b8b545accd
Add the server-side half of the brokered chat path (route B): the
opencode.Chat adapter forwards chat_prompt to POST
/api/session/{id}/prompt through the owned Connection.Call translation
and derives chat_event frames from the authenticated /api/event SSE
firehose (text and reasoning deltas, tool parts with their status
transitions, step and usage). Every pushed frame carries a stable
session and message id, with a last-message-id fallback so step/tool
events that omit the assistant message id stay attributable.
The /connect handler keeps per-session retained state (chatHub) capped at
1 MiB with least-recently-updated eviction at 128 turns, so a
reconnecting client re-syncs through chat_messages instead of replaying
deltas the OpenCode per-session log never resends. Prompt dispatch
subscribes before posting and buffers pre-ack frames to preserve order,
and cancels the stream on disconnect so no goroutine leaks. chat_sessions
and chat_create are declared but not yet brokered: they answer with an
explicit protocol error rather than the generic Pong.
Reuse the existing transport, validation and normalization; no second
OpenCode transport or credential path is introduced.
Tests: protocol golden round-trip, adapter normalization + v1/v2 Call
translation, and handler ack/stream/validation/resume/disconnect, plus a
real-adapter no-tool evidence harness.
459 lines
15 KiB
Go
459 lines
15 KiB
Go
package server
|
|
|
|
import (
|
|
"context"
|
|
"log"
|
|
"strings"
|
|
"sync"
|
|
"time"
|
|
|
|
"git.a-wan.net/alan/pipelines/internal/opencode"
|
|
"git.a-wan.net/alan/pipelines/internal/protocol"
|
|
)
|
|
|
|
// chatBroker is the server-side OpenCode chat seam. *opencode.Chat implements
|
|
// it; tests substitute a scripted broker. Keeping it an interface means the
|
|
// /connect handler never builds a second OpenCode transport.
|
|
type chatBroker interface {
|
|
Prompt(ctx context.Context, sessionID, text string) error
|
|
// Stream subscribes to the session's OpenCode event stream. ready is called
|
|
// once the subscription is attached, before any frame is read, so the caller
|
|
// can order the prompt after the subscription.
|
|
Stream(ctx context.Context, sessionID string, ready func(), emit func(protocol.ChatEventFrame) error) error
|
|
Interrupt(ctx context.Context, sessionID string) error
|
|
Messages(ctx context.Context, sessionID string, limit int) ([]protocol.ChatMessage, error)
|
|
Reply(ctx context.Context, request protocol.ChatReplyRequest) error
|
|
}
|
|
|
|
// chatRetainedLimit bounds the state one session may retain so a stalled reader
|
|
// or a runaway turn cannot grow memory without limit. It matches the OpenCode
|
|
// response bound (1 MiB).
|
|
const chatRetainedLimit = opencode.MaxResponse
|
|
|
|
// chatMaxTurns bounds how many sessions retain turn state. A long-lived server
|
|
// cannot accumulate state for every session ever prompted, so the
|
|
// least-recently-updated turn is evicted once the cap is exceeded.
|
|
const chatMaxTurns = 128
|
|
|
|
// chatHub retains the accumulated turn state per session, independently of any
|
|
// single connection, so a reconnecting client can re-sync what it missed. State
|
|
// is keyed by session and then by message id (a session holds many turns), and
|
|
// is capped at chatRetainedLimit.
|
|
type chatHub struct {
|
|
mu sync.Mutex
|
|
turns map[string]*chatTurn
|
|
now func() time.Time
|
|
}
|
|
|
|
// chatPart is the retained form of one part. Text and reasoning deltas append to
|
|
// text (amortized O(1)); a tool frame replaces part outright.
|
|
type chatPart struct {
|
|
part protocol.ChatPart
|
|
text []byte
|
|
}
|
|
|
|
// chatMessage is one retained assistant turn. Its parts keep the exact ids the
|
|
// supporting chat_event frames carried, so a resume preserves attribution.
