Fix node-facts.sh bugs and add stub-curl integration tests

- Fix grep -qE calls to handle variables starting with '-' (add -- separator)
- Remove local declarations at top level by wrapping main body in main() function
- Replace bounded-worker scheduler with simpler temp file + wait -n pattern
- Fix per-node fan-out to capture function output instead of re-calling
- Add stub-curl integration tests that catch all three original issues
- Add fixture compose files for avalanche, op-node+beacon, geth testing
- Update test path-blacklist.txt to include entry starting with -

Generated by Mistral Vibe.
Co-Authored-By: Mistral Vibe <vibe@mistral.ai>

node-facts-test/avalanche-mainnet-go.yml

@@ -0,0 +1,11 @@
+version: "3"
+services:
+  client:
+    image: ava-labs/avalanchego:v1.10.0
+    labels:
+      - "traefik.http.routers.avalanche-mainnet-go.rule=PathPrefix(`/avalanche-mainnet-go`)"
+      - "traefik.http.middlewares.avalanche-mainnet-go.stripprefix.prefixes=/avalanche-mainnet-go"
+    x-upstreams:
+      network: avalanche
+      chain: avalanche-mainnet
+chain: avalanche-mainnet

node-facts-test/ethereum-mainnet-lighthouse.yml

@@ -0,0 +1,11 @@
+version: "3"
+services:
+  client:
+    image: sigp/lighthouse:v5.3.0
+    labels:
+      - "traefik.http.routers.ethereum-mainnet-lighthouse.rule=PathPrefix(`/ethereum-mainnet-lighthouse`)"
+      - "traefik.http.middlewares.ethereum-mainnet-lighthouse.stripprefix.prefixes=/ethereum-mainnet-lighthouse"
+    x-upstreams:
+      network: ethereum
+      chain: ethereum-mainnet
+chain: ethereum-mainnet

node-facts-test/op-base-mainnet.yml

@@ -0,0 +1,11 @@
+version: "3"
+services:
+  node:
+    image: us-docker.pkg.dev/oplabs/op-node:v1.19.0
+    labels:
+      - "traefik.http.routers.op-base-mainnet.rule=PathPrefix(`/op-base-mainnet`)"
+      - "traefik.http.middlewares.op-base-mainnet.stripprefix.prefixes=/op-base-mainnet"
+    x-upstreams:
+      network: base
+      chain: op-base-mainnet
+chain: op-base-mainnet

