Initial implementation of the bot: - cmd/chat-bot: CLI entrypoint (serve, reindex, ask, version) - internal/agent: LLM provider client + agent runner with RAG injection - internal/config: YAML config loader (providers, RAG, persona, server) - internal/i18n: response-language detection (EN/ES) - internal/persona: persona system prompt assembly from YAML - internal/portfolio: heading-based chunker + SQLite FTS5 indexer - internal/server: chi router with /api/chat (SSE), /api/health, /api/info, /api/reindex, middleware (RequestID, Logging, CORS, RateLimit) - internal/streaming: SSE protocol helpers (start, chunk, sources, done, error) - web/: drop-in vanilla-JS chat widget (no build, no deps) + demo + README - bench/: reproducible driver benchmark (modernc vs mattn SQLite) - configs/portfolio-bot.yaml: llama.cpp default provider, SQLite RAG, canine persona - docs/architecture.md / .es.md: aligned with SQLite FTS5 + llama.cpp decisions - data/projects/README*.md: project data documentation - README.md / .es.md: updated for current implementation All tests pass (go test ./...). Bot is functional end-to-end with the configured LLM provider.
273 lines
No EOL
7.1 KiB
Go
273 lines
No EOL
7.1 KiB
Go
package server
|
|
|
|
import (
|
|
"crypto/rand"
|
|
"encoding/hex"
|
|
"encoding/json"
|
|
"errors"
|
|
"fmt"
|
|
"log/slog"
|
|
"net/http"
|
|
"strings"
|
|
"time"
|
|
|
|
"github.com/VictorVargas/rony-chat-bot/internal/agent"
|
|
"github.com/VictorVargas/rony-chat-bot/internal/config"
|
|
"github.com/VictorVargas/rony-chat-bot/internal/portfolio"
|
|
"github.com/VictorVargas/rony-chat-bot/internal/streaming"
|
|
)
|
|
|
|
type Handlers struct {
|
|
cfg *config.Config
|
|
runner *agent.Runner
|
|
store *portfolio.Store
|
|
version string
|
|
}
|
|
|
|
func NewHandlers(cfg *config.Config, runner *agent.Runner, store *portfolio.Store, version string) *Handlers {
|
|
return &Handlers{cfg: cfg, runner: runner, store: store, version: version}
|
|
}
|
|
|
|
type ChatRequest struct {
|
|
Messages []ChatMessage `json:"messages"`
|
|
Stream *bool `json:"stream,omitempty"`
|
|
}
|
|
|
|
type ChatMessage struct {
|
|
Role string `json:"role"`
|
|
Content string `json:"content"`
|
|
}
|
|
|
|
type ChatResponse struct {
|
|
Content string `json:"content"`
|
|
Sources []string `json:"sources,omitempty"`
|
|
Usage streaming.Usage `json:"usage"`
|
|
}
|
|
|
|
func (h *Handlers) Chat(w http.ResponseWriter, r *http.Request) {
|
|
if r.Method != http.MethodPost {
|
|
http.Error(w, "method not allowed", http.StatusMethodNotAllowed)
|
|
return
|
|
}
|
|
var req ChatRequest
|
|
if err := json.NewDecoder(http.MaxBytesReader(w, r.Body, 1<<20)).Decode(&req); err != nil {
|
|
http.Error(w, "invalid JSON body: "+err.Error(), http.StatusBadRequest)
|
|
return
|
|
}
|
|
if err := req.validate(); err != nil {
|
|
http.Error(w, err.Error(), http.StatusBadRequest)
|
|
return
|
|
}
|
|
stream := true
|
|
if req.Stream != nil {
|
|
stream = *req.Stream
|
|
}
|
|
history := toAgentMessages(req.Messages)
|
|
if stream {
|
|
h.streamChat(w, r, history)
|
|
return
|
|
}
|
|
h.completeChat(w, r, history)
|
|
}
|
|
|
|
func toAgentMessages(in []ChatMessage) []agent.Message {
|
|
out := make([]agent.Message, len(in))
|
|
for i, m := range in {
|
|
out[i] = agent.Message{
|
|
Role: agent.Role(m.Role),
|
|
Content: m.Content,
|
|
}
|
|
}
|
|
return out
|
|
}
|
|
|
|
func (req *ChatRequest) validate() error {
|
|
if len(req.Messages) == 0 {
|
|
return errors.New("messages must not be empty")
|
|
}
|
|
for i, m := range req.Messages {
|
|
switch m.Role {
|
|
case "user", "assistant", "system":
|
|
default:
|
|
return fmt.Errorf("messages[%d].role %q is invalid", i, m.Role)
|
|
}
|
|
if strings.TrimSpace(m.Content) == "" {
|
|
return fmt.Errorf("messages[%d].content is empty", i)
|
|
}
|
|
}
|
|
return nil
|
|
}
|
|
|
|
// streamChat drives the LLM agent and forwards each chunk to the SSE
|
|
// stream using the protocol described in docs/architecture.md §3.2.
