scraping-rtn 0.0.8.0__tar.gz → 0.0.8.2__tar.gz

{scraping_rtn-0.0.8.0 → scraping_rtn-0.0.8.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scraping_rtn
-Version: 0.0.8.0
+Version: 0.0.8.2
 Summary: package to scrape gymnastics data from Road To Nationals
 Author-email: Claire Harmon <ceharmon220@gmail.com>
 License: MIT

{scraping_rtn-0.0.8.0 → scraping_rtn-0.0.8.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "scraping_rtn"
-version = "0.0.8.0"
+version = "0.0.8.2"
 authors = [
     { name = "Claire Harmon", email = "ceharmon220@gmail.com" },
 ]

{scraping_rtn-0.0.8.0 → scraping_rtn-0.0.8.2}/src/scraping_rtn/RtnSingleTeamYear.py

@@ -1,4 +1,4 @@
-from .src import EVENT_MAP, EVENTS, get_session, get_data_from_api, fix_opponents, normalize_date, merge_dicts, \
+from .src import EVENT_MAP, EVENTS, get_data_from_api, fix_opponents, normalize_date, merge_dicts, \
     get_extra_cols, SCHEDULE_COLS, RESULTS_COLS, IND_RESULTS_COLS, ROSTER_COLS
 import pandas as pd
 import numpy as np
@@ -15,16 +15,11 @@ class RtnSingleTeamYear(object):
             else:
                 self.team_id = team_id
 
-        if session is None:
-            self.session = get_session()
-        else:
-            self.session = session
-
     def get_team_mapping(self, force_update=False):
         if force_update:
             get_data_from_api.cache_clear()
 
-        all_teams_data = get_data_from_api(endpoint='gymnasts2', suffix=str(self.year) + '/1', session=self.session).json()
+        all_teams_data = get_data_from_api(endpoint='gymnasts2', suffix=str(self.year) + '/1').json()
         return {team['team_name']: team['id'] for team in all_teams_data['teams']}
 
     def get_team_id(self):
@@ -47,7 +42,7 @@ class RtnSingleTeamYear(object):
         if force_update:
             get_data_from_api.cache_clear()
 
-        roster_data = get_data_from_api(endpoint='rostermain', suffix=str(self.year)+'/'+str(self.team_id)+'/1', session=self.session).json()
+        roster_data = get_data_from_api(endpoint='rostermain', suffix=str(self.year)+'/'+str(self.team_id)+'/1').json()
 
         self._raw_roster = [{**{rename_map.get(k, k): v if k != 'school_year' else school_year_map.get(v, '') for k, v in data.items()},
                             **{'Name': data['fname'] + ' ' + data['lname'], 'Team': self.team_name}}
@@ -71,7 +66,7 @@ class RtnSingleTeamYear(object):
         if force_update:
             get_data_from_api.cache_clear()
 
-        meets = get_data_from_api(endpoint='dashboard', suffix=str(self.year)+'/'+str(self.team_id), session=self.session).json()
+        meets = get_data_from_api(endpoint='dashboard', suffix=str(self.year)+'/'+str(self.team_id)).json()
         name_map = {'team_id': 'Team ID', 'team_name': 'Team', 'meet_id': 'Team Meet ID',
                     'meet_date': 'Meet Date', 'team_score': 'Score', 'home': 'Home/Away',
                     'opponent': 'Opponents', 'meet_desc': 'Meet Name', 'linked_id': 'Meet ID'}
@@ -130,7 +125,7 @@ class RtnSingleTeamYear(object):
                 if force_update:
                     get_data_from_api.cache_clear()
 
-                meet_res = get_data_from_api(endpoint='meetresults', suffix=str(meet_id), session=self.session).json()
+                meet_res = get_data_from_api(endpoint='meetresults', suffix=str(meet_id)).json()
                 # This API call returns scores from all teams at this meet, not just this team. Need to pick out correct score
                 team_scores = [score for score in meet_res['teams'] if score['tname'] == self.team_name and score['mid'] == str(meet_id)]
                 assert len(team_scores) == 1, 'Multiple team scores??'
@@ -149,7 +144,7 @@ class RtnSingleTeamYear(object):
         if force_update:
             get_data_from_api.cache_clear()
 
-        res = get_data_from_api(endpoint='teamConsistency', suffix=f'{self.year}/{self.team_id}', session=self.session).json()
+        res = get_data_from_api(endpoint='teamConsistency', suffix=f'{self.year}/{self.team_id}').json()
         if len(res['labels']) == 0:
             print(f'No team consistency data found for {self.team_name} in {self.year}')
             for i in range(len(self._raw_season_results)):
@@ -199,7 +194,7 @@ class RtnSingleTeamYear(object):
                 if force_update:
                     get_data_from_api.cache_clear()
 
