diff --git a/docs/changes.md b/docs/changes.md
index c680ed25aa..87579b11d5 100644
--- a/docs/changes.md
+++ b/docs/changes.md
@@ -1,5 +1,8 @@
 # 最新动态
 
+2023/12/07
+  * csv file object 支持通过 MVVM 来查询。
+
 2023/12/06
   * fscript 增加 date\_time\_format
   * 完善 fscript，先找全局函数，再找用户自定义的函数。
diff --git a/src/csv/csv_file_object.c b/src/csv/csv_file_object.c
index c4aaa33460..016d5b2ca8 100644
--- a/src/csv/csv_file_object.c
+++ b/src/csv/csv_file_object.c
@@ -23,19 +23,12 @@
 #include "tkc/mem.h"
 #include "tkc/utils.h"
 #include "tkc/object.h"
+#include "tkc/object_default.h"
 #include "tkc/data_reader.h"
 
 #include "csv_file.h"
 #include "csv_file_object.h"
 
-typedef struct _csv_file_object_t {
-  tk_object_t object;
-
-  /*private*/
-  csv_file_t* csv;
-  str_t str;
-} csv_file_object_t;
-
 static csv_file_object_t* csv_file_object_cast(tk_object_t* obj);
 #define CSV_FILE_OBJECT(obj) csv_file_object_cast((tk_object_t*)obj)
 
@@ -45,15 +38,27 @@ typedef struct _csv_path_t {
   const char* col_name;
 } csv_path_t;
 
-static ret_t csv_path_parse(csv_path_t* path, csv_file_t* csv, const char* name) {
+static ret_t csv_path_parse_impl(csv_path_t* path, csv_file_t* csv, const char* name) {
   const char* p = name;
+  return_value_if_fail(path != NULL && csv != NULL && name != NULL, RET_BAD_PARAMS);
+
   memset(path, 0x00, sizeof(*path));
 
-  while (*p && *p != '[') p++;
-  return_value_if_fail(*p == '[', RET_FAIL);
+  while (tk_isspace(*p)) p++;
+
+  if (tk_isdigit(*p)) {
+    path->row = tk_atoi(p);
+  } else if (*p != '[') {
+    return RET_BAD_PARAMS;
+  } else {
+    p++; /* skip '[' */
+    if (tk_isdigit(*p)) {
+      path->row = tk_atoi(p);
+    } else {
+      return RET_BAD_PARAMS;
+    }
+  }
 
-  p++;
-  path->row = tk_atoi(p);
   p = strchr(p, '.');
   if (p == NULL) {
     return RET_OK;
@@ -70,6 +75,8 @@ static ret_t csv_path_parse(csv_path_t* path, csv_file_t* csv, const char* name)
   if (*p == '[') {
     return_value_if_fail(tk_isdigit(p[1]), RET_BAD_PARAMS);
     path->col = tk_atoi(p + 1);
+  } else if (tk_isdigit(*p)) {
+    path->col = tk_atoi(p);
   } else {
     path->col = csv_file_get_col_of_name(csv, p);
     if (path->col < 0) {
@@ -83,11 +90,57 @@ static ret_t csv_path_parse(csv_path_t* path, csv_file_t* csv, const char* name)
   return RET_OK;
 }
 
+static ret_t csv_file_object_remove_map(csv_file_object_t* o, uint32_t index) {
+  uint32_t i = 0;
+  uint32_t n = 0;
+  uint32_t* rows_map = NULL;
+  return_value_if_fail(o != NULL, RET_BAD_PARAMS);
+
+  rows_map = o->rows_map;
+  n = o->rows_map_size;
+  for (i = index; i < n - 1; i++) {
+    rows_map[i] = rows_map[i + 1];
+  }
+
+  o->rows_map_size--;
+
+  return RET_OK;
+}
+
+static ret_t csv_path_parse_ex(csv_file_object_t* o, csv_path_t* path, const char* name,
+                            bool_t for_remove) {
+  ret_t ret = RET_BAD_PARAMS;
+  return_value_if_fail(o != NULL && path != NULL && name != NULL, RET_BAD_PARAMS);
+
+  ret = csv_path_parse_impl(path, o->csv, name);
+  if (ret == RET_OK) {
+    if (o->rows_map != NULL) {
+      uint32_t index = path->row;
+      path->row = o->rows_map[path->row];
+      if (for_remove) {
+        csv_file_object_remove_map(o, index);
+      }
+    }
+    return RET_OK;
+  } else {
+    return ret;
+  }
+}
+
+static ret_t csv_path_parse(csv_file_object_t* o, csv_path_t* path, const char* name) {
+  return csv_path_parse_ex(o, path, name, FALSE);
+}
+
 static ret_t csv_file_object_remove_prop(tk_object_t* obj, const char* name) {
   csv_path_t p;
   csv_file_object_t* o = CSV_FILE_OBJECT(obj);
   return_value_if_fail(o != NULL, RET_BAD_PARAMS);
-  return_value_if_fail(csv_path_parse(&p, o->csv, name) == RET_OK, RET_FAIL);
+
+  if (tk_str_start_with(name, CSV_QUERY_PREFIX)) {
+    return tk_object_remove_prop(o->query_args, name);
+  }
+
+  return_value_if_fail(csv_path_parse_ex(o, &p, name, TRUE) == RET_OK, RET_FAIL);
 
