Merge pull request 'Reuse corpus' (#7) from reuse-corpus into main

Reviewed-on: https://git.ts.mattnite.net/mattnite/cairn/pulls/7

actions/cairn-zig-fuzz/action.yml

@@ -60,6 +60,15 @@ runs:
         set -eu
         command -v zig >/dev/null 2>&1 || { echo "ERROR: zig not found in PATH. Install Zig before using this action (e.g. https://codeberg.org/mlugg/setup-zig@v2)."; exit 1; }
         command -v afl-cc >/dev/null 2>&1 || { echo "ERROR: afl-cc not found in PATH after setup-afl step."; exit 1; }
+        if ! command -v jq >/dev/null 2>&1; then
+          if command -v apt-get >/dev/null 2>&1; then
+            apt-get update -qq
+            apt-get install -y -qq jq >/dev/null 2>&1
+          else
+            echo "ERROR: jq not found and apt-get unavailable"
+            exit 1
+          fi
+        fi
         echo "zig $(zig version), afl-cc found"
 
     - name: Setup Cairn CLI
@@ -86,11 +95,16 @@ runs:
 
         # ── Start Cairn campaign ──
         SHORT_SHA=$(printf '%.8s' "${COMMIT}")
+        if [ -n "${TARGET}" ]; then
+          CAMPAIGN_NAME="fuzz-${TARGET}"
+        else
+          CAMPAIGN_NAME="fuzz-${SHORT_SHA}"
+        fi
         CAMPAIGN_OUTPUT=$(cairn campaign start \
           -server "${CAIRN_SERVER}" \
           -repo "${REPO}" \
           -owner "${OWNER}" \
-          -name "fuzz-${SHORT_SHA}" \
+          -name "${CAMPAIGN_NAME}" \
           -type fuzzing)
         CAMPAIGN_ID="${CAMPAIGN_OUTPUT#Campaign started: }"
         echo "Campaign ${CAMPAIGN_ID} started"
@@ -115,6 +129,18 @@ runs:
           echo "Target ${TARGET_NUM}: zig build ${BUILD_ARGS}"
           echo "=========================================="
 
+          # Special-case Zig fuzz targets: if build args contain -Dfuzz-target=<name>,
+          # use that as the effective Cairn target for metadata/track keying.
+          LINE_FUZZ_TARGET=$(printf '%s' "${BUILD_ARGS}" | sed -n 's/.*-Dfuzz-target=\([^[:space:]]*\).*/\1/p')
+          EFFECTIVE_TARGET="${TARGET}"
+          if [ -n "${LINE_FUZZ_TARGET}" ]; then
+            EFFECTIVE_TARGET="${LINE_FUZZ_TARGET}"
+          fi
+
+          TRACK_SOURCE="${OWNER}/${REPO}|${EFFECTIVE_TARGET}|${BUILD_ARGS}"
+          TRACK_KEY=$(printf '%s' "${TRACK_SOURCE}" | sha256sum | awk '{print $1}')
+          echo "Track key: ${TRACK_KEY}"
+
           # ── Build ──
           rm -rf zig-out
           zig build ${BUILD_ARGS}
@@ -142,11 +168,27 @@ runs:
           echo "Fuzz binary: ${FUZZ_BIN}"
 
           # ── Seed corpus ──
+          SEEDS="afl-seeds-${TARGET_NUM}"
+          rm -rf "${SEEDS}"
+          mkdir -p "${SEEDS}"
+
           if [ -n "${CORPUS_DIR}" ] && [ -d "${CORPUS_DIR}" ]; then
-            SEEDS="${CORPUS_DIR}"
+            cp -a "${CORPUS_DIR}/." "${SEEDS}/" 2>/dev/null || true
-          else
+          fi
-            SEEDS="afl-seeds"
+
-            mkdir -p "${SEEDS}"
+          LATEST_CORPUS_ID=$(curl -fsSL "${CAIRN_SERVER}/api/v1/artifacts?type=fuzz&limit=200" \
+            | jq -r --arg track "${TRACK_KEY}" --arg repo "${REPO}" \
+              '.artifacts[]? | select(.repo_name == $repo and (.metadata.kind // "") == "corpus" and (.metadata.track_key // "") == $track) | .id' \
+            | head -n1)
+          if [ -n "${LATEST_CORPUS_ID}" ] && [ "${LATEST_CORPUS_ID}" != "null" ]; then
+            echo "Downloading prior corpus artifact: ${LATEST_CORPUS_ID}"
+            if cairn download -server "${CAIRN_SERVER}" -id "${LATEST_CORPUS_ID}" -o "prior-corpus-${TARGET_NUM}.tar.gz"; then
+              tar xzf "prior-corpus-${TARGET_NUM}.tar.gz" -C "${SEEDS}" || true
+              rm -f "prior-corpus-${TARGET_NUM}.tar.gz"
+            fi
+          fi
+
+          if [ "$(find "${SEEDS}" -maxdepth 1 -type f | wc -l)" -eq 0 ]; then
             printf 'A' > "${SEEDS}/seed-0"
           fi
 
