gitea-mirror/Pulse
1
0
Fork 0
mirror of https://github.com/rcourtman/Pulse.git synced 2026-02-18 00:17:39 +01:00
Code Issues Packages Projects Releases Wiki Activity

fix(ai): filter DeepSeek DSML internal function-call format from responses

Browse Source
This commit is contained in:
rcourtman
2026-02-01 18:07:41 +00:00
parent 78cd64338a
commit 71e00ee7df
4 changed files with 145 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

64
internal/ai/chat/agentic.go
View File

@@ -267,6 +267,18 @@ func (a *AgenticLoop) executeWithTools(ctx context.Context, sessionID string, me
switch event.Type {
case "content":
if data, ok := event.Data.(providers.ContentEvent); ok {
// Check for DeepSeek DSML marker - if detected, stop streaming this chunk
// The DSML format indicates the model is outputting internal function call
// formatting instead of using the proper tool calling API
if containsDeepSeekMarker(data.Text) {
// Don't append or stream this content
return
}
// Also check if the accumulated content already has the marker
// (in case it arrived in a previous chunk)
if containsDeepSeekMarker(contentBuilder.String()) {
return
}
contentBuilder.WriteString(data.Text)
// Forward to callback - send ContentData struct
jsonData, _ := json.Marshal(ContentData{Text: data.Text})
@@ -388,10 +400,12 @@ func (a *AgenticLoop) executeWithTools(ctx context.Context, sessionID string, me
}
// Create assistant message
// Clean DeepSeek artifacts from the content before storing
cleanedContent := cleanDeepSeekArtifacts(contentBuilder.String())
assistantMsg := Message{
ID: uuid.New().String(),
Role: "assistant",
Content: contentBuilder.String(),
Content: cleanedContent,
ReasoningContent: thinkingBuilder.String(),
Timestamp: time.Now(),
}
@@ -1225,7 +1239,7 @@ func (a *AgenticLoop) ensureFinalTextResponse(
summaryMsg := Message{
ID: uuid.New().String(),
Role: "assistant",
Content: summaryBuilder.String(),
Content: cleanDeepSeekArtifacts(summaryBuilder.String()),
Timestamp: time.Now(),
}
resultMessages = append(resultMessages, summaryMsg)
@@ -2226,3 +2240,49 @@ func formatKeyParams(input map[string]interface{}) string {
return strings.Join(parts, ", ")
}
// cleanDeepSeekArtifacts removes DeepSeek's internal tool call format leakage.
// When DeepSeek doesn't properly use the function calling API, it may output
// its internal markup like <｜DSML｜function_calls>, <｜DSML｜invoke>, etc.
// These patterns can appear with Unicode pipe (｜) or ASCII pipe (|).
// This is applied to chat responses to prevent the artifacts from being shown to users.
func cleanDeepSeekArtifacts(content string) string {
if content == "" {
return content
}
// DeepSeek internal function call format markers
markers := []string{
"<｜DSML｜", // Unicode pipe variant (opening)
"</｜DSML｜", // Unicode pipe variant (closing)
"<|DSML|", // ASCII pipe variant (opening)
"</|DSML|", // ASCII pipe variant (closing)
"<｜/DSML｜", // Alternative Unicode closing
"<|/DSML|", // Alternative ASCII closing
}
for _, marker := range markers {
if idx := strings.Index(content, marker); idx >= 0 {
// DeepSeek function call blocks typically appear at the end of responses
// Remove everything from the marker to the end
content = strings.TrimSpace(content[:idx])
}
}
return content
}
// containsDeepSeekMarker checks if the content contains any DeepSeek internal function call markers.
// This is used during streaming to detect when we should stop forwarding content.
func containsDeepSeekMarker(content string) bool {
markers := []string{
"<｜DSML｜", // Unicode pipe variant
"<|DSML|", // ASCII pipe variant
}
for _, marker := range markers {
if strings.Contains(content, marker) {
return true
}
}
return false
}

11
internal/ai/chat/service.go
View File

@@ -328,6 +328,12 @@ func (s *Service) ExecuteStream(ctx context.Context, req ExecuteRequest, callbac
configuredModel = strings.TrimSpace(s.cfg.GetChatModel())
}
s.mu.RUnlock()
// Per-request autonomous mode override (used by investigation to avoid
// mutating shared service state from concurrent goroutines).
if req.AutonomousMode != nil {
autonomousMode = *req.AutonomousMode
}
selectedModel = configuredModel
if overrideModel != "" {
selectedModel = overrideModel
@@ -481,6 +487,11 @@ func (s *Service) ExecuteStream(ctx context.Context, req ExecuteRequest, callbac
defer loop.SetMaxTurns(MaxAgenticTurns)
}
// Apply per-request autonomous mode to the loop. For investigation requests
// with AutonomousMode set, this uses the per-request value instead of
// mutating shared service state from concurrent goroutines.
loop.SetAutonomousMode(autonomousMode)
resultMessages, err := loop.ExecuteWithTools(ctx, session.ID, messages, filteredTools, callback)
log.Debug().

