Filter pills by backend tags (slug membership), drop keyword matching
Replace the keyword/source-majority Interest system with a StoryPill enum that matches stories by canonical category slugs. StorySummary gains a tolerant tags[] (optional, decode-safe); effectiveTags uses tags, falling back to topic during the backend transition. No more headline keyword-searching on the client — the backend decides classification; the client asks slug membership. Pills: All, F1, Sport, AI, Cloud, HomeLab, Tech, Uganda, South Africa, Canada, US. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
Robin Kutesa
@@ -172,115 +172,56 @@ extension StorySummary {
|
||||
}
|
||||
}
|
||||
|
||||
// MARK: - Pinned interest filters (client-side)
|
||||
// MARK: - Story pills (backend-tag driven)
|
||||
//
|
||||
// The backend only classifies topics as finance/tech/politics/africa, so these
|
||||
// personal filters match keywords in the headline, summary, and source names of
|
||||
// the loaded stories instead of calling the API.
|
||||
// The backend classifies each story into canonical category slugs in `tags[]`.
|
||||
// The client filters by pill -> slug membership only; no headline keyword
|
||||
// matching. `topic` is a fallback while the backend finishes emitting `tags[]`.
|
||||
|
||||
struct Interest: Identifiable, Hashable {
|
||||
let id: String
|
||||
let label: String
|
||||
var keywords: [String] = []
|
||||
/// If any of these appear, the story is NOT this interest — used to keep the
|
||||
/// region pills exclusive (East Africa excludes Uganda & South Africa).
|
||||
var excludeKeywords: [String] = []
|
||||
/// When set, matches the backend topic classification instead of keywords
|
||||
/// (e.g. finance, tech).
|
||||
var topicSlug: String? = nil
|
||||
func matches(_ s: StorySummary) -> Bool {
|
||||
if id == "f1" { return Interest.isFormula1(s) }
|
||||
if id == "sport" { return Interest.isSports(s) }
|
||||
if let topicSlug {
|
||||
return s.topic.caseInsensitiveCompare(topicSlug) == .orderedSame
|
||||
enum StoryPill: String, CaseIterable, Identifiable {
|
||||
case all = "All"
|
||||
case f1 = "F1"
|
||||
case sport = "Sport"
|
||||
case ai = "AI"
|
||||
case cloud = "Cloud"
|
||||
case homelab = "HomeLab"
|
||||
case tech = "Tech"
|
||||
case uganda = "Uganda"
|
||||
case southAfrica = "South Africa"
|
||||
case canada = "Canada"
|
||||
case unitedStates = "US"
|
||||
|
||||
var id: String { rawValue }
|
||||
var label: String { rawValue }
|
||||
|
||||
var slugs: Set<String> {
|
||||
switch self {
|
||||
case .all: return []
|
||||
case .f1: return ["formula-1"]
|
||||
case .sport: return ["sports", "esports"]
|
||||
case .ai: return ["artificial-intelligence", "machine-learning", "robotics"]
|
||||
case .cloud: return ["cloud-computing"]
|
||||
case .homelab: return ["homelab"]
|
||||
case .tech: return ["technology", "programming-and-software-development",
|
||||
"cybersecurity", "security", "privacy-and-data-protection",
|
||||
"web-design-and-ui-ux", "wordpress-and-web-development"]
|
||||
case .uganda: return ["uganda"]
|
||||
case .southAfrica: return ["south-africa"]
|
||||
case .canada: return ["canada"]
|
||||
case .unitedStates: return ["united-states"]
|
||||
}
|
||||
// Headline only — the backend's summaries/outlet names are polluted by
|
||||
// over-clustering (a Congo headline can carry a South-Africa summary or an
|
||||
// SA outlet), so they're unsafe to match on.
|
||||
let hay = s.headline.lowercased()
|
||||
if excludeKeywords.contains(where: { hay.contains($0) }) { return false }
|
||||
return keywords.contains { hay.contains($0) }
|
||||
}
|
||||
|
||||
/// F1 = a clear F1 headline, or a majority of F1-feed sources. A single F1
|
||||
/// source isn't enough — the over-clustered backend sometimes attaches one to
|
||||
/// an unrelated story. Avoids substring traps like "fia" matching "deFIAnt".
|
||||
static func isFormula1(_ s: StorySummary) -> Bool {
|
||||
let h = s.headline.lowercased()
|
||||
if h.contains("formula 1") || h.contains("formula one") || h.contains("grand prix")
|
||||
|| h.hasPrefix("f1 ") || h.hasSuffix(" f1") || h.contains(" f1 ") || h.contains("f1:") {
|
||||
return true
|
||||
}
|
||||
let feeds = ["f1", "racefans", "motorsport", "autosport", "planetf1", "the race", "formula 1"]
|
||||
let srcs = s.sources
|
||||
guard !srcs.isEmpty else { return false }
|
||||
let n = srcs.filter { src in feeds.contains { src.name.lowercased().contains($0) } }.count
|
||||
return Double(n) / Double(srcs.count) >= 0.5
|
||||
}
|
||||
|
||||
/// Sport (non-F1) = a clear sports headline, or a majority of sports-outlet
|
||||
/// sources. The backend mis-tags World Cup / soccer into finance & tech, so
|
||||
/// this pulls them out. F1 has its own pill and is excluded here.
