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/nfl/espn_scoreboard.json
vendored
Normal file
245
sportstime_parser/tests/fixtures/nfl/espn_scoreboard.json
vendored
Normal file
@@ -0,0 +1,245 @@
|
||||
{
|
||||
"leagues": [
|
||||
{
|
||||
"id": "28",
|
||||
"uid": "s:20~l:28",
|
||||
"name": "National Football League",
|
||||
"abbreviation": "NFL"
|
||||
}
|
||||
],
|
||||
"season": {
|
||||
"type": 2,
|
||||
"year": 2025
|
||||
},
|
||||
"week": {
|
||||
"number": 1
|
||||
},
|
||||
"events": [
|
||||
{
|
||||
"id": "401671801",
|
||||
"uid": "s:20~l:28~e:401671801",
|
||||
"date": "2025-09-07T20:00:00Z",
|
||||
"name": "Kansas City Chiefs at Baltimore Ravens",
|
||||
"shortName": "KC @ BAL",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401671801",
|
||||
"uid": "s:20~l:28~e:401671801~c:401671801",
|
||||
"date": "2025-09-07T20:00:00Z",
|
||||
"attendance": 71547,
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "3814",
|
||||
"fullName": "M&T Bank Stadium",
|
||||
"address": {
|
||||
"city": "Baltimore",
|
||||
"state": "MD"
|
||||
},
|
||||
"capacity": 71008,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "33",
|
||||
"uid": "s:20~l:28~t:33",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "33",
|
||||
"uid": "s:20~l:28~t:33",
|
||||
"location": "Baltimore",
|
||||
"name": "Ravens",
|
||||
"abbreviation": "BAL",
|
||||
"displayName": "Baltimore Ravens"
|
||||
},
|
||||
"score": "20",
|
||||
"winner": false
|
||||
},
|
||||
{
|
||||
"id": "12",
|
||||
"uid": "s:20~l:28~t:12",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "12",
|
||||
"uid": "s:20~l:28~t:12",
|
||||
"location": "Kansas City",
|
||||
"name": "Chiefs",
|
||||
"abbreviation": "KC",
|
||||
"displayName": "Kansas City Chiefs"
|
||||
},
|
||||
"score": "27",
|
||||
"winner": true
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 0,
|
||||
"displayClock": "0:00",
|
||||
"period": 4,
|
||||
"type": {
|
||||
"id": "3",
|
||||
"name": "STATUS_FINAL",
|
||||
"state": "post",
|
||||
"completed": true
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": "401671802",
|
||||
"uid": "s:20~l:28~e:401671802",
|
||||
"date": "2025-09-08T17:00:00Z",
|
||||
"name": "Philadelphia Eagles at Green Bay Packers",
|
||||
"shortName": "PHI @ GB",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401671802",
|
||||
"uid": "s:20~l:28~e:401671802~c:401671802",
|
||||
"date": "2025-09-08T17:00:00Z",
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "3798",
|
||||
"fullName": "Lambeau Field",
|
||||
"address": {
|
||||
"city": "Green Bay",
|
||||
"state": "WI"
|
||||
},
|
||||
"capacity": 81441,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "9",
|
||||
"uid": "s:20~l:28~t:9",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "9",
|
||||
"uid": "s:20~l:28~t:9",
|
||||
"location": "Green Bay",
|
||||
"name": "Packers",
|
||||
"abbreviation": "GB",
|
||||
"displayName": "Green Bay Packers"
|
||||
},
|
||||
"score": "34",
|
||||
"winner": true
|
||||
},
|
||||
{
|
||||
"id": "21",
|
||||
"uid": "s:20~l:28~t:21",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "21",
|
||||
"uid": "s:20~l:28~t:21",
|
||||
"location": "Philadelphia",
|
||||
"name": "Eagles",
|
||||
"abbreviation": "PHI",
|
||||
"displayName": "Philadelphia Eagles"
|
||||
},
|
||||
"score": "29",
|
||||
"winner": false
|
||||
}
|
||||
],
|
||||
"status": {
|
||||
"clock": 0,
|
||||
"displayClock": "0:00",
|
||||
"period": 4,
|
||||
"type": {
|
||||
"id": "3",
|
||||
"name": "STATUS_FINAL",
|
||||
"state": "post",
|
||||
"completed": true
|
||||
}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": "401671803",
|
||||
"uid": "s:20~l:28~e:401671803",
|
||||
"date": "2025-09-08T20:25:00Z",
|
||||
"name": "Dallas Cowboys at Cleveland Browns",
|
||||
"shortName": "DAL @ CLE",
|
||||
"competitions": [
|
||||
{
|
||||
"id": "401671803",
|
||||
"uid": "s:20~l:28~e:401671803~c:401671803",
|
||||
"date": "2025-09-08T20:25:00Z",
|
||||
"type": {
|
||||
"id": "1",
|
||||
"abbreviation": "STD"
|
||||
},
|
||||
"venue": {
|
||||
"id": "3653",
|
||||
"fullName": "Cleveland Browns Stadium",
|
||||
"address": {
|
||||
"city": "Cleveland",
|
||||
"state": "OH"
|
||||
},
|
||||
"capacity": 67431,
|
||||
"indoor": false
|
||||
},
|
||||
"competitors": [
|
||||
{
|
||||
"id": "5",
|
||||
"uid": "s:20~l:28~t:5",
|
||||
"type": "team",
|
||||
"order": 0,
|
||||
"homeAway": "home",
|
||||
"team": {
|
||||
"id": "5",
|
||||
"uid": "s:20~l:28~t:5",
|
||||
"location": "Cleveland",
|
||||
"name": "Browns",
|
||||
"abbreviation": "CLE",
|
||||
"displayName": "Cleveland Browns"
|
||||
},
|
||||
"score": null,
|
||||
"winner": null
|
||||
},
|
||||
{
|
||||
"id": "6",
|
||||
"uid": "s:20~l:28~t:6",
|
||||
"type": "team",
|
||||
"order": 1,
|
||||
"homeAway": "away",
|
||||
"team": {
|
||||
"id": "6",
|
||||
"uid": "s:20~l:28~t:6",
|
||||
"location": "Dallas",
|
||||
"name": "Cowboys",
|
||||
"abbreviation": "DAL",
|
||||
"displayName": "Dallas Cowboys"
|
||||
},
|
||||
"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