Mathis/llamactl
1
0
Fork 0
mirror of https://github.com/lordmathis/llamactl.git synced 2025-11-06 09:04:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

Refactor command parsing and building

Browse Source
This commit is contained in:
Matúš Námešný
2025-09-19 20:23:25 +02:00
parent 34a949d22e
commit 64842e74b0
14 changed files with 434 additions and 539 deletions
Download Patch File Download Diff File Expand all files Collapse all files

28
pkg/backends/llamacpp/llama.go
View File

@@ -328,3 +328,31 @@ func (o *LlamaServerOptions) BuildCommandArgs() []string {
}
return backends.BuildCommandArgs(o, multipleFlags)
}
// ParseLlamaCommand parses a llama-server command string into LlamaServerOptions
// Supports multiple formats:
// 1. Full command: "llama-server --model file.gguf"
// 2. Full path: "/usr/local/bin/llama-server --model file.gguf"
// 3. Args only: "--model file.gguf --gpu-layers 32"
// 4. Multiline commands with backslashes
func ParseLlamaCommand(command string) (*LlamaServerOptions, error) {
executableNames := []string{"llama-server"}
var subcommandNames []string // Llama has no subcommands
multiValuedFlags := map[string]bool{
"override_tensor": true,
"override_kv": true,
"lora": true,
"lora_scaled": true,
"control_vector": true,
"control_vector_scaled": true,
"dry_sequence_breaker": true,
"logit_bias": true,
}
var llamaOptions LlamaServerOptions
if err := backends.ParseCommand(command, executableNames, subcommandNames, multiValuedFlags, &llamaOptions); err != nil {
return nil, err
}
return &llamaOptions, nil
}

115
pkg/backends/llamacpp/llama_test.go
View File

@@ -378,6 +378,121 @@ func TestUnmarshalJSON_ArrayFields(t *testing.T) {
}
}
func TestParseLlamaCommand(t *testing.T) {
tests := []struct {
name string
command string
expectErr bool
}{
{
name: "basic command",
command: "llama-server --model /path/to/model.gguf --gpu-layers 32",
expectErr: false,
},
{
name: "args only",
command: "--model /path/to/model.gguf --ctx-size 4096",
expectErr: false,
},
{
name: "mixed flag formats",
command: "llama-server --model=/path/model.gguf --gpu-layers 16 --verbose",
expectErr: false,
},
{
name: "quoted strings",
command: `llama-server --model test.gguf --api-key "sk-1234567890abcdef"`,
expectErr: false,
},
{
name: "empty command",
command: "",
expectErr: true,
},
{
name: "unterminated quote",
command: `llama-server --model test.gguf --api-key "unterminated`,
expectErr: true,
},
{
name: "malformed flag",
command: "llama-server ---model test.gguf",
expectErr: true,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
result, err := llamacpp.ParseLlamaCommand(tt.command)
if tt.expectErr {
if err == nil {
t.Errorf("expected error but got none")
}
return
}
if err != nil {
t.Errorf("unexpected error: %v", err)
return
}
if result == nil {
t.Errorf("expected result but got nil")
}
})
}
}
func TestParseLlamaCommandValues(t *testing.T) {
command := "llama-server --model /test/model.gguf --gpu-layers 32 --temp 0.7 --verbose --no-mmap"
result, err := llamacpp.ParseLlamaCommand(command)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if result.Model != "/test/model.gguf" {
t.Errorf("expected model '/test/model.gguf', got '%s'", result.Model)
}
if result.GPULayers != 32 {
t.Errorf("expected gpu_layers 32, got %d", result.GPULayers)
}
if result.Temperature != 0.7 {
t.Errorf("expected temperature 0.7, got %f", result.Temperature)
}
if !result.Verbose {
t.Errorf("expected verbose to be true")
}
if !result.NoMmap {
t.Errorf("expected no_mmap to be true")
}
}
func TestParseLlamaCommandArrays(t *testing.T) {
command := "llama-server --model test.gguf --lora adapter1.bin --lora=adapter2.bin"
result, err := llamacpp.ParseLlamaCommand(command)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if len(result.Lora) != 2 {
t.Errorf("expected 2 lora adapters, got %d", len(result.Lora))
}
expected := []string{"adapter1.bin", "adapter2.bin"}
for i, v := range expected {
if result.Lora[i] != v {
t.Errorf("expected lora[%d]=%s got %s", i, v, result.Lora[i])
}
}
}
// Helper functions
func contains(slice []string, item string) bool {
return slices.Contains(slice, item)

