feat(functions): better free string matching, allow to expect strings after JSON (#2445)

Allow now any non-character, both as suffix and prefix when mixed grammars are enabled

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

core/http/endpoints/openai/chat.go

@@ -67,9 +67,10 @@ func ChatEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, startup
 			return true
 		})
 
+		textContentToReturn = functions.ParseTextContent(result, config.FunctionsConfig)
 		result = functions.CleanupLLMResult(result, config.FunctionsConfig)
 		results := functions.ParseFunctionCall(result, config.FunctionsConfig)
-		textContentToReturn = functions.ParseTextContent(result, config.FunctionsConfig)
+		log.Debug().Msgf("Text content to return: %s", textContentToReturn)
 		noActionToRun := len(results) > 0 && results[0].Name == noAction || len(results) == 0
 
 		switch {
@@ -136,7 +137,8 @@ func ChatEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, startup
 					Model:   req.Model, // we have to return what the user sent here, due to OpenAI spec.
 					Choices: []schema.Choice{{
 						Delta: &schema.Message{
-							Role: "assistant",
+							Role:    "assistant",
+							Content: &textContentToReturn,
 							ToolCalls: []schema.ToolCall{
 								{
 									Index: i,
@@ -477,9 +479,10 @@ func ChatEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, startup
 					return
 				}
 
+				textContentToReturn = functions.ParseTextContent(s, config.FunctionsConfig)
 				s = functions.CleanupLLMResult(s, config.FunctionsConfig)
 				results := functions.ParseFunctionCall(s, config.FunctionsConfig)
-				textContentToReturn = functions.ParseTextContent(s, config.FunctionsConfig)
+				log.Debug().Msgf("Text content to return: %s", textContentToReturn)
 				noActionsToRun := len(results) > 0 && results[0].Name == noActionName || len(results) == 0
 
 				switch {
@@ -507,6 +510,7 @@ func ChatEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, startup
 						if len(input.Tools) > 0 {
 							// If we are using tools, we condense the function calls into
 							// a single response choice with all the tools
+							toolChoice.Message.Content = textContentToReturn
 							toolChoice.Message.ToolCalls = append(toolChoice.Message.ToolCalls,
 								schema.ToolCall{
 									ID:   id,
@@ -522,7 +526,8 @@ func ChatEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, startup
 							*c = append(*c, schema.Choice{
 								FinishReason: "function_call",
 								Message: &schema.Message{
-									Role: "assistant",
+									Role:    "assistant",
+									Content: &textContentToReturn,
 									FunctionCall: map[string]interface{}{
 										"name":      name,
 										"arguments": args,

pkg/functions/grammar_json_schema.go

@@ -54,7 +54,7 @@ var (
 		// however, if we don't have it, the grammar will be ambiguous and
 		// empirically results are way worse.
 		"freestring": `(
-			[^"\\] |
+			[^\x00] |
 			"\\" (["\\/bfnrt] | "u" [0-9a-fA-F] [0-9a-fA-F] [0-9a-fA-F] [0-9a-fA-F])
 		  )* space`,
 		"null": `"null" space`,
@@ -131,7 +131,7 @@ func (sc *JSONSchemaConverter) finalizeGrammar(options ...func(*GrammarOption))
 	grammarOpts := &GrammarOption{}
 	grammarOpts.Apply(options...)
 
-	suffix := grammarOpts.Suffix
+	prefix := grammarOpts.Prefix
 	maybeArray := grammarOpts.MaybeArray
 	disableParallelNewLines := grammarOpts.DisableParallelNewLines
 	maybeString := grammarOpts.MaybeString
@@ -139,7 +139,7 @@ func (sc *JSONSchemaConverter) finalizeGrammar(options ...func(*GrammarOption))
 
 	var lines []string
 
-	swapRoot := maybeArray || maybeString || suffix != ""
+	swapRoot := maybeArray || maybeString || prefix != ""
 
 	// write down the computed rules.
 	// if maybeArray is true, we need to add the array rule and slightly tweak the root rule
@@ -164,9 +164,9 @@ func (sc *JSONSchemaConverter) finalizeGrammar(options ...func(*GrammarOption))
 		freestringRule = "freestring"
 	}
 