   return csv_file_remove_row(o->csv, p.row);
 }
@@ -98,12 +151,16 @@ static ret_t csv_file_object_set_prop(tk_object_t* obj, const char* name, const
   csv_file_object_t* o = CSV_FILE_OBJECT(obj);
   return_value_if_fail(o != NULL, RET_BAD_PARAMS);
 
+  if (tk_str_start_with(name, CSV_QUERY_PREFIX)) {
+    return tk_object_set_prop(o->query_args, name, v);
+  }
+
   rows = csv_file_get_rows(o->csv);
   if (rows <= 0) {
     return RET_NOT_FOUND;
   }
 
-  return_value_if_fail(csv_path_parse(&p, o->csv, name) == RET_OK, RET_FAIL);
+  return_value_if_fail(csv_path_parse(o, &p, name) == RET_OK, RET_FAIL);
   if (p.col_name != NULL && tk_str_ieq(p.col_name, TK_OBJECT_PROP_CHECKED)) {
     return csv_file_set_row_checked(o->csv, p.row, value_bool(v));
   }
@@ -112,7 +169,6 @@ static ret_t csv_file_object_set_prop(tk_object_t* obj, const char* name, const
   return csv_file_set(o->csv, p.row, p.col, o->str.str);
 }
 
-
 static ret_t csv_file_object_get_prop(tk_object_t* obj, const char* name, value_t* v) {
   csv_path_t p;
   uint32_t rows = 0;
@@ -120,11 +176,19 @@ static ret_t csv_file_object_get_prop(tk_object_t* obj, const char* name, value_
   csv_file_object_t* o = CSV_FILE_OBJECT(obj);
   return_value_if_fail(o != NULL, RET_BAD_PARAMS);
 
+  if (tk_str_start_with(name, CSV_QUERY_PREFIX)) {
+    return tk_object_get_prop(o->query_args, name, v);
+  }
+
   rows = csv_file_get_rows(o->csv);
   if (tk_str_ieq(name, TK_OBJECT_PROP_SIZE)) {
-    value_set_int(v, rows);
+    if (o->rows_map != NULL) {
+      value_set_int(v, o->rows_map_size);
+    } else {
+      value_set_int(v, rows);
+    }
     return RET_OK;
-  } else if(tk_str_eq(name, TK_OBJECT_PROP_SELECTED_INDEX)) {
+  } else if (tk_str_eq(name, TK_OBJECT_PROP_SELECTED_INDEX)) {
     value_set_int(v, csv_file_get_first_checked(o->csv));
     return RET_OK;
   }
@@ -133,7 +197,7 @@ static ret_t csv_file_object_get_prop(tk_object_t* obj, const char* name, value_
     return RET_NOT_FOUND;
   }
 
-  return_value_if_fail(csv_path_parse(&p, o->csv, name) == RET_OK, RET_FAIL);
+  return_value_if_fail(csv_path_parse(o, &p, name) == RET_OK, RET_FAIL);
   if (p.col_name != NULL && tk_str_ieq(p.col_name, TK_OBJECT_PROP_CHECKED)) {
     return_value_if_fail(p.row < rows, RET_FAIL);
 
