Created a softball scraper, edited unique sport constraint

.gitignore

@@ -4,4 +4,5 @@ __pycache__/
 .env
 .envrc
 .DS_Store
-ca-certificate.crt
+ca-certificate.crt
+firebase-service-account-key.json

app.py

@@ -23,6 +23,27 @@
 from src.utils.team_loader import TeamLoader
 from src.database import db
 
+import os
+import firebase_admin
+from firebase_admin import credentials, auth
+
+SERVICE_ACCOUNT_PATH = os.environ.get("GOOGLE_APPLICATION_CREDENTIALS")
+
+
+def initialize_firebase():
+    if not firebase_admin._apps:
+        if not SERVICE_ACCOUNT_PATH:
+            raise ValueError(
+                "GOOGLE_APPLICATION_CREDENTIALS is not set. Set it to your firebase-service-account-key.json path."
+            )
+        cred = credentials.Certificate(SERVICE_ACCOUNT_PATH)
+        firebase_admin.initialize_app(cred)
+        logging.info("Firebase app initialized.")
+    return firebase_admin.get_app()
+
+
+initialize_firebase()
+
 app = Flask(__name__)
 
 # CORS: allow frontend (different origin) to call this API

requirements.txt

@@ -11,3 +11,4 @@ Flask-APScheduler
 python-dotenv
 pytz
 gunicorn
+firebase-admin

src/database.py

@@ -70,6 +70,7 @@ def setup_database_indexes():
         game_collection.create_index([("date", -1)], background=True)
 
         try:
+            # Ensure doubleheaders on the same day remain distinct by including `time`.
             game_collection.create_index(
                 [
                     ("sport", 1),
@@ -79,8 +80,10 @@ def setup_database_indexes():
                     ("city", 1),
                     ("state", 1),
                     ("location", 1),
+                    ("time", 1),
                 ],
                 unique=True,
+                name="uniq_game_key_with_time",
                 background=True
             )
         except (DuplicateKeyError, OperationFailure) as e:

src/mutations/login_user.py

@@ -1,19 +1,26 @@
 from graphql import GraphQLError
-from graphene import Mutation, String, Field
+from graphene import Mutation, String
 
+from firebase_admin import auth as firebase_auth
 from flask_jwt_extended import create_access_token, create_refresh_token
 from src.database import db
 
 
 class LoginUser(Mutation):
     class Arguments:
-        net_id = String(required=True, description="User's net ID (e.g. Cornell netid).")
+        id_token = String(required=True, description="Firebase ID token from the client.")
 
     access_token = String()
     refresh_token = String()
 
-    def mutate(self, info, net_id):
-        user = db["users"].find_one({"net_id": net_id})
+    def mutate(self, info, id_token):
+        try:
+            decoded = firebase_auth.verify_id_token(id_token)
+        except Exception:
+            raise GraphQLError("Invalid or expired token.")
+
+        firebase_uid = decoded["uid"]
+        user = db["users"].find_one({"firebase_uid": firebase_uid})
         if not user:
             raise GraphQLError("User not found.")
         identity = str(user["_id"])

src/mutations/signup_user.py

@@ -1,30 +1,38 @@
 from graphql import GraphQLError
 from graphene import Mutation, String
 
+from firebase_admin import auth as firebase_auth
 from flask_jwt_extended import create_access_token, create_refresh_token
 from src.database import db
 
 
 class SignupUser(Mutation):
     class Arguments:
-        net_id = String(required=True, description="User's net ID (e.g. Cornell netid).")
+        id_token = String(required=True, description="Firebase ID token from the client.")
         name = String(required=False, description="Display name.")
-        email = String(required=False, description="Email address.")
+        email = String(required=False, description="Email (overrides token email if provided).")
 
     access_token = String()
     refresh_token = String()
 
-    def mutate(self, info, net_id, name=None, email=None):
-        if db["users"].find_one({"net_id": net_id}):
-            raise GraphQLError("Net ID already exists.")
+    def mutate(self, info, id_token, name=None, email=None):
+        try:
+            decoded = firebase_auth.verify_id_token(id_token)
+        except Exception:
+            raise GraphQLError("Invalid or expired token.")
+
+        firebase_uid = decoded["uid"]
+        if db["users"].find_one({"firebase_uid": firebase_uid}):
+            raise GraphQLError("User already exists.")
+
+        email = email or decoded.get("email")
         user_doc = {
-            "net_id": net_id,
+            "firebase_uid": firebase_uid,
+            "email": email,
             "favorite_game_ids": [],
         }
         if name is not None:
             user_doc["name"] = name
-        if email is not None:
-            user_doc["email"] = email
         result = db["users"].insert_one(user_doc)
         identity = str(result.inserted_id)
         return SignupUser(