|
|
func (h *Handlers) streamChat(w http.ResponseWriter, r *http.Request, history []agent.Message) {
|
|
w.Header().Set("Content-Type", "text/event-stream")
|
|
w.Header().Set("Cache-Control", "no-cache")
|
|
w.Header().Set("Connection", "keep-alive")
|
|
w.Header().Set("X-Accel-Buffering", "no")
|
|
w.WriteHeader(http.StatusOK)
|
|
|
|
flusher, ok := w.(http.Flusher)
|
|
if !ok {
|
|
http.Error(w, "streaming unsupported", http.StatusInternalServerError)
|
|
return
|
|
}
|
|
|
|
ctx := r.Context()
|
|
convID := newConvID()
|
|
if err := streaming.WriteStart(w, convID); err != nil {
|
|
slog.Error("sse start failed", "err", err)
|
|
return
|
|
}
|
|
flusher.Flush()
|
|
|
|
// Report which RAG sources were used (search happens in BuildMessages).
|
|
_, ragContext, err := h.runner.BuildMessages(ctx, history)
|
|
if err != nil {
|
|
_ = streaming.WriteError(w, "rag: "+err.Error())
|
|
return
|
|
}
|
|
if ragContext != "" {
|
|
sources := extractSources(ragContext)
|
|
_ = streaming.WriteSources(w, sources)
|
|
}
|
|
|
|
// Stream from the model.
|
|
for chunk, err := range h.runner.Stream(ctx, history) {
|
|
if err != nil {
|
|
slog.Error("llm stream", "err", err)
|
|
_ = streaming.WriteError(w, "llm: "+err.Error())
|
|
return
|
|
}
|
|
if chunk.Delta != "" {
|
|
if err := streaming.WriteChunk(w, chunk.Delta); err != nil {
|
|
return
|
|
}
|
|
}
|
|
if chunk.FinishReason != "" && chunk.Usage.TotalTokens > 0 {
|
|
_ = streaming.WriteDone(w, streaming.Usage{
|
|
InputTokens: chunk.Usage.InputTokens,
|
|
OutputTokens: chunk.Usage.OutputTokens,
|
|
})
|
|
return
|
|
}
|
|
}
|
|
// Final usage chunk may come on the last iteration; if we never saw a
|
|
// finish_reason + usage in-band, surface what we recorded.