@@ -165,11 +229,75 @@ static bool_t csv_file_object_can_exec(tk_object_t* obj, const char* name, const
     return csv_file_get_checked_rows(o->csv) > 0;
   } else if (tk_str_ieq(name, TK_OBJECT_CMD_ADD)) {
     return TRUE;
+  } else if (tk_str_eq(name, CSV_CMD_QUERY)) {
+    return TRUE;
   }
 
   return FALSE;
 }
 
+static ret_t csv_file_object_prepare(csv_file_object_t* o) {
+  o->rows_map_size = 0;
+  if (o->rows_map != NULL) {
+    if (o->rows_map_capacity < csv_file_get_rows(o->csv)) {
+      TKMEM_FREE(o->rows_map);
+      o->rows_map = NULL;
+      o->rows_map_capacity = 0;
+    }
+  }
+
+  if (o->rows_map == NULL) {
+    o->rows_map_capacity = csv_file_get_rows(o->csv);
+    o->rows_map = TKMEM_ZALLOCN(uint32_t, o->rows_map_capacity);
+  }
+
+  return o->rows_map != NULL ? RET_OK : RET_OOM;
+}
+
+static ret_t csv_file_object_clear_query(tk_object_t* obj) {
+  csv_file_object_t* o = CSV_FILE_OBJECT(obj);
+  return_value_if_fail(o != NULL, RET_BAD_PARAMS);
+
+  o->rows_map_size = 0;
+  TKMEM_FREE(o->rows_map);
+  o->rows_map_capacity = 0;
+
+  return RET_OK;
+}
+
+static ret_t csv_file_object_query(tk_object_t* obj) {
+  uint32_t i = 0;
+  uint32_t n = 0;
+  csv_file_object_t* o = CSV_FILE_OBJECT(obj);
+  return_value_if_fail(o != NULL, RET_BAD_PARAMS);
+  return_value_if_fail(o->filter != NULL, RET_BAD_PARAMS);
+  return_value_if_fail(csv_file_object_prepare(o) == RET_OK, RET_OOM);
+
+  n = csv_file_get_rows(o->csv);
+  for (i = 0; i < n; i++) {
+    csv_row_t* row = csv_file_get_row(o->csv, i);
+    ret_t ret = o->filter(o->filter_ctx, o->query_args, i, row);
+
+    if (ret == RET_OK) {
+      o->rows_map[o->rows_map_size++] = i;
+    } else if (ret == RET_STOP) {
+      break;
+    }
+  }
+
+  return RET_OK;
+}
+
+ret_t csv_file_object_set_filter(tk_object_t* obj, csv_file_object_filter_t filter, void* ctx) {
+  csv_file_object_t* o = CSV_FILE_OBJECT(obj);
+  return_value_if_fail(o != NULL, RET_BAD_PARAMS);
+
+  o->filter = filter;
+  o->filter_ctx = ctx;
+
+  return RET_OK;
+}
+
 static ret_t csv_file_object_exec(tk_object_t* obj, const char* name, const char* args) {
   ret_t ret = RET_NOT_IMPL;
   csv_file_object_t* o = CSV_FILE_OBJECT(obj);
@@ -180,23 +308,26 @@ static ret_t csv_file_object_exec(tk_object_t* obj, const char* name, const char
   } else if (tk_str_ieq(name, TK_OBJECT_CMD_RELOAD)) {
     csv_file_reload(o->csv);
     ret = RET_ITEMS_CHANGED;
+    csv_file_object_clear_query(obj);
   } else if (tk_str_ieq(name, TK_OBJECT_CMD_CLEAR)) {
     csv_file_clear(o->csv);
     ret = RET_ITEMS_CHANGED;
+    csv_file_object_clear_query(obj);
   } else if (tk_str_ieq(name, TK_OBJECT_CMD_REMOVE)) {
-    const char* index = strrchr(args, '[');
-    if (index != NULL) {
-      index++;
-    } else {
-      index = args;
-    }
-    return_value_if_fail(index != NULL, RET_FAIL);
-    ret = csv_file_remove_row(o->csv, tk_atoi(index)) == RET_OK ? RET_ITEMS_CHANGED : RET_FAIL;
+    ret = csv_file_object_remove_prop(obj, args) == RET_OK ? RET_ITEMS_CHANGED : RET_FAIL;
   } else if (tk_str_ieq(name, TK_OBJECT_CMD_REMOVE_CHECKED)) {
     ret = csv_file_remove_checked_rows(o->csv) == RET_OK ? RET_ITEMS_CHANGED : RET_FAIL;
+    csv_file_object_clear_query(obj);
   } else if (tk_str_ieq(name, TK_OBJECT_CMD_ADD)) {
     return_value_if_fail(args != NULL, RET_FAIL);
     ret = csv_file_append_row(o->csv, args) == RET_OK ? RET_ITEMS_CHANGED : RET_FAIL;
+  } else if (tk_str_eq(name, CSV_CMD_QUERY)) {
+    if (tk_str_eq(args, CSV_CMD_QUERY_ARG_CLEAR)) {
+      ret = csv_file_object_clear_query(obj);
+    } else {
+      ret = csv_file_object_query(obj);
+    }
+    ret = RET_ITEMS_CHANGED;
   } else {
     return RET_NOT_IMPL;
   }
@@ -215,6 +346,8 @@ static ret_t csv_file_object_destroy(tk_object_t* obj) {
 