@@ -192,11 +234,12 @@ runs:
 
               echo "Uploading crash: ${CRASH_NAME}"
               set -- -server "${CAIRN_SERVER}" -repo "${REPO}" -owner "${OWNER}" \
-                -commit "${COMMIT}" -type fuzz -file "${crash_file}" \
+                -commit "${COMMIT}" -campaign-id "${CAMPAIGN_ID}" -type fuzz -file "${crash_file}" \
+                -kind crash -track-key "${TRACK_KEY}" \
                 -crash-message "AFL++ crash (${BUILD_ARGS}): ${CRASH_NAME}"
 
-              if [ -n "${TARGET}" ]; then
+              if [ -n "${EFFECTIVE_TARGET}" ]; then
-                set -- "$@" -target "${TARGET}"
+                set -- "$@" -target "${EFFECTIVE_TARGET}"
               fi
               if [ -n "${SIG}" ]; then
                 set -- "$@" -signal "${SIG}"
@@ -216,9 +259,11 @@ runs:
               tar czf "corpus-${TARGET_NUM}.tar.gz" -C "${QUEUE_DIR}" .
 
               set -- -server "${CAIRN_SERVER}" -repo "${REPO}" -owner "${OWNER}" \
-                -commit "${COMMIT}" -type fuzz -file "corpus-${TARGET_NUM}.tar.gz"
+                -commit "${COMMIT}" -campaign-id "${CAMPAIGN_ID}" -type fuzz \
-              if [ -n "${TARGET}" ]; then
+                -kind corpus -track-key "${TRACK_KEY}" \
-                set -- "$@" -target "${TARGET}"
+                -file "corpus-${TARGET_NUM}.tar.gz"
+              if [ -n "${EFFECTIVE_TARGET}" ]; then
+                set -- "$@" -target "${EFFECTIVE_TARGET}"
               fi
 
               cairn upload "$@"

cmd/cairn/integration_test.go

@@ -133,6 +133,20 @@ func TestCLIUploadAndDownloadRoundTrip(t *testing.T) {
 	serverURL, db, cleanup, _ := setupCLIServer(t, false)
 	defer cleanup()
 
+	if err := cmdCampaign("start", []string{
+		"-server", serverURL,
+		"-repo", "demo",
+		"-owner", "acme",
+		"-name", "seed-campaign",
+		"-type", "fuzz",
+	}); err != nil {
+		t.Fatalf("cmdCampaign start failed: %v", err)
+	}
+	var campaign models.Campaign
+	if err := db.First(&campaign).Error; err != nil {
+		t.Fatalf("querying campaign: %v", err)
+	}
+
 	artifactFile := filepath.Join(t.TempDir(), "artifact.bin")
 	original := []byte("artifact bytes")
 	if err := os.WriteFile(artifactFile, original, 0o644); err != nil {
@@ -144,10 +158,13 @@ func TestCLIUploadAndDownloadRoundTrip(t *testing.T) {
 		"-repo", "demo",
 		"-owner", "acme",
 		"-commit", "abcdef1234567890",
+		"-campaign-id", strconv.FormatUint(uint64(campaign.ID), 10),
 		"-type", "fuzz",
 		"-file", artifactFile,
 		"-crash-message", "boom",
 		"-signal", "11",
+		"-kind", "crash",
+		"-track-key", "track-123",
 	})
 	if err != nil {
 		t.Fatalf("cmdUpload failed: %v", err)
@@ -165,6 +182,15 @@ func TestCLIUploadAndDownloadRoundTrip(t *testing.T) {
 	if got := md["signal"]; got != "11" {
 		t.Fatalf("expected metadata.signal=11, got %#v", got)
 	}
+	if got := md["kind"]; got != "crash" {
+		t.Fatalf("expected metadata.kind=crash, got %#v", got)
+	}
+	if got := md["track_key"]; got != "track-123" {
+		t.Fatalf("expected metadata.track_key=track-123, got %#v", got)
+	}
+	if a.CampaignID == nil || *a.CampaignID != campaign.ID {
+		t.Fatalf("expected campaign_id=%d, got %#v", campaign.ID, a.CampaignID)
+	}
 