|
||||
static func isSports(_ s: StorySummary) -> Bool {
|
||||
if isFormula1(s) { return false }
|
||||
let h = s.headline.lowercased()
|
||||
let terms = ["world cup", "fifa", "premier league", "la liga", "champions league",
|
||||
"uefa", "ballon d'or", "knockout stage", "quarterfinal", "semifinal",
|
||||
"nba ", "nfl ", "nhl ", "mlb ", "super bowl", "playoff", "wimbledon",
|
||||
"grand slam", "olympic", "test match", "six nations", "messi", "ronaldo",
|
||||
"striker", "midfielder", "goalkeeper", "world cup match"]
|
||||
if terms.contains(where: { h.contains($0) }) { return true }
|
||||
let feeds = ["the athletic", "fox sports", "sportsnet", "espn", "sky sports",
|
||||
"bbc sport", "goal", "bleacher report", "sports illustrated", "sporting news"]
|
||||
let srcs = s.sources
|
||||
guard !srcs.isEmpty else { return false }
|
||||
let n = srcs.filter { src in feeds.contains { src.name.lowercased().contains($0) } }.count
|
||||
return Double(n) / Double(srcs.count) >= 0.5
|
||||
}
|
||||
|
||||
// Region keywords are place/person names found in the story text — never
|
||||
// outlet names (which are unreliable due to backend over-clustering).
|
||||
private static let ugandaKeywords = ["uganda", "ugandan", "kampala", "museveni",
|
||||
"entebbe", "jinja", "gulu"]
|
||||
private static let southAfricaKeywords = ["south africa", "south african", "johannesburg",
|
||||
"cape town", "pretoria", "durban", "ramaphosa",
|
||||
"eskom", "gauteng", "western cape", "soweto",
|
||||
"drakensberg", "mbalula", "zuma"]
|
||||
private static let canadaKeywords = ["canada", "canadian", "toronto", "ottawa", "vancouver",
|
||||
"montreal", "quebec", "alberta", "ontario", "manitoba",
|
||||
"calgary", "winnipeg", "trudeau", "carney"]
|
||||
private static let usKeywords = ["united states", "u.s.", "americans", "trump", "white house",
|
||||
"washington", "congress", "pentagon", "biden", "kamala",
|
||||
"harris", "wall street", "new york", "california", "texas",
|
||||
"florida", "supreme court", "senate", "capitol"]
|
||||
// Content topics — strong, unambiguous terms (no bare "ai"/"aws" substring traps).
|
||||
private static let aiKeywords = ["artificial intelligence", "machine learning", "deep learning",
|
||||
"generative ai", "chatgpt", "openai", "anthropic", "claude",
|
||||
"llm", "gpt", "gemini", "hugging face", "nvidia", "copilot",
|
||||
"agentic", "midjourney", "a.i."]
|
||||
private static let cloudKeywords = ["cloud", "azure", "google cloud", "kubernetes", "docker",
|
||||
"serverless", "cloudflare", "data center", "datacenter",
|
||||
"devops", "openstack", "terraform"]
|
||||
private static let homelabKeywords = ["homelab", "home lab", "self-host", "self-hosted",
|
||||
"selfhosted", "proxmox", "raspberry pi", "truenas",
|
||||
"unraid", "synology", "jellyfin", "home server", "nas ",
|
||||
"home assistant", "docker compose", "tailscale", "k3s",
|
||||
"mini pc"]
|
||||
|
||||
static let f1 = Interest(id: "f1", label: "F1") // matching via isFormula1
|
||||
static let sport = Interest(id: "sport", label: "Sport") // matching via isSports
|
||||
|
||||
/// Pinned filter pills, in order.
|
||||
static let pinned: [Interest] = [
|
||||
f1,
|
||||
sport,
|
||||
Interest(id: "uganda", label: "Uganda", keywords: ugandaKeywords),
|
||||
Interest(id: "south-africa", label: "South Africa", keywords: southAfricaKeywords),
|
||||
Interest(id: "canada", label: "Canada", keywords: canadaKeywords),
|
||||
Interest(id: "us", label: "US", keywords: usKeywords),
|
||||
Interest(id: "ai", label: "AI", keywords: aiKeywords),
|
||||
Interest(id: "cloud", label: "Cloud", keywords: cloudKeywords),
|
||||
Interest(id: "homelab", label: "HomeLab", keywords: homelabKeywords),
|
||||
Interest(id: "tech", label: "Tech", topicSlug: "tech"),
|
||||
]
|
||||
}
|
||||
|
||||
extension StorySummary {
|
||||
/// Backend `tags[]` are the source of truth; fall back to `topic` during the
|
||||
/// transition while the backend finishes emitting tags.
|
||||
var effectiveTags: Set<String> {
|
||||
let clean = Set((tags ?? []).filter { !$0.isEmpty })
|
||||
return clean.isEmpty ? [topic] : clean
|
||||
}
|
||||
|
||||
func matches(_ pill: StoryPill) -> Bool {
|
||||
pill == .all || !effectiveTags.isDisjoint(with: pill.slugs)
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user