   csv_file_destroy(o->csv);
   o->csv = NULL;
+  TK_OBJECT_UNREF(o->query_args);
+  TKMEM_FREE(o->rows_map);
   str_reset(&(o->str));
 
   return RET_OK;
@@ -247,6 +380,7 @@ tk_object_t* csv_file_object_create(csv_file_t* csv) {
   return_value_if_fail(o != NULL, NULL);
 
   o->csv = csv;
+  o->query_args = object_default_create_ex(FALSE);
   str_init(&(o->str), 0);
 
   return obj;
diff --git a/src/csv/csv_file_object.h b/src/csv/csv_file_object.h
index 79fe511c7c..f6614744b4 100644
--- a/src/csv/csv_file_object.h
+++ b/src/csv/csv_file_object.h
@@ -1,5 +1,5 @@
 /**
- * File:   object_csv_file.h
+ * File:   csv_file_object.h
  * Author: AWTK Develop Team
  * Brief:  csv file object
  *
@@ -27,10 +27,18 @@
 
 BEGIN_C_DECLS
 
+
+/**
+ * 返回值：
+   * RET_OK: 保留
+   * RET_STOP: 停止解析
+   * RET_FAIL: 忽略此行
+ */
+typedef ret_t (*csv_file_object_filter_t)(void* ctx, tk_object_t* args, uint32_t index, csv_row_t* row);
+
 /**
  * @class csv_file_object_t
  * @parent tk_object_t
- * @annotation["fake"]
  * 将cvs file包装成object对象。
  * 
  * 示例
@@ -85,6 +93,21 @@ BEGIN_C_DECLS
  * ```
  */
 
+typedef struct _csv_file_object_t {
+  tk_object_t object;
+
+  /*private*/
+  csv_file_t* csv;
+  str_t str;
+  
+  tk_object_t* query_args;
+  uint32_t* rows_map;
+  uint32_t rows_map_size;
+  uint32_t rows_map_capacity;
+  csv_file_object_filter_t filter;
+  void* filter_ctx;
+} csv_file_object_t;
+
 /**
  * @method csv_file_object_create
  *
@@ -144,6 +167,18 @@ tk_object_t* csv_file_object_load_from_buff(const void* buff, uint32_t size, cha
  */
 ret_t csv_file_object_save_to_buff(tk_object_t* obj, wbuffer_t* wb);
 
+/**
+ * @method csv_file_object_set_filter
+ * 设置过滤器。
+ * 
+ * @param {tk_object_t*} obj doc对象。
+ * @param {csv_file_object_filter_t} filter 过滤器。
+ * @param {void*} ctx 上下文。
+ * 
+ * @return {ret_t} 返回RET_OK表示成功，否则表示失败
+ */
+ret_t csv_file_object_set_filter(tk_object_t* obj, csv_file_object_filter_t filter, void* ctx);
+
 /**
  * @method csv_file_object_save_as
  * 将doc对象保存到指定文件。
@@ -156,6 +191,10 @@ ret_t csv_file_object_save_to_buff(tk_object_t* obj, wbuffer_t* wb);
  */
 ret_t csv_file_object_save_as(tk_object_t* obj, const char* filename);
 
+#define CSV_QUERY_PREFIX "query."
+#define CSV_CMD_QUERY "query"
+#define CSV_CMD_QUERY_ARG_CLEAR "clear"
+
 END_C_DECLS
 
 #endif /*TK_CSV_FILE_OBJECT_H*/