 	outFile := filepath.Join(t.TempDir(), "downloaded.bin")
 	if err := cmdDownload([]string{"-server", serverURL, "-id", strconv.FormatUint(uint64(a.ID), 10), "-o", outFile}); err != nil {

cmd/cairn/main.go

@@ -9,6 +9,7 @@ import (
 	"net/http"
 	"os"
 	"path/filepath"
+	"strconv"
 )
 
 var version = "dev"
@@ -69,10 +70,13 @@ Upload flags:
   -repo NAME           Repository name (required)
   -owner OWNER         Repository owner (required)
   -commit SHA          Commit SHA (required)
+  -campaign-id ID      Campaign ID (optional)
   -type TYPE           Artifact type: coredump, fuzz, sanitizer, simulation (required)
   -file PATH           Path to artifact file (required)
   -crash-message MSG   Crash message (optional)
   -stack-trace TRACE   Stack trace text (optional)
+  -kind VALUE          Artifact kind label (optional, stored in metadata)
+  -track-key VALUE     Fuzz track key (optional, stored in metadata)
   -signal VALUE        Crash signal number/name (optional, stored in metadata)
   -seed VALUE          Simulation seed for reproducibility (optional, stored in metadata)
   -target NAME         Target name/platform (optional, stored in metadata)
@@ -85,10 +89,13 @@ func cmdUpload(args []string) error {
 		repo         string
 		owner        string
 		commitSHA    string
+		campaignID   string
 		artifactType string
 		filePath     string
 		crashMessage string
 		stackTrace   string
+		kind         string
+		trackKey     string
 		signal       string
 		seed         string
 		target       string
@@ -108,6 +115,9 @@ func cmdUpload(args []string) error {
 		case "-commit":
 			i++
 			commitSHA = args[i]
+		case "-campaign-id":
+			i++
+			campaignID = args[i]
 		case "-type":
 			i++
 			artifactType = args[i]
@@ -120,6 +130,12 @@ func cmdUpload(args []string) error {
 		case "-stack-trace":
 			i++
 			stackTrace = args[i]
+		case "-kind":
+			i++
+			kind = args[i]
+		case "-track-key":
+			i++
+			trackKey = args[i]
 		case "-signal":
 			i++
 			signal = args[i]
@@ -144,6 +160,13 @@ func cmdUpload(args []string) error {
 		"commit_sha": commitSHA,
 		"type":       artifactType,
 	}
+	if campaignID != "" {
+		id, err := strconv.ParseUint(campaignID, 10, 64)
+		if err != nil {
+			return fmt.Errorf("invalid campaign id: %w", err)
+		}
+		meta["campaign_id"] = id
+	}
 	if crashMessage != "" {
 		meta["crash_message"] = crashMessage
 	}
@@ -151,6 +174,12 @@ func cmdUpload(args []string) error {
 		meta["stack_trace"] = stackTrace
 	}
 	md := map[string]any{}
+	if kind != "" {
+		md["kind"] = kind
+	}
+	if trackKey != "" {
+		md["track_key"] = trackKey
+	}
 	if signal != "" {
 		md["signal"] = signal
 	}

internal/handler/ingest.go

@@ -25,6 +25,7 @@ type IngestRequest struct {
 	Repository   string          `json:"repository"`
 	Owner        string          `json:"owner"`
 	CommitSHA    string          `json:"commit_sha"`
+	CampaignID   *uint           `json:"campaign_id,omitempty"`
 	Type         string          `json:"type"`
 	CrashMessage string          `json:"crash_message,omitempty"`
 	StackTrace   string          `json:"stack_trace,omitempty"`
@@ -93,6 +94,7 @@ func (h *IngestHandler) Create(c *gin.Context) {
 	artifact, err := models.CreateArtifact(ctx, h.DB, models.CreateArtifactParams{
 		RepositoryID: repo.ID,
 		CommitID:     commit.ID,
+		CampaignID:   req.CampaignID,
 		Type:         req.Type,
 		BlobKey:      blobKey,
 		BlobSize:     header.Size,