|
|
type chatMessage struct {
|
|
id string
|
|
order []string
|
|
parts map[string]*chatPart
|
|
}
|
|
|
|
type chatTurn struct {
|
|
sessionID string
|
|
order []string
|
|
messages map[string]*chatMessage
|
|
step *protocol.ChatStep
|
|
usage *protocol.ChatUsage
|
|
size int
|
|
overflow bool
|
|
updated time.Time
|
|
}
|
|
|
|
func newChatHub() *chatHub {
|
|
return &chatHub{turns: map[string]*chatTurn{}, now: time.Now}
|
|
}
|
|
|
|
func (h *chatHub) clock() time.Time {
|
|
if h.now != nil {
|
|
return h.now()
|
|
}
|
|
return time.Now()
|
|
}
|
|
|
|
// apply folds one pushed frame into the retained state. Text and reasoning
|
|
// deltas append to their message's part; a tool frame upserts the full part
|
|
// (accounting for its whole size) so the latest lifecycle status, input,
|
|
// content, metadata, error and time are retained under the frame's message id.
|
|
// Step and usage frames update the session's latest progress.
|
|
func (h *chatHub) apply(frame protocol.ChatEventFrame) {
|
|
if frame.SessionID == "" {
|
|
return
|
|
}
|
|
h.mu.Lock()
|
|
defer h.mu.Unlock()
|
|
turn := h.turns[frame.SessionID]
|
|
if turn == nil {
|
|
h.evictLocked()
|
|
turn = &chatTurn{sessionID: frame.SessionID, messages: map[string]*chatMessage{}}
|
|
h.turns[frame.SessionID] = turn
|
|
}
|
|
turn.updated = h.clock()
|
|
switch frame.Kind {
|
|
case protocol.ChatEventStep:
|
|
turn.step = frame.Step
|
|
return
|
|
case protocol.ChatEventUsage:
|
|
turn.usage = frame.Usage
|
|
return
|
|
}
|
|
if frame.MessageID == "" {
|
|
return
|
|
}
|
|
message := turn.messages[frame.MessageID]
|
|
if message == nil {
|
|
message = &chatMessage{id: frame.MessageID, parts: map[string]*chatPart{}}
|
|
turn.order = append(turn.order, frame.MessageID)
|
|
turn.messages[frame.MessageID] = message
|
|
}
|
|
switch frame.Kind {
|
|
case protocol.ChatEventText, protocol.ChatEventReasoning:
|
|
id := ""
|
|
if frame.Part != nil {
|
|
id = frame.Part.ID
|
|
}
|
|
if id == "" || frame.Delta == "" {
|
|
return
|
|
}
|
|
if !turn.retain(len(frame.Delta)) {
|
|
return
|
|
}
|
|
part, ok := message.parts[id]
|
|
if !ok {
|
|
part = &chatPart{part: protocol.ChatPart{Type: frame.Kind, ID: id}}
|
|
message.order = append(message.order, id)
|
|
message.parts[id] = part
|
|
}
|
|
part.text = append(part.text, frame.Delta...)
|
|
case protocol.ChatEventTool:
|
|
if frame.Part == nil || frame.Part.ID == "" || turn.overflow {
|
|
return
|
|
}
|
|
// The tool part is retained whole, so size the complete part (input,
|
|
// content, metadata, error and time) and charge the difference against
|
|
// the cap on every upsert, not just the first-seen name.
|
|
incoming := chatPartBytes(frame.Part)
|
|
part, ok := message.parts[frame.Part.ID]
|
|
if !ok {
|
|
if !turn.retain(incoming) {
|
|
return
|
|
}
|
|
part = &chatPart{}
|
|
message.order = append(message.order, frame.Part.ID)
|
|
message.parts[frame.Part.ID] = part
|
|
} else if previous := chatPartBytes(&part.part); incoming > previous {
|
|
if !turn.retain(incoming - previous) {
|
|
return
|
|
}
|
|
} else {
|
|
turn.shrink(previous - incoming)
|
|
}
|
|
part.part = *frame.Part
|
|
}
|
|
}
|
|
|
|
// chatPartBytes estimates the retained size of a part so the per-session cap
|
|
// accounts for its whole state instead of only the first-seen name. It walks the
|
|
// fields rather than re-marshalling, so the cost matches the bytes it stores.
