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/nwsl/espn_scoreboard.json
vendored
Normal file
245
sportstime_parser/tests/fixtures/nwsl/espn_scoreboard.json
vendored
Normal file
@@ -0,0 +1,245 @@
|
||||
{
|
||||
"leagues": [
|
||||
{
|
||||
"id": "761",
|
||||
"uid": "s:600~l:761",
|
||||
"name": "National Women's Soccer League",
|
||||
"abbreviation": "NWSL"
|
||||
}
|
||||
],
|
||||
"season": {
|
||||
"type": 2,
|
||||
"year": 2026
|
||||
},
|
||||
"day": {
|
||||
"date": "2026-04-10T00:00:00Z"
|
||||
},
|
||||
"events": [
|
||||
{
|
||||
"id": "401672201",
|
||||
"uid": "s:600~l:761~e:401672201",
|
||||
"date": "2026-04-10T23:00:00Z",
|
||||
"name": "Angel City FC at Portland Thorns",
|
||||
"shortName": "LA @ POR",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401672201",
|
||||
"uid": "s:600~l:761~e:401672201~c:401672201",
|
||||
"date": "2026-04-10T23:00:00Z",
|
||||
"attendance": 22000,
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "8070",
|
||||
"fullName": "Providence Park",
|
||||
"address": {
|
||||
"city": "Portland",
|
||||
"state": "OR"
|
||||
},
|
||||
"capacity": 25218,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "15625",
|
||||
"uid": "s:600~l:761~t:15625",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "15625",
|
||||
"uid": "s:600~l:761~t:15625",
|
||||
"location": "Portland",
|
||||
"name": "Thorns FC",
|
||||
"abbreviation": "POR",
|
||||
"displayName": "Portland Thorns FC"
|
||||
},
|
||||
"score": "2",
|
||||
"winner": true
|
||||
},
|
||||
{
|
||||
"id": "19934",
|
||||
"uid": "s:600~l:761~t:19934",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "19934",
|
||||
"uid": "s:600~l:761~t:19934",
|
||||
"location": "Los Angeles",
|
||||
"name": "Angel City",
|
||||
"abbreviation": "LA",
|
||||
"displayName": "Angel City FC"
|
||||
},
|
||||
"score": "1",
|
||||
"winner": false
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 90,
|
||||
"displayClock": "90'",
|
||||
"period": 2,
|
||||
"type": {
|
||||
"id": "3",
|
||||
"name": "STATUS_FINAL",
|
||||
"state": "post",
|
||||
"completed": true
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": "401672202",
|
||||
"uid": "s:600~l:761~e:401672202",
|
||||
"date": "2026-04-11T00:00:00Z",
|
||||
"name": "Orlando Pride at North Carolina Courage",
|
||||
"shortName": "ORL @ NC",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401672202",
|
||||
"uid": "s:600~l:761~e:401672202~c:401672202",
|
||||
"date": "2026-04-11T00:00:00Z",
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "8073",
|
||||
"fullName": "WakeMed Soccer Park",
|
||||
"address": {
|
||||
"city": "Cary",
|
||||
"state": "NC"
|
||||
},
|
||||
"capacity": 10000,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "15618",
|
||||
"uid": "s:600~l:761~t:15618",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "15618",
|
||||
"uid": "s:600~l:761~t:15618",
|
||||
"location": "North Carolina",
|
||||
"name": "Courage",
|
||||
"abbreviation": "NC",
|
||||
"displayName": "North Carolina Courage"
|
||||
},
|
||||
"score": "3",
|
||||
"winner": true
|
||||
},
|
||||
{
|
||||
"id": "15626",
|
||||
"uid": "s:600~l:761~t:15626",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "15626",
|
||||
"uid": "s:600~l:761~t:15626",
|
||||
"location": "Orlando",
|
||||
"name": "Pride",
|
||||
"abbreviation": "ORL",
|
||||
"displayName": "Orlando Pride"
|
||||
},
|
||||
"score": "1",
|
||||
"winner": false
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 90,
|
||||
"displayClock": "90'",
|
||||
"period": 2,
|
||||
"type": {
|
||||
"id": "3",
|
||||
"name": "STATUS_FINAL",
|
||||
"state": "post",
|
||||
"completed": true
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": "401672203",
|
||||
"uid": "s:600~l:761~e:401672203",
|
||||
"date": "2026-04-11T02:00:00Z",
|
||||
"name": "San Diego Wave at Bay FC",
|
||||
"shortName": "SD @ BAY",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401672203",
|
||||
"uid": "s:600~l:761~e:401672203~c:401672203",
|
||||
"date": "2026-04-11T02:00:00Z",
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "3945",
|
||||
"fullName": "PayPal Park",
|
||||
"address": {
|
||||
"city": "San Jose",
|
||||
"state": "CA"
|
||||
},
|
||||
"capacity": 18000,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "25645",
|
||||
"uid": "s:600~l:761~t:25645",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "25645",
|
||||
"uid": "s:600~l:761~t:25645",
|
||||
"location": "Bay Area",
|
||||
"name": "FC",
|
||||
"abbreviation": "BAY",
|
||||
"displayName": "Bay FC"
|
||||
},
|
||||
"score": null,
|
||||
"winner": null
|
||||
},
|
||||
{
|
||||
"id": "22638",
|
||||
"uid": "s:600~l:761~t:22638",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "22638",
|
||||
"uid": "s:600~l:761~t:22638",
|
||||
"location": "San Diego",
|
||||
"name": "Wave FC",
|
||||
"abbreviation": "SD",
|
||||
"displayName": "San Diego Wave FC"
|
||||
},
|
||||
"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