fix: remplace cv url by cv binary (#220)

* fix:add some error handling for jobology connector * fix:jobology flake8 connector * fix:some type * fix:regarding jamal review * fix: remplace cv url by cv binary * docs: update docs * fix: flake8 outputs * fix: jobology catch profile * docs: update docs * fix: regarding jamal review * fix: handle possible error binasciii * fix: flake8 and docs * fix: some flake8 output
Riminder · Feb 12, 2024 · b41676e · b41676e
1 parent 4785086
commit b41676e
Show file tree

Hide file tree

Showing 3 changed files with 33 additions and 18 deletions.
diff --git a/README.md b/README.md
@@ -135,7 +135,7 @@ We invite developers to join us in our mission to bring AI and data integration
 | **Indeed**                                                                                                                            | Job Board           | 🎯                  |                |               |
 | **Inzojob**                                                                                                                           | Job Board           | 🎯                  |                |               |
 | **Jobijoba**                                                                                                                          | Job Board           | 🎯                  |                |               |
-| [**Jobology**](./src/hrflow_connectors/connectors/jobology/README.md) | Job Board | :white_check_mark: | *21/12/2022* | *08/01/2024* | :x: | :x: | :x: | :x: |
+| [**Jobology**](./src/hrflow_connectors/connectors/jobology/README.md) | Job Board | :white_check_mark: | *21/12/2022* | *10/02/2024* | :x: | :x: | :x: | :x: |
 | **Jobrapido**                                                                                                                         | Job Board           | 🎯                  |                |               |
 | **JobTeaser**                                                                                                                         | Job Board           | 🎯                  |                |               |
 | **Jobtransport**                                                                                                                      | Job Board           | 🎯                  |                |               |

diff --git a/src/hrflow_connectors/connectors/jobology/test-config.yaml b/src/hrflow_connectors/connectors/jobology/test-config.yaml
@@ -9,7 +9,7 @@ warehouse:
             "lastName":"Mezid",
             "phone":"0611423374",
             "email":"pierre.dupont@mailprovider.com",
-            "cvUrl":"https://riminder-documents-eu-2019-12.s3-eu-west-1.amazonaws.com/teams/fc9d40fd60e679119130ea74ae1d34a3e22174f2/sources/e6159e65395995ae9f69a90ce916f7b6f90823dd/profiles/ab1c521d49b73d321bd54f6f6d3a4d81a31835eb/parsing/original.pdf",
+            "cvBase64": b"JVBERi0xLjcNCiW1tb...",
             "coverText":"Ceci est le texte de motivation.\nCe champ peut contenir des retours à la ligne.",
             "profilecountry": "France",
             "profileregions": "Ile-de-france",
@@ -29,7 +29,7 @@ actions:
             "lastName":"Mezid",
             "phone":"0611423374",
             "email":"pierre.dupont@mailprovider.com",
-            "cvUrl":"https://riminder-documents-eu-2019-12.s3-eu-west-1.amazonaws.com/teams/fc9d40fd60e679119130ea74ae1d34a3e22174f2/sources/e6159e65395995ae9f69a90ce916f7b6f90823dd/profiles/ab1c521d49b73d321bd54f6f6d3a4d81a31835eb/parsing/original.pdf",
+            "cvBase64": b"JVBERi0xLjcNCiW1tb...",
             "coverText":"Ceci est le texte de motivation.\nCe champ peut contenir des retours à la ligne.",
             "profilecountry": "France",
             "profileregions": "Ile-de-france",
@@ -53,7 +53,7 @@ actions:
             "lastName":"Mezid",
             "phone":"0611423374",
             "email":"pierre.dupont@mailprovider.com",
-            "cvUrl":"https://riminder-documents-eu-2019-12.s3-eu-west-1.amazonaws.com/teams/fc9d40fd60e679119130ea74ae1d34a3e22174f2/sources/e6159e65395995ae9f69a90ce916f7b6f90823dd/profiles/ab1c521d49b73d321bd54f6f6d3a4d81a31835eb/parsing/original.pdf",
+            "cvBase64": b"JVBERi0xLjcNCiW1tb...",
             "coverText":"Ceci est le texte de motivation.\nCe champ peut contenir des retours à la ligne.",
             "profilecountry": "France",
             "profileregions": "Ile-de-france",
@@ -78,7 +78,7 @@ actions:
             "lastName":"Mezid",
             "phone":"0611423374",
             "email":"pierre.dupont@mailprovider.com",
-            "cvUrl":"https://riminder-documents-eu-2019-12.s3-eu-west-1.amazonaws.com/teams/fc9d40fd60e679119130ea74ae1d34a3e22174f2/sources/e6159e65395995ae9f69a90ce916f7b6f90823dd/profiles/ab1c521d49b73d321bd54f6f6d3a4d81a31835eb/parsing/original.pdf",
+            "cvBase64": b"JVBERi0xLjcNCiW1tb...",
             "coverText":"Ceci est le texte de motivation.\nCe champ peut contenir des retours à la ligne.",
             "profilecountry": "France",
             "profileregions": "Ile-de-france",

diff --git a/src/hrflow_connectors/connectors/jobology/warehouse.py b/src/hrflow_connectors/connectors/jobology/warehouse.py
@@ -1,7 +1,8 @@
+import base64
 import typing as t
 from logging import LoggerAdapter
 
-import requests
+import magic
 from pydantic import Field
 
 from hrflow_connectors.core import (
@@ -22,25 +23,39 @@ class ReadProfilesParameters(ParametersModel):
     )
 
 
+def get_content_type(binary_data: bytes):
+    mime = magic.Magic(mime=True)
+    content_type = mime.from_buffer(binary_data)
+    if not content_type:
+        return "application/octet-stream"
+    return content_type
+
+
 def read(
     adapter: LoggerAdapter,
     parameters: ReadProfilesParameters,
     read_mode: t.Optional[ReadMode] = None,
     read_from: t.Optional[str] = None,
 ) -> t.Iterable[t.Dict]:
     result = {**parameters.profile}
-    cv_url = result["cvUrl"]
-    response = requests.get(cv_url)
-    if response.status_code == 200:
-        result["cv"] = response.content
-        result["content_type"] = response.headers["Content-Type"]
-    elif response.status_code == 400:
-        raise Exception(f"Bad Request {response.text}")
-    else:
-        raise Exception(
-            f"request failed with status code {response.status_code} and message"
-            f" {response.text}"
-        )
+    cv_base64 = result.get("cvBase64")
+
+    if cv_base64 is None:
+        raise ValueError("No base64 string provided for CV.")
+
+    try:
+        binary_data = base64.b64decode(cv_base64)
+    except base64.binascii.Error:
+        padding_needed = 4 - (len(cv_base64) % 4)
+        if padding_needed != 4:
+            cv_base64 += "=" * padding_needed
+            binary_data = base64.b64decode(cv_base64)
+
+    if not binary_data:
+        raise Exception("Error decoding base64 string")
+    content_type = get_content_type(binary_data)
+    result["cv"] = binary_data
+    result["content_type"] = content_type
 
     return [result]