|
|
func chatPartBytes(part *protocol.ChatPart) int {
|
|
if part == nil {
|
|
return 0
|
|
}
|
|
n := len(part.Type) + len(part.ID) + len(part.Name)
|
|
if part.State == nil {
|
|
return n
|
|
}
|
|
n += len(part.State.Status) + len(part.State.Input)
|
|
for _, content := range part.State.Content {
|
|
n += len(content.Type) + len(content.Text) + len(content.URI) + len(content.Mime) + len(content.Name)
|
|
}
|
|
if part.State.Error != nil {
|
|
n += len(part.State.Error.Type) + len(part.State.Error.Message)
|
|
}
|
|
for key, value := range part.State.Metadata {
|
|
n += len(key)
|
|
if text, ok := value.(string); ok {
|
|
n += len(text)
|
|
} else {
|
|
n += 8
|
|
}
|
|
}
|
|
return n
|
|
}
|
|
|
|
// evictLocked drops the least-recently-updated turn when the session cap is
|
|
// reached, keeping retained memory bounded across session churn.
|
|
func (h *chatHub) evictLocked() {
|
|
if len(h.turns) < chatMaxTurns {
|
|
return
|
|
}
|
|
oldestID := ""
|
|
var oldest time.Time
|
|
for id, turn := range h.turns {
|
|
if oldestID == "" || turn.updated.Before(oldest) {
|
|
oldestID, oldest = id, turn.updated
|
|
}
|
|
}
|
|
if oldestID != "" {
|
|
delete(h.turns, oldestID)
|
|
}
|
|
}
|
|
|
|
// retain records n more bytes and reports whether the turn is still within the
|
|
// bound. The first overflow records one diagnostic and then drops further state
|
|
// instead of growing without limit.
|
|
func (t *chatTurn) retain(n int) bool {
|
|
if t.overflow {
|
|
return false
|
|
}
|
|
if t.size+n > chatRetainedLimit {
|
|
t.overflow = true
|
|
log.Printf("chat session %s retained state exceeds %d bytes; dropping further state", t.sessionID, chatRetainedLimit)
|
|
return false
|
|
}
|
|
t.size += n
|
|
return true
|
|
}
|
|
|
|
// shrink releases n bytes when a tool upsert replaces a larger retained part.
|
|
func (t *chatTurn) shrink(n int) {
|
|
t.size -= n
|
|
if t.size < 0 {
|
|
t.size = 0
|
|
}
|
|
}
|
|
|
|
// snapshot returns the accumulated session as one assistant message per
|
|
// retained message id, with the same stable part ids, or nil when nothing is
|
|
// retained. It is the resume source: a reconnecting client re-syncs the whole
|
|
// session once instead of replaying deltas, and each turn keeps its own id.
|
|
func (h *chatHub) snapshot(sessionID string) []protocol.ChatMessage {
|
|
h.mu.Lock()
|
|
defer h.mu.Unlock()
|
|
turn := h.turns[sessionID]
|
|
if turn == nil {
|
|
return nil
|
|
}
|
|
turn.updated = h.clock()
|
|
var messages []protocol.ChatMessage
|
|
for _, id := range turn.order {
|
|
message := turn.messages[id]
|
|
if message == nil || len(message.order) == 0 {
|
|
continue
|
|
}
|
|
out := protocol.ChatMessage{ID: id, SessionID: sessionID, Role: protocol.ChatRoleAssistant}
|
|
for _, partID := range message.order {
|
|
if part, ok := message.parts[partID]; ok {
|
|
piece := part.part
|
|
piece.Text = string(part.text)
|
|
out.Parts = append(out.Parts, piece)
|
|
}
|
|
}
|
|
messages = append(messages, out)
|
|
}
|
|
return messages
|
|
}
|
|
|
|
// handleChat dispatches one chat_* request. It returns true when the request was
|
|
// a chat type and has been fully answered (the caller must not fall through to
|
|
// the generic response). Frames are pushed after the ack, each through the
|
|
// connection-scoped write function so the read loop and the stream goroutine
|
|
// never interleave partial websocket writes.