-	if suffix != "" {
+	if prefix != "" {
 		// quote newlines in suffix
-		suffix = utils.EscapeNewLines(suffix)
+		prefix = utils.EscapeNewLines(prefix)
 
 		if maybeArray && maybeString {
 			newRoot = "(" + newRoot + ")"
@@ -174,9 +174,9 @@ func (sc *JSONSchemaConverter) finalizeGrammar(options ...func(*GrammarOption))
 
 		if maybeString {
 			//newRoot = "( (\"" + suffix + "\" " + newRoot + ") | freestring ) "
-			newRoot = "( \"" + suffix + "\" " + newRoot + " | " + freestringRule + " ) "
+			newRoot = "( \"" + prefix + "\" " + newRoot + " | " + freestringRule + " ) "
 		} else {
-			newRoot = "\"" + suffix + "\" " + "" + newRoot + ""
+			newRoot = "\"" + prefix + "\" " + "" + newRoot + ""
 		}
 	} else if maybeString {
 		if maybeArray {
@@ -194,9 +194,17 @@ func (sc *JSONSchemaConverter) finalizeGrammar(options ...func(*GrammarOption))
 	}
 
 	if maybeArray {
-		lines = append(lines, `mixedstring ::= freestring | freestring arr | freestring realvalue | realvalue | arr`)
+		if grammarOpts.ExpectStringsAfterJSON {
+			lines = append(lines, `mixedstring ::= freestring | freestring arr freestring | (freestring realvalue freestring)* | realvalue | arr`)
+		} else {
+			lines = append(lines, `mixedstring ::= freestring | freestring arr | freestring realvalue | realvalue | arr`)
+		}
 	} else {
-		lines = append(lines, `mixedstring ::= freestring | freestring realvalue | realvalue`)
+		if grammarOpts.ExpectStringsAfterJSON {
+			lines = append(lines, `mixedstring ::= freestring | (freestring realvalue freestring)* | realvalue`)
+		} else {
+			lines = append(lines, `mixedstring ::= freestring | freestring realvalue | realvalue`)
+		}
 	}
 
 	return strings.Join(lines, "\n")

pkg/functions/options.go

@@ -2,11 +2,12 @@ package functions
 
 type GrammarOption struct {
 	PropOrder               string
-	Suffix                  string
+	Prefix                  string
 	MaybeArray              bool
 	DisableParallelNewLines bool
 	MaybeString             bool
 	NoMixedFreeString       bool
+	ExpectStringsAfterJSON  bool
 }
 
 func (o *GrammarOption) Apply(options ...func(*GrammarOption)) {
@@ -31,8 +32,13 @@ var NoMixedFreeString func(*GrammarOption) = func(o *GrammarOption) {
 	o.NoMixedFreeString = true
 }
 
+// ExpectStringsAfterJSON enables mixed string suffix
+var ExpectStringsAfterJSON func(*GrammarOption) = func(o *GrammarOption) {
+	o.ExpectStringsAfterJSON = true
+}
+
 func SetPrefix(suffix string) func(*GrammarOption) {
 	return func(o *GrammarOption) {
-		o.Suffix = suffix
+		o.Prefix = suffix
 	}
 }

pkg/functions/parse.go

@@ -29,6 +29,9 @@ type GrammarConfig struct {
 	// Prefix is the suffix to append to the grammar when being generated
 	// This is useful when models prepend a tag before returning JSON
 	Prefix string `yaml:"prefix"`
+
+	// ExpectStringsAfterJSON enables mixed string suffix
+	ExpectStringsAfterJSON bool `yaml:"expect_strings_after_json"`
 }
 
 // FunctionsConfig is the configuration for the tool/function call.
@@ -98,6 +101,9 @@ func (g GrammarConfig) Options() []func(o *GrammarOption) {
 	if g.NoMixedFreeString {
 		opts = append(opts, NoMixedFreeString)
 	}
+	if g.ExpectStringsAfterJSON {
+		opts = append(opts, ExpectStringsAfterJSON)
+	}
 	return opts
 }
 
@@ -116,6 +122,9 @@ func CleanupLLMResult(llmresult string, functionConfig FunctionsConfig) string {
 }
 
 func ParseTextContent(llmresult string, functionConfig FunctionsConfig) string {
+	log.Debug().Msgf("ParseTextContent: %s", llmresult)
+	log.Debug().Msgf("CaptureLLMResult: %s", functionConfig.CaptureLLMResult)
+
 	for _, r := range functionConfig.CaptureLLMResult {
 		// We use a regex to extract the JSON object from the response
 		var respRegex = regexp.MustCompile(r)