Hide symbol to reduce package size

CMakeLists.txt

@@ -2,6 +2,11 @@ cmake_minimum_required(VERSION 3.25)
 
 project(ClickHouse LANGUAGES C CXX ASM)
 
+# chDB: set visibility to hidden for smaller chDB package
+set(CMAKE_CXX_VISIBILITY_PRESET hidden)
+set(CMAKE_C_VISIBILITY_PRESET hidden)
+set(CMAKE_VISIBILITY_INLINES_HIDDEN ON)
+
 # If turned off: e.g. when ENABLE_FOO is ON, but FOO tool was not found, the CMake will continue.
 option(FAIL_ON_UNSUPPORTED_OPTIONS_COMBINATION
    "Stop/Fail CMake configuration if some ENABLE_XXX option is defined (either ON or OFF)

chdb/build.sh

@@ -22,7 +22,6 @@ if [ "$(uname)" == "Darwin" ]; then
     GLIBC_COMPATIBILITY="-DGLIBC_COMPATIBILITY=0"
     UNWIND="-DUSE_UNWIND=0"
     JEMALLOC="-DENABLE_JEMALLOC=0"
-    PYINIT_ENTRY="-Wl,-exported_symbol,_PyInit_${CHDB_PY_MOD}"
     HDFS="-DENABLE_HDFS=0 -DENABLE_GSASL_LIBRARY=0 -DENABLE_KRB5=0"
     MYSQL="-DENABLE_MYSQL=0"
     ICU="-DENABLE_ICU=0"
@@ -54,7 +53,6 @@ elif [ "$(uname)" == "Linux" ]; then
     GLIBC_COMPATIBILITY="-DGLIBC_COMPATIBILITY=1"
     UNWIND="-DUSE_UNWIND=1"
     JEMALLOC="-DENABLE_JEMALLOC=1"
-    PYINIT_ENTRY="-Wl,-ePyInit_${CHDB_PY_MOD}"
     ICU="-DENABLE_ICU=1"
     SED_INPLACE="sed -i"
     # only x86_64, enable AVX, enable embedded compiler
@@ -153,16 +151,13 @@ if [ ! "${USING_RESPONSE_FILE}" == "" ]; then
     ${SED_INPLACE} 's/ '${CHDB_PY_MODULE}'/ '${LIBCHDB_SO}'/g' CMakeFiles/libchdb.rsp
 fi
 
-if [ "$(uname)" == "Linux" ]; then
-    LIBCHDB_CMD=$(echo ${LIBCHDB_CMD} | sed 's/ '${PYINIT_ENTRY}'/ /g')
-    if [ ! "${USING_RESPONSE_FILE}" == "" ]; then
-        ${SED_INPLACE} 's/ '${PYINIT_ENTRY}'/ /g' CMakeFiles/libchdb.rsp
-    fi
-fi
-
+# Control exported symbols for libchdb.so
 if [ "$(uname)" == "Darwin" ]; then
-    LIBCHDB_CMD=$(echo ${LIBCHDB_CMD} | sed 's/ '${PYINIT_ENTRY}'/ -Wl,-exported_symbol,_query_stable -Wl,-exported_symbol,_free_result -Wl,-exported_symbol,_query_stable_v2 -Wl,-exported_symbol,_free_result_v2/g')
-    # ${SED_INPLACE} 's/ '${PYINIT_ENTRY}'/ -Wl,-exported_symbol,_query_stable -Wl,-exported_symbol,_free_result -Wl,-exported_symbol,_query_stable_v2 -Wl,-exported_symbol,_free_result_v2/g' CMakeFiles/libchdb.rsp
+    # macOS: use exported_symbols_list file
+    LIBCHDB_CMD="${LIBCHDB_CMD} -Wl,-exported_symbols_list,${CHDB_DIR}/libchdb_export_macos.txt"
+else
+    # Linux: use version script
+    LIBCHDB_CMD="${LIBCHDB_CMD} -Wl,--version-script=${CHDB_DIR}/libchdb_export.map"
 fi
 
