Skip to content

RAGFlow is an agent-native database: a Retrieval-Augmented Generation (RAG) engine that grew into context infrastructure.

Go: fix context and warnings (#18097)

Jin Haid7661b676d42
diff --git a/internal/agent/audio/tts_design.md b/internal/agent/audio/tts_design.md
index 4d5b779ac7..410b71b0b5 100644
--- a/internal/agent/audio/tts_design.md
+++ b/internal/agent/audio/tts_design.md
@@ -16,15 +16,15 @@ and creates an `LLMBundle(tenant, tts_model_config)`. The TTS
factory in `rag/llm/tts_model.py` dispatches to one of several
**HTTP-based** providers:
-| Provider | Backend | Pure Go? |
-|----------|---------|----------|
-| FishAudioTTS | HTTP POST to `api.fish.audio/v1/tts` (msgpack body) | feasible |
-| QwenTTS | DashScope SDK (over WebSocket) | heavy |
-| OpenAITTS | HTTP POST to OpenAI-compatible `/audio/tts` | feasible |
-| StepFunTTS | HTTP POST to vendor endpoint | feasible |
-| RAGconTTS | LiteLLM proxy (HTTP) | feasible |
-| XinferenceTTS | HTTP POST to xinference | feasible |
-| TongyiTTS | DashScope SDK | heavy |
+| Provider | Backend | Pure Go? |
+|---------------|-----------------------------------------------------|----------|
+| FishAudioTTS | HTTP POST to `api.fish.audio/v1/tts` (msgpack body) | feasible |
+| QwenTTS | DashScope SDK (over WebSocket) | heavy |
+| OpenAITTS | HTTP POST to OpenAI-compatible `/audio/tts` | feasible |
+| StepFunTTS | HTTP POST to vendor endpoint | feasible |
+| RAGconTTS | LiteLLM proxy (HTTP) | feasible |
+| XinferenceTTS | HTTP POST to xinference | feasible |
+| TongyiTTS | DashScope SDK | heavy |
None of the production providers are gtts / edge-tts. The "gtts or
edge-tts shell-out" wording in the original plan was a placeholder
diff --git a/internal/agent/canvas/cancel.go b/internal/agent/canvas/cancel.go
index 6afcda9626..20a107d900 100644
--- a/internal/agent/canvas/cancel.go
+++ b/internal/agent/canvas/cancel.go
@@ -14,10 +14,6 @@
// limitations under the License.
//
-// cancel.go implements the cross-process cancel signal for ordinary Agent
-// sessions. A canvas run polls Redis for "{sessionID}-cancel"; when the HTTP
-// handler sets the key, the watcher cancels the same context used by the
-// workflow.
package canvas
import (
@@ -38,8 +34,8 @@ const cancelKeySuffix = "-cancel"
func cancelKey(sessionID string) string { return sessionID + cancelKeySuffix }
// cancelPollInterval is the gap between Redis Get polls. 500ms keeps
-// cancel latency p99 ≤ 500ms while staying cheap (one GET every half-
-// second per active run). Tunable later if a tenant needs lower latency.
+// cancel latency p99 ≤ 500ms while staying cheap (one GET every half-second per active run).
+// Tunable later if a tenant needs lower latency.
const cancelPollInterval = 500 * time.Millisecond
// RequestCancelTTL is the lifetime of the cancel flag in Redis. Long
@@ -62,7 +58,7 @@ var cancelClientFn = func() (*redis.Client, error) {
return c, nil
}
-// WatchCancel blocks until either ctx is cancelled or the Redis
+// WatchCancel blocks until either ctx is canceled or the Redis
// "{sessionID}-cancel" key contains a non-empty value. When fired, it calls
// onCancel exactly once and returns. Polling interval is fixed
// at 500ms (see plan §4.9 — revised 2026-06-03 from 1s to 500ms).
diff --git a/internal/agent/canvas/canvas.go b/internal/agent/canvas/canvas.go
index f3becc8586..38aa6aedb8 100644
--- a/internal/agent/canvas/canvas.go
+++ b/internal/agent/canvas/canvas.go
@@ -15,7 +15,7 @@ import (
)
// legacyNoOpNames is the set of component names that the Go port
-// recognises for DSL v1 compatibility but does not ship a real
+// recognizes for DSL v1 compatibility but does not ship a real
// implementation for. Encountering one of these in a DSL is mapped to
// the same no-op echo lambda used for placeholder bodies by the
// BuildWorkflow in scheduler.go. New DSLs should not use these names —
diff --git a/internal/agent/canvas/compile.go b/internal/agent/canvas/compile.go
index 6a29622f17..4bf528740e 100644
--- a/internal/agent/canvas/compile.go
+++ b/internal/agent/canvas/compile.go
@@ -2,7 +2,7 @@
//
// Compile turns a Canvas (DSL) into a CompiledCanvas: a compiled
// compose.Runnable plus the CheckPointID used at this compile. The
-// compile-time wiring (state pre/post handlers, checkpoint store,
+// compile-time wiring (state pre- / post-handlers, checkpoint store,
// serializer) is configured here; the actual run path lives in
// runner.go and the HTTP handler / SSE / RunTracker are wired in
// internal/service and internal/handler.
@@ -88,12 +88,12 @@ type CompileOptions struct {
// CompileOption mutates a CompileOptions before the compile runs.
type CompileOption func(*CompileOptions)
-// WithCheckPointStore attaches a CheckPointStore to the compile.
+// WithCheckPointStore attaches a CheckPointStore to compile.
func WithCheckPointStore(s CheckPointStore) CompileOption {
return func(o *CompileOptions) { o.Store = s }
}
-// WithStateSerializer attaches a StateSerializer to the compile.
+// WithStateSerializer attaches a StateSerializer to compile.
func WithStateSerializer(s StateSerializer) CompileOption {
return func(o *CompileOptions) { o.Serializer = s }
}
@@ -130,7 +130,7 @@ func WithInterruptAfterNonTerminalCpn() CompileOption {
}
// WithOverrideParams attaches a run-level override map (keyed by
-// cpnID) to the compile. Each component's params are merged with
+// cpnID) to compile. Each component's params are merged with
// its own entry at compile time (run-level wins on key collision, see
// node_body.go applyOverrideParams). Passing nil is a no-op.
func WithOverrideParams(m map[string]any) CompileOption {
@@ -138,7 +138,7 @@ func WithOverrideParams(m map[string]any) CompileOption {
}
// Compile builds the eino Workflow from the Canvas and returns the
-// compiled Runnable. State pre/post handlers are wired inside BuildWorkflow
+// compiled Runnable. State pre- / post-handlers are wired inside BuildWorkflow
// (see scheduler.go). Checkpoint store + serializer are wired here as
// compile-time options (compose.GraphCompileOption).
//
@@ -314,9 +314,8 @@ func dedupeStrings(in []string) []string {
}
// checkPointAdapter drops the Delete method that compose.CheckPointStore
-// does not declare. The RedisCheckPointStore in this package has
-// Delete; eino
-// doesn't, so the adapter is a thin passthrough.
+// does not declare. The RedisCheckPointStore in this package has deleted;
+// the adapter is a thin passthrough.
type checkPointAdapter struct{ inner CheckPointStore }
func (a checkPointAdapter) Get(ctx context.Context, id string) ([]byte, bool, error) {
diff --git a/internal/agent/canvas/loop_semantics_test.go b/internal/agent/canvas/loop_semantics_test.go
index 4183598523..94097a9529 100644
--- a/internal/agent/canvas/loop_semantics_test.go
+++ b/internal/agent/canvas/loop_semantics_test.go
@@ -32,7 +32,6 @@
package canvas
import (
- "context"
"encoding/json"
"io"
"strings"
@@ -257,7 +256,7 @@ func TestLoop_MessageEmitsEveryIteration(t *testing.T) {
Upstream: []string{"bump"},
}
- cc, err := Compile(context.Background(), dsl)
+ cc, err := Compile(ctx, dsl)
if err != nil {
t.Fatalf("Compile: %v", err)
}
diff --git a/internal/agent/canvas/loop_subgraph.go b/internal/agent/canvas/loop_subgraph.go
index bf420eb9b8..c3e40410e6 100644
--- a/internal/agent/canvas/loop_subgraph.go
+++ b/internal/agent/canvas/loop_subgraph.go
@@ -46,7 +46,7 @@ import (
"github.com/cloudwego/eino/compose"
)
-// loopExpansion holds the two artefacts produced by buildLoopExpansion
+// loopExpansion holds the two artifacts produced by buildLoopExpansion
// and consumed by BuildWorkflow to install the loop node.
type loopExpansion struct {
Sub *compose.Workflow[map[string]any, map[string]any]
@@ -124,7 +124,7 @@ func collectLoopMembers(c *Canvas, loopID string) map[string]bool {
func collectDescendants(c *Canvas, root string) map[string]bool {
visited := make(map[string]bool)
- queue := []string{}
+ var queue []string
for _, child := range c.Components[root].Downstream {
if child == root {
continue
@@ -179,8 +179,7 @@ func buildSubWorkflow(
// body's mutations accumulate across iterations — otherwise a
// VariableAssigner that increments `counter` would be clobbered
// back to its initial value at the top of every iteration and
- // the loop could never terminate on a condition that watches the
- // counter.
+ // the loop could never terminate on a condition that watches the counter.
//
// "First time" is detected by checking whether the loop's state
// bucket already holds the variable: a missing bucket entry
@@ -535,7 +534,7 @@ func translateLoopCondition(loopID string, params map[string]any) (workflowx.Loo
return false, fmt.Errorf("loop %q: condition eval: no canvas state in context", loopID)
}
if len(conditions) == 0 {
- // No conditions means the loop only stops at max count
+ // No conditions mean the loop only stops at max count
// — never quit on conditions. Mirrors Python fallback.
return false, nil
}
diff --git a/internal/agent/canvas/node_body_timeout_test.go b/internal/agent/canvas/node_body_timeout_test.go
index 03926e8cc1..fee39df7fc 100644
--- a/internal/agent/canvas/node_body_timeout_test.go
+++ b/internal/agent/canvas/node_body_timeout_test.go
@@ -59,7 +59,7 @@ func TestRealComponentBody_RespectsParentCancellation(t *testing.T) {
comp := &blockingComponent{}
body := realComponentBody("test-cpn", "TestBlocking", comp)
- parentCtx, cancel := context.WithCancel(context.Background())
+ parentCtx, cancel := context.WithCancel(t.Context())
cancel() // pre-cancel
_, err := body(parentCtx, map[string]any{"x": 1})
@@ -77,7 +77,7 @@ func TestRealComponentBody_NoTimeoutWhenFast(t *testing.T) {
comp := &echoComponent{}
body := realComponentBody("test-cpn", "TestEcho", comp)
- out, err := body(context.Background(), map[string]any{"x": 1})
+ out, err := body(t.Context(), map[string]any{"x": 1})
if err != nil {
t.Fatalf("unexpected error: %v", err)
}
diff --git a/internal/agent/canvas/parallel_batch_test.go b/internal/agent/canvas/parallel_batch_test.go
index 6b394b7cd0..1685c0ea05 100644
--- a/internal/agent/canvas/parallel_batch_test.go
+++ b/internal/agent/canvas/parallel_batch_test.go
@@ -50,7 +50,7 @@ func TestBuildWorkflow_ParallelBatchStructure(t *testing.T) {
Path: []string{"begin", "a", "b", "final"},
}
- ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+ ctx, cancel := context.WithTimeout(t.Context(), 5*time.Second)
defer cancel()
type compileResult struct {
diff --git a/internal/agent/canvas/parallel_subgraph_test.go b/internal/agent/canvas/parallel_subgraph_test.go
index 1399e6cf13..3a198162d7 100644
--- a/internal/agent/canvas/parallel_subgraph_test.go
+++ b/internal/agent/canvas/parallel_subgraph_test.go
@@ -1,7 +1,6 @@
package canvas
import (
- "context"
"testing"
)
@@ -75,7 +74,7 @@ func TestBuildParallelExpansion_PrefersGroupedMembersOverDescendants(t *testing.
},
}
- exp, err := buildParallelExpansion(context.Background(), c, "Parallel:IterateList")
+ exp, err := buildParallelExpansion(t.Context(), c, "Parallel:IterateList")
if err != nil {
t.Fatalf("buildParallelExpansion: %v", err)
}
diff --git a/internal/agent/canvas/parallel_timing_test.go b/internal/agent/canvas/parallel_timing_test.go
index d8d9392486..ed8d833db2 100644
--- a/internal/agent/canvas/parallel_timing_test.go
+++ b/internal/agent/canvas/parallel_timing_test.go
@@ -40,7 +40,6 @@
package canvas
import (
- "context"
"testing"
)
@@ -73,7 +72,7 @@ func TestCanvas_ParallelExecution_StaticAnalysis(t *testing.T) {
Path: []string{"begin", "a", "b", "c", "final"},
}
- cc, err := Compile(context.Background(), c)
+ cc, err := Compile(t.Context(), c)
if err != nil {
t.Fatalf("Compile: %v", err)
}
diff --git a/internal/agent/canvas/run_tracker_test.go b/internal/agent/canvas/run_tracker_test.go
index 35ffd7d142..53a51e6134 100644
--- a/internal/agent/canvas/run_tracker_test.go
+++ b/internal/agent/canvas/run_tracker_test.go
@@ -42,7 +42,7 @@ func newTestTracker(t *testing.T, ttl time.Duration) (*RunTracker, *miniredis.Mi
func TestRunTracker_StateTransitions(t *testing.T) {
tracker, mr := newTestTracker(t, 30*24*time.Hour)
- ctx := context.Background()
+ ctx := t.Context()
// B1: Start
if err := tracker.Start(ctx, "run_1", "canvas_42", "tenant_a", ""); err != nil {
@@ -73,7 +73,7 @@ func TestRunTracker_StateTransitions(t *testing.T) {
}
// AttachCheckpoint
- if err := tracker.AttachCheckpoint(ctx, "run_1", "cpn_xyz"); err != nil {
+ if err = tracker.AttachCheckpoint(ctx, "run_1", "cpn_xyz"); err != nil {
t.Fatalf("AttachCheckpoint: %v", err)
}
got, _ = tracker.Get(ctx, "run_1")
@@ -82,14 +82,14 @@ func TestRunTracker_StateTransitions(t *testing.T) {
}
// B2: MarkSucceeded
- if err := tracker.MarkSucceeded(ctx, "run_1"); err != nil {
+ if err = tracker.MarkSucceeded(ctx, "run_1"); err != nil {
t.Fatalf("MarkSucceeded: %v", err)
}
got, _ = tracker.Get(ctx, "run_1")
if got["status"] != "1" {
t.Fatalf("status = %q, want 1 (succeeded)", got["status"])
}
- if _, err := strconv.ParseInt(got["finished_at"], 10, 64); err != nil {
+ if _, err = strconv.ParseInt(got["finished_at"], 10, 64); err != nil {
t.Fatalf("finished_at %q is not an int: %v", got["finished_at"], err)
}
// All previous fields preserved.