src/schema.py

@@ -31,9 +31,11 @@ class Mutation(ObjectType):
     create_team = CreateTeam.Field(description="Creates a new team.")
     create_youtube_video = CreateYoutubeVideo.Field(description="Creates a new youtube video.")
     create_article = CreateArticle.Field(description="Creates a new article.")
-    login_user = LoginUser.Field(description="Login by net_id; returns access_token and refresh_token.")
+    login_user = LoginUser.Field(
+        description="Login with Firebase ID token; returns access_token and refresh_token.",
+    )
     signup_user = SignupUser.Field(
-        description="Create a new user by net_id; returns access_token and refresh_token (no separate login needed).",
+        description="Create a new user with Firebase ID token; returns access_token and refresh_token.",
     )
     refresh_access_token = RefreshAccessToken.Field(
         description="Exchange a valid refresh token (in Authorization header) for a new access_token.",

src/scrapers/game_details_scrape.py

@@ -53,6 +53,34 @@ def extract_teams_and_scores(box_score_section, sport):
 
     return team_names, period_scores
 
+def softball_summary(box_score_section):
+    summary = []
+    scoring_section = box_score_section.find(TAG_SECTION, {ATTR_ARIA_LABEL: LABEL_SCORING_SUMMARY})
+    if scoring_section:
+        scoring_rows = scoring_section.find(TAG_TBODY)
+        if scoring_rows:
+            for row in scoring_rows.find_all(TAG_TR):
+                team = row.find_all(TAG_TD)[0].find(TAG_IMG)[ATTR_ALT]
+                inning = row.find_all(TAG_TD)[3].text.strip()
+                desc_cell = row.find_all(TAG_TD)[4]
+                span = desc_cell.find(TAG_SPAN)
+                if span:
+                    span.extract()
+                desc = desc_cell.get_text(strip=True)
+                cornell_score = int(row.find_all(TAG_TD)[5].get_text(strip=True) or 0)
+                opp_score = int(row.find_all(TAG_TD)[6].get_text(strip=True) or 0)
+                summary.append({
+                        'team': team, 
+                        'inning': inning,
+                        'description': desc,
+                        'cor_score': cornell_score,
+                        'opp_score': opp_score
+                    })
+    if not summary:
+        summary = [{"message": "No scoring events in this game."}]
+    return summary 
+
+
 def soccer_summary(box_score_section):
     summary = []
     scoring_section = box_score_section.find(TAG_SECTION, {ATTR_ARIA_LABEL: LABEL_SCORING_SUMMARY})
@@ -124,14 +152,13 @@ def hockey_summary(box_score_section):
                 scorer = row.find_all(TAG_TD)[4].text.strip()
                 assist = row.find_all(TAG_TD)[5].text.strip()
 
-                if team == "COR" or team == "CU" or team == "Cornell":
+                if team == "COR" or team == "CU" or team == "Cornell" or team == "CORNELL":
                     cornell_score += 1
                 else:
                     opp_score += 1
 
                 summary.append({
                     'team': team,
-                    'period': period,
                     'time': time,
                     'scorer': scorer,
                     'assist': assist,
@@ -272,6 +299,7 @@ def scrape_game(url, sport):
         'field hockey': (lambda: extract_teams_and_scores(box_score_section, 'field hockey'), field_hockey_summary),
         'lacrosse': (lambda: extract_teams_and_scores(box_score_section, 'lacrosse'), lacrosse_summary),
         'baseball': (lambda: extract_teams_and_scores(box_score_section, 'baseball'), baseball_summary),
+        'softball': (lambda: extract_teams_and_scores(box_score_section, 'softball'), softball_summary),
         'basketball': (lambda: extract_teams_and_scores(box_score_section, 'basketball'), lambda _: []),
     }
 

src/scrapers/games_scraper.py

@@ -125,7 +125,8 @@ def parse_schedule_page(url, sport, gender):
 
         result_tag = game_item.select_one(RESULT_TAG)
         if result_tag:
-            game_data["result"] = result_tag.text.strip().replace("\n", "")
+            #game_data["result"] = result_tag.get_text(" ", strip=True)
+            game_data["result"] = result_tag.text.strip().replace("\n", " ")
         else:
             game_data["result"] = None
 
@@ -241,17 +242,16 @@ def process_game_data(game_data):
             if str(final_box_cor_score) != str(cor_final) or str(final_box_opp_score) != str(opp_final):
                 game_data["score_breakdown"] = game_data["score_breakdown"][::-1]
 
-    # Try to find by tournament key fields to handle placeholder teams
+    # Try to find an existing game record to update.
     curr_game = GameService.get_game_by_tournament_key_fields(
         city,
         game_data["date"],
         game_data["gender"],
         location,
         game_data["sport"],
-        state
+        state,
     )
-
-    # If no tournament game found, try the regular lookup with opponent_id
+
     if not curr_game:
         curr_game = GameService.get_game_by_key_fields(
             city,
@@ -260,14 +260,15 @@ def process_game_data(game_data):
             location,
             team.id,
             game_data["sport"],
-            state
+            state,
         )
 
     if isinstance(curr_game, list):
         if curr_game:
             curr_game = curr_game[0]
         else:
             curr_game = None
+
     if curr_game:
         updates = {
             "time": game_time,

src/types.py

@@ -54,6 +54,7 @@ class BoxScoreEntryType(ObjectType):
 
     team = String(required=False)
     period = String(required=False)
+    inning = String(required=False)
     time = String(required=False)
     description = String(required=False)
     scorer = String(required=False)