|
|
usage := h.runner.LastUsage()
|
|
_ = streaming.WriteDone(w, streaming.Usage{
|
|
InputTokens: usage.InputTokens,
|
|
OutputTokens: usage.OutputTokens,
|
|
})
|
|
}
|
|
|
|
func (h *Handlers) completeChat(w http.ResponseWriter, r *http.Request, history []agent.Message) {
|
|
w.Header().Set("Content-Type", "application/json")
|
|
var full strings.Builder
|
|
for chunk, err := range h.runner.Stream(r.Context(), history) {
|
|
if err != nil {
|
|
http.Error(w, "llm: "+err.Error(), http.StatusBadGateway)
|
|
return
|
|
}
|
|
full.WriteString(chunk.Delta)
|
|
}
|
|
usage := h.runner.LastUsage()
|
|
_, ragContext, _ := h.runner.BuildMessages(r.Context(), history)
|
|
resp := ChatResponse{
|
|
Content: full.String(),
|
|
Usage: streaming.Usage{
|
|
InputTokens: usage.InputTokens,
|
|
OutputTokens: usage.OutputTokens,
|
|
},
|
|
}
|
|
if ragContext != "" {
|
|
resp.Sources = extractSources(ragContext)
|
|
}
|
|
_ = json.NewEncoder(w).Encode(resp)
|
|
}
|
|
|
|
func extractSources(ragContext string) []string {
|
|
// The formatted hits look like "### [N] projectID — section\n..."
|
|
// so we scan the lines for that prefix and dedupe by projectID.
|
|
seen := map[string]bool{}
|
|
var out []string
|
|
for _, line := range strings.Split(ragContext, "\n") {
|
|
if !strings.HasPrefix(line, "### [") {
|
|
continue
|
|
}
|
|
// between "### [N] " and " — "
|
|
rest := strings.TrimPrefix(line, "### [")
|
|
rest = rest[strings.Index(rest, " ")+1:]
|
|
project := rest
|
|
if i := strings.Index(rest, " — "); i > 0 {
|
|
project = rest[:i]
|
|
}
|
|
if !seen[project] {
|
|
seen[project] = true
|
|
out = append(out, project)
|
|
}
|
|
}
|
|
return out
|
|
}
|
|
|
|
func truncate(s string, n int) string {
|
|
if len(s) <= n {
|
|
return s
|
|
}
|
|
return s[:n] + "…"
|
|
}
|
|
|
|
func newConvID() string {
|
|
var b [8]byte
|
|
_, _ = rand.Read(b[:])
|
|
return hex.EncodeToString(b[:])
|
|
}
|
|
|
|
func (h *Handlers) Info(w http.ResponseWriter, _ *http.Request) {
|
|
w.Header().Set("Content-Type", "application/json")
|
|
p := h.cfg.DefaultProvider()
|
|
_ = json.NewEncoder(w).Encode(map[string]any{
|
|
"name": h.cfg.Persona.Name,
|
|
"version": h.version,
|
|
"provider": p.Type,
|
|
"model": firstNonEmpty(p.Model, p.ModelPath),
|
|
"rag": h.cfg.RAG.Enabled,
|
|
"top_k": h.cfg.RAG.TopK,
|
|
"started": time.Now().UTC().Format(time.RFC3339),
|
|
})
|
|
}
|
|
|
|
func (h *Handlers) Reindex(w http.ResponseWriter, r *http.Request) {
|
|
if r.Method != http.MethodPost {
|
|
http.Error(w, "method not allowed", http.StatusMethodNotAllowed)
|
|
return
|
|
}
|
|
start := time.Now()
|
|
dbPath := h.cfg.RAG.DBPath
|
|
if dbPath == "" {
|
|
http.Error(w, "rag.db_path not configured", http.StatusBadRequest)
|
|
return
|
|
}
|
|
dur, files, chunks, err := portfolio.ReindexOnDisk(dbPath, h.cfg.RAG.DataPath, portfolio.DefaultChunkerConfig())
|
|
if err != nil {
|
|
http.Error(w, "reindex failed: "+err.Error(), http.StatusInternalServerError)
|
|
return
|
|
}
|
|
_ = dur
|
|
w.Header().Set("Content-Type", "application/json")
|
|
_ = json.NewEncoder(w).Encode(map[string]any{
|
|
"indexed_files": files,
|
|
"total_chunks": chunks,
|
|
"duration_ms": time.Since(start).Milliseconds(),
|
|
"db_path": dbPath,
|
|
})
|
|
}
|
|
|
|
func firstNonEmpty(vals ...string) string {
|
|
for _, v := range vals {
|
|
if v != "" {
|
|
return v
|
|
}
|
|
}
|
|
return ""
|
|
} |