|
|
func handleChat(ctx context.Context, request protocol.Message, broker chatBroker, hub *chatHub, write func(protocol.Message) error) bool {
|
|
switch request.Type {
|
|
case protocol.ChatPrompt:
|
|
handleChatPrompt(ctx, request, broker, hub, write)
|
|
return true
|
|
case protocol.ChatMessages:
|
|
handleChatMessages(ctx, request, broker, hub, write)
|
|
return true
|
|
case protocol.ChatInterrupt:
|
|
handleChatSimple(ctx, request, broker, hub, write)
|
|
return true
|
|
case protocol.ChatReply:
|
|
handleChatSimple(ctx, request, broker, hub, write)
|
|
return true
|
|
case protocol.ChatSessions, protocol.ChatCreate:
|
|
// Declared in the protocol for the client but not yet brokered. Answer
|
|
// explicitly so the client never mistakes the generic Pong for a chat
|
|
// reply, and the boundary stays visible instead of silently dropping.
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "chat_sessions and chat_create are not yet brokered"})
|
|
return true
|
|
}
|
|
return false
|
|
}
|
|
|
|
func handleChatPrompt(ctx context.Context, request protocol.Message, broker chatBroker, hub *chatHub, write func(protocol.Message) error) {
|
|
if broker == nil {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "chat is not configured on this server"})
|
|
return
|
|
}
|
|
prompt := request.ChatPrompt
|
|
if prompt == nil {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "a chat prompt payload is required"})
|
|
return
|
|
}
|
|
if !opencode.ValidSessionID(prompt.SessionID) {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "a valid OpenCode session id is required"})
|
|
return
|
|
}
|
|
if strings.TrimSpace(prompt.Text) == "" {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "a nonempty chat prompt is required"})
|
|
return
|
|
}
|
|
// Reject an oversized prompt before any OpenCode call; the bound matches the
|
|
// OpenCode request limit.
|
|
if len(prompt.Text) > opencode.MaxRequest {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "chat prompt exceeds 64 KiB"})
|
|
return
|
|
}
|
|
|
|
// Subscribe to the OpenCode firehose before posting the prompt: the stream
|
|
// has no replay cursor (research 2.4), so a delta emitted between the POST
|
|
// and the subscription would otherwise be lost. Frames that arrive before the
|
|
// ack is written are buffered and flushed in order after it.
|
|
sessionID := prompt.SessionID
|
|
streamCtx, streamCancel := context.WithCancel(ctx)
|
|
attached := make(chan struct{})
|
|
var attachOnce sync.Once
|
|
streamResult := make(chan error, 1)
|
|
var bufferMu sync.Mutex
|
|
buffering := true
|
|
var buffered []protocol.ChatEventFrame
|
|
go func() {
|
|
streamResult <- broker.Stream(streamCtx, sessionID, func() {
|
|
attachOnce.Do(func() { close(attached) })
|
|
}, func(frame protocol.ChatEventFrame) error {
|
|
bufferMu.Lock()
|
|
if buffering {
|
|
buffered = append(buffered, frame)
|
|
bufferMu.Unlock()
|
|
return nil
|
|
}
|
|
bufferMu.Unlock()
|
|
hub.apply(frame)
|
|
return write(protocol.Message{Type: protocol.ChatEvent, ChatEvent: &frame})
|
|
})
|
|
}()
|
|
|
|
select {
|
|
case <-attached:
|
|
case err := <-streamResult:
|
|
streamCancel()
|
|
if err != nil && ctx.Err() == nil {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "chat stream: " + err.Error()})
|
|
}
|
|
return
|
|
case <-ctx.Done():
|
|
streamCancel()
|
|
return
|
|
}
|
|
|
|
if err := broker.Prompt(ctx, sessionID, prompt.Text); err != nil {
|
|
streamCancel()
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: err.Error()})
|
|
return
|
|
}
|
|
// Acknowledge the accepted prompt with the echoed request id, then stream the
|
|
// reply. The prompt is sent exactly once and never re-sent on resume.