11
internal/ai/chat/types.go
View File

@@ -67,11 +67,12 @@ type StructuredMention struct {
// ExecuteRequest represents a chat execution request
type ExecuteRequest struct {
Prompt string `json:"prompt"`
SessionID string `json:"session_id,omitempty"`
Model string `json:"model,omitempty"`
Mentions []StructuredMention `json:"mentions,omitempty"`
MaxTurns int `json:"max_turns,omitempty"` // Override max agentic turns (0 = use default)
Prompt string `json:"prompt"`
SessionID string `json:"session_id,omitempty"`
Model string `json:"model,omitempty"`
Mentions []StructuredMention `json:"mentions,omitempty"`
MaxTurns int `json:"max_turns,omitempty"` // Override max agentic turns (0 = use default)
AutonomousMode *bool `json:"autonomous_mode,omitempty"` // Per-request autonomous override (nil = use service default)
}
// QuestionAnswer represents a user's answer to a question

87
internal/ai/patrol_ai_test.go
View File

@@ -8,7 +8,7 @@ import (
func TestCleanThinkingTokens_DeepSeek(t *testing.T) {
input := "Some analysis\n<｜end▁of▁thinking｜>\nActual content here"
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if containsSubstr(result, "end▁of▁thinking") {
t.Errorf("DeepSeek thinking marker should be removed, got: %s", result)
}
@@ -34,22 +34,22 @@ After comprehensive analysis of your infrastructure, I identified several issues
1. **Critical CPU overload on Tower host**`
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if containsSubstr(result, "<｜end▁of▁thinking｜>") {
t.Errorf("cleanThinkingTokens() should have removed DeepSeek thinking markers")
t.Errorf("CleanThinkingTokens() should have removed DeepSeek thinking markers")
}
if containsSubstr(result, "Now, also consider") || containsSubstr(result, "Let's add an info") {
t.Errorf("cleanThinkingTokens() should have removed internal reasoning")
t.Errorf("CleanThinkingTokens() should have removed internal reasoning")
}
if !containsSubstr(result, "## Analysis Summary") {
t.Errorf("cleanThinkingTokens() removed header")
t.Errorf("CleanThinkingTokens() removed header")
}
if !containsSubstr(result, "### Key Findings") {
t.Errorf("cleanThinkingTokens() removed findings section")
t.Errorf("CleanThinkingTokens() removed findings section")
}
if !containsSubstr(result, "Critical CPU overload") {
t.Errorf("cleanThinkingTokens() removed actual finding")
t.Errorf("CleanThinkingTokens() removed actual finding")
}
}
@@ -60,10 +60,10 @@ Now, let's check something.
## Real Content`
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if result != "## Real Content" {
t.Errorf("cleanThinkingTokens() failed for ASCII variant: got %q", result)
t.Errorf("CleanThinkingTokens() failed for ASCII variant: got %q", result)
}
}
@@ -71,7 +71,7 @@ Now, let's check something.
func TestCleanThinkingTokens_ThinkBlock(t *testing.T) {
input := "Before content\n<think>This is internal reasoning\nthat spans multiple lines</think>\nAfter content"
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if containsSubstr(result, "internal reasoning") {
t.Errorf("<think> block content should be removed, got: %s", result)
}
@@ -85,7 +85,7 @@ func TestCleanThinkingTokens_ThinkBlock(t *testing.T) {
func TestCleanThinkingTokens_ThoughtBlock(t *testing.T) {
input := "Start\n<thought>Some deep thought process here</thought>\nEnd"
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if containsSubstr(result, "deep thought") {
t.Errorf("<thought> block content should be removed, got: %s", result)
}
@@ -99,7 +99,7 @@ func TestCleanThinkingTokens_ThoughtBlock(t *testing.T) {
func TestCleanThinkingTokens_ReasoningBlock(t *testing.T) {
input := "Start\n<|reasoning|>Internal reasoning here<|/reasoning|>\nEnd"
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if containsSubstr(result, "Internal reasoning") {
t.Errorf("<|reasoning|> block content should be removed, got: %s", result)
}
@@ -113,7 +113,7 @@ func TestCleanThinkingTokens_ReasoningBlock(t *testing.T) {
func TestCleanThinkingTokens_CaseInsensitiveBlocks(t *testing.T) {
input := "Before\n<THINK>uppercase thinking</THINK>\nAfter"
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if containsSubstr(result, "uppercase thinking") {
