mirror of
https://github.com/ollama/ollama.git
synced 2026-06-03 13:59:06 -04:00
9db4bdbad6
* broad lint fixes to sidestep CI scope glitch * runner: Remove CGO engines, use llama-server exclusively for GGML models Remove the vendored GGML and llama.cpp backend, CGO runner, Go model implementations, and sample. llama-server (built from upstream llama.cpp via FetchContent) is now the sole inference engine for GGUF-based models. (Safetensor based models continue to run on the new MLX engine.) This allows us to more rapidly pick up new capabilities and fixes from llama.cpp as they come out. On windows this now requires recent AMD driver versions to support ROCm v7 as llama.cpp currently does not support building against v6. * llama/compat: load Ollama-format GGUFs in llama-server Squashed from upstream/jmorganca/llama-compat on 2026-04-29. Source tip:0c33775d37. Original source commits: -25223160dllama/compat: add in-memory shim so llama-server can load Ollama-format GGUFs -7449b539allm,server: route Ollama-format gemma3 blobs through llama/compat -436f2e2b1llama/compat: make patch-apply idempotent -8c2c9d4c8llama/compat: extend gemma3 handler to cover 1B and 270M blobs -021389f7bllama/compat: shrink clip.cpp injection from 18 lines to 1 -61b367ec2llama/compat: shrink patch to pure call-site hooks (34 -> 20 lines) -36049361cllama/compat: simplify shim (gemma3-tested) -8fa664865llama/compat: add qwen35moe text handler -db0c74530llama/compat: add qwen35moe vision (clip) support -2a388da77llama/compat: split shared infra into a util TU -9a69a17dcllama/compat: document non-public API dependencies -d0f38a915llama/compat: add gpt-oss and lfm2 handlers -086071822llama/compat: add mistral3 text handler (vision TODO) -63bde9ff7llama/compat: add mistral3 vision (clip) support -3a57b89d5llama/compat: apply LLaMA RoPE permute to mistral3 vision Q/K -99cb87439llama/compat: add qwen35, gemma4, deepseek-ocr handlers -2c7850dballama/compat: add nemotron_h_moe handler (latent FFN + MTP skip) -9e3b54225llama/compat: add llama4 text + clip handlers -034fee349llama/compat: add gemma4 clip handler (gemma4v projector) -9945c5a93server: remove dhiltgen/* compat redirect table -5d4539101llama/compat: rewrite gemma4 tokenizer model to BPE -7e0765327llama/compat: add glm-ocr text handler + text-loader load-op hook -f1bd1a25allama/compat: add glm-ocr clip handler (glm4v projector) -4b5cf3420llama/compat: collapse text-loader hook back to one new patch line -eb4ecf4fcllama/compat: extend gemma4 clip handler to gemma4a (audio) -a23a5e76fllama/compat: fix gemma4a per-block norm tensor mapping -cd2dcaff4llama/compat: add embeddinggemma handler -1ce8a6b26llama/compat: add qwen3-vl + qwen2.5-vl handlers -fd98ffa1ellama/compat: add gemma3n + glm4moelite handlers -cc7bdf0bcllama/compat: handle null buft in maybe_load_tensor -0c33775d3llama/compat: disable mmap when load_op transforms text-side tensors * refine implementation * ci: fix windows MLX build * ci: fix windows llama-server build * ci: fix windows rocm build * ci: windows mlx tuning Shorten long-tail on build, and get OllamaSetup.exe back under 2g limit * ci: fix windows dependencies * win: fix dependency gathering * disable openmp * win: arm64 cross-compile