingestion/services/tablebuilder_service.py

@@ -7,25 +7,27 @@
 logger = logging.getLogger(__name__)
 
 
-def fetch_key_stat(statistic: dict, release_id: str, i: int) -> str:
+def fetch_data_block(release_id: str, data_block_id: str, key_statistic: dict[str, str], index: int) -> str:
     try:
-        data_block_id = statistic["dataBlockId"]
-        res = requests.get(f"{settings.ees_url_api_data}/tablebuilder/release/{release_id}/data-block/{data_block_id}")
-        response_json = res.json()
+        response = requests.get(
+            url=f"{settings.ees_url_api_data}/tablebuilder/release/{release_id}/data-block/{data_block_id}"
+        )
+        response.raise_for_status()
+        response_json = response.json()
         label = response_json["subjectMeta"]["indicators"][0]["label"]
         measure = list(response_json["results"][0]["measures"].values())[0]
         try:
             unit = response_json["subjectMeta"]["indicators"][0]["unit"]
-            measure = f"{measure} {unit}"
+            measure = f"{measure}{unit}"
         except KeyError:
             logger.error("No unit found")
     except Exception:
-        label = statistic["title"]
-        measure = statistic["statistic"]
+        label = key_statistic["title"]
+        measure = key_statistic["statistic"]
     try:
-        trend = statistic["trend"]
-        data_string = f"{i + 1}: {label}-{measure} {trend}."
+        trend = key_statistic["trend"]
+        data_string: str = f"{index + 1}: {label}-{measure} {trend}."
     except Exception:
-        data_string = f"{i +1}: {label}-{measure}."
+        data_string = f"{index + 1}: {label}-{measure}."
 
     return data_string