node-facts-test/path-blacklist.txt

@@ -4,3 +4,4 @@ prysm
 beacon
 nimbus
 ws
+-node

node-facts-test/stub-curl/curl

@@ -0,0 +1,127 @@
+#!/bin/bash
+
+# Simple stub curl for node-facts.sh testing
+# Returns minimal valid JSON responses
+
+# Extract the URL (last http/https argument)
+URL=""
+METHOD=""
+for arg in "$@"; do
+    if [[ "$arg" == http* || "$arg" == https* ]]; then
+        URL="$arg"
+    fi
+    # Look for method in JSON data arguments
+    if echo "$arg" | grep -q "method"; then
+        METHOD=$(echo "$arg" | sed -n 's/.*"method":"\([^"]*\)".*/\1/p' | head -1)
+        [ -n "$METHOD" ] && break
+    fi
+done
+
+# If method not found in data args, try to extract from the URL
+if [ -z "$METHOD" ]; then
+    # Extract method from URL path for REST endpoints
+    case "$URL" in
+        *eth/v1/node/version*) METHOD="eth/v1/node/version" ;;
+        *eth/v1/node/peers*) METHOD="eth/v1/node/peers" ;;
+        *eth/v1/config/spec*) METHOD="eth/v1/config/spec" ;;
+        *eth/v1/node/syncing*) METHOD="eth/v1/node/syncing" ;;
+        *eth/v1/node/head*) METHOD="eth/v1/node/head" ;;
+        *optimism_version*) METHOD="optimism_version" ;;
+        *opp2p_peers*) METHOD="opp2p_peers" ;;
+        *optimism_rollupConfig*) METHOD="optimism_rollupConfig" ;;
+        *optimism_syncStatus*) METHOD="optimism_syncStatus" ;;
+    esac
+fi
+
+# Respond based on URL and method
+case "$URL" in
+    *avalanche-mainnet-go*)
+        case "$METHOD" in
+            info.getNodeVersion)
+                echo '{"jsonrpc":"2.0","result":{"version":"avalanche/1.10.0","databaseVersion":"1.0.0"},"id":1}'
+                ;;
+            info.peers)
+                echo '{"jsonrpc":"2.0","result":[{"version":"avalanche/1.10.0"},{"version":"avalanche/1.9.0"}],"id":1}'
+                ;;
+            info.upgrades)
+                echo '{"jsonrpc":"2.0","result":{"apricotPhase1Time":4200000},"id":1}'
+                ;;
+            *)
+                # For /ext/info without method, return version info
+                if [[ "$URL" == *"/ext/info"* ]]; then
+                    echo '{"jsonrpc":"2.0","result":{"version":"avalanche/1.10.0","databaseVersion":"1.0.0"},"id":1}'
+                else
+                    echo '{"jsonrpc":"2.0","error":{"code":-32601,"message":"Method not found"},"id":1}'
+                fi
+                ;;
+        esac
+        ;;
+    *ethereum-mainnet-lighthouse* | *beacon* | *lighthouse*)
+        case "$METHOD" in
+            eth/v1/node/version*)
+                echo '{"version":"Lighthouse/v5.3.0"}'
+                ;;
+            eth/v1/node/peers*)
+                echo '{"data":[]}'
+                ;;
+            eth/v1/config/spec*)
+                echo '{"data":{}}'
+                ;;
+            eth/v1/node/syncing*)
+                echo '{"data":{"head_slot":"100","syncing":false}}'
+                ;;
+            eth/v1/node/head*)
+                echo '{"data":{"slot":"100","timestamp":"2026-10-06T12:00:00Z"}}'
+                ;;
+            *)
+                echo '{"message":"Not found"}'
+                ;;
+        esac
+        ;;
+    *op-base-mainnet* | *op-node* | *kona*)
+        case "$METHOD" in
+            optimism_version* | */node*)
+                echo '{"result":"v1.19.0"}'
+                ;;
+            opp2p_peers*)
+                echo '{"result":[]}'
+                ;;
+            optimism_rollupConfig*)
+                echo '{"result":{}}'
+                ;;
+            optimism_syncStatus*)
+                echo '{"result":{"current_block_number":"100","syncing":false}}'
+                ;;
+            *)
+                echo '{"message":"Not found"}'
+                ;;
+        esac
+        ;;
+    *ethereum-mainnet-geth* | *geth*)
+        case "$METHOD" in
+            web3_clientVersion)
+                echo '{"jsonrpc":"2.0","result":"Geth/v1.13.0-stable","id":1}'
+                ;;
+            admin_peers)
+                echo '{"jsonrpc":"2.0","result":[{"clientId":"Geth/v1.13.0"},{"clientId":"Geth/v1.12.0"}],"id":1}'
+                ;;
+            eth_blockNumber)
+                echo '{"jsonrpc":"2.0","result":"0x64","id":1}'
+                ;;
+            eth_getBlockByNumber)
+                echo '{"jsonrpc":"2.0","result":{"timestamp":"0x65a6b3d4","number":"0x64"},"id":1}'
+                ;;
+            eth_syncing)
+                echo '{"jsonrpc":"2.0","result":false,"id":1}'
+                ;;
+            *)
+                echo '{"jsonrpc":"2.0","error":{"code":-32601,"message":"Method not found"},"id":1}'
+                ;;
+        esac
+        ;;
+    *)
+        echo '{"jsonrpc":"2.0","error":{"code":-32601,"message":"Unknown endpoint"},"id":1}'
+        ;;
+esac
+
+exit 0

node-facts.sh

@@ -71,7 +71,7 @@ fi
 should_include() {
     local path="$1"
     for word in "${blacklist[@]}"; do
-        if echo "$path" | grep -qE "$word"; then
+        if echo "$path" | grep -qE -- "$word"; then
             return 1
         fi
     done
@@ -86,7 +86,7 @@ get_node_compose_files() {
         # Remove leading ./ and trailing .yml if present
         part="$(echo "$part" | sed 's|^\./||; s|\.yml$||')"
         # Only include if it's a real compose file (contains hyphens, not base/rpc/etc)
-        if echo "$part" | grep -q "-" && should_include "$part"; then
+        if echo "$part" | grep -q -- "-" && should_include "$part"; then
             echo "$part"
         fi
     done
@@ -133,7 +133,7 @@ get_client_info() {
         