@@ -100,7 +100,7 @@ func TestRunTracker_StateTransitions(t *testing.T) {
func TestRunTracker_FailedAndCancelled(t *testing.T) {
tracker, _ := newTestTracker(t, time.Hour)
- ctx := context.Background()
+ ctx := t.Context()
// B3: MarkFailed
if err := tracker.Start(ctx, "run_fail", "c", "t", "run_parent"); err != nil {
@@ -138,7 +138,7 @@ func TestRunTracker_FailedAndCancelled(t *testing.T) {
func TestRunTracker_TTLRefresh(t *testing.T) {
tracker, mr := newTestTracker(t, 2*time.Second)
- ctx := context.Background()
+ ctx := t.Context()
if err := tracker.Start(ctx, "run_ttl", "c", "t", ""); err != nil {
t.Fatalf("Start: %v", err)
@@ -168,7 +168,7 @@ func TestRunTracker_TTLRefresh(t *testing.T) {
// A wait-for-user pause must also keep the run metadata alive for the
// full tracker TTL so the next request can resume it.
mr.FastForward(500 * time.Millisecond)
- if err := tracker.MarkWaiting(ctx, "run_ttl"); err != nil {
+ if err = tracker.MarkWaiting(ctx, "run_ttl"); err != nil {
t.Fatalf("MarkWaiting: %v", err)
}
if d := mr.TTL(runKey("run_ttl")); d < 1500*time.Millisecond {
@@ -185,7 +185,7 @@ func TestRunTracker_TTLRefresh(t *testing.T) {
func TestRunTracker_NilClient(t *testing.T) {
tracker := &RunTracker{client: nil, ttl: time.Minute}
- ctx := context.Background()
+ ctx := t.Context()
if err := tracker.Start(ctx, "x", "c", "t", ""); err == nil {
t.Fatal("Start with nil client: err = nil, want error")
}
@@ -211,7 +211,7 @@ func TestRunTracker_NilClient(t *testing.T) {
func TestRunTrackerActiveSessionRegistration(t *testing.T) {
tracker, mr := newTestTracker(t, time.Hour)
- ctx := context.Background()
+ ctx := t.Context()
if err := tracker.client.Set(ctx, cancelKey("session-1"), "x", time.Hour).Err(); err != nil {
t.Fatalf("seed cancel marker: %v", err)
}
@@ -259,7 +259,7 @@ func TestRunTrackerActiveSessionRegistration(t *testing.T) {
func TestRunTrackerActiveSessionTokenProtectsRefreshAndRelease(t *testing.T) {
tracker, _ := newTestTracker(t, time.Hour)
- ctx := context.Background()
+ ctx := t.Context()
registered, err := tracker.RegisterActiveSession(ctx, ActiveSession{SessionID: "session-1", Token: "owner-a"})
if err != nil || !registered {
t.Fatalf("RegisterActiveSession = %v, %v", registered, err)
@@ -275,7 +275,7 @@ func TestRunTrackerActiveSessionTokenProtectsRefreshAndRelease(t *testing.T) {
if got, _ := tracker.GetActiveSession(ctx, "session-1"); got == nil {
t.Fatal("foreign release removed active session")
}
- if err := tracker.client.Set(ctx, cancelKey("session-1"), "x", time.Hour).Err(); err != nil {
+ if err = tracker.client.Set(ctx, cancelKey("session-1"), "x", time.Hour).Err(); err != nil {
t.Fatalf("seed cancel marker: %v", err)
}
released, err = tracker.ReleaseActiveSession(ctx, "session-1", "owner-a")
@@ -292,7 +292,7 @@ func TestRunTrackerActiveSessionTokenProtectsRefreshAndRelease(t *testing.T) {
func TestRunTrackerActiveSessionTokenProtectsCancel(t *testing.T) {
tracker, _ := newTestTracker(t, time.Hour)
- ctx := context.Background()
+ ctx := t.Context()
registered, err := tracker.RegisterActiveSession(ctx, ActiveSession{SessionID: "session-1", Token: "owner-a"})
if err != nil || !registered {
t.Fatalf("RegisterActiveSession = %v, %v", registered, err)
diff --git a/internal/agent/canvas/scheduler_test.go b/internal/agent/canvas/scheduler_test.go
index fe4d86f625..bd1dca9adc 100644
--- a/internal/agent/canvas/scheduler_test.go
+++ b/internal/agent/canvas/scheduler_test.go
@@ -2,7 +2,6 @@
package canvas
import (
- "context"
"encoding/json"
"strings"
"testing"
@@ -32,7 +31,7 @@ func TestBuildWorkflow_3NodeLinear(t *testing.T) {
Path: []string{"begin_0", "llm_0", "message_0"},
}
- wf, err := BuildWorkflow(context.Background(), c)
+ wf, err := BuildWorkflow(t.Context(), c)
if err != nil {
t.Fatalf("BuildWorkflow: %v", err)
}
@@ -41,7 +40,7 @@ func TestBuildWorkflow_3NodeLinear(t *testing.T) {
}
// Compile to a Runnable to confirm the topology is internally consistent.
- cc, err := Compile(context.Background(), c)
+ cc, err := Compile(t.Context(), c)
if err != nil {
t.Fatalf("Compile: %v", err)
}
@@ -84,7 +83,7 @@ func TestBuildWorkflow_5NodeDiamond(t *testing.T) {
Path: []string{"begin_0", "a_0", "b_0", "c_0", "d_0"},
}
- cc, err := Compile(context.Background(), c)
+ cc, err := Compile(t.Context(), c)
if err != nil {
t.Fatalf("Compile diamond: %v", err)
}
@@ -124,7 +123,7 @@ func TestBuildWorkflow_MultiTerminalSucceeds(t *testing.T) {
Path: []string{"begin_0", "a_0", "b_0", "c_0"},
}
- cc, err := Compile(context.Background(), c)
+ cc, err := Compile(t.Context(), c)
if err != nil {
t.Fatalf("Compile multi-terminal canvas: %v", err)
}
@@ -192,7 +191,7 @@ func TestBuildWorkflow_ParallelGroupWithOuterFollowerSucceeds(t *testing.T) {
},
}
- cc, err := Compile(context.Background(), c)
+ cc, err := Compile(t.Context(), c)
if err != nil {
t.Fatalf("Compile parallel canvas: %v", err)
}
@@ -218,7 +217,7 @@ func TestBuildWorkflow_ErrorsOnUnknownUpstream(t *testing.T) {
},
},
}
- _, err := BuildWorkflow(context.Background(), c)
+ _, err := BuildWorkflow(t.Context(), c)
if err == nil {
t.Fatal("expected error for unknown upstream")
}
@@ -238,7 +237,7 @@ func TestBuildWorkflow_ErrorsOnSelfEdge(t *testing.T) {
},
},
}
- _, err := BuildWorkflow(context.Background(), c)
+ _, err := BuildWorkflow(t.Context(), c)
if err == nil {
t.Fatal("expected error for self-edge")
}
@@ -249,7 +248,7 @@ func TestBuildWorkflow_ErrorsOnSelfEdge(t *testing.T) {
func TestNodeLifecycleEventsSafeMarshalNonSerializableInputs(t *testing.T) {
events := make(chan RunEvent, 2)
- ctx := WithRunMeta(context.Background(), &RunMeta{
+ ctx := WithRunMeta(t.Context(), &RunMeta{
Events: events,
MessageID: "msg-1",
SessionID: "session-1",
@@ -280,7 +279,7 @@ func TestNodeLifecycleEventsSafeMarshalNonSerializableInputs(t *testing.T) {
if !ok {
t.Fatalf("%s inputs = %T %v, want object", wantType, payload["inputs"], payload["inputs"])
}
- if _, ok := in["callback"]; !ok {
+ if _, ok = in["callback"]; !ok {
t.Fatalf("%s inputs missing sanitized callback key: %v", wantType, in)
}
if in["callback"] != nil {
@@ -294,7 +293,7 @@ func TestNodeLifecycleEventsSafeMarshalNonSerializableInputs(t *testing.T) {
func TestNodeLifecycleEventsFallbackMarshalErrorPreservesIdentity(t *testing.T) {
events := make(chan RunEvent, 2)
- ctx := WithRunMeta(context.Background(), &RunMeta{
+ ctx := WithRunMeta(t.Context(), &RunMeta{
Events: events,
MessageID: "msg-2",
SessionID: "session-2",
diff --git a/internal/agent/component/agent.go b/internal/agent/component/agent.go
index b36d85d23d..1ac7847dc6 100644
--- a/internal/agent/component/agent.go
+++ b/internal/agent/component/agent.go
@@ -1195,7 +1195,7 @@ func toolMessageTextContent(msg *schema.Message) string {
return ""
}
-// formatArtifactMarkdown renders a slice of artifacts as markdown
+// formatArtifactMarkdown renders a slice of artifacts as Markdown
// links, omitting URLs already present in the existing text (Python's
// `_collect_tool_artifact_markdown` does the same de-duplication).
//
diff --git a/internal/agent/component/agent_test.go b/internal/agent/component/agent_test.go
index 642ba8c61f..89634f7880 100644
--- a/internal/agent/component/agent_test.go
+++ b/internal/agent/component/agent_test.go
@@ -787,7 +787,7 @@ func TestAgent_Invoke_RespectsParentCancellation(t *testing.T) {
})
c := NewAgentComponent(AgentParam{ModelID: "echo", MaxRounds: 1})
- ctx, cancel := context.WithCancel(context.Background())
+ ctx, cancel := context.WithCancel(t.Context())
cancel() // pre-cancel
_, err := c.Invoke(ctx, nil, map[string]any{"user_prompt": "hi"})
@@ -1065,15 +1065,16 @@ func TestAgent_CanvasSubAgentToolBuildsDynamicTool(t *testing.T) {
if len(agent.param.SubAgents) != 1 {
t.Fatalf("sub agents = %d, want 1", len(agent.param.SubAgents))
}
+ ctx := t.Context()
- tools, err := buildAgentTools(t.Context(), agent.param)
+ tools, err := buildAgentTools(ctx, agent.param)
if err != nil {
t.Fatalf("buildAgentTools: %v", err)
}
if len(tools) != 1 {
t.Fatalf("len(tools) = %d, want 1", len(tools))
}
- info, err := tools[0].Info(context.Background())
+ info, err := tools[0].Info(ctx)
if err != nil {
t.Fatalf("tool.Info: %v", err)
}
@@ -1125,7 +1126,7 @@ func TestAgent_CanvasSubAgentToolNamesAreUniqueAfterNormalization(t *testing.T)
}
var names []string
for _, tool := range tools {
- info, err := tool.Info(context.Background())
+ info, err := tool.Info(t.Context())
if err != nil {
t.Fatalf("tool.Info: %v", err)
}
diff --git a/internal/agent/component/artifact_markdown_test.go b/internal/agent/component/artifact_markdown_test.go
index 6505a54f0a..4e5f88b65b 100644
--- a/internal/agent/component/artifact_markdown_test.go
+++ b/internal/agent/component/artifact_markdown_test.go
@@ -31,7 +31,7 @@ func TestFormatArtifactMarkdown_Empty(t *testing.T) {
}
}
-// TestFormatArtifactMarkdown_ImageLink: image URL → markdown image syntax.
+// TestFormatArtifactMarkdown_ImageLink: image URL → Markdown image syntax.
func TestFormatArtifactMarkdown_ImageLink(t *testing.T) {
arts := []artifactEntry{
{Name: "chart", URL: "https://example.com/chart.png"},
diff --git a/internal/agent/component/browser_test.go b/internal/agent/component/browser_test.go
index 3e6322b7d6..d6ef4c5afb 100644
--- a/internal/agent/component/browser_test.go
+++ b/internal/agent/component/browser_test.go
@@ -84,7 +84,7 @@ func stateWith(t *testing.T, sys map[string]any) context.Context {
t.Helper()
state := canvas.NewCanvasState("run-1", "task-1")
state.Sys = sys
- return canvas.WithState(context.Background(), state)
+ return canvas.WithState(t.Context(), state)
}
// TestBrowser_AcceptsPythonSchema: the v1 fixture's param surface
@@ -282,7 +282,7 @@ func TestBrowser_MissingTenant(t *testing.T) {
})
// state with no tenant_id
state := canvas.NewCanvasState("run-1", "task-1")
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
_, err := c.Invoke(ctx, nil, nil)
if err == nil || !strings.Contains(err.Error(), "tenant_id") {
diff --git a/internal/agent/component/canvas_var_resolution_test.go b/internal/agent/component/canvas_var_resolution_test.go
index e4747a5c4b..a318e0988d 100644
--- a/internal/agent/component/canvas_var_resolution_test.go
+++ b/internal/agent/component/canvas_var_resolution_test.go
@@ -17,7 +17,6 @@
package component
import (
- "context"
"testing"
"ragflow/internal/agent/runtime"
@@ -38,7 +37,7 @@ func TestLLM_Invoke_ResolvesTemplateRefs(t *testing.T) {
SystemPrompt: "Use the following context:",
UserPrompt: "What does {{retrieval:0@content}} say?",
})
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
_, err := c.Invoke(ctx, nil, map[string]any{})
if err != nil {
t.Fatalf("Invoke: %v", err)
@@ -64,7 +63,7 @@ func TestLLM_Invoke_NoState_LeavesPromptsUnchanged(t *testing.T) {
ModelID: "echo",
UserPrompt: "What does {{retrieval:0@content}} say?",
})
- _, err := c.Invoke(context.Background(), nil, map[string]any{})
+ _, err := c.Invoke(t.Context(), nil, map[string]any{})
if err != nil {
t.Fatalf("Invoke: %v", err)
}
@@ -89,7 +88,7 @@ func TestLLM_Invoke_UnresolvedRef_LeavesPromptIntact(t *testing.T) {
ModelID: "echo",
UserPrompt: "Use {{retrieval:0@content}} please",
})
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
_, err := c.Invoke(ctx, nil, map[string]any{})
if err != nil {
t.Fatalf("Invoke should not error on unresolved ref: %v", err)
diff --git a/internal/agent/component/chat_template_kwargs_test.go b/internal/agent/component/chat_template_kwargs_test.go
index 74b95e0d4b..147af3c673 100644
--- a/internal/agent/component/chat_template_kwargs_test.go
+++ b/internal/agent/component/chat_template_kwargs_test.go
@@ -17,7 +17,6 @@
package component
import (
- "context"
"testing"
)
@@ -76,7 +75,7 @@ func TestLLM_Invoke_ChatTemplateKwargsDoesNotBreak(t *testing.T) {
ModelID: "echo",
ChatTemplateKwargs: map[string]any{"seed": 1},
})
- _, err := c.Invoke(context.Background(), nil, map[string]any{"user_prompt": "hi"})
+ _, err := c.Invoke(t.Context(), nil, map[string]any{"user_prompt": "hi"})
if err != nil {
t.Fatalf("Invoke: %v", err)
}
diff --git a/internal/agent/component/citation_test.go b/internal/agent/component/citation_test.go
index cec92ad073..8a03a6b72a 100644
--- a/internal/agent/component/citation_test.go
+++ b/internal/agent/component/citation_test.go
@@ -17,7 +17,6 @@
package component
import (
- "context"
"strings"
"testing"
@@ -132,7 +131,7 @@ func TestLLM_Invoke_AppendsCitationPrompt(t *testing.T) {
withStubInvoker(t, stub)
c := NewLLMComponent(LLMParam{ModelID: "echo", SystemPrompt: "You are a bot.", Cite: true})
- if _, err := c.Invoke(context.Background(), nil, map[string]any{
+ if _, err := c.Invoke(t.Context(), nil, map[string]any{
"user_prompt": "hi",
}); err != nil {
t.Fatalf("Invoke: %v", err)
@@ -159,7 +158,7 @@ func TestLLM_Invoke_CiteFalseDisablesInjection(t *testing.T) {
withStubInvoker(t, stub)
c := NewLLMComponent(LLMParam{ModelID: "echo", SystemPrompt: "sys"})
- _, _ = c.Invoke(context.Background(), nil, map[string]any{
+ _, _ = c.Invoke(t.Context(), nil, map[string]any{
"user_prompt": "hi",
"cite": false,
})
diff --git a/internal/agent/component/docs_generator.go b/internal/agent/component/docs_generator.go
index ae75bd39bb..ce01f52441 100644
--- a/internal/agent/component/docs_generator.go
+++ b/internal/agent/component/docs_generator.go
@@ -73,7 +73,7 @@ var validOutputFormats = map[string]bool{
"txt": true,
"markdown": true,
"html": true,
- "md": true, // alias for markdown
+ "md": true, // alias for Markdown
}
// docsGeneratorParam is the static DSL param surface.
