prisma/schema.prisma

// This is your Prisma schema file,
// learn more about it in the docs: https://pris.ly/d/prisma-schema

generator client {
  provider = "prisma-client-js"
}

// Python generator disabled — prisma-client-py is optional and not installed in this env.
// generator python {
//   provider = "prisma-client-py"
// }

datasource db {
  provider = "sqlite" // Start with SQLite for MVP, easy migration to Postgres later
  url      = env("DATABASE_URL")
}

// Core Deals table - matches your provided table structure + temporal metadata
model Deal {
  id     String  @id @default(uuid()) // UUID for new deals, custom IDs allowed in seed
  date   String? // "2024-05-22" or "2024" or "2021-2024"
  period String? // For ranges like "2021-2024"

  // Temporal metadata
  announcementDate      DateTime? @map("announcement_date")
  effectiveStartDate    DateTime? @map("effective_start_date")
  effectiveEndDate      DateTime? @map("effective_end_date")
  discoveryDate         DateTime? @map("discovery_date")
  lastVerified          DateTime? @map("last_verified")
  sourcePublicationDate DateTime? @map("source_publication_date")

  modality            String // "Text", "Image", "Audio", "Video", "Satellite", etc.
  provider            String
  buyer               String // Can be comma-separated for multiple buyers
  dataType            String   @map("data_type") // Short description
  reportedTerms       String?  @map("reported_terms") // Headline terms
  creatorsCompensated Boolean? @map("creators_compensated") // Yes/No/Unclear
  exclusive           Boolean? // Yes/No/N/A
  pricingMechanism    String   @map("pricing_mechanism")

  // Enhanced content metadata
  contentType             String?  @map("content_type") // text, image, audio, etc.
  volumeDescription       String?  @map("volume_description")
  updateFrequency         String?  @map("update_frequency") // streaming, batch, one-time
  historicalArchiveAccess Boolean? @default(false) @map("historical_archive_access")

  // Pricing structure (JSON)
  pricingStructure  String? @map("pricing_structure") // JSON: {headline_price, currency, billing_interval, estimated_unit_cost}
  billingInterval   String? @map("billing_interval") // annual, monthly, one-time
  estimatedUnitCost String? @map("estimated_unit_cost") // per 1000 API calls, etc.

  // Linkages metadata (JSON)
  linkagesMetadata String? @map("linkages_metadata") // JSON: {likely_used_by_models, impact_on_training_data, modality_integration, feed_type}

  // Extended fields from your schema plan
  dealType         String?   @map("deal_type") // aggregate, per-unit, commissioning, etc.
  priceUsd         Float?    @map("price_usd")
  priceRangeMinUsd Float?    @map("price_range_min_usd")
  priceRangeMaxUsd Float?    @map("price_range_max_usd")
  priceCurrency    String?   @default("USD") @map("price_currency")
  durationYears    Float?    @map("duration_years")
  startDate        DateTime? @map("start_date")
  endDate          DateTime? @map("end_date")

  // Rights & restrictions
  trainingAllowed       Boolean? @map("training_allowed")
  finetuningAllowed     Boolean? @map("finetuning_allowed")
  inferenceAllowed      Boolean? @map("inference_allowed")
  redistributionAllowed Boolean? @map("redistribution_allowed")
  deletionRequired      Boolean? @map("deletion_required")

  // Compensation details
  creatorSplitPercentage Float?   @map("creator_split_percentage")
  revenueShare           Boolean? @map("revenue_share")

  // Provenance & Discovery
  sources        String // JSON array of URLs
  sourcePrimary  String?   @map("source_primary") // "Reuters", "SEC filing", etc.
  discoveredVia  String?   @map("discovered_via") // exa, manual, filing, etc.
  exaQuery       String?   @map("exa_query")
  exaScore       Float?    @map("exa_score")
  exaRetrievedAt DateTime? @map("exa_retrieved_at")

  // Extraction metadata
  extractionMetadata String?   @map("extraction_metadata") // JSON
  rawTextSnippets    String?   @map("raw_text_snippets") // JSON array
  regexConfidence    String?   @map("regex_confidence") // high, medium, low
  llmConfidence      String?   @map("llm_confidence") // high, medium, low
  lastExtracted      DateTime? @map("last_extracted")

  notes           String?
  dealStage       String? @default("confirmed") @map("deal_stage") // announced, rumored, confirmed, settled
  confidenceScore Float?  @default(1.0) @map("confidence_score") // 0-1

  // Linkages to models
  modelLinkages DealModelLinkage[]

  // Relations
  providerRelation Provider? @relation(fields: [providerId], references: [id])
  providerId       String?   @map("provider_id")

  buyerRelations DealBuyer[]

  // Pricing normalization
  pricingNormalizations PricingNormalization[]

  // Versioning
  version String? @default("1.0")

  // Timestamps
  createdAt DateTime @default(now()) @map("created_at")
  updatedAt DateTime @updatedAt @map("updated_at")

  @@index([provider])
  @@index([buyer])
  @@index([modality])
  @@index([date])
  @@index([dealStage])
  @@index([discoveryDate])
  @@index([announcementDate])
  @@map("deals")
}

// Model Registry table
model ModelRegistry {
  id          String    @id @default(uuid())
  modelId     String    @unique @map("model_id")
  family      String?
  provider    String
  releaseDate DateTime? @map("release_date")
  lastUpdated DateTime? @map("last_updated")

  // Parameters
  params       Float? // in billions
  paramsActive Float? @map("params_active") // for MoE

