feat(scripts): add sportstime-parser data pipeline
Complete Python package for scraping, normalizing, and uploading sports schedule data to CloudKit. Includes: - Multi-source scrapers for NBA, MLB, NFL, NHL, MLS, WNBA, NWSL - Canonical ID system for teams, stadiums, and games - Fuzzy matching with manual alias support - CloudKit uploader with batch operations and deduplication - Comprehensive test suite with fixtures - WNBA abbreviation aliases for improved team resolution - Alias validation script to detect orphan references All 5 phases of data remediation plan completed: - Phase 1: Alias fixes (team/stadium alias additions) - Phase 2: NHL stadium coordinate fixes - Phase 3: Re-scrape validation - Phase 4: iOS bundle update - Phase 5: Code quality improvements (WNBA aliases) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Trey t
245
sportstime_parser/tests/fixtures/mlb/espn_scoreboard.json
vendored
Normal file
245
sportstime_parser/tests/fixtures/mlb/espn_scoreboard.json
vendored
Normal file
@@ -0,0 +1,245 @@
|
||||
{
|
||||
"leagues": [
|
||||
{
|
||||
"id": "10",
|
||||
"uid": "s:1~l:10",
|
||||
"name": "Major League Baseball",
|
||||
"abbreviation": "MLB"
|
||||
}
|
||||
],
|
||||
"season": {
|
||||
"type": 2,
|
||||
"year": 2026
|
||||
},
|
||||
"day": {
|
||||
"date": "2026-04-15T00:00:00Z"
|
||||
},
|
||||
"events": [
|
||||
{
|
||||
"id": "401584801",
|
||||
"uid": "s:1~l:10~e:401584801",
|
||||
"date": "2026-04-15T23:05:00Z",
|
||||
"name": "New York Yankees at Boston Red Sox",
|
||||
"shortName": "NYY @ BOS",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401584801",
|
||||
"uid": "s:1~l:10~e:401584801~c:401584801",
|
||||
"date": "2026-04-15T23:05:00Z",
|
||||
"attendance": 37435,
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "3",
|
||||
"fullName": "Fenway Park",
|
||||
"address": {
|
||||
"city": "Boston",
|
||||
"state": "MA"
|
||||
},
|
||||
"capacity": 37755,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "2",
|
||||
"uid": "s:1~l:10~t:2",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "2",
|
||||
"uid": "s:1~l:10~t:2",
|
||||
"location": "Boston",
|
||||
"name": "Red Sox",
|
||||
"abbreviation": "BOS",
|
||||
"displayName": "Boston Red Sox"
|
||||
},
|
||||
"score": "5",
|
||||
"winner": true
|
||||
},
|
||||
{
|
||||
"id": "10",
|
||||
"uid": "s:1~l:10~t:10",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "10",
|
||||
"uid": "s:1~l:10~t:10",
|
||||
"location": "New York",
|
||||
"name": "Yankees",
|
||||
"abbreviation": "NYY",
|
||||
"displayName": "New York Yankees"
|
||||
},
|
||||
"score": "3",
|
||||
"winner": false
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 0,
|
||||
"displayClock": "0:00",
|
||||
"period": 9,
|
||||
"type": {
|
||||
"id": "3",
|
||||
"name": "STATUS_FINAL",
|
||||
"state": "post",
|
||||
"completed": true
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": "401584802",
|
||||
"uid": "s:1~l:10~e:401584802",
|
||||
"date": "2026-04-15T20:10:00Z",
|
||||
"name": "Chicago Cubs at St. Louis Cardinals",
|
||||
"shortName": "CHC @ STL",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401584802",
|
||||
"uid": "s:1~l:10~e:401584802~c:401584802",
|
||||
"date": "2026-04-15T20:10:00Z",
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "87",
|
||||
"fullName": "Busch Stadium",
|
||||
"address": {
|
||||
"city": "St. Louis",
|
||||
"state": "MO"
|
||||
},
|
||||
"capacity": 45538,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "24",
|
||||
"uid": "s:1~l:10~t:24",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "24",
|
||||
"uid": "s:1~l:10~t:24",
|
||||
"location": "St. Louis",
|
||||
"name": "Cardinals",
|
||||
"abbreviation": "STL",
|
||||
"displayName": "St. Louis Cardinals"
|
||||
},
|
||||
"score": "7",
|
||||
"winner": true
|
||||
},
|
||||
{
|
||||
"id": "16",
|
||||
"uid": "s:1~l:10~t:16",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "16",
|
||||
"uid": "s:1~l:10~t:16",
|
||||
"location": "Chicago",
|
||||
"name": "Cubs",
|
||||
"abbreviation": "CHC",
|
||||
"displayName": "Chicago Cubs"
|
||||
},
|
||||
"score": "4",
|
||||
"winner": false
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 0,
|
||||
"displayClock": "0:00",
|
||||
"period": 9,
|
||||
"type": {
|
||||
"id": "3",
|
||||
"name": "STATUS_FINAL",
|
||||
"state": "post",
|
||||
"completed": true
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": "401584803",
|
||||
"uid": "s:1~l:10~e:401584803",
|
||||
"date": "2026-04-16T00:10:00Z",
|
||||
"name": "Los Angeles Dodgers at San Francisco Giants",
|
||||
"shortName": "LAD @ SF",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401584803",
|
||||
"uid": "s:1~l:10~e:401584803~c:401584803",
|
||||
"date": "2026-04-16T00:10:00Z",
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "116",
|
||||
"fullName": "Oracle Park",
|
||||
"address": {
|
||||
"city": "San Francisco",
|
||||
"state": "CA"
|
||||
},
|
||||
"capacity": 41915,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "26",
|
||||
"uid": "s:1~l:10~t:26",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "26",
|
||||
"uid": "s:1~l:10~t:26",
|
||||
"location": "San Francisco",
|
||||
"name": "Giants",
|
||||
"abbreviation": "SF",
|
||||
"displayName": "San Francisco Giants"
|
||||
},
|
||||
"score": null,
|
||||
"winner": null
|
||||
},
|
||||
{
|
||||
"id": "19",
|
||||
"uid": "s:1~l:10~t:19",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "19",
|
||||
"uid": "s:1~l:10~t:19",
|
||||
"location": "Los Angeles",
|
||||
"name": "Dodgers",
|
||||
"abbreviation": "LAD",
|
||||
"displayName": "Los Angeles Dodgers"
|
||||
},
|
||||
"score": null,
|
||||
"winner": null
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 0,
|
||||
"displayClock": "0:00",
|
||||
"period": 0,
|
||||
"type": {
|
||||
"id": "1",
|
||||
"name": "STATUS_SCHEDULED",
|
||||
"state": "pre",
|
||||
"completed": false
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
}
|
||||
Reference in New Issue
Block a user