build also DRY out CI steps * scheduler improvements * ci: improvements from #15982 * win: favor ninja for faster developer builds * win: fix build * win: fix arm64 cross-compile * win: avoid spaces in compiler path * misc discovery fixes, and bos handling * lint fixes * win: fix arm cross-compile build/CI bugs * llama.cpp update * win: handle multiple CRT dirs * vulkan: add windows iGPU detection * fix creation bugs for patched models, other refactoring work * tune batch size for better performance * ci and lint fixes * fix repeat_last_n bug * build: revamp build for better developer UX * amd, sampler, qwen3next fixes * version bump * fix mlx build * revamp GPU discovery Scanning the output of llama-server is turning out to be too error prone across llama.cpp updates, so this switches to a thin dynamic library load against the bundled GGML libraries so more details can be gathered from the API. * version bump * missing file * ci: fix cache miss on rocm build * refine vulkan dep handling * fix ps reporting bug on full GPU load * improve cmake wiring for customized local builds * version bump * docker build arg cleanup * improve windows exit error logs * fix community gemma4 support and ci flakes * fix mlx unit test * tighten up ps logic to avoid double counting fit log lines * version bump * fix ps view for full gpu layer offload * add MTP wiring for llama-server and create with GGUFs * pick best template by capabilities * version bump * ci: harden apt repos * remove unused cpu core discovery * adjust batch default logic to reduce OOMs * support larger tool calls * fix audio support, template show * qwen35 mtp patch support * flesh out dtypes * rocm deps * version bump * lint fix * block broken gfx1150 on windows * fix qwen3.5 moe mtp tensors in patch * mmproj oom fallback and vulkan on by default * qwen MTP compat fix * version bump * ci: fix WoA cross-compile * ci: workaround ui tool in cross-compile * version bump * win: enable OpenMP for CPU builds * build: improve developer UX * ci: windows path workaround for CPU build * win: fix WoA dependencies * win: fix large offset reads for mmproj patched loads * version bump * fix vulkan dup detection * add OLLAMA_IGPU_ENABLE and largely disable iGPUs by default * opt-in MTP, win large offset, integraton fixes * fix unit test scheduler interaction hang * fix multi-gpu filtering * version bump * review comments * fix thinking level * fix linux rocm ordering and granite 3.3 template * version bump * ci fix - non-shallow MLX checkout * bypass linux sysfs unit test on windows --------- Co-authored-by: jmorganca <jmorganca@gmail.com>
373 lines
9.9 KiB
Go
373 lines
9.9 KiB
Go
package api
|
|
|
|
import (
|
|
"encoding/json"
|
|
"fmt"
|
|
"io"
|
|
"net/http"
|
|
"net/http/httptest"
|
|
"net/url"
|
|
"strings"
|
|
"testing"
|
|
)
|
|
|
|
func TestClientFromEnvironment(t *testing.T) {
|
|
type testCase struct {
|
|
value string
|
|
expect string
|
|
err error
|
|
}
|
|
|
|
testCases := map[string]*testCase{
|
|
"empty": {value: "", expect: "http://127.0.0.1:11434"},
|
|
"only address": {value: "1.2.3.4", expect: "http://1.2.3.4:11434"},
|
|
"only port": {value: ":1234", expect: "http://:1234"},
|
|