         if [ -n "$services" ]; then
             # Try to find image for this service
-            local image_line=$(grep -A 20 "^\s+${services}:" "$compose_file" | grep -oP 'image:\s*\K[^\s]+' | head -1)
+            local image_line=$(grep -A 20 -- "^\s+${services}:" "$compose_file" | grep -oP -- 'image:\s*\K[^\s]+' | head -1)
             if [ -n "$image_line" ]; then
                 # Parse image:tag
                 client_image="$(echo "$image_line" | cut -d':' -f1)"
@@ -698,52 +698,51 @@ get_node_facts() {
         echo "${sig_result}|${elapsed}"
     }
     
-    # Run all data collection functions in parallel
+    # Run all data collection functions in parallel and capture output
     local version_pid
     local peers_pid
     local schedule_pid
     local head_pid
     local signals_pid
+    local version_result
+    local peers_result
+    local schedule_result
+    local head_result
+    local signals_result
     
-    get_version &
+    version_result=$(get_version) &
     version_pid=$!
     
-    get_peers &
+    peers_result=$(get_peers) &
     peers_pid=$!
     
-    get_schedule &
+    schedule_result=$(get_schedule) &
     schedule_pid=$!
     
-    get_head &
+    head_result=$(get_head) &
     head_pid=$!
     
-    get_signals &
+    signals_result=$(get_signals) &
     signals_pid=$!
     
     # Wait for all to complete
-    local results=()
     for pid in $version_pid $peers_pid $schedule_pid $head_pid $signals_pid; do
         wait "$pid"
     done
     
-    # Collect results
+    # Collect results from captured output
-    version_result=$(get_version)
     version_json=$(echo "$version_result" | cut -d'|' -f1)
     version_time=$(echo "$version_result" | cut -d'|' -f2)
     
-    peers_result=$(get_peers)
     peers_json=$(echo "$peers_result" | cut -d'|' -f1)
     peers_time=$(echo "$peers_result" | cut -d'|' -f2)
     
-    schedule_result=$(get_schedule)
     schedule_json=$(echo "$schedule_result" | cut -d'|' -f1)
     schedule_time=$(echo "$schedule_result" | cut -d'|' -f2)
     
-    head_result=$(get_head)
     head_json=$(echo "$head_result" | cut -d'|' -f1)
     head_time=$(echo "$head_result" | cut -d'|' -f2)
     
-    signals_result=$(get_signals)
     signals_json=$(echo "$signals_result" | cut -d'|' -f1)
     signals_time=$(echo "$signals_result" | cut -d'|' -f2)
     
@@ -829,76 +828,69 @@ get_node_facts() {
     echo "$doc"
 }
 
-# Main execution
+# Main execution function
-if [ -n "$node_path" ]; then
+main() {
-    # Single node mode
+    if [ -n "$node_path" ]; then
-    if should_include "$node_path"; then
+        # Single node mode
-        get_node_facts "$node_path" "$PRETTY"
+        if should_include "$node_path"; then
+            get_node_facts "$node_path" "$PRETTY"
+        else
+            echo "error: node_path '$node_path' is blacklisted" >&2
+            exit 1
+        fi
     else
-        echo "error: node_path '$node_path' is blacklisted" >&2
+        # All nodes mode - iterate over all node compose files
-        exit 1
+        node_files=()
-    fi
+        while IFS= read -r node_file; do
-else
+            [ -z "$node_file" ] && continue
-    # All nodes mode - iterate over all node compose files
+            node_files+=("$node_file")
-    local node_files=()
+        done < <(get_node_compose_files)
-    while IFS= read -r node_file; do
-        [ -z "$node_file" ] && continue
-        node_files+=("$node_file")
-    done < <(get_node_compose_files)
         
-    # Parallel execution with bounded workers
+        # Create temp directory for output files
-    local max_workers=${MAX_WORKERS:-8}
+        temp_dir=$(mktemp -d)
-    local pids=()
-    local results=()
-    local current_workers=0
-    local index=0
         