34
pkg/backends/llamacpp/parser.go
View File

@@ -1,34 +0,0 @@
package llamacpp
import (
"llamactl/pkg/backends"
)
// ParseLlamaCommand parses a llama-server command string into LlamaServerOptions
// Supports multiple formats:
// 1. Full command: "llama-server --model file.gguf"
// 2. Full path: "/usr/local/bin/llama-server --model file.gguf"
// 3. Args only: "--model file.gguf --gpu-layers 32"
// 4. Multiline commands with backslashes
func ParseLlamaCommand(command string) (*LlamaServerOptions, error) {
executableNames := []string{"llama-server"}
var subcommandNames []string // Llama has no subcommands
multiValuedFlags := map[string]bool{
"override_tensor": true,
"override_kv": true,
"lora": true,
"lora_scaled": true,
"control_vector": true,
"control_vector_scaled": true,
"dry_sequence_breaker": true,
"logit_bias": true,
}
var llamaOptions LlamaServerOptions
if err := backends.ParseCommand(command, executableNames, subcommandNames, multiValuedFlags, &llamaOptions); err != nil {
return nil, err
}
return &llamaOptions, nil
}

121
pkg/backends/llamacpp/parser_test.go
View File

@@ -1,121 +0,0 @@
package llamacpp_test
import (
"llamactl/pkg/backends/llamacpp"
"testing"
)
func TestParseLlamaCommand(t *testing.T) {
tests := []struct {
name string
command string
expectErr bool
}{
{
name: "basic command",
command: "llama-server --model /path/to/model.gguf --gpu-layers 32",
expectErr: false,
},
{
name: "args only",
command: "--model /path/to/model.gguf --ctx-size 4096",
expectErr: false,
},
{
name: "mixed flag formats",
command: "llama-server --model=/path/model.gguf --gpu-layers 16 --verbose",
expectErr: false,
},
{
name: "quoted strings",
command: `llama-server --model test.gguf --api-key "sk-1234567890abcdef"`,
expectErr: false,
},
{
name: "empty command",
command: "",
expectErr: true,
},
{
name: "unterminated quote",
command: `llama-server --model test.gguf --api-key "unterminated`,
expectErr: true,
},
{
name: "malformed flag",
command: "llama-server ---model test.gguf",
expectErr: true,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
result, err := llamacpp.ParseLlamaCommand(tt.command)
if tt.expectErr {
if err == nil {
t.Errorf("expected error but got none")
}
return
}
if err != nil {
t.Errorf("unexpected error: %v", err)
return
}
if result == nil {
t.Errorf("expected result but got nil")
}
})
}
}
func TestParseLlamaCommandValues(t *testing.T) {
command := "llama-server --model /test/model.gguf --gpu-layers 32 --temp 0.7 --verbose --no-mmap"
result, err := llamacpp.ParseLlamaCommand(command)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if result.Model != "/test/model.gguf" {
t.Errorf("expected model '/test/model.gguf', got '%s'", result.Model)
}
if result.GPULayers != 32 {
t.Errorf("expected gpu_layers 32, got %d", result.GPULayers)
}
if result.Temperature != 0.7 {
t.Errorf("expected temperature 0.7, got %f", result.Temperature)
}
if !result.Verbose {
t.Errorf("expected verbose to be true")
}
if !result.NoMmap {
t.Errorf("expected no_mmap to be true")
}
}
func TestParseLlamaCommandArrays(t *testing.T) {
command := "llama-server --model test.gguf --lora adapter1.bin --lora=adapter2.bin"
result, err := llamacpp.ParseLlamaCommand(command)
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
if len(result.Lora) != 2 {
t.Errorf("expected 2 lora adapters, got %d", len(result.Lora))
}
expected := []string{"adapter1.bin", "adapter2.bin"}
for i, v := range expected {
if result.Lora[i] != v {
t.Errorf("expected lora[%d]=%s got %s", i, v, result.Lora[i])
}
}
}