"address and port": {value: "1.2.3.4:1234", expect: "http://1.2.3.4:1234"},
|
|
"scheme http and address": {value: "http://1.2.3.4", expect: "http://1.2.3.4:80"},
|
|
"scheme https and address": {value: "https://1.2.3.4", expect: "https://1.2.3.4:443"},
|
|
"scheme, address, and port": {value: "https://1.2.3.4:1234", expect: "https://1.2.3.4:1234"},
|
|
"hostname": {value: "example.com", expect: "http://example.com:11434"},
|
|
"hostname and port": {value: "example.com:1234", expect: "http://example.com:1234"},
|
|
"scheme http and hostname": {value: "http://example.com", expect: "http://example.com:80"},
|
|
"scheme https and hostname": {value: "https://example.com", expect: "https://example.com:443"},
|
|
"scheme, hostname, and port": {value: "https://example.com:1234", expect: "https://example.com:1234"},
|
|
"trailing slash": {value: "example.com/", expect: "http://example.com:11434"},
|
|
"trailing slash port": {value: "example.com:1234/", expect: "http://example.com:1234"},
|
|
}
|
|
|
|
for k, v := range testCases {
|
|
t.Run(k, func(t *testing.T) {
|
|
t.Setenv("OLLAMA_HOST", v.value)
|
|
|
|
client, err := ClientFromEnvironment()
|
|
if err != v.err {
|
|
t.Fatalf("expected %s, got %s", v.err, err)
|
|
}
|
|
|
|
if client.base.String() != v.expect {
|
|
t.Fatalf("expected %s, got %s", v.expect, client.base.String())
|
|
}
|
|
})
|
|
}
|
|
}
|
|
|
|
// testError represents an internal error type with status code and message
|
|
// this is used since the error response from the server is not a standard error struct
|
|
type testError struct {
|
|
message string
|
|
statusCode int
|
|
raw bool // if true, write message as-is instead of JSON encoding
|
|
}
|
|
|
|
func (e testError) Error() string {
|
|
return e.message
|
|
}
|
|
|
|
func TestClientStream(t *testing.T) {
|
|
testCases := []struct {
|
|
name string
|
|
responses []any
|
|
wantErr string
|
|
}{
|
|
{
|
|
name: "immediate error response",
|
|
responses: []any{
|
|
testError{
|
|
message: "test error message",
|
|
statusCode: http.StatusBadRequest,
|
|
},
|
|
},
|
|
wantErr: "test error message",
|
|
},
|
|
{
|
|
name: "error after successful chunks, ok response",
|
|
responses: []any{
|
|
ChatResponse{Message: Message{Content: "partial response 1"}},
|
|
ChatResponse{Message: Message{Content: "partial response 2"}},
|
|
testError{
|
|
message: "mid-stream error",
|
|
statusCode: http.StatusOK,
|
|
},
|
|
},
|
|
wantErr: "mid-stream error",
|
|
},
|
|
{
|
|
name: "http status error takes precedence over general error",
|
|
responses: []any{
|
|
testError{
|
|
message: "custom error message",
|
|
statusCode: http.StatusInternalServerError,
|
|
},
|
|
},
|
|
wantErr: "500",
|
|
},
|
|
{
|
|
name: "successful stream completion",
|
|
responses: []any{
|
|
ChatResponse{Message: Message{Content: "chunk 1"}},
|
|
ChatResponse{Message: Message{Content: "chunk 2"}},
|
|
ChatResponse{
|
|
Message: Message{Content: "final chunk"},
|
|
Done: true,
|
|
DoneReason: "stop",
|
|
},
|
|
},
|
|
},
|
|
{
|
|
name: "plain text error response",
|
|
responses: []any{
|
|
"internal server error",
|
|
},
|
|
wantErr: "internal server error",
|
|
},
|
|
{
|
|
name: "HTML error page",
|
|
responses: []any{
|
|
"<html><body>404 Not Found</body></html>",
|
|
},
|
|