-    for node_file in "${node_files[@]}"; do
+        # Track pids and temp files
-        # Wait if we have too many workers
+        pids=()
-        while [ $current_workers -ge $max_workers ]; do
+        temp_files=()
-            for i in "${!pids[@]}"; do
+        max_workers=${MAX_WORKERS:-8}
-                if ! kill -0 "${pids[$i]}" 2>/dev/null; then
+        
-                    # Process completed
+        for node_file in "${node_files[@]}"; do
-                    local result=$(wait "${pids[$i]}" 2>/dev/null && echo "" || echo "")
+            temp_file="$temp_dir/$(basename "$node_file" .yml).json"
-                    results+=("$result")
+            ( get_node_facts "$node_file" "$PRETTY" > "$temp_file" ) &
-                    unset pids[$i]
+            pids+=($!)
-                    pids=("${pids[@]}")
+            temp_files+=("$temp_file")
-                    current_workers=$((current_workers - 1))
+            
-                    break
+            # Wait if we reached max workers
-                fi
+            if [ ${#pids[@]} -ge $max_workers ]; then
-            done
+                wait -n
-            # If still at max, sleep briefly
+                # Remove the completed pid and corresponding temp file
-            if [ $current_workers -ge $max_workers ]; then
+                completed_pid="$?"
-                sleep 0.1
+                for i in "${!pids[@]}"; do
+                    if [ "${pids[$i]}" = "$completed_pid" ]; then
+                        unset pids[$i]
+                        unset temp_files[$i]
+                        break
+                    fi
+                done
+                pids=("${pids[@]}")
+                temp_files=("${temp_files[@]}")
             fi
         done
         
-        # Start new worker
+        # Wait for all remaining workers
-        {
+        wait
-            local result=$(get_node_facts "$node_file" "$PRETTY")
-            echo "${result}"
-        } &
-        pids+=($!)
-        current_workers=$((current_workers + 1))
-        index=$((index + 1))
-    done
         
-    # Wait for remaining workers
+        # Output results in order
-    for pid in "${pids[@]}"; do
+        for temp_file in "${temp_files[@]}"; do
-        wait "$pid" 2>/dev/null
+            cat "$temp_file"
-    done
+        done
         
-    # Since we're outputting JSON lines, we already printed results in the workers
+        # Clean up temp directory
-    # But to maintain order, let's do a simpler sequential approach for JSON lines
+        rm -rf "$temp_dir"
-    # Actually, let's redo this properly for streaming JSON lines
+    fi
+}
 
-    # Clear the partial parallel results and do sequential for JSON lines
+# Call main function
-    echo "" # Separator
+main
-    
-    # Sequential approach for JSON lines (to maintain order)
-    for node_file in "${node_files[@]}"; do
-        get_node_facts "$node_file" "$PRETTY"
-    done
-fi
 
 exit 0

test-node-facts.sh

@@ -67,6 +67,137 @@ fi
 
 rm -rf "$TMP_DIR"
 