-                meet_res = get_data_from_api(endpoint='meetresults', suffix=str(meet_id), session=self.session).json()
+                meet_res = get_data_from_api(endpoint='meetresults', suffix=str(meet_id)).json()
                 if len(meet_res) == 0 or len(meet_res['scores']) == 0 or len(meet_res['scores'][0]) == 0:
                     print(f'No data found for meet {meet_id}')
                     continue
@@ -241,7 +236,7 @@ class RtnSingleTeamYear(object):
                 if force_update:
                     get_data_from_api.cache_clear()
 
-                res = get_data_from_api(endpoint='indConsistency', suffix=f"{self.year}/{gymnast['Gymnast ID']}", session=self.session).json()
+                res = get_data_from_api(endpoint='indConsistency', suffix=f"{self.year}/{gymnast['Gymnast ID']}").json()
                 ind_consistency = [{'Meet Date': normalize_date(res['labels'][i][:7] + str(self.year), dt_format='%b-%d-%Y'),
                                     'VT': round(float(res['vts'][i]), 4) if res['vts'][i] is not None else np.nan,
                                     'UB': round(float(res['ubs'][i]), 4) if res['ubs'][i] is not None else np.nan,
@@ -281,7 +276,7 @@ class RtnSingleTeamYear(object):
         if force_update:
             get_data_from_api.cache_clear()
 
-        nqsData = get_data_from_api(endpoint='rostermain', suffix=f'{self.year}/{self.team_id}/4', session=self.session).json()
+        nqsData = get_data_from_api(endpoint='rostermain', suffix=f'{self.year}/{self.team_id}/4').json()
         ind_nqs = [{name_map[k]: round(float(v), 4) if k != 'gid' and v != '' else (np.nan if k != 'gid' else v)
                     for k, v in data.items() if k in name_map.keys()} for data in nqsData['ind']]
 
@@ -297,7 +292,7 @@ class RtnSingleTeamYear(object):
             if force_update:
                 get_data_from_api.cache_clear()
 
-            week_data = get_data_from_api(endpoint='currentweek', suffix=str(self.year), session=self.session).json()
+            week_data = get_data_from_api(endpoint='currentweek', suffix=str(self.year)).json()
             return min(int(week_data['week']), int(week_data['max']))
 
     def _get_raw_rankings(self, team_vs_ind, event, week, force_update=False):
@@ -310,7 +305,7 @@ class RtnSingleTeamYear(object):
         if force_update:
             get_data_from_api.cache_clear()
 
-        res = get_data_from_api(endpoint='results', suffix=f'{self.year}/{week}/{team_ind_map[team_vs_ind]}/{event_api_map[event]}', session=self.session).json()
+        res = get_data_from_api(endpoint='results', suffix=f'{self.year}/{week}/{team_ind_map[team_vs_ind]}/{event_api_map[event]}').json()
         if team_vs_ind == 'ind':
             self._raw_rankings[team_vs_ind][event] = [{**{rename_map.get(k): float(v) if k in ['rqs', 'ave', 'high'] else v for k, v in data.items() if k in rename_map},
                                                        **{'Name': data['fname'] + ' ' + data['lname'], 'Event': event}}

{scraping_rtn-0.0.8.0 → scraping_rtn-0.0.8.2}/src/scraping_rtn/__init__.py

@@ -1,5 +1,5 @@
 from .RtnSingleTeamYear import RtnSingleTeamYear
-from .src import get_session, validate_input, get_extra_cols, SCHEDULE_COLS, RESULTS_COLS, IND_RESULTS_COLS, EVENTS, ROSTER_COLS
+from .src import validate_input, get_extra_cols, SCHEDULE_COLS, RESULTS_COLS, IND_RESULTS_COLS, EVENTS, ROSTER_COLS
 import pandas as pd
 
 BLANK_SPACES = ' '*30
@@ -16,13 +16,12 @@ def all_teams(year, force_update=False):
 
 def roster(year, teams, include_hometowns=False, include_class=False, include_events=False, verbose=False, force_update=False):
     teams = validate_input(teams)
-    session = get_session()
 
     all_rosters = []
     for i, team in enumerate(teams):
         if verbose:
             print(f'Getting roster for {team}{BLANK_SPACES}', end='\r' if team != teams[-1] else None)
-        rtn = RtnSingleTeamYear(year=year, team_name=team, session=session)
+        rtn = RtnSingleTeamYear(year=year, team_name=team)
         res = rtn.get_roster(include_hometowns=include_hometowns, include_class=include_class,
                              include_events=include_events, force_update=force_update if i == 0 else False)
         if verbose and len(res) == 0:
@@ -35,13 +34,12 @@ def roster(year, teams, include_hometowns=False, include_class=False, include_ev
 