@@ -141,7 +141,7 @@ func (p *docsGeneratorParam) Update(conf map[string]any) error {
}
// Check validates the param. FontSize must be ≥ 12; output_format must
-// be one of pdf / docx / txt / markdown / html.
+// be one of pdf / docx / txt / Markdown / html.
func (p *docsGeneratorParam) Check() error {
if !validOutputFormats[strings.ToLower(strings.TrimSpace(p.OutputFormat))] {
return &ParamError{
@@ -510,7 +510,7 @@ func sanitizeFilename(raw, ext string) string {
}
// renderTXT / renderMarkdown / renderHTML live in
-// internal/agent/component/io/{txt,markdown,html}_writer.go so
+// internal/agent/component/io/{txt,Markdown,html}_writer.go so
// every per-format writer lives in its own file under io/,
// matching the pdf_writer.go / docx_writer.go packaging.
diff --git a/internal/agent/component/docs_generator_test.go b/internal/agent/component/docs_generator_test.go
index 6e2443be40..60511ca1c3 100644
--- a/internal/agent/component/docs_generator_test.go
+++ b/internal/agent/component/docs_generator_test.go
@@ -17,7 +17,6 @@
package component
import (
- "context"
"encoding/json"
"strings"
"testing"
@@ -96,7 +95,7 @@ func TestDocsGenerator_Invoke_HappyPath(t *testing.T) {
if err != nil {
t.Fatalf("New(DocsGenerator): %v", err)
}
- _, _ = c.Invoke(context.Background(), nil, map[string]any{})
+ _, _ = c.Invoke(t.Context(), nil, map[string]any{})
// We do not assert err == nil here because txt output requires
// the internal writer (which may not be available in this
// checkout). The test pins that the call doesn't panic.
@@ -110,7 +109,7 @@ func TestDocsGenerator_Invoke_UsesQueryWhenContentEmpty(t *testing.T) {
if err != nil {
t.Fatalf("New(DocsGenerator): %v", err)
}
- out, err := c.Invoke(context.Background(), nil, map[string]any{
+ out, err := c.Invoke(t.Context(), nil, map[string]any{
"query": "Hello from begin query",
})
if err != nil {
@@ -134,10 +133,11 @@ func TestDocsGenerator_Invoke_AcceptsDecorationParamNames(t *testing.T) {
"footer_text": "Canonical Footer",
"watermark_text": "Canonical Watermark",
})
+ ctx := t.Context()
if err != nil {
t.Fatalf("New(DocsGenerator): %v", err)
}
- out, err := c.Invoke(context.Background(), nil, map[string]any{})
+ out, err := c.Invoke(ctx, nil, map[string]any{})
if err != nil {
t.Fatalf("Invoke canonical params: %v", err)
}
@@ -161,7 +161,7 @@ func TestDocsGenerator_Invoke_AcceptsDecorationParamNames(t *testing.T) {
if err != nil {
t.Fatalf("New(DocsGenerator alias params): %v", err)
}
- out, err = c.Invoke(context.Background(), nil, map[string]any{})
+ out, err = c.Invoke(ctx, nil, map[string]any{})
if err != nil {
t.Fatalf("Invoke alias params: %v", err)
}
diff --git a/internal/agent/component/downstream_message_test.go b/internal/agent/component/downstream_message_test.go
index d00e23b609..516964fbd5 100644
--- a/internal/agent/component/downstream_message_test.go
+++ b/internal/agent/component/downstream_message_test.go
@@ -17,7 +17,6 @@
package component
import (
- "context"
"testing"
"time"
)
@@ -48,7 +47,7 @@ func TestLLM_Stream_ExposesThinkingAndContentForDownstreamMessage(t *testing.T)
withStubInvoker(t, stub)
c := NewLLMComponent(LLMParam{ModelID: "echo"})
- ch, err := c.Stream(context.Background(), nil, map[string]any{"user_prompt": "go"})
+ ch, err := c.Stream(t.Context(), nil, map[string]any{"user_prompt": "go"})
if err != nil {
t.Fatalf("Stream: %v", err)
}
diff --git a/internal/agent/component/excel_processor_test.go b/internal/agent/component/excel_processor_test.go
index d4f4664815..82980b012a 100644
--- a/internal/agent/component/excel_processor_test.go
+++ b/internal/agent/component/excel_processor_test.go
@@ -32,7 +32,7 @@ import (
func excelCtx(t *testing.T) context.Context {
t.Helper()
state := canvas.NewCanvasState("run-xlsx", "task-xlsx")
- return canvas.WithState(context.Background(), state)
+ return canvas.WithState(t.Context(), state)
}
// TestExcelProcessor_WriteThenRead: write a 2x2 grid, then read it
diff --git a/internal/agent/component/fillup_test.go b/internal/agent/component/fillup_test.go
index c6c97c5c7d..0bcd1a39af 100644
--- a/internal/agent/component/fillup_test.go
+++ b/internal/agent/component/fillup_test.go
@@ -17,7 +17,6 @@
package component
import (
- "context"
"testing"
"ragflow/internal/agent/canvas"
@@ -30,7 +29,7 @@ import (
func TestFillup_PassesThroughInputs(t *testing.T) {
c, _ := New(componentNameFillup, nil)
state := canvas.NewCanvasState("run-1", "task-1")
- ctx := withStateForTest(context.Background(), state)
+ ctx := withStateForTest(t.Context(), state)
out, err := c.Invoke(ctx, nil, map[string]any{
"inputs": map[string]any{
@@ -60,7 +59,7 @@ func TestFillup_NoTipsKey(t *testing.T) {
"tips": "should be ignored",
})
state := canvas.NewCanvasState("run-2", "task-2")
- ctx := withStateForTest(context.Background(), state)
+ ctx := withStateForTest(t.Context(), state)
out, err := c.Invoke(ctx, nil, map[string]any{
"inputs": map[string]any{
@@ -86,7 +85,7 @@ func TestFillup_NoTipsKey(t *testing.T) {
func TestFillup_NonDictInput(t *testing.T) {
c, _ := New(componentNameFillup, nil)
state := canvas.NewCanvasState("run-3", "task-3")
- ctx := withStateForTest(context.Background(), state)
+ ctx := withStateForTest(t.Context(), state)
out, err := c.Invoke(ctx, nil, map[string]any{
"inputs": map[string]any{
@@ -117,7 +116,7 @@ func TestFillup_NonDictInput(t *testing.T) {
func TestFillup_FileInputStub(t *testing.T) {
c, _ := New(componentNameFillup, nil)
state := canvas.NewCanvasState("run-4", "task-4")
- ctx := withStateForTest(context.Background(), state)
+ ctx := withStateForTest(t.Context(), state)
out, err := c.Invoke(ctx, nil, map[string]any{
"inputs": map[string]any{
diff --git a/internal/agent/component/grounding_test.go b/internal/agent/component/grounding_test.go
index 204c3baa73..69194cf667 100644
--- a/internal/agent/component/grounding_test.go
+++ b/internal/agent/component/grounding_test.go
@@ -77,7 +77,7 @@ func TestGrounding_Applied(t *testing.T) {
state.SetRetrievalChunks([]map[string]any{
{"id": "0", "content": "the source content"},
})
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
c := NewAgentComponent(AgentParam{ModelID: "stub", Cite: true})
out, err := c.Invoke(ctx, nil, map[string]any{"user_prompt": "q"})
@@ -116,7 +116,7 @@ func TestGrounding_NoChunks(t *testing.T) {
state := canvas.NewCanvasState("r1", "t1")
// No SetRetrievalChunks — state has no chunks recorded.
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
c := NewAgentComponent(AgentParam{ModelID: "stub", Cite: true})
out, err := c.Invoke(ctx, nil, map[string]any{"user_prompt": "q"})
@@ -147,7 +147,7 @@ func TestGrounding_CiteFalse(t *testing.T) {
})
c := NewAgentComponent(AgentParam{ModelID: "stub", Cite: false})
- out, err := c.Invoke(context.Background(), nil, map[string]any{"user_prompt": "q"})
+ out, err := c.Invoke(t.Context(), nil, map[string]any{"user_prompt": "q"})
if err != nil {
t.Fatalf("Invoke: %v", err)
}
@@ -173,7 +173,7 @@ func TestGrounding_LLMError(t *testing.T) {
state := canvas.NewCanvasState("r1", "t1")
state.SetRetrievalChunks([]map[string]any{{"id": "0", "content": "x"}})
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
c := NewAgentComponent(AgentParam{ModelID: "stub", Cite: true})
out, err := c.Invoke(ctx, nil, map[string]any{"user_prompt": "q"})
@@ -203,7 +203,7 @@ func TestGrounding_EmptyContent(t *testing.T) {
state := canvas.NewCanvasState("r1", "t1")
state.SetRetrievalChunks([]map[string]any{{"id": "0", "content": "x"}})
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
c := NewAgentComponent(AgentParam{ModelID: "stub", Cite: true})
out, err := c.Invoke(ctx, nil, map[string]any{"user_prompt": "q"})
diff --git a/internal/agent/component/history_window_test.go b/internal/agent/component/history_window_test.go
index aa4c13e528..803caf5de8 100644
--- a/internal/agent/component/history_window_test.go
+++ b/internal/agent/component/history_window_test.go
@@ -17,7 +17,6 @@
package component
import (
- "context"
"testing"
"github.com/cloudwego/eino/schema"
@@ -130,7 +129,7 @@ func TestLLM_Invoke_HistoryWindow_PrependsFromState(t *testing.T) {
UserPrompt: "now",
MessageHistoryWindowSize: window,
})
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
_, err := c.Invoke(ctx, nil, map[string]any{})
if err != nil {
t.Fatalf("Invoke: %v", err)
@@ -166,7 +165,7 @@ func TestLLM_Invoke_HistoryWindow_DoesNotDuplicateCurrentUser(t *testing.T) {
UserPrompt: "now",
MessageHistoryWindowSize: 5,
})
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
if _, err := c.Invoke(ctx, nil, map[string]any{}); err != nil {
t.Fatalf("Invoke: %v", err)
}
@@ -200,7 +199,7 @@ func TestLLM_Invoke_HistoryWindow_ZeroIsNoop(t *testing.T) {
UserPrompt: "now",
// MessageHistoryWindowSize: 0 (default)
})
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
_, _ = c.Invoke(ctx, nil, map[string]any{})
if len(stub.captured.Messages) != 1 {
t.Errorf("expected only 1 message (no history), got %d", len(stub.captured.Messages))
@@ -260,7 +259,7 @@ func TestBuildAgentInputMessagesIncludesPriorConversation(t *testing.T) {
state.AppendHistory("user", "第 8 章下面的第一个副标题是什么?")
state.AppendHistory("assistant", map[string]any{"content": "第一个是解析器工厂。"})
state.AppendCurrentUser("第二个呢?")
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
messages := buildAgentInputMessages(ctx, AgentParam{
SystemPrompt: "请根据知识库回答。",
@@ -291,7 +290,7 @@ func TestBuildAgentInputMessagesUsesConversationTurnWindow(t *testing.T) {
state.AppendHistory("user", "recent question")
state.AppendHistory("assistant", "recent answer")
state.AppendCurrentUser("current question")
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
messages := buildAgentInputMessages(ctx, AgentParam{
UserPrompt: "current question",
@@ -310,7 +309,7 @@ func TestBuildAgentInputMessagesZeroWindowDisablesHistory(t *testing.T) {
state.AppendHistory("user", "ignored question")
state.AppendHistory("assistant", "ignored answer")
state.AppendCurrentUser("current question")
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
messages := buildAgentInputMessages(ctx, AgentParam{UserPrompt: "current question"})
if len(messages) != 1 || messages[0].Content != "current question" {
@@ -322,7 +321,7 @@ func TestBuildAgentInputMessagesReplacesTrailingUserLikePython(t *testing.T) {
state := runtime.NewCanvasState("run-agent-trailing-user", "task-agent-trailing-user")
state.AppendHistory("user", "unanswered previous input")
state.AppendCurrentUser("current question")
- ctx := runtime.WithState(context.Background(), state)
+ ctx := runtime.WithState(t.Context(), state)
messages := buildAgentInputMessages(ctx, AgentParam{
UserPrompt: "current question",
diff --git a/internal/agent/component/invoke_test.go b/internal/agent/component/invoke_test.go
index 13a8c3d956..5c19ee1792 100644
--- a/internal/agent/component/invoke_test.go
+++ b/internal/agent/component/invoke_test.go
@@ -63,7 +63,7 @@ func TestInvoke_GET(t *testing.T) {
defer srv.Close()
c, _ := NewInvokeComponent(nil)
- out, err := c.Invoke(context.Background(), nil, map[string]any{
+ out, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "GET",
"url": srv.URL,
})
@@ -101,7 +101,7 @@ func TestInvoke_POST(t *testing.T) {
defer srv.Close()
c, _ := NewInvokeComponent(nil)
- out, err := c.Invoke(context.Background(), nil, map[string]any{
+ out, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "POST",
"url": srv.URL,
"body": `{"k":"v"}`,
@@ -139,7 +139,7 @@ func TestInvoke_UsesNodeParams(t *testing.T) {
if err != nil {
t.Fatalf("NewInvokeComponent: %v", err)
}
- out, err := c.Invoke(context.Background(), nil, map[string]any{})
+ out, err := c.Invoke(t.Context(), nil, map[string]any{})
if err != nil {
t.Fatalf("Invoke: %v", err)
}
@@ -176,7 +176,7 @@ func TestInvoke_BadMethod(t *testing.T) {
setupAllowAnyHost(t, true)
c, _ := NewInvokeComponent(nil)
- _, err := c.Invoke(context.Background(), nil, map[string]any{
+ _, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "PATCH",
"url": "http://localhost:1",
})
@@ -193,7 +193,7 @@ func TestInvoke_MissingURL(t *testing.T) {
setupAllowAnyHost(t, true)
c, _ := NewInvokeComponent(nil)
- _, err := c.Invoke(context.Background(), nil, map[string]any{
+ _, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "GET",
})
if err == nil {
@@ -222,7 +222,7 @@ func TestInvoke_SSRFGuard_BlocksLoopback(t *testing.T) {
defer srv.Close()
c, _ := NewInvokeComponent(nil)
- out, err := c.Invoke(context.Background(), nil, map[string]any{
+ out, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "GET",
"url": srv.URL,
})
@@ -252,7 +252,7 @@ func TestInvoke_SSRFGuard_BlocksMetadataIP(t *testing.T) {
defer srv.Close()
c, _ := NewInvokeComponent(nil)
- out, err := c.Invoke(context.Background(), nil, map[string]any{
+ out, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "GET",
"url": "http://169.254.169.254/latest/meta-data/",
})
@@ -288,7 +288,7 @@ func TestInvoke_SSRFGuard_BlocksProxy(t *testing.T) {
// (without a side-effect on the local server) is still
// caught. PR review round 5 / duplicate fix from the
// serverHit-only assertion in the earlier review.
