Support duckdb (#22)

* support embedded ducdb with fixed version

* aupport test different version of duckdb by modifying pom.xml

* add customzied dockerfile to duckdb/

* add customized dckerfile

* return to original dockerfile

* modify config

Author: CelestaLiu

build.py

@@ -4,10 +4,29 @@
 import logging
 from utils import run_command
 
-def build_sqlancer_image(script_log, docker_log, force_rebuild=False):
+def build_sqlancer_image(script_log, docker_log, embedded, dbms, force_rebuild=False):
+    context_dir = "./sqlancer"
+    if embedded == "yes":
+        dockerfile_path = f"./{dbms}/Dockerfile"
+        script_log.info(f"Using embedded DBMS Dockerfile: {dockerfile_path}")
+        script_log.info("Embedded mode: rebuilding SQLancer image unconditionally ...")
+        run_command(
+            ["docker", "build", "--no-cache", "-f", dockerfile_path, "-t", "sqlancer:latest", context_dir],
+             docker_log
+        )
+        script_log.info("SQLancer image built: sqlancer:latest")
+        return
+    else:
+        dockerfile_path = "./sqlancer/Dockerfile"
+        context_dir = "./sqlancer"
+        script_log.info(f"Using default SQLancer Dockerfile: {dockerfile_path}")
+
     if force_rebuild:
         script_log.info("Rebuilding SQLancer image: sqlancer:latest ...")
-        run_command(["docker", "build", "--no-cache", "-t", "sqlancer:latest", "./sqlancer"], docker_log)
+         run_command(
+            ["docker", "build", "--no-cache", "-f", dockerfile_path, "-t", "sqlancer:latest", context_dir],
+            docker_log
+        )
         script_log.info("SQLancer image built: sqlancer:latest")
         return
 
@@ -19,10 +38,16 @@ def build_sqlancer_image(script_log, docker_log, force_rebuild=False):
     if "sqlancer:latest" in images:
         script_log.info("SQLancer image exists: sqlancer:latest")
     else:
-        script_log.info("Building SQLancer image from cache: sqlancer:latest...")
-        run_command(["docker", "build", "-t", "sqlancer:latest", "./sqlancer"], docker_log)
+        script_log.info("Building SQLancer image from cache: sqlancer:latest ...")
+        run_command(
+            ["docker", "build", "-f", dockerfile_path, "-t", "sqlancer:latest", context_dir],
+            docker_log
+        )
         script_log.info("SQLancer image built: sqlancer:latest")
 
+
+
+
 def build_network(script_log, docker_log, network_name="sqlancer-net"):
     try:
         output = subprocess.check_output([
@@ -63,7 +88,7 @@ def build_db_image(cfg, use_cache, script_log, docker_log, custom=False, dockerf
 def build_environment(cfg, use_cache, script_log, docker_log, custom=False, dockerfile_path=""):
     script_log.info("==============================Building environment==============================")
     build_network(script_log, docker_log)
-    build_sqlancer_image(script_log, docker_log, force_rebuild=not use_cache)
+    build_sqlancer_image(script_log, docker_log, cfg["embedded"], cfg["dbms"], force_rebuild=False)
     if cfg["embedded"] == "no":
         build_db_image(cfg, use_cache, script_log, docker_log, custom, dockerfile_path)
     script_log.info("==============================Building environment==============================")

config.json

@@ -1,5 +1,5 @@
 {
-  "dbms_list": ["mysql", "postgres", "sqlite", "tidb", "cockroachdb"],
+  "dbms_list": ["mysql", "postgres", "sqlite", "tidb", "cockroachdb", "duckdb"],
   "dbms": "mysql",
   "container_name": "mysql-custom",
   "image": "mysql-custom",

duckdb/Dockerfile

@@ -0,0 +1,20 @@
+FROM ubuntu:20.04
+ENV DEBIAN_FRONTEND=noninteractive
+RUN apt-get update && \
+    apt-get install -y openjdk-17-jdk maven git jq netcat mysql-client && \
+    rm -rf /var/lib/apt/lists/*
+
+WORKDIR /root/sqlancer
+
+ARG DUCKDB_JDBC_VERSION=1.2.2.0 
+
+RUN git clone https://github.com/sqlancer/sqlancer.git . && \
+    mvn -q versions:use-dep-version \
+        -Dincludes=org.duckdb:duckdb_jdbc \
+        -DdepVersion=${DUCKDB_JDBC_VERSION} \
+        -DforceVersion=true && \
+    mvn -q clean package -DskipTests
+
+COPY entrypoint.sh /root/entrypoint.sh
+RUN chmod +x /root/entrypoint.sh
+CMD ["/root/entrypoint.sh"]

duckdb/__init__.py

@@ -0,0 +1,10 @@
+{
+  "embedded": "yes",
+  "dbms": "duckdb",
+  "username": "N/A",
+  "password": "N/A",
+  "container_name": "localhost",
+  "oracle": "NoREC",
+  "num_threads": 4,
+  "timeout_seconds": 60
+}

duckdb/config.json

@@ -16,4 +16,3 @@ RUN chmod +x /root/entrypoint.sh
 
 CMD ["/root/entrypoint.sh"]
 
-

sqlancer/Dockerfile

@@ -18,12 +18,22 @@ for i in {1..60}; do
   sleep 1
 done
 
-CMD="java -jar sqlancer-*.jar --num-threads \"$SQLANCER_THREADS\" --timeout-seconds \"$SQLANCER_TIMEOUT\" --username \"$SQLANCER_USERNAME\" --password \"$SQLANCER_PASSWORD\" --host \"$SQLANCER_HOST\" \"$SQLANCER_DBMS\" --oracle \"$SQLANCER_ORACLE\""
 
-# Show command to console + log file
+CMD="java -jar sqlancer-*.jar --num-threads \"$SQLANCER_THREADS\" --timeout-seconds \"$SQLANCER_TIMEOUT\" --host \"$SQLANCER_HOST\" \"$SQLANCER_DBMS\" --oracle \"$SQLANCER_ORACLE\""
+
+
+if [ "$(printf '%s' "${SQLANCER_USERNAME:-}" | tr '[:lower:]' '[:upper:]')" != "N/A" ]; then
+  CMD="$CMD --username \"$SQLANCER_USERNAME\""
+fi
+
+
+if [ "$(printf '%s' "${SQLANCER_PASSWORD:-}" | tr '[:lower:]' '[:upper:]')" != "N/A" ]; then
+  CMD="$CMD --password \"$SQLANCER_PASSWORD\""
+fi
+
+
 echo "Running: $CMD" | tee -a "$LOG_FILE"
 
-# Run command: log everything, show only stats to console
 eval "$CMD" 2>&1 | tee -a "$LOG_FILE" | grep --line-buffered "Executed"
 
 echo "[INFO] SQLancer finished. Logs saved to $LOG_DIR"