feat(import): enrich baidu and bytedance release metadata

This commit is contained in:
phamnazage-jpg
2026-05-13 22:37:37 +08:00
parent bb5a1ff9e5
commit 92c9a40f4b
6 changed files with 279 additions and 21 deletions

View File

@@ -7,6 +7,7 @@ import (
"encoding/json"
"log"
"os"
"strings"
"time"
_ "github.com/lib/pq"
@@ -26,21 +27,65 @@ type ModelPricing struct {
ContextLength int
IsFree bool
SourceURL string
ModelSourceURL string
ReleaseDate string
Modality string
}
func releaseDateValue(raw string) time.Time {
func releaseDateValue(raw string) any {
if raw == "" {
return time.Now()
return nil
}
parsed, err := time.Parse("2006-01-02", raw)
if err != nil {
return time.Now()
return nil
}
return parsed
}
type bytedanceModelMetadata struct {
Prefix string
ReleaseDate string
ModelSourceURL string
}
var bytedanceModelMetadataRules = []bytedanceModelMetadata{
{
Prefix: "bytedance-doubao-1.5-thinking",
ReleaseDate: "2025-04-17",
ModelSourceURL: "https://developer.volcengine.com/articles/7496718897794039827",
},
{
Prefix: "bytedance-doubao-seed-1.6",
ReleaseDate: "2025-06-11",
ModelSourceURL: "https://developer.volcengine.com/articles/7517188354606104612",
},
{
Prefix: "bytedance-seedance-1.0-lite",
ReleaseDate: "2025-05-13",
ModelSourceURL: "https://developer.volcengine.com/articles/7504284064976502823",
},
}
func enrichBytedanceModelMetadata(model ModelPricing) ModelPricing {
normalizedID := strings.ToLower(model.ModelID)
for _, metadata := range bytedanceModelMetadataRules {
if strings.HasPrefix(normalizedID, metadata.Prefix) {
if metadata.ReleaseDate != "" {
model.ReleaseDate = metadata.ReleaseDate
}
if metadata.ModelSourceURL != "" {
model.ModelSourceURL = metadata.ModelSourceURL
}
return model
}
}
if model.ModelSourceURL == "" {
model.ModelSourceURL = model.SourceURL
}
return model
}
func main() {
dsn := os.Getenv("DATABASE_URL")
if dsn == "" {
@@ -79,7 +124,7 @@ func main() {
batchID := "manual-seed"
for _, b := range raw.Bytedance {
p := ModelPricing{
p := enrichBytedanceModelMetadata(ModelPricing{
ModelID: "bytedance-" + b.Model,
ModelName: b.Model,
ProviderName: "ByteDance",
@@ -94,7 +139,7 @@ func main() {
IsFree: b.InputPrice == 0,
SourceURL: "https://www.volcengine.com/docs/82379/1099320",
Modality: "text",
}
})
// Find or create provider
var providerID int64
@@ -131,7 +176,7 @@ func main() {
err = db.QueryRow(
`INSERT INTO models (external_id, name, provider_id, modality, context_length, status, source, batch_id, source_url, release_date)
VALUES ($1, $2, $3, $4, $5, 'active', $6, $7, $8, $9) RETURNING id`,
p.ModelID, p.ModelName, providerID, p.Modality, p.ContextLength, p.OperatorName, batchID, p.SourceURL, releaseDateValue(p.ReleaseDate),
p.ModelID, p.ModelName, providerID, p.Modality, p.ContextLength, p.OperatorName, batchID, firstNonEmpty(p.ModelSourceURL, p.SourceURL), releaseDateValue(p.ReleaseDate),
).Scan(&modelID)
}
if err != nil {
@@ -144,7 +189,7 @@ func main() {
release_date = COALESCE(release_date, $3),
updated_at = CURRENT_TIMESTAMP
WHERE id = $1`,
modelID, p.SourceURL, releaseDateValue(p.ReleaseDate),
modelID, firstNonEmpty(p.ModelSourceURL, p.SourceURL), releaseDateValue(p.ReleaseDate),
)
// Insert pricing
@@ -181,3 +226,12 @@ func main() {
log.Printf("Successfully imported %d ByteDance models", len(raw.Bytedance))
}
func firstNonEmpty(values ...string) string {
for _, value := range values {
if value != "" {
return value
}
}
return ""
}