- out, err := c.Invoke(context.Background(), nil, map[string]any{
+ out, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "GET",
"url": "https://example.com/api",
"proxy": "http://127.0.0.1:8080",
@@ -320,7 +320,7 @@ func TestInvoke_NoRedirects_NotFollowed(t *testing.T) {
defer srv.Close()
c, _ := NewInvokeComponent(nil)
- out, err := c.Invoke(context.Background(), nil, map[string]any{
+ out, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "GET",
"url": srv.URL,
})
@@ -388,7 +388,7 @@ func TestInvoke_ProxyDNSPin(t *testing.T) {
// resolver below). The SSRF guard validates against
// the validated public IP, then the dialer is
// expected to use that IP — even if the hostname
- // "rebinds" to a different answer afterwards.
+ // "rebinds" to a different answer afterward.
// We achieve "rebinding" by stubbing the DNS lookup
// to return a different IP on a second call.
originalLookup := utility.LookupHost
@@ -401,7 +401,7 @@ func TestInvoke_ProxyDNSPin(t *testing.T) {
t.Cleanup(func() { utility.LookupHost = originalLookup })
c, _ := NewInvokeComponent(nil)
- ctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+ ctx, cancel := context.WithTimeout(t.Context(), 2*time.Second)
defer cancel()
_, err = c.Invoke(ctx, nil, map[string]any{
"method": "GET",
@@ -429,7 +429,7 @@ func TestInvoke_ProxyRejectsHostnameTarget(t *testing.T) {
setupAllowAnyHost(t, false)
c, _ := NewInvokeComponent(nil)
- out, err := c.Invoke(context.Background(), nil, map[string]any{
+ out, err := c.Invoke(t.Context(), nil, map[string]any{
"method": "GET",
"url": "http://example.com/api",
"proxy": "http://proxy.example.invalid:9999",
diff --git a/internal/agent/component/io/markdown_writer.go b/internal/agent/component/io/markdown_writer.go
index cb9ae57077..949c732fd3 100644
--- a/internal/agent/component/io/markdown_writer.go
+++ b/internal/agent/component/io/markdown_writer.go
@@ -16,7 +16,7 @@
// Package io — Markdown writer.
//
-// Round-trips markdown content: header / footer become HTML
+// Round-trips Markdown content: header / footer become HTML
// comments (so they don't affect rendering), and an optional
// front-matter-style timestamp comment goes at the top.
diff --git a/internal/agent/component/io/txt_writer.go b/internal/agent/component/io/txt_writer.go
index 7fa2e3b3c0..4c6a808154 100644
--- a/internal/agent/component/io/txt_writer.go
+++ b/internal/agent/component/io/txt_writer.go
@@ -18,9 +18,9 @@
//
// The TXT writer is the trivial plain-text path: header / footer /
// timestamp are wrapped as plain text lines around the body. No
-// encoding negotiation (UTF-8 only); no markdown stripping
-// (downstream tools that don't understand markdown can consume TXT
-// when the source content is markdown-flavoured — the writer does
+// encoding negotiation (UTF-8 only); no Markdown stripping
+// (downstream tools that don't understand Markdown can consume TXT
+// when the source content is Markdown-flavored — the writer does
// not rewrite the body).
package io
diff --git a/internal/agent/component/list_operations_test.go b/internal/agent/component/list_operations_test.go
index b9c13376bf..1a2e671288 100644
--- a/internal/agent/component/list_operations_test.go
+++ b/internal/agent/component/list_operations_test.go
@@ -17,7 +17,6 @@
package component
import (
- "context"
"reflect"
"sort"
"strings"
@@ -38,7 +37,7 @@ func TestListOperations_Head(t *testing.T) {
}
state := canvas.NewCanvasState("run-1", "task-1")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{1, 2, 3, 4, 5}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -71,7 +70,7 @@ func TestListOperations_TopNLegacyAlias(t *testing.T) {
}
state := canvas.NewCanvasState("run-topn", "task-topn")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{1, 2, 3, 4}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -94,7 +93,7 @@ func TestListOperations_Filter(t *testing.T) {
})
state := canvas.NewCanvasState("run-2", "task-2")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{"foo", "bar", "foobar"}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -119,7 +118,7 @@ func TestListOperations_DropDuplicates(t *testing.T) {
map[string]any{"k": 1},
map[string]any{"k": 2},
}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -146,7 +145,7 @@ func TestListOperations_Tail(t *testing.T) {
})
state := canvas.NewCanvasState("run-4", "task-4")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{1, 2, 3, 4, 5}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -168,7 +167,7 @@ func TestListOperations_NthPositive(t *testing.T) {
})
state := canvas.NewCanvasState("run-5", "task-5")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{"a", "b", "c", "d"}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -190,7 +189,7 @@ func TestListOperations_SortDesc(t *testing.T) {
})
state := canvas.NewCanvasState("run-6", "task-6")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{3, 1, 4, 1, 5, 9, 2, 6}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -227,7 +226,7 @@ func TestListOperations_SortByFieldList(t *testing.T) {
map[string]any{"id": 3, "score": 0.76, "title": "Gamma"},
},
}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
// sort_by="score", desc → Alpha(0.91), Beta(0.88), Gamma(0.76)
cSortDesc, err := NewListOperationsComponent(map[string]any{
@@ -312,7 +311,7 @@ func TestListOperations_NotAList(t *testing.T) {
})
state := canvas.NewCanvasState("run-7", "task-7")
state.Outputs["cpn_0"] = map[string]any{"x": "not-a-list"}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
_, err := c.Invoke(ctx, nil, nil)
if err == nil {
@@ -362,7 +361,7 @@ func TestListOperations_StrictMode_ReturnsError(t *testing.T) {
}
state := canvas.NewCanvasState("run-strict", "task-strict")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{1, 2, 3}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
_, err = c.Invoke(ctx, nil, nil)
if err == nil {
@@ -389,7 +388,7 @@ func TestListOperations_StrictStringCoercion(t *testing.T) {
}
state := canvas.NewCanvasState("run-str-"+v, "task-str-"+v)
state.Outputs["cpn_0"] = map[string]any{"xs": []any{1, 2, 3}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
_, err = c.Invoke(ctx, nil, nil)
if err == nil {
@@ -414,7 +413,7 @@ func TestListOperations_StrictFalseStringsIgnored(t *testing.T) {
}
state := canvas.NewCanvasState("run-strf-"+v, "task-strf-"+v)
state.Outputs["cpn_0"] = map[string]any{"xs": []any{1, 2, 3}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -452,7 +451,7 @@ func TestListOperations_FilterEqBool(t *testing.T) {
}
state := canvas.NewCanvasState("run-feq", "task-feq")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{true, false, true, "True"}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
out, err := c.Invoke(ctx, nil, nil)
if err != nil {
@@ -480,7 +479,7 @@ func TestListOperations_UnknownOp_ReturnsError(t *testing.T) {
}
state := canvas.NewCanvasState("run-bogus", "task-bogus")
state.Outputs["cpn_0"] = map[string]any{"xs": []any{1, 2, 3}}
- ctx := canvas.WithState(context.Background(), state)
+ ctx := canvas.WithState(t.Context(), state)
_, err := c.Invoke(ctx, nil, nil)
if err == nil {
diff --git a/internal/agent/component/llm_retry_test.go b/internal/agent/component/llm_retry_test.go
index c55298e969..8a263ae804 100644
--- a/internal/agent/component/llm_retry_test.go
+++ b/internal/agent/component/llm_retry_test.go
@@ -61,7 +61,7 @@ func TestRetryInvoker_SucceedsOnSecondAttempt(t *testing.T) {
inner := &scriptedInvoker{failTimes: 1, err: errors.New("transient"), resp: want}
r := newRetryInvoker(inner, 3, time.Millisecond)
- resp, err := r.Invoke(context.Background(), nil, ChatInvokeRequest{ModelName: "m"})
+ resp, err := r.Invoke(t.Context(), nil, ChatInvokeRequest{ModelName: "m"})
if err != nil {
t.Fatalf("Invoke: unexpected err: %v", err)
}
@@ -81,7 +81,7 @@ func TestRetryInvoker_ExhaustsRetries(t *testing.T) {
inner := &alwaysFailInvoker{err: sentinel}
r := newRetryInvoker(inner, 3, time.Millisecond)
- _, err := r.Invoke(context.Background(), nil, ChatInvokeRequest{ModelName: "m"})
+ _, err := r.Invoke(t.Context(), nil, ChatInvokeRequest{ModelName: "m"})
if err == nil {
t.Fatal("expected error after exhaustion")
}
@@ -121,7 +121,7 @@ func TestRetryInvoker_HonorsContextCancellation(t *testing.T) {
// honored. The cancellation lands within milliseconds.
r := newRetryInvoker(inner, 5, 30*time.Second)
- ctx, cancel := context.WithCancel(context.Background())
+ ctx, cancel := context.WithCancel(t.Context())
// Cancel after a short delay, while the retry is sleeping
// through the first backoff.
go func() {
@@ -164,7 +164,7 @@ func TestRetryInvoker_ExponentialBackoff(t *testing.T) {
r := newRetryInvoker(inner, 3, initial)
start := time.Now()
- _, _ = r.Invoke(context.Background(), nil, ChatInvokeRequest{ModelName: "m"})
+ _, _ = r.Invoke(t.Context(), nil, ChatInvokeRequest{ModelName: "m"})
elapsed := time.Since(start)
// 20 + 40 + 80 = 140ms of backoff (3 retries, 4 attempts total).
@@ -185,7 +185,7 @@ func TestRetryInvoker_NoRetriesWhenZero(t *testing.T) {
inner := &alwaysFailInvoker{err: errors.New("nope")}
r := newRetryInvoker(inner, 0, 50*time.Millisecond)
- _, err := r.Invoke(context.Background(), nil, ChatInvokeRequest{ModelName: "m"})
+ _, err := r.Invoke(t.Context(), nil, ChatInvokeRequest{ModelName: "m"})
if err == nil {
t.Fatal("expected error")
}
@@ -198,7 +198,7 @@ func TestRetryInvoker_NoRetriesWhenZero(t *testing.T) {
// should not panic when constructed with nil inner.
func TestRetryInvoker_NilInner(t *testing.T) {
r := newRetryInvoker(nil, 3, time.Millisecond)
- _, err := r.Invoke(context.Background(), nil, ChatInvokeRequest{ModelName: "m"})
+ _, err := r.Invoke(t.Context(), nil, ChatInvokeRequest{ModelName: "m"})
if err == nil {
t.Fatal("expected error for nil inner")
}
@@ -220,7 +220,7 @@ func TestLLMParam_RespectsMaxRetries(t *testing.T) {
// zero-value default is 2s, which would make this test slow.
c.param.DelayAfterError = time.Millisecond
- _, err := c.Invoke(context.Background(), nil, map[string]any{"user_prompt": "x"})
+ _, err := c.Invoke(t.Context(), nil, map[string]any{"user_prompt": "x"})
if err == nil {
t.Fatal("expected error from exhausted retries")
}
@@ -249,7 +249,7 @@ func TestLLMParam_ZeroRetriesMeansOneAttempt(t *testing.T) {
// with maxRetries=0.
c.param.DelayAfterError = time.Millisecond
- _, err := c.Invoke(context.Background(), nil, map[string]any{"user_prompt": "x"})
+ _, err := c.Invoke(t.Context(), nil, map[string]any{"user_prompt": "x"})
if err == nil {
t.Fatal("expected error")
}
@@ -280,7 +280,7 @@ func TestLLMParam_DefaultRetries(t *testing.T) {
SetDefaultChatInvoker(newRetryInvoker(fastInner, 2, time.Millisecond))
t.Cleanup(func() { SetDefaultChatInvoker(nil) })
- _, err := c.Invoke(context.Background(), nil, map[string]any{"user_prompt": "x"})
+ _, err := c.Invoke(t.Context(), nil, map[string]any{"user_prompt": "x"})
if err == nil {
t.Fatal("expected error after default retries")
}
@@ -364,7 +364,7 @@ func TestLLM_ParamOverride_AbsoluteCount_NotStacked(t *testing.T) {
// Force a tiny delay so the test runs fast.
c.param.DelayAfterError = time.Millisecond
- _, err := c.Invoke(context.Background(), nil, map[string]any{"user_prompt": "x"})
+ _, err := c.Invoke(t.Context(), nil, map[string]any{"user_prompt": "x"})
if err == nil {
t.Fatal("expected error from exhausted retries")
}
@@ -394,7 +394,7 @@ func TestLLM_NoParamOverride_StackingPreserved(t *testing.T) {
// No param override: MaxRetries=0 AND DelayAfterError=0.
c := NewLLMComponent(LLMParam{ModelID: "m"})
- _, err := c.Invoke(context.Background(), nil, map[string]any{"user_prompt": "x"})
+ _, err := c.Invoke(t.Context(), nil, map[string]any{"user_prompt": "x"})
if err == nil {
t.Fatal("expected error from exhausted retries")
}
diff --git a/internal/agent/component/message.go b/internal/agent/component/message.go
index abcf552ac0..578fe762ef 100644
--- a/internal/agent/component/message.go
+++ b/internal/agent/component/message.go
@@ -21,7 +21,7 @@
// and (optionally) emits the result as a single SSE chunk.
//
// Capabilities:
-// - output_format rendering (html / markdown / plain) via render.go
+// - output_format rendering (html / Markdown / plain) via render.go
// - auto_play → TTS engine dispatch via internal/agent/audio
// - download extraction from inputs (the {doc_id, filename,
// mime_type} walk from Python's _extract_downloads)
diff --git a/internal/agent/component/prompts/citation.go b/internal/agent/component/prompts/citation.go
index 96b26279b7..229ab36603 100644
--- a/internal/agent/component/prompts/citation.go
+++ b/internal/agent/component/prompts/citation.go
@@ -27,7 +27,7 @@ import (
// citationPromptText is the citation-instruction template, mirrored
// from rag/prompts/citation_prompt.md.
//
-// The full markdown (123 lines) is preserved so the LLM receives the
+// The full Markdown (123 lines) is preserved so the LLM receives the
// same instruction set as the Python engine. Trim or extend here only
// after a corresponding change in citation_prompt.md.
//
diff --git a/internal/agent/component/render.go b/internal/agent/component/render.go
index bfdd2f6095..e07f2b9b36 100644
--- a/internal/agent/component/render.go
+++ b/internal/agent/component/render.go
@@ -26,7 +26,7 @@
// become <a href="..." download="...">filename</a>
// - "markdown" — pass through verbatim; downloads become
// [filename](url) links. Python also passes
-// markdown through with light normalization; we
+// Markdown through with light normalization; we
// match that for parity.
// - "plain" — strip HTML tags (best-effort) and present
// downloads as "filename (url)" lines. Default
@@ -173,7 +173,7 @@ func joinPlain(body, dlBlock string) string {
// plain renderer. It removes paired and unpaired tags without
// attempting to keep attribute content. This matches the
// pragmatic behaviour in Python's `_stringify_message_value`
-// fallback path: "if markdown → use as-is, if plain → strip tags".