wantErr: "404 Not Found",
|
|
},
|
|
}
|
|
|
|
for _, tc := range testCases {
|
|
t.Run(tc.name, func(t *testing.T) {
|
|
ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
|
flusher, ok := w.(http.Flusher)
|
|
if !ok {
|
|
t.Fatal("expected http.Flusher")
|
|
}
|
|
|
|
w.Header().Set("Content-Type", "application/x-ndjson")
|
|
|
|
for _, resp := range tc.responses {
|
|
if errResp, ok := resp.(testError); ok {
|
|
w.WriteHeader(errResp.statusCode)
|
|
err := json.NewEncoder(w).Encode(map[string]string{
|
|
"error": errResp.message,
|
|
})
|
|
if err != nil {
|
|
t.Fatal("failed to encode error response:", err)
|
|
}
|
|
return
|
|
}
|
|
|
|
if str, ok := resp.(string); ok {
|
|
fmt.Fprintln(w, str)
|
|
flusher.Flush()
|
|
continue
|
|
}
|
|
|
|
if err := json.NewEncoder(w).Encode(resp); err != nil {
|
|
t.Fatalf("failed to encode response: %v", err)
|
|
}
|
|
flusher.Flush()
|
|
}
|
|
}))
|
|
defer ts.Close()
|
|
|
|
client := NewClient(&url.URL{Scheme: "http", Host: ts.Listener.Addr().String()}, http.DefaultClient)
|
|
|
|
var receivedChunks []ChatResponse
|
|
err := client.stream(t.Context(), http.MethodPost, "/v1/chat", nil, func(chunk []byte) error {
|
|
var resp ChatResponse
|
|
if err := json.Unmarshal(chunk, &resp); err != nil {
|
|
return fmt.Errorf("failed to unmarshal chunk: %w", err)
|
|
}
|
|
receivedChunks = append(receivedChunks, resp)
|
|
return nil
|
|
})
|
|
|
|
if tc.wantErr != "" {
|
|
if err == nil {
|
|
t.Fatal("expected error but got nil")
|
|
}
|
|
if !strings.Contains(err.Error(), tc.wantErr) {
|
|
t.Errorf("expected error containing %q, got %v", tc.wantErr, err)
|
|
}
|
|
return
|
|
}
|
|
if err != nil {
|
|
t.Errorf("unexpected error: %v", err)
|
|
}
|
|
})
|
|
}
|
|
}
|
|
|
|
func TestClientStreamReportsReadErrors(t *testing.T) {
|
|
client := NewClient(
|
|
&url.URL{Scheme: "http", Host: "example.com"},
|
|
&http.Client{Transport: roundTripFunc(func(*http.Request) (*http.Response, error) {
|
|
body := failingReader{
|
|
data: []byte(`{"message":{"content":"partial"}}` + "\n"),
|
|
err: io.ErrUnexpectedEOF,
|
|
}
|
|
|
|
return &http.Response{
|
|
StatusCode: http.StatusOK,
|
|
Status: "200 OK",
|
|
Body: io.NopCloser(&body),
|
|
Header: make(http.Header),
|
|
}, nil
|
|
})},
|
|
)
|
|
|
|
err := client.stream(t.Context(), http.MethodPost, "/api/chat", nil, func([]byte) error {
|
|
return nil
|
|
})
|
|
if err == nil {
|
|
t.Fatal("expected stream read error")
|
|
}
|
|
if !strings.Contains(err.Error(), io.ErrUnexpectedEOF.Error()) {
|
|
t.Fatalf("expected unexpected EOF, got %v", err)
|
|
}
|
|
}
|
|
|
|
func TestClientDo(t *testing.T) {
|
|
testCases := []struct {
|
|
name string
|
|
response any
|
|
wantErr string
|
|
wantStatusCode int
|
|
}{
|
|
{
|
|
name: "immediate error response",
|
|
response: testError{
|
|
message: "test error message",
|
|
statusCode: http.StatusBadRequest,
|
|
},
|
|
wantErr: "test error message",
|
|
wantStatusCode: http.StatusBadRequest,
|
|
},
|
|
{
|
|
name: "server error response",
|
|
response: testError{
|
|
message: "internal error",
|
|
statusCode: http.StatusInternalServerError,
|
|
},
|
|
wantErr: "internal error",
|
|
wantStatusCode: http.StatusInternalServerError,
|
|
},
|
|
{
|
|
name: "successful response",