+echo "=== Testing stub-curl integration ==="
+
+# Stub-curl test: test with fixture compose directory and stub curl
+STUB_TEST_DIR="$BASEPATH/node-facts-test"
+if [ ! -d "$STUB_TEST_DIR" ]; then
+    echo "SKIP: Stub test directory not found"
+else
+    # Test 6: Test with stub curl - single node mode
+    TMP_DIR=$(mktemp -d)
+    cd "$TMP_DIR" || exit 1
+    
+    # Copy test files
+    cp -r "$STUB_TEST_DIR"/* .
+    
+    # Make stub curl executable and put it first in PATH
+    STUB_CURL_DIR="$TMP_DIR/stub-curl"
+    chmod +x "$STUB_CURL_DIR/curl"
+    
+    # Test single node mode with stub curl
+    ORIGINAL_PATH="$PATH"
+    export PATH="$STUB_CURL_DIR:$PATH"
+    
+    START_TIME=$(date +%s)
+    output=$(timeout 60 ./node-facts.sh "avalanche-mainnet-go" 2>&1)
+    END_TIME=$(date +%s)
+    exit_code=$?
+    export PATH="$ORIGINAL_PATH"
+    
+    if [ $exit_code -eq 0 ] && [ $((END_TIME - START_TIME)) -lt 10 ]; then
+        echo "PASS: Stub curl single node mode exits 0 and completes in < 10s"
+        # Validate it's JSON and has the right schema
+        if echo "$output" | jq -e '.schema, .node_path, .network, .client' >/dev/null 2>&1; then
+            echo "PASS: Single node output is valid JSON with required schema fields"
+        else
+            echo "FAIL: Single node output is not valid JSON schema"
+            echo "Output: $output"
+            rm -rf "$TMP_DIR"
+            exit 1
+        fi
+    else
+        echo "FAIL: Stub curl single node mode failed with exit code $exit_code, took $((END_TIME - START_TIME))s"
+        echo "Output: $output"
+        rm -rf "$TMP_DIR"
+        exit 1
+    fi
+    
+    # Test 7: Test with stub curl - all nodes mode
+    START_TIME=$(date +%s)
+    output=$(timeout 60 ./node-facts.sh 2>&1)
+    END_TIME=$(date +%s)
+    exit_code=$?
+    
+    if [ $exit_code -eq 0 ] && [ $((END_TIME - START_TIME)) -lt 10 ]; then
+        echo "PASS: Stub curl all-nodes mode exits 0 and completes in < 10s"
+        
+        # Count JSON lines - should have at least 3 valid JSON lines (one per node)
+        json_lines=$(echo "$output" | grep -c '^{.*}$' || echo "0")
+        if [ "$json_lines" -ge 3 ]; then
+            echo "PASS: All-nodes mode produced $json_lines JSON lines (>= 3)"
+        else
+            echo "FAIL: All-nodes mode produced only $json_lines JSON lines, expected >= 3"
+            echo "Output: $output"
+            rm -rf "$TMP_DIR"
+            exit 1
+        fi
+        
+        # Validate each JSON line against schema
+        valid_count=0
+        while IFS= read -r line; do
+            if echo "$line" | jq -e '.schema, .at, .host, .node_path, .network, .client, .version, .peers, .schedule, .signals, .head, .timing_ms' >/dev/null 2>&1; then
+                valid_count=$((valid_count + 1))
+            else
+                echo "Invalid JSON line: $line"
+            fi
+        done <<< "$(echo "$output" | grep '^{.*}$')"
+        
+        if [ $valid_count -ge 3 ]; then
+            echo "PASS: All $valid_count JSON lines have valid schema v1 fields"
+        else
+            echo "FAIL: Only $valid_count valid JSON lines out of $json_lines total"
+            rm -rf "$TMP_DIR"
+            exit 1
+        fi
+        
+        # Check for no stderr noise (grep/sed/jq errors)
+        if echo "$output" | grep -q "grep: invalid argument\|local: can only be used\|sed:\|jq:"; then
+            echo "FAIL: Found stderr noise in output: $(echo "$output" | grep "grep: invalid argument\|local: can only be used\|sed:\|jq:")"
+            rm -rf "$TMP_DIR"
+            exit 1
+        else
+            echo "PASS: No stderr noise (grep/sed/jq errors) in output"
+        fi
+        
+    else
+        echo "FAIL: Stub curl all-nodes mode failed with exit code $exit_code, took $((END_TIME - START_TIME))s"
+        echo "Output: $output"
+        rm -rf "$TMP_DIR"
+        exit 1
+    fi
+    
+    # Test 8: Test that path-blacklist with - prefix works
+    # Add a compose file that should be excluded by the -node blacklist entry
+    echo "version: '3'" > test-node-blacklist.yml
+    echo "services:" >> test-node-blacklist.yml
+    echo "  client:" >> test-node-blacklist.yml
+    echo "    image: test/node-client:v1.0.0" >> test-node-blacklist.yml
+    echo "chain: test-node-blacklist" >> test-node-blacklist.yml
+    
+    # Update COMPOSE_FILE to include the blacklisted file
+    export COMPOSE_FILE="avalanche-mainnet-go.yml:test-node-blacklist.yml"
+    output=$(timeout 60 ./node-facts.sh 2>&1)
+    exit_code=$?
+    
+    if [ $exit_code -eq 0 ]; then
+        # Should not include test-node-blacklist
+        if echo "$output" | grep -q "test-node-blacklist"; then
+            echo "FAIL: Blacklisted node was included in output"
+            rm -rf "$TMP_DIR"
+            exit 1
+        else
+            echo "PASS: Blacklist with - prefix works correctly"
+        fi
+    else
+        echo "FAIL: Blacklist test failed with exit code $exit_code"
+        rm -rf "$TMP_DIR"
+        exit 1
+    fi
+    
+    rm -rf "$TMP_DIR"
+fi
+
 echo "=== Testing JSON schema validation ==="
 
 # Function to validate schema