+// fallback path: "if Markdown → use as-is, if plain → strip tags".
var htmlTagRe = regexp.MustCompile(`<[^>]*>`)
// StripHTMLTags removes HTML tags from s. Used by callers that
diff --git a/internal/agent/component/render_test.go b/internal/agent/component/render_test.go
index a93086fb3f..740f939905 100644
--- a/internal/agent/component/render_test.go
+++ b/internal/agent/component/render_test.go
@@ -63,7 +63,7 @@ func TestRender_HTMLEscapesBody(t *testing.T) {
}
}
-// TestRender_Markdown: markdown passthrough + markdown link list
+// TestRender_Markdown: Markdown passthrough + Markdown link list
// for downloads.
func TestRender_Markdown(t *testing.T) {
got := Render(RenderRequest{
diff --git a/internal/agent/harness/route.go b/internal/agent/harness/route.go
index 9e5ca228c1..dff1219b78 100644
--- a/internal/agent/harness/route.go
+++ b/internal/agent/harness/route.go
@@ -140,7 +140,7 @@ var (
)
// unmarshalModelJSON mirrors Python's _extract_json: strip thinking preamble and
-// markdown fences, then parse JSON.
+// Markdown fences, then parse JSON.
func unmarshalModelJSON(text string, out interface{}) error {
text = reThinkTag.ReplaceAllString(text, "")
text = reFence.ReplaceAllString(text, "")
diff --git a/internal/agent/tool/code_exec.go b/internal/agent/tool/code_exec.go
index ec883be04a..f2336b4f97 100644
--- a/internal/agent/tool/code_exec.go
+++ b/internal/agent/tool/code_exec.go
@@ -70,7 +70,7 @@ type codeExecArgs struct {
// shape mirrors the Python tool's `content` / `_ERROR` / `actual_type`
// fields so downstream nodes can pattern-match unchanged. Artifacts and
// Attachments are surfaced for the model and downstream component
-// consumption (e.g. Message component's artifact markdown formatter).
+// consumption (e.g. Message component's artifact Markdown formatter).
type codeExecResult struct {
Content string `json:"content,omitempty"`
ActualType string `json:"actual_type,omitempty"`
@@ -184,7 +184,7 @@ func (c *CodeExecTool) InvokableRun(ctx context.Context, argumentsInJSON string,
// surface it as `_ARTIFACTS` to match the Python envelope).
// - Metadata["attachments"] → Attachments (rendered into
// downstream Markdown by Message via the same path the Agent
-// tool artifact markdown uses).
+// tool artifact Markdown uses).
//
// Artifacts / Attachments with the wrong element type (anything
// other than map[string]any) are silently dropped with a log
diff --git a/internal/agent/tool/code_exec_test.go b/internal/agent/tool/code_exec_test.go
index 369d34378a..c0e308e503 100644
--- a/internal/agent/tool/code_exec_test.go
+++ b/internal/agent/tool/code_exec_test.go
@@ -159,7 +159,7 @@ func TestCodeExec_Info(t *testing.T) {
// collection: SandboxResponse.Metadata["artifacts"] must be
// surfaced as `_ARTIFACTS` in the tool's JSON envelope so the
// Message
-// component's artifact markdown formatter can render them.
+// component's artifact Markdown formatter can render them.
func TestCodeExec_ResultExtractsArtifacts(t *testing.T) {
t.Parallel()
@@ -222,7 +222,7 @@ func TestCodeExec_ResultDropsBadArtifactShape(t *testing.T) {
}
// TestCodeExec_ResultExtractsAttachments pins the attachments
-// (rendered to downstream Message markdown) path. Distinct from
+// (rendered to downstream Message Markdown) path. Distinct from
// artifacts so renderers can route them differently.
func TestCodeExec_ResultExtractsAttachments(t *testing.T) {
t.Parallel()
diff --git a/internal/agent/tool/wiki_query.go b/internal/agent/tool/wiki_query.go
index bd02c9a165..7b356ad6cd 100644
--- a/internal/agent/tool/wiki_query.go
+++ b/internal/agent/tool/wiki_query.go
@@ -30,7 +30,7 @@ import (
// WikiQueryTool is the wiki_query agent tool (Python harness/tools/exploration.py
// wiki_query). It hybrid-searches the compiled wiki/artifact pages of the bound
-// datasets and returns each page's rendered markdown as a chunk, narrowed by
+// datasets and returns each page's rendered Markdown as a chunk, narrowed by
// keywords. Input keeps query + keywords so the LLM's tool schema matches the
// other search tools.
//
diff --git a/internal/cli/filesystem/skill.go b/internal/cli/filesystem/skill.go
index 9710075e98..4388031026 100644
--- a/internal/cli/filesystem/skill.go
+++ b/internal/cli/filesystem/skill.go
@@ -1606,7 +1606,7 @@ func readSkillFiles(skillPath string) ([]*SkillFile, error) {
return files, err
}
-// parseFrontmatter extracts YAML frontmatter from markdown content
+// parseFrontmatter extracts YAML frontmatter from Markdown content
func parseFrontmatter(content string) (*SkillMetadata, error) {
lines := strings.Split(content, "\n")
diff --git a/internal/common/kg_query_rewrite.go b/internal/common/kg_query_rewrite.go
index 8bee01e596..a37e04b4e3 100644
--- a/internal/common/kg_query_rewrite.go
+++ b/internal/common/kg_query_rewrite.go
@@ -133,7 +133,7 @@ func ParseQueryRewriteResponse(response string) (*QueryRewriteResult, error) {
return result, nil
}
- // Fallback: try to extract JSON from markdown code blocks
+ // Fallback: try to extract JSON from Markdown code blocks
cleaned := strings.TrimSpace(response)
if idx := strings.Index(cleaned, "```"); idx >= 0 {
rest := cleaned[idx+3:]
diff --git a/internal/deepdoc/parser/pdf/layout/boxes_sections.go b/internal/deepdoc/parser/pdf/layout/boxes_sections.go
index 5b5cc73f35..a3424030c7 100644
--- a/internal/deepdoc/parser/pdf/layout/boxes_sections.go
+++ b/internal/deepdoc/parser/pdf/layout/boxes_sections.go
@@ -101,13 +101,13 @@ func NormalizeSectionPositions(sections []pdf.Section) {
}
}
-// SectionsToMarkdown converts Sections to a markdown string.
+// SectionsToMarkdown converts Sections to a Markdown string.
//
// Title sections get a "## " prefix.
// Figure sections produce an "![Image](data:image/png;base64,...)" tag.
// Text and all other sections are appended verbatim.
//
-// This mirrors the Python parser.py:665-671 markdown output path.
+// This mirrors the Python parser.py:665-671 Markdown output path.
func SectionsToMarkdown(sections []pdf.Section) string {
var b strings.Builder
for _, s := range sections {
diff --git a/internal/deepdoc/parser/pdf/parse_cgo.go b/internal/deepdoc/parser/pdf/parse_cgo.go
index 7299ddcae2..d48e9e4e9b 100644
--- a/internal/deepdoc/parser/pdf/parse_cgo.go
+++ b/internal/deepdoc/parser/pdf/parse_cgo.go
@@ -15,7 +15,7 @@ import (
// transfers ownership of it to the returned *pdf.ParseResult (result.Engine is
// populated inside ParseRaw/processPages). The engine is intentionally kept
// alive past Parse so the serialization step can still use it — most
-// importantly the markdown path (cropMarkdownFigures) renders and crops figure
+// importantly the Markdown path (cropMarkdownFigures) renders and crops figure
// images on demand. Parse must therefore NOT close the engine itself.
//
// The caller (the adapter layer: pdfParseResultToJSONWithOptions /
diff --git a/internal/deepdoc/parser/type/types.go b/internal/deepdoc/parser/type/types.go
index 486d331424..5c9f79d56f 100644
--- a/internal/deepdoc/parser/type/types.go
+++ b/internal/deepdoc/parser/type/types.go
@@ -38,12 +38,12 @@ type ParseResult struct {
DLARegions []DLAPageRegions
// Engine is the native PDF backend used to lazily crop section
- // images on demand (e.g. for markdown figure embeds or downstream
+ // images on demand (e.g. for Markdown figure embeds or downstream
// chunk-time cropping). It is populated by ParseRaw and carries
// ownership of the engine: Parse does NOT close the engine, so the
// caller must release it via Close once the result is fully
// serialized. The JSON parse path closes it immediately after
- // serialization; the markdown path crops figure images first, then
+ // serialization; the Markdown path crops figure images first, then
// closes. Close is idempotent and safe to call on a nil result.
Engine PDFEngine
}
diff --git a/internal/engine/elasticsearch/kg_test.go b/internal/engine/elasticsearch/kg_test.go
index 6f83a4fa33..9e062a2c76 100644
--- a/internal/engine/elasticsearch/kg_test.go
+++ b/internal/engine/elasticsearch/kg_test.go
@@ -19,7 +19,6 @@
package elasticsearch
import (
- "context"
"testing"
"ragflow/internal/common"
@@ -51,8 +50,9 @@ func TestKGSearchSelectFields(t *testing.T) {
SelectFields: []string{"entity_kwd", "entity_type_kwd", "rank_flt"},
Limit: 10,
}
+ ctx := t.Context()
- result, err := engine.Search(context.Background(), req)
+ result, err := engine.Search(ctx, req)
if err != nil {
t.Fatalf("search failed: %v", err)
}
diff --git a/internal/entity/models/mineru_local.go b/internal/entity/models/mineru_local.go
index 18545c303a..da18d8cb59 100644
--- a/internal/entity/models/mineru_local.go
+++ b/internal/entity/models/mineru_local.go
@@ -247,7 +247,7 @@ func (m *MinerULocalModel) ShowTask(ctx context.Context, taskID string, apiConfi
return nil, fmt.Errorf("missing results field")
}
- // Get markdown
+ // Get Markdown
for _, fileObj := range results {
fileMap, ok := fileObj.(map[string]interface{})
diff --git a/internal/harness/core/middlewares/skill/doc.go b/internal/harness/core/middlewares/skill/doc.go
index ddcf980a43..b026e762e6 100644
--- a/internal/harness/core/middlewares/skill/doc.go
+++ b/internal/harness/core/middlewares/skill/doc.go
@@ -6,6 +6,6 @@
// - ForkWithContext: skill tools are loaded with context injection
//
// Skills can be loaded from:
-// - FileSystemBackend: read skill definitions from markdown files
+// - FileSystemBackend: read skill definitions from Markdown files
// - Embedded content: inline skill definitions
package skill
diff --git a/internal/ingestion/component/chunker/common.go b/internal/ingestion/component/chunker/common.go
index 69fbfa5b97..da6699e6fc 100644
--- a/internal/ingestion/component/chunker/common.go
+++ b/internal/ingestion/component/chunker/common.go
@@ -90,8 +90,8 @@ func compileDelimPattern(delims []string) *regexp.Regexp {
// (token_chunker.py:79-90). The captured delimiter is DISCARDED rather than
// glued to a segment: re.split with a captured group keeps delimiters at odd
// indices, and only the even-index (text) parts are kept. This is the
-// behaviour every delimiter path (primary and children, text/markdown/html
-// and json) must reproduce so a split chunk reads "first sentence here"
+// behavior every delimiter path (primary and children, text/markdown/html
+// and JSON) must reproduce so a split chunk reads "first sentence here"
// without the trailing delimiter.
func splitDroppingDelim(text string, pattern *regexp.Regexp) []string {
if pattern == nil {
diff --git a/internal/ingestion/component/chunker/custom_delim_test.go b/internal/ingestion/component/chunker/custom_delim_test.go
index 9ed9701b9a..341a8428e1 100644
--- a/internal/ingestion/component/chunker/custom_delim_test.go
+++ b/internal/ingestion/component/chunker/custom_delim_test.go
@@ -5,15 +5,15 @@ import (
"testing"
)
-// custom_delim_test pins the backtick-wrapped newline delimiter behaviour of
+// custom_delim_test pins the backtick-wrapped newline delimiter behavior of
// TokenChunker against Python's rag/flow/chunker/token_chunker.py.
//
-// All delimiter paths (primary and children, text/markdown/html and json)
+// All delimiter paths (primary and children, text/markdown/html and JSON)
// must DROP the captured delimiter from each chunk's text, matching Python's
// _split_text_by_pattern (token_chunker.py:79-90, used by both _build_json_chunks
// and _split_chunk_docs_by_children). Go's splitDroppingDelim reproduces this.
//
-// The json primary path is the one most prone to regress: Python's
+// The JSON primary path is the one most prone to regress: Python's
// _build_json_chunks (token_chunker.py:121) splits each item through
// _split_text_by_pattern, which keeps only the even-index (text) parts and
// DISCARDS the captured delimiter. So a "first segment line one\n" item yields
@@ -119,7 +119,7 @@ func TestCustomDelimMarkdownDropsDelimiter(t *testing.T) {
"markdown": "# Title\n\nParagraph one.\n\nParagraph two.",
}
chunks := invokeTokenChunks(t, params, input)
- // The upstream decode normalizes markdown block boundaries into the
+ // The upstream decode normalizes Markdown block boundaries into the
// backtick-newline delimiter, so the text path must split into exactly
// three trimmed chunks with the delimiter dropped (no trailing newline).
want := []string{"# Title", "Paragraph one.", "Paragraph two."}
diff --git a/internal/ingestion/component/chunker/group.go b/internal/ingestion/component/chunker/group.go
index 51ba9d4691..6c87507300 100644
--- a/internal/ingestion/component/chunker/group.go
+++ b/internal/ingestion/component/chunker/group.go
@@ -378,7 +378,7 @@ func mergePositionMatrix(sources ...json.RawMessage) [][]float64 {
// python BaseTitleChunker.extract_line_records uses:
//
// 1. If upstream emitted chunks (output_format == "chunks") OR
-// upstream emitted JSON, normalise from the list payload.
+// upstream emitted JSON, normalize from the list payload.
// 2. Otherwise, treat text/markdown/html as a "one record per line"
// stream (preserving indentation for non-text formats, strip-only
// for the text format).
diff --git a/internal/ingestion/component/chunker/image_upload.go b/internal/ingestion/component/chunker/image_upload.go
index 4c27f3a366..60a5928460 100644
--- a/internal/ingestion/component/chunker/image_upload.go
+++ b/internal/ingestion/component/chunker/image_upload.go
@@ -122,7 +122,7 @@ func uploadOneImage(ctx context.Context, up component.ImageUploader, kbID, chunk
// decodeChunkImage strips an optional data-URL prefix and base64-decodes the
// payload. Chunker image payloads are "data:image/...;base64,<b64>" (pdfcrop,
-// markdown, docx). A bare base64 string is also accepted.
+// Markdown, docx). A bare base64 string is also accepted.
func decodeChunkImage(s string) ([]byte, error) {
if i := strings.Index(s, ";base64,"); i >= 0 {
s = s[i+len(";base64,"):]
diff --git a/internal/ingestion/component/chunker/title_test.go b/internal/ingestion/component/chunker/title_test.go
index 1ebce286c6..338748fd02 100644
--- a/internal/ingestion/component/chunker/title_test.go
+++ b/internal/ingestion/component/chunker/title_test.go
@@ -106,7 +106,7 @@ func TestTitleChunker_InvokeEmptyInput(t *testing.T) {
}
// TestTitleChunker_Headings_ASCII is the golden-file parity check
-// for the markdown detector: a # heading + body + ## subheading + body
+// for the Markdown detector: a # heading + body + ## subheading + body
// is recognized at least one chunked partition boundary.
