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/mls/espn_scoreboard.json
vendored
Normal file
245
sportstime_parser/tests/fixtures/mls/espn_scoreboard.json
vendored
Normal file
@@ -0,0 +1,245 @@
|
||||
{
|
||||
"leagues": [
|
||||
{
|
||||
"id": "19",
|
||||
"uid": "s:600~l:19",
|
||||
"name": "Major League Soccer",
|
||||
"abbreviation": "MLS"
|
||||
}
|
||||
],
|
||||
"season": {
|
||||
"type": 2,
|
||||
"year": 2026
|
||||
},
|
||||
"day": {
|
||||
"date": "2026-03-15T00:00:00Z"
|
||||
},
|
||||
"events": [
|
||||
{
|
||||
"id": "401672001",
|
||||
"uid": "s:600~l:19~e:401672001",
|
||||
"date": "2026-03-15T22:00:00Z",
|
||||
"name": "LA Galaxy at LAFC",
|
||||
"shortName": "LA @ LAFC",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401672001",
|
||||
"uid": "s:600~l:19~e:401672001~c:401672001",
|
||||
"date": "2026-03-15T22:00:00Z",
|
||||
"attendance": 22000,
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "8909",
|
||||
"fullName": "BMO Stadium",
|
||||
"address": {
|
||||
"city": "Los Angeles",
|
||||
"state": "CA"
|
||||
},
|
||||
"capacity": 22000,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "21295",
|
||||
"uid": "s:600~l:19~t:21295",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "21295",
|
||||
"uid": "s:600~l:19~t:21295",
|
||||
"location": "Los Angeles",
|
||||
"name": "FC",
|
||||
"abbreviation": "LAFC",
|
||||
"displayName": "Los Angeles FC"
|
||||
},
|
||||
"score": "3",
|
||||
"winner": true
|
||||
},
|
||||
{
|
||||
"id": "3610",
|
||||
"uid": "s:600~l:19~t:3610",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "3610",
|
||||
"uid": "s:600~l:19~t:3610",
|
||||
"location": "Los Angeles",
|
||||
"name": "Galaxy",
|
||||
"abbreviation": "LA",
|
||||
"displayName": "LA Galaxy"
|
||||
},
|
||||
"score": "2",
|
||||
"winner": false
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 90,
|
||||
"displayClock": "90'",
|
||||
"period": 2,
|
||||
"type": {
|
||||
"id": "3",
|
||||
"name": "STATUS_FINAL",
|
||||
"state": "post",
|
||||
"completed": true
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": "401672002",
|
||||
"uid": "s:600~l:19~e:401672002",
|
||||
"date": "2026-03-15T23:00:00Z",
|
||||
"name": "Seattle Sounders at Portland Timbers",
|
||||
"shortName": "SEA @ POR",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401672002",
|
||||
"uid": "s:600~l:19~e:401672002~c:401672002",
|
||||
"date": "2026-03-15T23:00:00Z",
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "8070",
|
||||
"fullName": "Providence Park",
|
||||
"address": {
|
||||
"city": "Portland",
|
||||
"state": "OR"
|
||||
},
|
||||
"capacity": 25218,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "5282",
|
||||
"uid": "s:600~l:19~t:5282",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "5282",
|
||||
"uid": "s:600~l:19~t:5282",
|
||||
"location": "Portland",
|
||||
"name": "Timbers",
|
||||
"abbreviation": "POR",
|
||||
"displayName": "Portland Timbers"
|
||||
},
|
||||
"score": "2",
|
||||
"winner": false
|
||||
},
|
||||
{
|
||||
"id": "4687",
|
||||
"uid": "s:600~l:19~t:4687",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "4687",
|
||||
"uid": "s:600~l:19~t:4687",
|
||||
"location": "Seattle",
|
||||
"name": "Sounders FC",
|
||||
"abbreviation": "SEA",
|
||||
"displayName": "Seattle Sounders FC"
|
||||
},
|
||||
"score": "2",
|
||||
"winner": false
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 90,
|
||||
"displayClock": "90'",
|
||||
"period": 2,
|
||||
"type": {
|
||||
"id": "3",
|
||||
"name": "STATUS_FINAL",
|
||||
"state": "post",
|
||||
"completed": true
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": "401672003",
|
||||
"uid": "s:600~l:19~e:401672003",
|
||||
"date": "2026-03-16T00:00:00Z",
|
||||
"name": "New York Red Bulls at Atlanta United",
|
||||
"shortName": "NY @ ATL",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401672003",
|
||||
"uid": "s:600~l:19~e:401672003~c:401672003",
|
||||
"date": "2026-03-16T00:00:00Z",
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "8904",
|
||||
"fullName": "Mercedes-Benz Stadium",
|
||||
"address": {
|
||||
"city": "Atlanta",
|
||||
"state": "GA"
|
||||
},
|
||||
"capacity": 42500,
|
||||
"indoor": true
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "18626",
|
||||
"uid": "s:600~l:19~t:18626",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "18626",
|
||||
"uid": "s:600~l:19~t:18626",
|
||||
"location": "Atlanta",
|
||||
"name": "United FC",
|
||||
"abbreviation": "ATL",
|
||||
"displayName": "Atlanta United FC"
|
||||
},
|
||||
"score": null,
|
||||
"winner": null
|
||||
},
|
||||
{
|
||||
"id": "399",
|
||||
"uid": "s:600~l:19~t:399",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "399",
|
||||
"uid": "s:600~l:19~t:399",
|
||||
"location": "New York",
|
||||
"name": "Red Bulls",
|
||||
"abbreviation": "NY",
|
||||
"displayName": "New York Red Bulls"
|
||||
},
|
||||
"score": null,
|
||||
"winner": null
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 0,
|
||||
"displayClock": "0'",
|
||||
"period": 0,
|
||||
"type": {
|
||||
"id": "1",
|
||||
"name": "STATUS_SCHEDULED",
|
||||
"state": "pre",
|
||||
"completed": false
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
}
|
||||
Reference in New Issue
Block a user