|
|
response: struct {
|
|
ID string `json:"id"`
|
|
Success bool `json:"success"`
|
|
}{
|
|
ID: "msg_123",
|
|
Success: true,
|
|
},
|
|
},
|
|
{
|
|
name: "plain text error response",
|
|
response: testError{
|
|
message: "internal server error",
|
|
statusCode: http.StatusInternalServerError,
|
|
raw: true,
|
|
},
|
|
wantErr: "internal server error",
|
|
wantStatusCode: http.StatusInternalServerError,
|
|
},
|
|
{
|
|
name: "HTML error page",
|
|
response: testError{
|
|
message: "<html><body>404 Not Found</body></html>",
|
|
statusCode: http.StatusNotFound,
|
|
raw: true,
|
|
},
|
|
wantErr: "<html><body>404 Not Found</body></html>",
|
|
wantStatusCode: http.StatusNotFound,
|
|
},
|
|
}
|
|
|
|
for _, tc := range testCases {
|
|
t.Run(tc.name, func(t *testing.T) {
|
|
ts := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
|
|
if errResp, ok := tc.response.(testError); ok {
|
|
w.WriteHeader(errResp.statusCode)
|
|
if !errResp.raw {
|
|
err := json.NewEncoder(w).Encode(map[string]string{
|
|
"error": errResp.message,
|
|
})
|
|
if err != nil {
|
|
t.Fatal("failed to encode error response:", err)
|
|
}
|
|
} else {
|
|
// Write raw message (simulates non-JSON error responses)
|
|
fmt.Fprint(w, errResp.message)
|
|
}
|
|
return
|
|
}
|
|
|
|
w.Header().Set("Content-Type", "application/json")
|
|
if err := json.NewEncoder(w).Encode(tc.response); err != nil {
|
|
t.Fatalf("failed to encode response: %v", err)
|
|
}
|
|
}))
|
|
defer ts.Close()
|
|
|
|
client := NewClient(&url.URL{Scheme: "http", Host: ts.Listener.Addr().String()}, http.DefaultClient)
|
|
|
|
var resp struct {
|
|
ID string `json:"id"`
|
|
Success bool `json:"success"`
|
|
}
|
|
err := client.do(t.Context(), http.MethodPost, "/v1/messages", nil, &resp)
|
|
|
|
if tc.wantErr != "" {
|
|
if err == nil {
|
|
t.Fatalf("got nil, want error %q", tc.wantErr)
|
|
}
|
|
if err.Error() != tc.wantErr {
|
|
t.Errorf("error message mismatch: got %q, want %q", err.Error(), tc.wantErr)
|
|
}
|
|
if tc.wantStatusCode != 0 {
|
|
if statusErr, ok := err.(StatusError); ok {
|
|
if statusErr.StatusCode != tc.wantStatusCode {
|
|
t.Errorf("status code mismatch: got %d, want %d", statusErr.StatusCode, tc.wantStatusCode)
|
|
}
|
|
} else {
|
|
t.Errorf("expected StatusError, got %T", err)
|
|
}
|
|
}
|
|
return
|
|
}
|
|
|
|
if err != nil {
|
|
t.Fatalf("got error %q, want nil", err)
|
|
}
|
|
|
|
if expectedResp, ok := tc.response.(struct {
|
|
ID string `json:"id"`
|
|
Success bool `json:"success"`
|
|
}); ok {
|
|
if resp.ID != expectedResp.ID {
|
|
t.Errorf("response ID mismatch: got %q, want %q", resp.ID, expectedResp.ID)
|
|
}
|
|
if resp.Success != expectedResp.Success {
|
|
t.Errorf("response Success mismatch: got %v, want %v", resp.Success, expectedResp.Success)
|
|
}
|
|
}
|
|
})
|
|
}
|
|
}
|
|
|
|
type roundTripFunc func(*http.Request) (*http.Response, error)
|
|
|
|
func (f roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
|
|
return f(req)
|
|
}
|
|
|
|
type failingReader struct {
|
|
data []byte
|
|
err error
|
|
}
|
|
|
|
func (r *failingReader) Read(p []byte) (int, error) {
|
|
if len(r.data) > 0 {
|
|
n := copy(p, r.data)
|
|
r.data = r.data[n:]
|
|
return n, nil
|
|
}
|
|
return 0, r.err
|
|
}
|