//
// Note: the title strategy is dispatched to the underlying strategy
@@ -149,7 +149,7 @@ func TestTitleChunker_Headings_ASCII(t *testing.T) {
}
// TestTitleChunker_NoHeadings_FallsBack feeds plain text without any
-// markdown heading; the chunker should still produce a chunk
+// Markdown heading; the chunker should still produce a chunk
// containing the body text (single chunk fallback).
func TestTitleChunker_NoHeadings_FallsBack(t *testing.T) {
c, err := NewTitleChunker(map[string]any{
diff --git a/internal/ingestion/component/chunker/token.go b/internal/ingestion/component/chunker/token.go
index 9b2b925411..de85340dae 100644
--- a/internal/ingestion/component/chunker/token.go
+++ b/internal/ingestion/component/chunker/token.go
@@ -310,7 +310,7 @@ func cropTitleChunks(ctx context.Context, engine deepdoctype.PDFEngine, chunks [
}
// invokeTextPayload handles plain-text input (output_format in
-// {markdown,text,html} on the python side).
+// {Markdown,text,html} on the python side).
func (c *TokenChunkerComponent) invokeTextPayload(_ context.Context, text string, delimPattern, childrenPattern *regexp.Regexp) map[string]any {
if text == "" {
return emptyOutputs()
diff --git a/internal/ingestion/component/docx_vision_dispatch.go b/internal/ingestion/component/docx_vision_dispatch.go
index 24a71a0297..795ff1bb7e 100644
--- a/internal/ingestion/component/docx_vision_dispatch.go
+++ b/internal/ingestion/component/docx_vision_dispatch.go
@@ -22,8 +22,8 @@
// Unlike the PDF vision path (which replaces dispatchParse entirely),
// DOCX vision is a post-processing step. It mirrors Python exactly:
// vision enrichment happens ONLY on the JSON output path, where each
-// item carries a doc_type_kwd and an optional image. The markdown path
-// performs no vision enrichment in Python, so it must not here either.
+// item carries a doc_type_kwd and an optional image. The Markdown path
+// performs no vision enrichment in Python, so it must not be here either.
package component
@@ -73,8 +73,8 @@ var (
// without an image (e.g. DOCX tables) are left untouched, exactly as Python
// skips them via `if item.get("image") is None: continue`.
//
-// The markdown output path receives no vision enrichment — Python's DOCX
-// markdown branch only concatenates text and never calls the vision model.
+// The Markdown output path receives no vision enrichment — Python's DOCX
+// Markdown branch only concatenates text and never calls the vision model.
func maybeDispatchDOCXVision(
ctx context.Context,
db *gorm.DB,
@@ -141,7 +141,7 @@ func maybeDispatchDOCXVision(
return
}
// DOCX JSON items have no surrounding context (unlike the
- // former markdown path), so use the bare figure prompt —
+ // former Markdown path), so use the bare figure prompt —
// matching Python's VisionFigureParser(context_size=0).
prompt, perr := docxVisionPromptBuilder("", "")
if perr != nil {
diff --git a/internal/ingestion/component/docx_vision_dispatch_test.go b/internal/ingestion/component/docx_vision_dispatch_test.go
index 11d76ab4f3..4b38f74d7f 100644
--- a/internal/ingestion/component/docx_vision_dispatch_test.go
+++ b/internal/ingestion/component/docx_vision_dispatch_test.go
@@ -135,9 +135,9 @@ func TestMaybeDispatchDOCXVision_EnhancesJSONImages(t *testing.T) {
}
}
-// TestMaybeDispatchDOCXVision_JSONOnly verifies Diff 2.4: the markdown output
+// TestMaybeDispatchDOCXVision_JSONOnly verifies Diff 2.4: the Markdown output
// path must NOT be enhanced (Python's markdown/docx branch performs no vision
-// enrichment). A markdown result with embedded figures is returned untouched.
+// enrichment). A Markdown result with embedded figures is returned untouched.
func TestMaybeDispatchDOCXVision_JSONOnly(t *testing.T) {
origResolver := resolveTenantModelByType
origInvoker := visionChatInvoker
diff --git a/internal/ingestion/component/extractor.go b/internal/ingestion/component/extractor.go
index 88f2eb9a89..177b31f8f1 100644
--- a/internal/ingestion/component/extractor.go
+++ b/internal/ingestion/component/extractor.go
@@ -1510,11 +1510,11 @@ func substituteChunkPlaceholders(prompt string, ck map[string]any, chunkText str
// tryParseJSONObject tries to parse s as a JSON object. Returns
// (parsed, true) on success; (nil, false) on parse error or when
-// s is not a JSON object. Trims common markdown code fences
+// s is not a JSON object. Trims common Markdown code fences
// (```json ... ```) before parsing.
func tryParseJSONObject(s string) (map[string]any, bool) {
trimmed := strings.TrimSpace(s)
- // Strip a surrounding markdown code fence. Models commonly wrap JSON in
+ // Strip a surrounding Markdown code fence. Models commonly wrap JSON in
// ```json ... ``` (language tag on the opening line) but some emit the tag
// on its own line (```\njson\n{...}); Python's json_repair tolerates both,
// encoding/json does not, so we drop the fence and any bare leading
diff --git a/internal/ingestion/component/extractor_test.go b/internal/ingestion/component/extractor_test.go
index 221cab3284..d10236d2b2 100644
--- a/internal/ingestion/component/extractor_test.go
+++ b/internal/ingestion/component/extractor_test.go
@@ -317,7 +317,7 @@ func TestExtractorComponent_Invoke_KeepsJSONAsString(t *testing.T) {
}
// TestExtractorComponent_Invoke_KeepsJSONStringInFence verifies a JSON
-// response wrapped in a markdown code fence is stored as the raw string —
+// response wrapped in a Markdown code fence is stored as the raw string —
// the code fence is not stripped on the field-extraction path (Python's
// _generate_async returns the raw text untouched).
func TestExtractorComponent_Invoke_KeepsJSONStringInFence(t *testing.T) {
diff --git a/internal/ingestion/component/knowledge_compiler/mindmap/mindmap.go b/internal/ingestion/component/knowledge_compiler/mindmap/mindmap.go
index eea837dd5d..b643f03f53 100644
--- a/internal/ingestion/component/knowledge_compiler/mindmap/mindmap.go
+++ b/internal/ingestion/component/knowledge_compiler/mindmap/mindmap.go
@@ -6,8 +6,8 @@
// and shaped into the {"id","children"} mind-map tree. The tree emits as one
// product per node with parent links.
//
-// Per PORT_PLAN.md the markdown source is the LLM's reply, NOT the source
-// document markdown, so the Parser component is not reused.
+// Per PORT_PLAN.md the Markdown source is the LLM's reply, NOT the source
+// document Markdown, so the Parser component is not reused.
package mindmap
import (
diff --git a/internal/ingestion/component/knowledge_compiler/wiki/page.go b/internal/ingestion/component/knowledge_compiler/wiki/page.go
index d9e9df45bd..28cc60f203 100644
--- a/internal/ingestion/component/knowledge_compiler/wiki/page.go
+++ b/internal/ingestion/component/knowledge_compiler/wiki/page.go
@@ -81,7 +81,7 @@ func refineFromPlan(ctx context.Context, deps common.Deps, llmID, plan string) (
return resp.Content, nil
}
-// Section is one heading block parsed out of a wiki page's markdown body. The
+// Section is one heading block parsed out of a wiki page's Markdown body. The
// page itself is a separate product; sections live as children linked via
// parent_id so the retrieval-side can scope queries to a section (e.g. user
// asks "what does the page say about X?" and the SIDE picks the page + the X
@@ -92,7 +92,7 @@ type Section struct {
Content string // section body until the next heading of equal-or-lower depth
}
-// ParseOutline splits a wiki page's markdown body into a flat section list.
+// ParseOutline splits a wiki page's Markdown body into a flat section list.
// Headings are ATX style ("#", "##", ...). The page is treated as a single
// implicit "preamble" section when it starts with non-heading text. The page
// itself is *not* in the result; buildPageProducts emits it separately so
@@ -171,7 +171,7 @@ func buildPageProducts(tenantID, docID, page string, sourceChunkIDs []string) []
// buildWikiPageProducts turns multiple wiki page drafts into page + section
// products. Each page result becomes one page product and one section product
-// per heading in its markdown body.
+// per heading in its Markdown body.
func buildWikiPageProducts(tenantID, docID string, pages []wikiPageResult) []common.Product {
if len(pages) == 0 {
appcommon.Info("wiki: buildWikiPageProducts skipped (no refined pages)",
@@ -273,7 +273,7 @@ func buildWikiPageProducts(tenantID, docID string, pages []wikiPageResult) []com
return out
}
-// firstHeading returns the first markdown "# "-prefixed line, trimmed.
+// firstHeading returns the first Markdown "# "-prefixed line, trimmed.
func firstHeading(md string) string {
for _, line := range strings.Split(md, "\n") {
s := strings.TrimSpace(line)
diff --git a/internal/ingestion/component/knowledge_compiler/wiki/wiki.go b/internal/ingestion/component/knowledge_compiler/wiki/wiki.go
index eeaa66779b..1a2097ddd6 100644
--- a/internal/ingestion/component/knowledge_compiler/wiki/wiki.go
+++ b/internal/ingestion/component/knowledge_compiler/wiki/wiki.go
@@ -2307,7 +2307,7 @@ func transformWikiLinks(content, kbID string, pageTitles, slugToPageType map[str
})
// rawLinks counts the model-authored wikilinks in the source content. It
// must be counted against content, not out: by this point every [[...]] has
- // been rewritten to markdown, so counting on out would always report zero.
+ // been rewritten to Markdown, so counting on out would always report zero.
rawLinks := wikiWikilinkSimpleRe.FindAllString(content, -1)
appcommon.Info("knowledge_compiler: transformWikiLinks resolved outlinks",
zap.Int("raw_wikilinks", len(rawLinks)),
diff --git a/internal/ingestion/component/knowledge_compiler/wiki/wiki_refine_test.go b/internal/ingestion/component/knowledge_compiler/wiki/wiki_refine_test.go
index a8bf19bedf..67710cf020 100644
--- a/internal/ingestion/component/knowledge_compiler/wiki/wiki_refine_test.go
+++ b/internal/ingestion/component/knowledge_compiler/wiki/wiki_refine_test.go
@@ -11,7 +11,7 @@ import (
"ragflow/internal/ingestion/component/knowledge_compiler/common"
)
-// refineChatStub returns per-page markdown keyed by the page title in the
+// refineChatStub returns per-page Markdown keyed by the page title in the
// writer prompt so each page's result is distinct and deterministic.
type refineChatStub struct{}
diff --git a/internal/ingestion/component/markdown_vision_dispatch.go b/internal/ingestion/component/markdown_vision_dispatch.go
index 4eff26ec6f..a70a3e846d 100644
--- a/internal/ingestion/component/markdown_vision_dispatch.go
+++ b/internal/ingestion/component/markdown_vision_dispatch.go
@@ -14,13 +14,13 @@
// limitations under the License.
//
-// Markdown vision figure dispatch: enriches parsed markdown JSON
+// Markdown vision figure dispatch: enriches parsed Markdown JSON
// items with LLM-generated descriptions of embedded images,
// mirroring Python's enhance_media_sections_with_vision in
// rag/flow/parser/utils.py, called from the _markdown path.
//
// Unlike the DOCX vision path (which processes a separate figures
-// array), markdown vision iterates over the JSON items produced by
+// array), Markdown vision iterates over the JSON items produced by
// MarkdownParser.ParseWithResult and enhances items whose
// doc_type_kwd == "image" and whose "image" field contains a
// base64-encoded image.
@@ -43,7 +43,7 @@ var (
markdownVisionConcurrency uint = 10
)
-// maybeDispatchMarkdownVision checks whether the markdown parse result
+// maybeDispatchMarkdownVision checks whether the Markdown parse result
// contains JSON items with embedded images and, when a vision model is
// available, enriches those items with AI-generated figure descriptions.
//
diff --git a/internal/ingestion/component/media_dispatch_test.go b/internal/ingestion/component/media_dispatch_test.go
index 61dfdd4bf2..46213b657d 100644
--- a/internal/ingestion/component/media_dispatch_test.go
+++ b/internal/ingestion/component/media_dispatch_test.go
@@ -357,7 +357,7 @@ func TestMaybeDispatchAudio_DefaultOutputFormatJson(t *testing.T) {
}
}
-// TestMaybeDispatchMarkdownVision_EnhancesTables pins diff 2.5: markdown
+// TestMaybeDispatchMarkdownVision_EnhancesTables pins diff 2.5: Markdown
// vision enhancement must also process items whose doc_type_kwd is "table"
// (Python checks {"image","table"} in parser/utils.py:181), not only "image".
// Before the fix the table item was skipped and never sent to the VLM.
diff --git a/internal/ingestion/component/parser.go b/internal/ingestion/component/parser.go
index 8155c2f028..18e3e1f73a 100644
--- a/internal/ingestion/component/parser.go
+++ b/internal/ingestion/component/parser.go
@@ -42,7 +42,7 @@
// - WHAT IS NOT YET PORTED:
//
// - The Python component dispatches to 13 file-format branches
-// (pdf, markdown, text&code, html, spreadsheet, slides, doc,
+// (pdf, Markdown, text&code, html, spreadsheet, slides, doc,
// docx, image, audio, video, email, epub) — see parser.py
// function_map at line ~1273. The Go counterparts in
// internal/parser/parser/ are SKELETONS that print to
@@ -427,7 +427,7 @@ func (c *ParserComponent) Invoke(ctx context.Context, db *gorm.DB, inputs map[st
// off the python family identifiers in schema.ParserParam.
//
// For most families the two forms coincide; the divergence
- // exists for markdown ("md" vs "markdown") and slides
+ // exists for Markdown ("md" vs "markdown") and slides
// ("ppt"/"pptx" vs "slides") and is intentional — the python
// ParserParam collapses the slide family into a single key.
fileTypeExt := fileTypeFromInputs(inputs)
diff --git a/internal/ingestion/component/parser_dispatch.go b/internal/ingestion/component/parser_dispatch.go
index f69ce9637f..50129474f5 100644
--- a/internal/ingestion/component/parser_dispatch.go
+++ b/internal/ingestion/component/parser_dispatch.go
@@ -191,7 +191,7 @@ func dispatchParse(ctx context.Context, fileType utility.FileType, filename stri
//
// 1. inputs["file_type"] — explicit family hint from the upstream
// File component. We accept either the extension ("md", "docx")
-// or the python family name ("markdown"); both are normalised
+// or the python family name ("markdown"); both are normalized
// to the extension form via the pythonFamilyName / familyToExt
// lookup tables below.
// 2. inputs["file"].name — fall back to the filename so a caller
@@ -382,7 +382,7 @@ func pagesFromDispatch(pages []schema.Page) [][]byte {
// - output_format string — the dispatch's OutputFormat,
// or "text" for the raw-text
// fallback
-// - json | markdown | text | html — the dispatched payload on
+// - json | Markdown | text | html — the dispatched payload on
// the matching family key (only
// populated on a structured
// dispatch)
diff --git a/internal/ingestion/component/parser_dispatch_test.go b/internal/ingestion/component/parser_dispatch_test.go
index 3a355682bd..99fd7cb764 100644
--- a/internal/ingestion/component/parser_dispatch_test.go
+++ b/internal/ingestion/component/parser_dispatch_test.go
@@ -61,7 +61,7 @@ func (c *captureSetupConfigurer) ConfigureFromSetup(setup map[string]any) {
func TestDispatch_OutputFormatValidation_Allowed(t *testing.T) {
param := schema.ParserParam{}.Defaults()
setups := defaultSetups()
- // Defaults already include markdown → {text, json}.