t.Errorf("Case-insensitive <THINK> block should be removed, got: %s", result)
}
@@ -124,7 +124,7 @@ func TestCleanThinkingTokens_CaseInsensitiveBlocks(t *testing.T) {
func TestCleanThinkingTokens_MultipleBlocks(t *testing.T) {
input := "<think>first block</think>content between<think>second block</think>final content"
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if containsSubstr(result, "first block") || containsSubstr(result, "second block") {
t.Errorf("All think blocks should be removed, got: %s", result)
}
@@ -135,7 +135,7 @@ func TestCleanThinkingTokens_MultipleBlocks(t *testing.T) {
func TestCleanThinkingTokens_UnclosedBlock(t *testing.T) {
input := "Start content<think>unclosed block with no end"
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if !containsSubstr(result, "Start content") {
t.Errorf("Content before unclosed block should be preserved, got: %s", result)
}
@@ -157,14 +157,14 @@ Now, I need to look at memory.
- Issue 1`
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if !containsSubstr(result, "## Analysis") || !containsSubstr(result, "### Findings") || !containsSubstr(result, "- Issue 1") {
t.Errorf("cleanThinkingTokens() removed too much: got %q", result)
t.Errorf("CleanThinkingTokens() removed too much: got %q", result)
}
if containsSubstr(result, "Let's check") || containsSubstr(result, "Now, I need") {
t.Errorf("cleanThinkingTokens() should have removed reasoning: got %q", result)
t.Errorf("CleanThinkingTokens() should have removed reasoning: got %q", result)
}
}
@@ -180,14 +180,14 @@ This is a normal response without any thinking tokens.
1. Issue one
2. Issue two`
result := cleanThinkingTokens(input)
result := CleanThinkingTokens(input)
if result != input {
t.Errorf("cleanThinkingTokens() modified clean content:\nGot: %q\nExpected: %q", result, input)
t.Errorf("CleanThinkingTokens() modified clean content:\nGot: %q\nExpected: %q", result, input)
}
}
func TestCleanThinkingTokens_EmptyContent(t *testing.T) {
result := cleanThinkingTokens("")
result := CleanThinkingTokens("")
if result != "" {
t.Errorf("Empty string should return empty, got: %q", result)
}
@@ -203,3 +203,48 @@ func containsSubstr(s, substr string) bool {
}
return false
}
func TestCleanThinkingTokens_DeepSeekDSMLFormat(t *testing.T) {
// This test verifies that DeepSeek's internal function call format is removed.
// When DeepSeek doesn't properly use the function calling API, it outputs
// its internal markup like <｜DSML｜function_calls> into visible responses.
input := `I've successfully updated the services.yaml file to remove the InfluxDB entry. Let me verify the change:
<｜DSML｜function_calls>
<｜DSML｜invoke name="pulse_read">
<｜DSML｜parameter name="action" string="true">exec</｜DSML｜parameter>
<｜DSML｜parameter name="command" string="true">grep -A 10 "Monitoring:" /opt/homepage/config/services.yaml</｜DSML｜parameter>
<｜DSML｜parameter name="target_host" string="true">homepage-docker</｜DSML｜parameter>
</｜DSML｜invoke>
</｜DSML｜function_calls>`
result := CleanThinkingTokens(input)
expected := "I've successfully updated the services.yaml file to remove the InfluxDB entry. Let me verify the change:"
if result != expected {
t.Errorf("CleanThinkingTokens() failed to remove DeepSeek DSML format:\nGot: %q\nExpected: %q", result, expected)
}
if containsSubstr(result, "DSML") {
t.Errorf("CleanThinkingTokens() should have removed all DSML markers, got: %s", result)
}
if containsSubstr(result, "pulse_read") {
t.Errorf("CleanThinkingTokens() should have removed tool call content, got: %s", result)
}
}
func TestCleanThinkingTokens_DeepSeekDSMLFormatASCII(t *testing.T) {
// Test the ASCII variant of DSML markers
input := `Command completed successfully.
<|DSML|function_calls>
<|DSML|invoke name="pulse_query">
<|DSML|parameter name="action">health</|DSML|parameter>
</|DSML|invoke>
</|DSML|function_calls>`
result := CleanThinkingTokens(input)
expected := "Command completed successfully."
if result != expected {
t.Errorf("CleanThinkingTokens() failed to remove ASCII DSML format:\nGot: %q\nExpected: %q", result, expected)
}
}