Use jaro_winkler similarity instead of rapidfuzz (#491)

### Description

Use jaro_winkler similarity from jellyfish module instead of rapidfuzz.
It has the same speed and better results

app/core/users/endpoints_users.py

@@ -34,7 +34,11 @@
 )
 from app.types.content_type import ContentType
 from app.utils.mail.mailworker import send_email
-from app.utils.tools import fuzzy_search_user, get_file_from_data, save_file_as_data
+from app.utils.tools import (
+    get_file_from_data,
+    save_file_as_data,
+    sort_user,
+)
 
 router = APIRouter(tags=["Users"])
 
@@ -95,7 +99,7 @@ async def search_users(
     user: models_core.CoreUser = Depends(is_user_an_ecl_member),
 ):
     """
-    Search for a user using Fuzzy String Matching
+    Search for a user using Jaro_Winkler distance algorithm. The
 
     `query` will be compared against users name, firstname and nickname
 
@@ -108,7 +112,7 @@ async def search_users(
         excluded_groups=excludedGroups,
     )
 
-    return fuzzy_search_user(query, users)
+    return sort_user(query, users)
 
 
 @router.get(

app/utils/tools.py

@@ -10,8 +10,8 @@
 import fitz
 from fastapi import HTTPException, UploadFile
 from fastapi.responses import FileResponse
+from jellyfish import jaro_winkler_similarity
 from pydantic import ValidationError
-from rapidfuzz import process
 from sqlalchemy.ext.asyncio import AsyncSession
 
 from app.core import cruds_core, models_core
@@ -60,7 +60,7 @@ def is_user_external(
     return user.external is True
 
 
-def fuzzy_search_user(
+def sort_user(
     query: str,
     users: Sequence[models_core.CoreUser],
     limit: int = 10,
@@ -71,26 +71,48 @@ def fuzzy_search_user(
     `query` will be compared against `users` name, firstname and nickname.
     The size of the answer can be limited using `limit` parameter.
 
-    Use RapidFuzz library
+    Use Jellyfish library
     """
 
-    # We can give a dictionary of {object: string used for the comparison} to the extract function
-    # https://maxbachmann.github.io/RapidFuzz/Usage/process.html#extract
-
     # TODO: we may want to cache this object. Its generation may take some time if there is a big user base
-    choices = []
-
-    for user in users:
-        choices.append(f"{user.firstname} {user.name} {user.nickname}")
-
-    results: list[tuple[str, int | float, int]] = process.extract(
-        query,
-        choices,
-        limit=limit,
-    )
-
-    # results has the format : (string used for the comparison, similarity score, index of the object in the choices collection)
-    return [users[res[2]] for res in results]
+    names = [f"{user.firstname} {user.name}" for user in users]
+    nicknames = [user.nickname for user in users]
+    scored: list[
+        tuple[CoreUser, float, float, int]
+    ] = [  # (user, name_score, nickname_score, index)
+        (
+            user,
+            jaro_winkler_similarity(query, name),
+            jaro_winkler_similarity(query, nickname) if nickname else 0,
+            index,
+        )
+        for index, (user, name, nickname) in enumerate(
+            zip(users, names, nicknames, strict=True),
+        )
+    ]
+
+    results = []
+    for _ in range(min(limit, len(scored))):
+        maximum_name = max(scored, key=lambda r: r[1])
+        maximum_nickname = max(scored, key=lambda r: r[2])
+        if maximum_name[1] > maximum_nickname[1]:
+            results.append(maximum_name)
+            scored[maximum_name[3]] = (  # We don't want to use this user again
+                maximum_name[0],
+                -1,
+                -1,
+                maximum_name[3],
+            )
+        else:
+            results.append(maximum_nickname)
+            scored[maximum_nickname[3]] = (  # We don't want to use this user again
+                maximum_nickname[0],
+                -1,
+                -1,
+                maximum_nickname[3],
+            )
+
+    return [result[0] for result in results]
 
 
 async def is_group_id_valid(group_id: str, db: AsyncSession) -> bool:

requirements-common.txt

@@ -7,14 +7,14 @@ fastapi==0.111.0
 firebase-admin==6.5.0               # Firebase is used for push notification
 HelloAssoAPIWrapper==1.0.0
 icalendar==5.0.13
+jellyfish==1.0.4                    # String Matching
 Jinja2==3.1.4                       # template engine for html files
 pydantic-settings==2.3.4
 pydantic==2.7.4
 pyjwt[crypto]==2.8.0                # generate and verify the JWT tokens, imported as `jwt`
 PyMuPDF==1.24.9                     # PDF processing
 python-dotenv==1.0.1                # load environment variables from .env file
 python-multipart==0.0.9             # a form data parser, as oauth flow requires form-data parameters
-rapidfuzz==3.9.5                    # Fuzzy String Matching
 redis==5.0.8
 requests==2.32.3
 SQLAlchemy[asyncio]==2.0.32         # [asyncio] allows greenlet to be installed on Apple M1 devices.