+ // Defaults already include Markdown → {text, json}.
c := &ParserComponent{Param: param, Setups: setups}
out, err := c.Invoke(t.Context(), nil, map[string]any{
@@ -100,14 +100,14 @@ func TestDispatch_OutputFormatValidation_Allowed(t *testing.T) {
// TestDispatch_OutputFormatValidation_Rejection pins the
// whitelist enforcement: a request for output_format=html on the
-// markdown family is rejected because markdown's allowed list is
+// Markdown family is rejected because Markdown's allowed list is
// {text, json}. The component must surface this as a hard error
// before any fallback so a misconfigured template cannot silently
// degrade.
func TestDispatch_OutputFormatValidation_Rejection(t *testing.T) {
param := schema.ParserParam{}.Defaults()
setups := defaultSetups()
- // Override the markdown setup to ask for an unsupported format.
+ // Override the Markdown setup to ask for an unsupported format.
// The key is "markdown" (the python-side family identifier),
// NOT "md" — utility.FileTypeMarkdown happens to be the string
// "md" but the setup key is the family name. resolveOutputFormat
@@ -347,7 +347,7 @@ func TestDispatch_PDFMarkdown_UsesConfiguredOutputFormat(t *testing.T) {
if !ok || md == "" {
t.Fatalf("markdown payload missing or empty: %T", out["markdown"])
}
- if _, ok := out["json"]; ok {
+ if _, ok = out["json"]; ok {
t.Fatalf("json payload must be absent for markdown output: %+v", out["json"])
}
}
diff --git a/internal/ingestion/component/schema/chunker.go b/internal/ingestion/component/schema/chunker.go
index 782906b6ca..0e1ee3e629 100644
--- a/internal/ingestion/component/schema/chunker.go
+++ b/internal/ingestion/component/schema/chunker.go
@@ -85,7 +85,7 @@ type ChunkerFromUpstream struct {
// Python). Set when OutputFormat == "json".
JSONResult []ChunkDoc `json:"json,omitempty"`
- // MarkdownResult is the upstream markdown payload (alias "markdown").
+ // MarkdownResult is the upstream Markdown payload (alias "markdown").
// Set when OutputFormat == "markdown".
MarkdownResult *string `json:"markdown,omitempty"`
diff --git a/internal/ingestion/component/schema/parser.go b/internal/ingestion/component/schema/parser.go
index b74f59b4b7..d47ed57df9 100644
--- a/internal/ingestion/component/schema/parser.go
+++ b/internal/ingestion/component/schema/parser.go
@@ -116,7 +116,7 @@ type ParserOutputs struct {
// JSON holds the list of structured sections when output_format == "json".
JSON []map[string]any `json:"json,omitempty"`
- // Markdown holds the rendered markdown when output_format == "markdown".
+ // Markdown holds the rendered Markdown when output_format == "markdown".
Markdown string `json:"markdown,omitempty"`
// Text holds the rendered plain text when output_format == "text".
diff --git a/internal/ingestion/component/schema/schema_test.go b/internal/ingestion/component/schema/schema_test.go
index d9559b0e66..f3f7509e65 100644
--- a/internal/ingestion/component/schema/schema_test.go
+++ b/internal/ingestion/component/schema/schema_test.go
@@ -150,7 +150,7 @@ func TestParserParamJSONRoundTrip(t *testing.T) {
t.Errorf("expected allowed_output_format in JSON, got %s", data)
}
var decoded ParserParam
- if err := json.Unmarshal(data, &decoded); err != nil {
+ if err = json.Unmarshal(data, &decoded); err != nil {
t.Fatalf("unmarshal: %v", err)
}
if got := decoded.AllowedOutputFormat["pdf"]; len(got) != 2 || got[0] != "json" || got[1] != "markdown" {
@@ -364,11 +364,11 @@ func TestTokenizerFromUpstreamValidate(t *testing.T) {
if err := (&TokenizerFromUpstream{OutputFormat: PayloadFormatChunks}).Validate(); err != nil {
t.Fatalf("output_format=chunks should be valid, got %v", err)
}
- // output_format=markdown with no MarkdownResult -> error.
+ // output_format=Markdown with no MarkdownResult -> error.
if err := (&TokenizerFromUpstream{OutputFormat: PayloadFormatMarkdown}).Validate(); err == nil {
t.Fatal("expected error for markdown without payload")
}
- // output_format=markdown with payload -> OK.
+ // output_format=Markdown with payload -> OK.
md := "# title"
if err := (&TokenizerFromUpstream{OutputFormat: PayloadFormatMarkdown, MarkdownResult: &md}).Validate(); err != nil {
t.Fatalf("markdown with payload should be valid, got %v", err)
@@ -434,7 +434,7 @@ func TestTokenizerFromUpstreamJSONRoundTrip(t *testing.T) {
if decoded.MarkdownResult == nil || *decoded.MarkdownResult != "# title" {
t.Errorf("markdown round-trip mismatch: got %v", decoded.MarkdownResult)
}
- // Re-marshal the decoded and confirm we can read markdown back.
+ // Re-marshal the decoded and confirm we can read Markdown back.
data2, err := json.Marshal(decoded)
if err != nil {
t.Fatalf("re-marshal: %v", err)
diff --git a/internal/ingestion/component/schema/tokenizer.go b/internal/ingestion/component/schema/tokenizer.go
index 429189c7f6..7f4e9de562 100644
--- a/internal/ingestion/component/schema/tokenizer.go
+++ b/internal/ingestion/component/schema/tokenizer.go
@@ -60,7 +60,7 @@ type TokenizerFromUpstream struct {
// JSONResult is the upstream structured JSON list (alias "json").
JSONResult []ChunkDoc `json:"json,omitempty"`
- // MarkdownResult is the upstream markdown payload (alias "markdown").
+ // MarkdownResult is the upstream Markdown payload (alias "markdown").
MarkdownResult *string `json:"markdown,omitempty"`
// TextResult is the upstream plain-text payload (alias "text").
diff --git a/internal/ingestion/task/debug_result_dsl_test.go b/internal/ingestion/task/debug_result_dsl_test.go
index 15ea52cd6e..90b812af5f 100644
--- a/internal/ingestion/task/debug_result_dsl_test.go
+++ b/internal/ingestion/task/debug_result_dsl_test.go
@@ -401,7 +401,7 @@ func TestDeepCopy_MapSliceIsDeep(t *testing.T) {
}
// TestDetectFormat_Priority locks the format selection order
-// (chunks > json > text > html > markdown) used to pick a component's payload
+// (chunks > json > text > html > Markdown) used to pick a component's payload
// key. A component that emits multiple recognized keys must surface the
// highest-priority one, matching NormalizeChunks and the front-end tab render.
func TestDetectFormat_Priority(t *testing.T) {
diff --git a/internal/parser/parser/align_test.go b/internal/parser/parser/align_test.go
index d031f4f738..ac4f991c07 100644
--- a/internal/parser/parser/align_test.go
+++ b/internal/parser/parser/align_test.go
@@ -60,7 +60,7 @@ func CollapseWhitespace() Normalizer {
}
// htmlTagRE matches an HTML tag so table/HTML markup can be ignored when
-// comparing content across the two markdown libraries (goldmark vs
+// comparing content across the two Markdown libraries (goldmark vs
// Python-Markdown serialize tables differently but the cell text is the same).
var htmlTagRE = regexp.MustCompile(`(?is)<[^>]+>`)
@@ -68,7 +68,7 @@ var htmlTagRE = regexp.MustCompile(`(?is)<[^>]+>`)
// visible text. Tags are replaced with a single space (not deleted) so
// adjacent cell text does not fuse — e.g. "<td>A</td><td>B</td>" becomes
// "A B" rather than "AB". CollapseWhitespace then folds the extra space.
-// Used so table-markup differences between markdown libraries don't mask the
+// Used so table-markup differences between Markdown libraries don't mask the
// underlying content equivalence.
func StripHTMLTags() Normalizer {
return func(s string) string {
@@ -77,17 +77,17 @@ func StripHTMLTags() Normalizer {
}
// Markdown-syntax regexes removed by StripMarkdownSyntax. The Python flow
-// parser keeps raw markdown in its section text ("# Title", "- item",
+// parser keeps raw Markdown in its section text ("# Title", "- item",
// ``` fenced ```); the Go parser emits clean per-block text. These are
// representation differences (PARSER_ALIGNMENT_HANDOFF.md §3.1), not content
-// divergences, so the markdown alignment strips them before comparing.
+// divergences, so the Markdown alignment strips them before comparing.
var (
mdHeaderRE = regexp.MustCompile(`(?m)^#{1,6}\s+`)
mdListRE = regexp.MustCompile(`(?m)^\s*[-*+]\s+`)
mdFenceRE = regexp.MustCompile("(?s)```[^\n]*\n(.*?)```")
)
-// StripMarkdownSyntax returns a Normalizer that removes markdown presentation
+// StripMarkdownSyntax returns a Normalizer that removes Markdown presentation
// characters (ATX headings, list bullets, fenced-code fences) from a section,
// leaving the bare text. It must run before CollapseWhitespace because the
// fence regex relies on the surrounding newlines.
@@ -203,7 +203,7 @@ func LoadGolden(t *testing.T, path string) []map[string]any {
return items
}
-// MarkdownAlignOptions returns the normalizer preset for markdown. The order
+// MarkdownAlignOptions returns the normalizer preset for Markdown. The order
// matters:
// - StripMarkdownSyntax first: drops "#"/"-"/fenced-code markup that Python
// keeps inline but Go parses out (relies on the surrounding newlines, so it
@@ -218,7 +218,7 @@ func LoadGolden(t *testing.T, path string) []map[string]any {
// inter-tag gaps of an HTML table, the space from delimiter replacement,
// the newlines inside a fenced code block) into single spaces.
//
-// Reused by every markdown alignment test; other formats define their own
+// Reused by every Markdown alignment test; other formats define their own
// preset and share CompareAlignment.
func MarkdownAlignOptions(delimiter string) AlignOptions {
return AlignOptions{
@@ -232,6 +232,6 @@ func MarkdownAlignOptions(delimiter string) AlignOptions {
}
}
-// DefaultMarkdownDelimiter is the flow parser's default markdown delimiter
+// DefaultMarkdownDelimiter is the flow parser's default Markdown delimiter
// set, used when generating/loading the golden baseline.
const DefaultMarkdownDelimiter = "\n!?;。;!?"
diff --git a/internal/parser/parser/docx_ir.go b/internal/parser/parser/docx_ir.go
index 36fae673a8..5e28829f8d 100644
--- a/internal/parser/parser/docx_ir.go
+++ b/internal/parser/parser/docx_ir.go
@@ -36,7 +36,7 @@ type DOCXFigure struct {
Image string `json:"image"` // base64-encoded image bytes
ContextAbove string `json:"context_above"` // text before the image block
ContextBelow string `json:"context_below"` // text after the image block
- Marker string `json:"marker"` // substring to locate image position in markdown
+ Marker string `json:"marker"` // substring to locate image position in Markdown
}
// --- office_oxide IR types (local copy, independent of deepdoc) ---
@@ -349,7 +349,7 @@ func extractDOCXFiguresFromIR(irJSON string) []DOCXFigure {
// Marker: text of the immediately preceding flat block,
// used by the vision dispatcher to locate the image position
- // in the rendered markdown for inline insertion.
+ // in the rendered Markdown for inline insertion.
for j := i - 1; j >= 0; j-- {
if flat[j].text != "" {
fig.Marker = flat[j].text
diff --git a/internal/parser/parser/docx_parser.go b/internal/parser/parser/docx_parser.go
index 4405b17096..e6c0972c53 100644
--- a/internal/parser/parser/docx_parser.go
+++ b/internal/parser/parser/docx_parser.go
@@ -44,7 +44,7 @@ func NewDOCXParser() *DOCXParser {
// ConfigureFromSetup implements parserSetupConfigurer, receiving the
// DSL "docx" family setup map. The output_format key drives whether
-// ParseWithResult produces JSON items (structured) or markdown.
+// ParseWithResult produces JSON items (structured) or Markdown.
func (p *DOCXParser) ConfigureFromSetup(setup map[string]any) {
if p == nil || setup == nil {
return
@@ -120,12 +120,12 @@ func (p *DOCXParser) ParseWithResult(ctx context.Context, filename string, data
}
}
- // Default / markdown path.
+ // Default / Markdown path.
md, err := doc.ToMarkdown()
if err != nil {
return ParseResult{Err: fmt.Errorf("docx to-markdown: %w", err)}
}
- // remove_header_footer on markdown: filter lines by exact match
+ // remove_header_footer on Markdown: filter lines by exact match
// (mirrors Python parser.py:923-926 split lines → filter → rejoin).
if p.RemoveHeaderFooter {
hfTexts := extractDOCXHeaderFooterTexts(data)
@@ -141,8 +141,8 @@ func (p *DOCXParser) ParseWithResult(ctx context.Context, filename string, data
}
md = strings.Join(rebuilt, "\n")
}
- // remove_toc on markdown: split lines, filter, rejoin
- // (mirrors Python parser.py:927-928 remove_toc_word on markdown).
+ // remove_toc on Markdown: split lines, filter, rejoin
+ // (mirrors Python parser.py:927-928 remove_toc_word on Markdown).
if p.RemoveTOC && irErr == nil {
outlines := extractDOCXOutlines(irJSON)
lines := strings.Split(md, "\n")
diff --git a/internal/parser/parser/markdown_parser.go b/internal/parser/parser/markdown_parser.go
index e41897d152..54175449f6 100644
--- a/internal/parser/parser/markdown_parser.go
+++ b/internal/parser/parser/markdown_parser.go
@@ -37,7 +37,7 @@ import (
// dataURIPrefix is the MIME prefix for data URI images.
const dataURIPrefix = "data:image/"
-// GoMarkdown is the lib_type identifier for the pure-Go markdown backend.
+// GoMarkdown is the lib_type identifier for the pure-Go Markdown backend.
const GoMarkdown = "go_markdown"
// ssrfAllowLoopback lets tests exercise the HTTP image fetch path against a
@@ -83,10 +83,10 @@ func (p *MarkdownParser) ConfigureFromSetup(setup map[string]any) {
}
// ParseWithResult implements ParseResultProducer (plan §6.5) and
-// returns a structured markdown payload that mirrors the Python
+// returns a structured Markdown payload that mirrors the Python
// parser's `output_format == "json"` shape. Each top-level block
// emits one item with `text` + `doc_type_kwd: "text"`. When the
-// block contains a markdown image reference (![alt](src)), the image
+// block contains a Markdown image reference (![alt](src)), the image
// data is resolved and the item carries `doc_type_kwd: "image"` with
// the base64-encoded image payload. The legacy debug-print path has
// been removed; callers consume ParseResult directly.
@@ -162,7 +162,7 @@ func renderMarkdownTablesInline(text string) (string, bool) {
// Wrap the inlined <table> HTML in blank lines so gomarkdown
// keeps it as a single HTML block (one item) instead of
// re-parsing it into scattered cell text. See
- // PARSER_ALIGNMENT_HANDOFF.md §3.1 (markdown session A, 方案 Y).