 LIBCHDB_CMD=$(echo ${LIBCHDB_CMD} | sed 's/@CMakeFiles\/clickhouse.rsp/@CMakeFiles\/libchdb.rsp/g')
@@ -212,20 +207,13 @@ fi
 
 # extract the command to generate CHDB_PY_MODULE
 PYCHDB_CMD=$(grep -m 1 'clang++.*-o programs/clickhouse .*' build.log \
-    | sed "s/-o programs\/clickhouse/-fPIC -Wl,-undefined,dynamic_lookup -shared ${PYINIT_ENTRY} -o ${CHDB_PY_MODULE}/" \
+    | sed "s/-o programs\/clickhouse/-fPIC -Wl,-undefined,dynamic_lookup -shared -o ${CHDB_PY_MODULE}/" \
     | sed 's/^[^&]*&& //' | sed 's/&&.*//' \
     | sed 's/ -Wl,-undefined,error/ -Wl,-undefined,dynamic_lookup/g' \
     | sed 's/ -Xlinker --no-undefined//g' \
     | sed 's/@CMakeFiles\/clickhouse.rsp/@CMakeFiles\/pychdb.rsp/g' \
      )
 
-
-# # inplace modify the CMakeFiles/pychdb.rsp
-# ${SED_INPLACE} 's/-o programs\/clickhouse/-fPIC -Wl,-undefined,dynamic_lookup -shared ${PYINIT_ENTRY} -o ${CHDB_PY_MODULE}/' CMakeFiles/pychdb.rsp
-# ${SED_INPLACE} 's/ -Wl,-undefined,error/ -Wl,-undefined,dynamic_lookup/g' CMakeFiles/pychdb.rsp
-# ${SED_INPLACE} 's/ -Xlinker --no-undefined//g' CMakeFiles/pychdb.rsp
-
-
 if [ "$(uname)" == "Linux" ]; then
     # remove src/CMakeFiles/clickhouse_malloc.dir/Common/stubFree.c.o
     PYCHDB_CMD=$(echo ${PYCHDB_CMD} | sed 's/ src\/CMakeFiles\/clickhouse_malloc.dir\/Common\/stubFree.c.o//g')
@@ -239,8 +227,10 @@ fi
 
 if [ "$(uname)" == "Darwin" ]; then
     PYCHDB_CMD=$(echo ${PYCHDB_CMD} | sed 's|-Wl,-rpath,/[^[:space:]]*/pybind11-cmake|-Wl,-rpath,@loader_path|g')
+    PYCHDB_CMD="${PYCHDB_CMD} -Wl,-exported_symbols_list,${CHDB_DIR}/pychdb_export_macos.txt"
 else
     PYCHDB_CMD=$(echo ${PYCHDB_CMD} | sed 's|-Wl,-rpath,/[^[:space:]]*/pybind11-cmake|-Wl,-rpath,\$ORIGIN|g')
+    PYCHDB_CMD="${PYCHDB_CMD} -Wl,--undefined=PyInit__chdb -Wl,--version-script=${CHDB_DIR}/pychdb_export.map"
 fi
 
 # save the command to a file for debug
@@ -294,20 +284,7 @@ ${NM} ${LIBCHDB} | grep query_stable || true
 
 echo -e "\nAfter copy:"
 cd ${PROJ_DIR} && pwd
-# ls -lh ${PROJ_DIR}
-
-# strip the binary (no debug info at all)
-# strip ${CHDB_DIR}/${CHDB_PY_MODULE} || true
-
-# echo -e "\nAfter strip:"
-# echo -e "\nLIBCHDB: ${PYCHDB}"
-# ls -lh ${CHDB_DIR}
-# echo -e "\nfile info of ${PYCHDB}"
-# file ${CHDB_DIR}/${CHDB_PY_MODULE}
 