 def schedule(year, teams, verbose=False, force_update=False):
     teams = validate_input(teams)
-    session = get_session()
 
     all_schedules = []
     for i, team in enumerate(teams):
         if verbose:
             print(f'Getting schedule for {team}{BLANK_SPACES}', end='\r' if team != teams[-1] else None)
-        rtn = RtnSingleTeamYear(year=year, team_name=team, session=session)
+        rtn = RtnSingleTeamYear(year=year, team_name=team)
         res = rtn.get_schedule(force_update=force_update if i == 0 else False)
         if verbose and len(res) == 0:
             print(f'\tNo schedule found for {team}')
@@ -52,13 +50,12 @@ def schedule(year, teams, verbose=False, force_update=False):
 
 def team_results(year, teams, method='team_consistency', force_update=False, verbose=False):
     teams = validate_input(teams)
-    session = get_session()
     
     all_results = []
     for i, team in enumerate(teams):
         if verbose:
             print(f'Getting schedule and results for {team}{BLANK_SPACES}', end='\r' if team != teams[-1] else None)
-        rtn = RtnSingleTeamYear(year=year, team_name=team, session=session)
+        rtn = RtnSingleTeamYear(year=year, team_name=team)
         res = rtn.get_team_scores(method=method, force_update=force_update if i == 0 else False)
         if verbose and len(res) == 0:
             print(f'\tNo schedule and results found for {team}')
@@ -69,13 +66,12 @@ def team_results(year, teams, method='team_consistency', force_update=False, ver
 
 def individual_results(year, teams, method='by_meet', force_update=False, verbose=False):
     teams = validate_input(teams)
-    session = get_session()
     
     all_scores = []
     for i, team in enumerate(teams):
         if verbose:
             print(f'Getting scores for {team}{BLANK_SPACES}', end='\r' if team != teams[-1] else None)
-        rtn = RtnSingleTeamYear(year=year, team_name=team, session=session)
+        rtn = RtnSingleTeamYear(year=year, team_name=team)
         res = rtn.get_individual_scores(method=method, force_update=force_update if i == 0 else False)
         if verbose and len(res) == 0:
             print(f'\tNo scores found for {team}')
@@ -86,13 +82,12 @@ def individual_results(year, teams, method='by_meet', force_update=False, verbos
 
 def individual_nqs(year, teams, verbose=False, force_update=False):
     teams = validate_input(teams)
-    session = get_session()
     
     all_nqs = []
     for i, team in enumerate(teams):
         if verbose:
             print(f'Getting individual NQS for {team}{BLANK_SPACES}', end='\r' if team != teams[-1] else None)
-        rtn = RtnSingleTeamYear(year=year, team_name=team, session=session)
+        rtn = RtnSingleTeamYear(year=year, team_name=team)
         res = rtn.get_individual_nqs(force_update=force_update if i == 0 else False)
         if verbose and len(res) == 0:
             print(f'\tNo individual NQS found for {team}')
@@ -102,6 +97,5 @@ def individual_nqs(year, teams, verbose=False, force_update=False):
 
 
 def rankings(year, team_vs_ind='team', event='AA', week=None, force_update=False):
-    session = get_session()
-    rtn = RtnSingleTeamYear(year=year, team_name=None, session=session)
+    rtn = RtnSingleTeamYear(year=year, team_name=None)
     return rtn.get_overall_rankings(team_vs_ind=team_vs_ind, event=event, week=week, force_update=force_update)

{scraping_rtn-0.0.8.0 → scraping_rtn-0.0.8.2}/src/scraping_rtn/src.py

@@ -21,19 +21,13 @@ def validate_input(teams):
     return teams
 
 
-def get_session():
+@lru_cache(maxsize=1000000)
+def get_data_from_api(endpoint, suffix):
     session = requests.Session()
     retry = Retry(connect=3, backoff_factor=0.5)
     adapter = HTTPAdapter(max_retries=retry)
     session.mount('http://', adapter)
     session.mount('https://', adapter)
-    return session
-
-
-@lru_cache(maxsize=1000000)
-def get_data_from_api(endpoint, suffix, session=None):
-    if not session:    
-        session = get_session()
 
     url = 'https://www.roadtonationals.com/api/women/' + endpoint
     if suffix:

{scraping_rtn-0.0.8.0 → scraping_rtn-0.0.8.2}/src/scraping_rtn.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scraping_rtn
-Version: 0.0.8.0
+Version: 0.0.8.2
 Summary: package to scrape gymnastics data from Road To Nationals
 Author-email: Claire Harmon <ceharmon220@gmail.com>
 License: MIT