+ // PARSER_ALIGNMENT_HANDOFF.md §3.1 (Markdown session A, 方案 Y).
ensureTrailingBlankLine(&buf)
buf.WriteString(strings.TrimRight(string(tableHTML), "\r\n"))
buf.WriteString("\n\n")
@@ -259,7 +259,7 @@ func markdownTableCells(line string) []string {
// walkMarkdownBlocksWithImages emits one normalized item per
// top-level block. Headings, paragraphs, lists, and code blocks are
-// emitted with their text. When a block contains a markdown image
+// emitted with their text. When a block contains a Markdown image
// reference (![alt](src)), the image data is resolved via
// findBlockImage (per-block AST walk) and the item carries
// `doc_type_kwd: "image"` together with the base64-encoded image
@@ -345,7 +345,7 @@ func walkMarkdownBlocksWithImages(doc ast.Node, out *[]map[string]any, flatten b
item["ck_type"] = ckType
}
- // Resolve markdown images from the AST node of THIS block only, so
+ // Resolve Markdown images from the AST node of THIS block only, so
// the image payload (and doc_type_kwd:"image") is attached to the
// single block that actually contains the ![alt](src) reference.
// Scanning the whole document (the old approach) wrongly tagged
@@ -399,7 +399,7 @@ func findBlockImage(n ast.Node) (string, bool) {
return url, found
}
-// resolveImageURL resolves a markdown image URL to its base64-encoded data.
+// resolveImageURL resolves a Markdown image URL to its base64-encoded data.
// Supports:
// - data:image/... URIs → decoded directly
// - http:// / https:// URLs → fetched (with basic SSRF filtering)
@@ -536,7 +536,7 @@ func resolveAndValidateHost(host string) (net.IP, error) {
return nil, fmt.Errorf("markdown: cannot resolve image host: %s", hostname)
}
for _, addr := range addrs {
- ip := addr.IP
+ ip = addr.IP
if (ip.IsLoopback() && !ssrfAllowLoopback) || ip.IsLinkLocalUnicast() || ip.IsLinkLocalMulticast() ||
ip.IsPrivate() || ip.IsUnspecified() {
return nil, fmt.Errorf("markdown: rejected image URL resolving to internal address: %s (%s)", host, ip)
diff --git a/internal/parser/parser/markdown_parser_test.go b/internal/parser/parser/markdown_parser_test.go
index 34ab5693a3..82891932dd 100644
--- a/internal/parser/parser/markdown_parser_test.go
+++ b/internal/parser/parser/markdown_parser_test.go
@@ -297,7 +297,7 @@ func TestFetchImageAsBase64_InvalidURL(t *testing.T) {
}
// TestMarkdownParser_TableNotCollapsed is the core regression guard for the
-// markdown table fix (方案 Y): a document containing a GFM table must keep one
+// Markdown table fix (方案 Y): a document containing a GFM table must keep one
// item per top-level block instead of collapsing into one giant item, AND the
// table must be emitted as a single raw <table>…</table> HTML block (not
// scattered cell text), appearing both as an inlined text copy and as a
@@ -526,8 +526,8 @@ func TestMarkdownParser_MultipleTablesOrdering(t *testing.T) {
// TestMarkdownParser_AlignmentGolden verifies Go's ParseWithResult output is
// content-equivalent to Python's _markdown on the shared sample, using the
// shared concatenation-normalization alignment tool (align_test.go). Python
-// keeps raw markdown and splits on the delimiter set; Go emits clean per-block
-// text. The comparison normalizes both (markdown syntax, html tags, delimiters
+// keeps raw Markdown and splits on the delimiter set; Go emits clean per-block
+// text. The comparison normalizes both (Markdown syntax, html tags, delimiters
// stripped; whitespace collapsed) and ignores "table"/"image" items, which are
// accepted representation differences (PARSER_ALIGNMENT_HANDOFF.md §3.1).
//
diff --git a/internal/parser/parser/paddleocr.go b/internal/parser/parser/paddleocr.go
index 7ec242a231..633019a225 100644
--- a/internal/parser/parser/paddleocr.go
+++ b/internal/parser/parser/paddleocr.go
@@ -358,7 +358,7 @@ func (c *PaddleOCRClient) extractImageText(resultLines []map[string]any) string
}
content, _ := blockMap["block_content"].(string)
content = strings.TrimSpace(content)
- // Remove markdown image blocks
+ // Remove Markdown image blocks
content = removeMarkdownImages(content)
if content != "" {
texts = append(texts, content)
diff --git a/internal/parser/parser/parse_result.go b/internal/parser/parser/parse_result.go
index 6dc642cdf8..a8b9cedd2a 100644
--- a/internal/parser/parser/parse_result.go
+++ b/internal/parser/parser/parse_result.go
@@ -61,7 +61,7 @@ type ParseResult struct {
// Shape depends on the parser family: PDF emits
// `[]map[string]any` with `text` + `doc_type_kwd` keys (and
// optional `image` / `layout` / `positions` fields);
- // markdown / html / text emit normalized
+ // Markdown / HTML / text emit normalized
// `{text, doc_type_kwd}` items; image emits OCR/VLM result
// items. Exactly one payload family is populated on success.
JSON []map[string]any
diff --git a/internal/parser/parser/pdf_parser_common.go b/internal/parser/parser/pdf_parser_common.go
index 7de46ede7b..05b08d352f 100644
--- a/internal/parser/parser/pdf_parser_common.go
+++ b/internal/parser/parser/pdf_parser_common.go
@@ -473,8 +473,8 @@ func sectionsToMarkdown(sections []deepdoctype.Section) string {
}
// cropMarkdownFigures crops inline images for figure sections so the
-// markdown output can embed them. It mirrors the figure branch of the
-// former Parse.fillSectionImages, but runs only for the markdown path
+// Markdown output can embed them. It mirrors the figure branch of the
+// former Parse.fillSectionImages, but runs only for the Markdown path
// (the JSON path defers cropping to the chunker). The engine is read
// from result.Engine; callers must Close the result afterwards.
func cropMarkdownFigures(result *deepdoctype.ParseResult) {
diff --git a/internal/parser/parser/pdf_postprocess.go b/internal/parser/parser/pdf_postprocess.go
index 969324d21b..7a0ff99b2f 100644
--- a/internal/parser/parser/pdf_postprocess.go
+++ b/internal/parser/parser/pdf_postprocess.go
@@ -76,7 +76,7 @@ func assignPDFDocTypeKeywords(result *deepdoctype.ParseResult, flatten bool) {
default:
// doc_type_kwd is derived from layout, not from whether a
// section image was cropped. Cropping happens lazily at
- // markdown serialization / chunk time, so it must not
+ // Markdown serialization / chunk time, so it must not
// influence classification here (otherwise every positioned
// text box would be mislabeled "image").
section.DocTypeKwd = "text"
diff --git a/internal/parser/parser/pdf_postprocess_test.go b/internal/parser/parser/pdf_postprocess_test.go
index f95bbd2128..dbb62bb12c 100644
--- a/internal/parser/parser/pdf_postprocess_test.go
+++ b/internal/parser/parser/pdf_postprocess_test.go
@@ -50,7 +50,7 @@ func TestApplyPDFPostProcess_AssignsDocTypeKeywords(t *testing.T) {
applyPDFPostProcess(result, pdfPostProcessOptions{})
// doc_type_kwd is derived from layout type only. A pre-set Image no
// longer reclassifies a section as "image" — cropping happens lazily
- // at markdown serialization / chunk time (see pdf_parser_common.go).
+ // at Markdown serialization / chunk time (see pdf_parser_common.go).
want := []string{"table", "image", "text", "text"}
for i, s := range result.Sections {
if s.DocTypeKwd != want[i] {
diff --git a/internal/service/chat_pipeline.go b/internal/service/chat_pipeline.go
index de24f100ec..29d46142e9 100644
--- a/internal/service/chat_pipeline.go
+++ b/internal/service/chat_pipeline.go
@@ -2895,10 +2895,10 @@ func (s *ChatPipelineService) decorateAnswer(
Answer: think + ans,
Reference: refs,
AudioBinary: audioBinary,
- // Fix 7: Apply the markdown line-break substitution
+ // Fix 7: Apply the Markdown line-break substitution
// re.sub(r"\n", " \n", prompt) at the very end, matching
// dialog_service.py:865. This converts single \n to " \n"
- // so multi-line prompt text renders as a single markdown
+ // so multi-line prompt text renders as a single Markdown
// paragraph instead of being broken into separate lines.
Prompt: strings.ReplaceAll(timeStats, "\n", " \n"),
CreatedAt: float64(finishChatTs.Unix()),
diff --git a/internal/service/chat_pipeline_test.go b/internal/service/chat_pipeline_test.go
index b47f974e26..e7c940babc 100644
--- a/internal/service/chat_pipeline_test.go
+++ b/internal/service/chat_pipeline_test.go
@@ -830,7 +830,7 @@ func TestStripISOTimestamps(t *testing.T) {
{"multiple", "x T01:02:03|y T04:05:06|z", "x |y |z"},
{"no space before T", "abcT13:24:55|def", "abc|def"},
{"empty", "", ""},
- // Realistic markdown cell: |2024-01-15T13:24:55| → |2024-01-15|
+ // Realistic Markdown cell: |2024-01-15T13:24:55| → |2024-01-15|
{"realistic cell", "|2024-01-15T13:24:55|", "|2024-01-15|"},
}
for _, tc := range cases {
diff --git a/internal/service/deep_researcher.go b/internal/service/deep_researcher.go
index 4593bd793c..6a8ae3f9e7 100644
--- a/internal/service/deep_researcher.go
+++ b/internal/service/deep_researcher.go
@@ -682,14 +682,14 @@ func (dr *DeepResearcher) chatOnce(
return *resp.Answer, nil
}
-// cleanLLMResponse strips think tags, markdown fences, and trailing backticks.
+// cleanLLMResponse strips think tags, Markdown fences, and trailing backticks.
var thinkTagRe = regexp.MustCompile(`(?s)^.*?</think>`)
var trailingCommaRe = regexp.MustCompile(`,\s*([}\]])`)
var cleanResponseRe = regexp.MustCompile(`(?s)(^.*?</think>|` + "```json\\n" + `|` + "```\\n*$" + `)`)
var trailingBacktickRe = regexp.MustCompile("```\\n*$")
func cleanLLMResponse(raw string) string {
- // Strip think tags, markdown fences
+ // Strip think tags, Markdown fences
raw = cleanResponseRe.ReplaceAllString(raw, "")
// Also handle trailing ```` in case any remain after the regex pass
diff --git a/internal/service/load_prompt.go b/internal/service/load_prompt.go
index 9da2e0e3c0..13df5570dd 100644
--- a/internal/service/load_prompt.go
+++ b/internal/service/load_prompt.go
@@ -49,7 +49,7 @@ var (
// rest of the response invisible to the caller.
var thinkBlockRE = regexp.MustCompile(`^[\s\S]*</think>`)
-// jsonFenceRE matches markdown code fences around JSON responses.
+// jsonFenceRE matches Markdown code fences around JSON responses.
// Mirrors Python's re.sub(r"(`{3}json\n|`{3}\n*$)", ..., flags=re.DOTALL).
// Note: `\n*` is intentionally narrower than Go's `\s*` — Python only
// matches newlines, not other whitespace, so a closing fence followed
diff --git a/internal/syncer/connector/github.go b/internal/syncer/connector/github.go
index ae48aebd24..7984c192c3 100644
--- a/internal/syncer/connector/github.go
+++ b/internal/syncer/connector/github.go
@@ -723,7 +723,7 @@ func configBoolDefault(value any, fallback bool) bool {
return fallback
}
-// sanitizeGitHubName mirrors Python's sanitized markdown filename intent.
+// sanitizeGitHubName mirrors Python's sanitized Markdown filename intent.
func sanitizeGitHubName(name, extension string) string {
name = strings.TrimSpace(name)
if name == "" {
diff --git a/internal/utility/markdown_to_json.go b/internal/utility/markdown_to_json.go
index 7ea56c7ed0..73cccdfe06 100644
--- a/internal/utility/markdown_to_json.go
+++ b/internal/utility/markdown_to_json.go
@@ -104,7 +104,7 @@ const (
otherBlock
)
-// mdBlock is one markdown block in dictify terms: heading (level+text), list
+// mdBlock is one Markdown block in dictify terms: heading (level+text), list
// (flattened items), paragraph/fenced/other (text).
type mdBlock struct {
typ blockType
@@ -117,7 +117,7 @@ type mdBlock struct {
// lines (``` or ```lang) are removed before parsing.
var fenceStripRE = regexp.MustCompile("```[^\\n]*")
-// StripFences removes markdown code-fence marker lines from an LLM reply.
+// StripFences removes Markdown code-fence marker lines from an LLM reply.
func StripFences(s string) string {
return fenceStripRE.ReplaceAllString(s, "")
}
@@ -422,7 +422,7 @@ func toBlock(n ast.Node, src []byte) mdBlock {
}
// headingRawText renders the heading's raw source line without the ATX
-// markers, keeping inline markdown (e.g. asterisks) intact like CommonMark's
+// markers, keeping inline Markdown (e.g. asterisks) intact like CommonMark's
// block.strings (goldmark's inline text would strip emphasis). Both leading
// and trailing '#' runs are removed (deviation #1 fixed: Python's CommonMark
// strips the closing '#' sequence too).
diff --git a/internal/utility/upload.go b/internal/utility/upload.go
index 585dc668b9..ea72cd1473 100644
--- a/internal/utility/upload.go
+++ b/internal/utility/upload.go
@@ -110,7 +110,7 @@ func FetchRemoteFileSafelyWithTimeout(ctx context.Context, rawURL string, maxSiz
// NormalizeUploadInfoContent normalizes an uploaded file's filename, content
// type, and content bytes: detects PDF by magic bytes, converts HTML to
-// readable markdown, and fixes the filename extension.
+// readable Markdown, and fixes the filename extension.
func NormalizeUploadInfoContent(filename, contentType string, data []byte) (string, string, []byte) {
lowerCT := strings.ToLower(strings.TrimSpace(strings.Split(contentType, ";")[0]))
if lowerCT == "" {
diff --git a/tools/gen-component-parity/main.go b/tools/gen-component-parity/main.go
index 8c644f283d..8f897a26ab 100644
--- a/tools/gen-component-parity/main.go
+++ b/tools/gen-component-parity/main.go
@@ -9,7 +9,7 @@
//
// gen-component-parity is a small developer tool that walks the live Go
-// component and tool registries and emits a markdown parity matrix for
+// component and tool registries and emits a Markdown parity matrix for
// docs/component-parity.md. It supersedes the manually-curated table
// previously maintained in that file.
//
@@ -21,7 +21,7 @@
// The tool imports the production packages (which trigger init()
// registration of every component / tool factory) and reflects on each
// registered entry's Name() / Inputs() / Outputs() (for components) or
-// Info() (for tools). It then writes a markdown table to stdout.
+// Info() (for tools). It then writes a Markdown table to stdout.
//
// Limitations:
//
@@ -31,7 +31,7 @@
// the structural inventory.
// - Component factories that require non-empty params to construct
// (e.g. Retrieval, ExeSQL) are skipped with a "⏭" annotation; their
-// parity is captured in the hand-written matrix.
+// parity is captured in the handwritten matrix.
// - The script does not import cmd/server_main.go to avoid pulling
// in the full boot path (Redis, MySQL, etc.).
package main
diff --git a/tools/gen-component-parity/main_test.go b/tools/gen-component-parity/main_test.go
index 7f86f61698..b5ce06a9f4 100644
--- a/tools/gen-component-parity/main_test.go
+++ b/tools/gen-component-parity/main_test.go
@@ -17,7 +17,7 @@ import (
"testing"
)
-// TestWriteMarkdown_SpotCheck pins the markdown output contract: it
+// TestWriteMarkdown_SpotCheck pins the Markdown output contract: it
// must enumerate both universes and surface at least one entry per
// universe. The exhaustive list is owned by RegisteredNames() and
// tool/registry.go — this test just guards against the structural