|
|
if err := write(protocol.Message{Type: protocol.ChatPrompt, ID: request.ID}); err != nil {
|
|
streamCancel()
|
|
return
|
|
}
|
|
// Flush frames buffered while the prompt was accepted, then resume live
|
|
// delivery. Holding bufferMu preserves order: the stream goroutine either
|
|
// buffered before this flush or waits and writes after it.
|
|
bufferMu.Lock()
|
|
buffering = false
|
|
pending := buffered
|
|
buffered = nil
|
|
for _, frame := range pending {
|
|
hub.apply(frame)
|
|
if err := write(protocol.Message{Type: protocol.ChatEvent, ChatEvent: &frame}); err != nil {
|
|
bufferMu.Unlock()
|
|
streamCancel()
|
|
return
|
|
}
|
|
}
|
|
bufferMu.Unlock()
|
|
go drainChatStream(ctx, streamResult, streamCancel, sessionID, write)
|
|
}
|
|
|
|
// drainChatStream reports the terminal error of a chat subscription and releases
|
|
// its context once the stream has ended.
|
|
func drainChatStream(ctx context.Context, result <-chan error, cancel context.CancelFunc, sessionID string, write func(protocol.Message) error) {
|
|
defer cancel()
|
|
if err := <-result; err != nil && ctx.Err() == nil {
|
|
log.Printf("chat stream session=%s: %v", sessionID, err)
|
|
_ = write(protocol.Message{Type: protocol.Error, Error: "chat stream: " + err.Error()})
|
|
}
|
|
}
|
|
|
|
func handleChatMessages(ctx context.Context, request protocol.Message, broker chatBroker, hub *chatHub, write func(protocol.Message) error) {
|
|
query := request.ChatQuery
|
|
if query == nil || !opencode.ValidSessionID(query.SessionID) {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "a valid OpenCode session id is required"})
|
|
return
|
|
}
|
|
response := protocol.Message{Type: protocol.ChatMessages, ID: request.ID}
|
|
// Resume from retained server state first: a reconnecting client gets the
|
|
// accumulated turns with the same stable ids without re-reading OpenCode.
|
|
if messages := hub.snapshot(query.SessionID); len(messages) > 0 {
|
|
response.ChatMessages = messages
|
|
_ = write(response)
|
|
return
|
|
}
|
|
if broker == nil {
|
|
_ = write(response)
|
|
return
|
|
}
|
|
messages, err := broker.Messages(ctx, query.SessionID, query.Limit)
|
|
if err != nil {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: err.Error()})
|
|
return
|
|
}
|
|
response.ChatMessages = messages
|
|
_ = write(response)
|
|
}
|
|
|
|
func handleChatSimple(ctx context.Context, request protocol.Message, broker chatBroker, hub *chatHub, write func(protocol.Message) error) {
|
|
if broker == nil {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "chat is not configured on this server"})
|
|
return
|
|
}
|
|
if request.Type == protocol.ChatInterrupt {
|
|
query := request.ChatQuery
|
|
if query == nil || !opencode.ValidSessionID(query.SessionID) {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "a valid OpenCode session id is required"})
|
|
return
|
|
}
|
|
if err := broker.Interrupt(ctx, query.SessionID); err != nil {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: err.Error()})
|
|
return
|
|
}
|
|
_ = write(protocol.Message{Type: protocol.ChatInterrupt, ID: request.ID})
|
|
return
|
|
}
|
|
reply := request.ChatReply
|
|
if reply == nil || !opencode.ValidSessionID(reply.SessionID) {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: "a valid OpenCode session id is required"})
|
|
return
|
|
}
|
|
if err := broker.Reply(ctx, *reply); err != nil {
|
|
_ = write(protocol.Message{Type: protocol.Error, ID: request.ID, Error: err.Error()})
|
|
return
|
|
}
|
|
_ = write(protocol.Message{Type: protocol.ChatReply, ID: request.ID})
|
|
}
|