  // Architecture
  architectureType String?  @map("architecture_type")
  isMoe            Boolean? @default(false) @map("is_moe")
  numExperts       Int?     @map("num_experts")
  activeExperts    Int?     @map("active_experts")
  longContext      Int?     @map("long_context")
  multimodal       Boolean? @default(false)

  // Training compute
  flopsReported  Float?  @map("flops_reported")
  flopsEstimated Float?  @map("flops_estimated")
  computeSources String? @map("compute_sources") // JSON array

  // Token estimates
  tokensEstMin           Float?    @map("tokens_est_min")
  tokensEstMax           Float?    @map("tokens_est_max")
  tokensEstMid           Float?    @map("tokens_est_mid")
  tokensRangeGeneratedAt DateTime? @map("tokens_range_generated_at")
  openDataTokensReported Float?    @map("open_data_tokens_reported")

  // Estimation metadata
  estimationMethod     String?   @map("estimation_method") // JSON array of methods used
  estimationConfidence Float?    @map("estimation_confidence") // 0-1 confidence score
  estimationDate       DateTime? @map("estimation_date")
  estimationVersion    String?   @default("1.0") @map("estimation_version")

  // Evidence profile
  evidenceTypes              String?   @map("evidence_types") // JSON array
  evidenceStrength           String?   @map("evidence_strength") // S-High, S-Medium, S-Low
  uncertaintySources         String?   @map("uncertainty_sources") // JSON array
  evidenceProfileGeneratedAt DateTime? @map("evidence_profile_generated_at")

  // Data composition
  compositionEstimates String? @map("composition_estimates") // JSON

  // Sources
  sources             String? // JSON array
  rawEvidenceSnippets String? @map("raw_evidence_snippets") // JSON array

  // Temporal
  trainingPeriodStart  DateTime? @map("training_period_start")
  trainingPeriodEnd    DateTime? @map("training_period_end")
  inferenceGeneratedAt DateTime? @map("inference_generated_at")

  // Versioning
  version String? @default("1.0")

  // Linkages
  modelLinkages DealModelLinkage[]

  // Timestamps
  createdAt DateTime @default(now()) @map("created_at")
  updatedAt DateTime @updatedAt @map("updated_at")

  @@index([modelId])
  @@index([provider])
  @@index([releaseDate])
  @@map("model_registry")
}

// Linkage between deals and models
model DealModelLinkage {
  id                String   @id @default(uuid())
  dealId            String   @map("deal_id")
  modelId           String   @map("model_id")
  linkageType       String   @map("linkage_type") // temporal_overlap, inferred, explicit
  linkageStrength   String   @map("linkage_strength") // high, medium, low
  impactInference   String?  @map("impact_inference")
  analysisTimestamp DateTime @default(now()) @map("analysis_timestamp")

  deal  Deal          @relation(fields: [dealId], references: [id], onDelete: Cascade)
  model ModelRegistry @relation(fields: [modelId], references: [id], onDelete: Cascade)

  @@unique([dealId, modelId])
  @@index([dealId])
  @@index([modelId])
  @@map("deal_model_linkages")
}

// Provider table (normalized)
model Provider {
  id                 String  @id @default(uuid())
  name               String  @unique
  industry           String?
  country            String?
  annualRevenue      Float?  @map("annual_revenue")
  contentType        String? @map("content_type")
  ownershipStructure String? @map("ownership_structure") // publisher, creator-owned, platform, commons

  deals Deal[]

  createdAt DateTime @default(now()) @map("created_at")
  updatedAt DateTime @updatedAt @map("updated_at")

  @@map("providers")
}

// Buyer table (normalized)
model Buyer {
  id             String  @id @default(uuid())
  name           String  @unique
  orgType        String? @map("org_type") // lab, cloud, aggregator, startup, research
  modelsAffected String? @map("models_affected") // JSON array

  deals DealBuyer[]

  createdAt DateTime @default(now()) @map("created_at")
  updatedAt DateTime @updatedAt @map("updated_at")

  @@map("buyers")
}

// Many-to-many relationship (deals can have multiple buyers)
model DealBuyer {
  id      String @id @default(uuid())
  dealId  String @map("deal_id")
  buyerId String @map("buyer_id")

  deal  Deal  @relation(fields: [dealId], references: [id], onDelete: Cascade)
  buyer Buyer @relation(fields: [buyerId], references: [id], onDelete: Cascade)

  @@unique([dealId, buyerId])
  @@map("deal_buyers")
}

// Pricing normalization table
model PricingNormalization {
  id                    String  @id @default(uuid())
  dealId                String  @map("deal_id")
  unitType              String  @map("unit_type") // token, record, dataset, corpus, stream, minute, image, track
  normalizedCostPerUnit Float   @map("normalized_cost_per_unit")
  normalizationMethod   String  @map("normalization_method") // manual, inferred, approximate
  assumptions           String? // JSON for assumptions used

  deal Deal @relation(fields: [dealId], references: [id], onDelete: Cascade)

  createdAt DateTime @default(now()) @map("created_at")

  @@index([dealId])
  @@map("pricing_normalizations")
}

// Source tracking for provenance
model DealSource {
  id              String    @id @default(uuid())
  dealId          String    @map("deal_id")
  url             String
  sourceType      String    @map("source_type") // news, filing, press_release, rss
  scrapedAt       DateTime? @map("scraped_at")
  htmlArchive     String?   @map("html_archive") // Path to stored HTML
  extractedFields String?   @map("extracted_fields") // JSON of what was extracted

  createdAt DateTime @default(now()) @map("created_at")

  @@index([dealId])
  @@map("deal_sources")
}