 ccache -s || true
 
-# bash ${DIR}/build_bind.sh
-# bash ${DIR}/test_smoke.sh
-
 CMAKE_ARGS="${CMAKE_ARGS}" bash ${DIR}/build_pybind11.sh --all

chdb/libchdb_export.map

@@ -0,0 +1,46 @@
+{
+    global:
+        /* Deprecated API */
+        query_stable;
+        free_result;
+        query_stable_v2;
+        free_result_v2;
+        connect_chdb;
+        close_conn;
+        query_conn;
+        query_conn_n;
+        query_conn_streaming;
+        query_conn_streaming_n;
+        chdb_streaming_result_error;
+        chdb_streaming_fetch_result;
+        chdb_streaming_cancel_query;
+        chdb_destroy_result;
+
+        /* New API */
+        chdb_connect;
+        chdb_close_conn;
+        chdb_query;
+        chdb_query_n;
+        chdb_query_cmdline;
+        chdb_stream_query;
+        chdb_stream_query_n;
+        chdb_stream_fetch_result;
+        chdb_stream_cancel_query;
+        chdb_destroy_query_result;
+        chdb_result_buffer;
+        chdb_result_length;
+        chdb_result_elapsed;
+        chdb_result_rows_read;
+        chdb_result_bytes_read;
+        chdb_result_storage_rows_read;
+        chdb_result_storage_bytes_read;
+        chdb_result_error;
+
+        /* Arrow Integration */
+        chdb_arrow_scan;
+        chdb_arrow_array_scan;
+        chdb_arrow_unregister_table;
+
+    local:
+        *;
+};

chdb/libchdb_export_macos.txt

@@ -0,0 +1,40 @@
+# Deprecated API
+_query_stable
+_free_result
+_query_stable_v2
+_free_result_v2
+_connect_chdb
+_close_conn
+_query_conn
+_query_conn_n
+_query_conn_streaming
+_query_conn_streaming_n
+_chdb_streaming_result_error
+_chdb_streaming_fetch_result
+_chdb_streaming_cancel_query
+_chdb_destroy_result
+
+# New API
+_chdb_connect
+_chdb_close_conn
+_chdb_query
+_chdb_query_n
+_chdb_query_cmdline
+_chdb_stream_query
+_chdb_stream_query_n
+_chdb_stream_fetch_result
+_chdb_stream_cancel_query
+_chdb_destroy_query_result
+_chdb_result_buffer
+_chdb_result_length
+_chdb_result_elapsed
+_chdb_result_rows_read
+_chdb_result_bytes_read
+_chdb_result_storage_rows_read
+_chdb_result_storage_bytes_read
+_chdb_result_error
+
+# Arrow Integration
+_chdb_arrow_scan
+_chdb_arrow_array_scan
+_chdb_arrow_unregister_table

chdb/pychdb_export.map

@@ -0,0 +1,21 @@
+{
+    global:
+        PyInit__chdb;
+
+        /* C++ operator new/delete (mangled names) */
+        _Znwm;              /* operator new(size_t) */
+        _Znam;              /* operator new[](size_t) */
+        _ZdlPv;             /* operator delete(void*) */
+        _ZdaPv;             /* operator delete[](void*) */
+        _ZdlPvm;            /* operator delete(void*, size_t) */
+        _ZdaPvm;            /* operator delete[](void*, size_t) */
+        _ZnwmSt11align_val_t;   /* operator new(size_t, align_val_t) */
+        _ZnamSt11align_val_t;   /* operator new[](size_t, align_val_t) */
+        _ZdlPvSt11align_val_t;  /* operator delete(void*, align_val_t) */
+        _ZdaPvSt11align_val_t;  /* operator delete[](void*, align_val_t) */
+        _ZdlPvmSt11align_val_t; /* operator delete(void*, size_t, align_val_t) */
+        _ZdaPvmSt11align_val_t; /* operator delete[](void*, size_t, align_val_t) */
+
+    local:
+        *;
+};

chdb/pychdb_export_macos.txt

@@ -0,0 +1 @@
+_PyInit__chdb