BACKUP-105 use a regex to decide if source files should be excluded f…

…rom the backup

README.md

@@ -29,7 +29,7 @@ install plugin tokudb_backup soname 'tokudb_backup.so';
 
 # Run a backup
 
-1 Backup to the '/tmp/backup1047' directory.  This blocks until the backup is complete.
+Backup to the '/tmp/backup1047' directory.  This blocks until the backup is complete.
 ```
 set tokudb_backup_dir='/tmp/backup1047';
 ```
@@ -39,6 +39,15 @@ The ```set tokudb_backup_dir``` statement will succeed if the backup was taken.
 select @@tokudb_backup_last_error, @@tokudb_backup_last_error_string;
 ```
 
+# Exclude source files
+Lets suppose that you want to exclude all 'lost+found' directories from the backup.  The ```tokudb_backup_exclude``` session variable contains a regular expression that all source file names are compared with.  If the source file name matches the exclude regular expression, then the source file is excluded from the backup.
+```
+set tokudb_backup_exclude='/lost\\+found($|/)';
+```
+```
+set tokudb_backup_dir='/tmp/backup105';
+\\\
+
 # Monitor progress
 The Tokutek hot backup updates the processlist state with progress information while it is running.
 
@@ -101,6 +110,13 @@ The ```tokudb_backup_throttle``` variable imposes an upper bound on the write ra
 * type:str
 * comment:error string of the last backup
 
+## tokudb_backup_exclude
+* name:tokudb_backup_exclude
+* readonly:false
+* scope:session
+* type:str
+* comment:exclude source file regular expression
+
 # Build the hot backup plugin from source
 1 Checkout the Percona Server source
 

backup/backup.h

@@ -5,18 +5,25 @@
 #define BACKUP_H
 
 #ident "Copyright (c) 2012-2013 Tokutek Inc.  All rights reserved."
-#ident "$Id: f3e40716a68df061474b60d63cede2cfc99f7be9 $"
+#ident "$Id: 4edcb140608da110c9a225a606dc13e77301d274 $"
 
 extern "C" {
 
 // These public API's should be in C.
 
-typedef int (*backup_poll_fun_t)(float progress, const char *progress_string, void*poll_extra);
+typedef int (*backup_poll_fun_t)(float progress, const char *progress_string, void *poll_extra);
+
 typedef void (*backup_error_fun_t)(int error_number, const char *error_string, void *error_extra);
 
+// The exclude_copy callback if called for every file that will be copied.
+// When it returns 0, the file is copied.  Otherwise, the file copy is skipped.
+typedef int (*backup_exclude_copy_fun_t)(const char *source_file,void *extra);
+
 int tokubackup_create_backup(const char *source_dirs[], const char *dest_dirs[], int dir_count,
                              backup_poll_fun_t poll_fun, void *poll_extra,
-                             backup_error_fun_t error_fun, void *error_extra) throw() __attribute__((visibility("default")));
+                             backup_error_fun_t error_fun, void *error_extra,
+                             backup_exclude_copy_fun_t check_fun, void *exclude_copy_extra)
+    throw() __attribute__((visibility("default")));
 // Effect: Backup the directories in source_dirs into correspnding dest_dirs.
 // Periodically call poll_fun.
 //  If poll_fun returns 0, then the backup continues.
@@ -41,6 +48,8 @@ int tokubackup_create_backup(const char *source_dirs[], const char *dest_dirs[],
 //   poll_extra: a value passed to poll_fun.
 //   error_fun: a function to call if an error happens
 //   error_extra: a value passed to error_fun.
+//   exclude_copy_fun: a function to call for every file being copied
+//   exclude_copy_extra: a valuue passed to the exclude_copy_fun
 // Return value:  0 if the backup succeeded.  Nonzero if the backup
 //   was stopped early (returning the result from the poll_fun) or the
 //   error code.

tokudb_backup.cc

@@ -18,6 +18,7 @@
 #include <sql_acl.h>   // SUPER_ACL
 #include <sql_parse.h> // check_global_access
 #include "backup/backup.h"
+#include <regex.h>
 
 #ifdef TOKUDB_BACKUP_PLUGIN_VERSION
 #define stringify2(x) #x
@@ -40,7 +41,9 @@ static MYSQL_SYSVAR_STR(version, tokudb_backup_version, PLUGIN_VAR_NOCMDARG | PL
 static MYSQL_THDVAR_ULONG(last_error, PLUGIN_VAR_THDLOCAL, "error from the last backup. 0 is success",
                           NULL, NULL, 0 /*default*/, 0 /*min*/, ~0ULL /*max*/, 1 /*blocksize*/);
 
-static MYSQL_THDVAR_STR(last_error_string, PLUGIN_VAR_THDLOCAL | PLUGIN_VAR_MEMALLOC, "error string from the last backup", NULL, NULL, NULL);
+static MYSQL_THDVAR_STR(last_error_string, PLUGIN_VAR_THDLOCAL + PLUGIN_VAR_MEMALLOC, "error string from the last backup", NULL, NULL, NULL);
+
+static MYSQL_THDVAR_STR(exclude, PLUGIN_VAR_THDLOCAL + PLUGIN_VAR_MEMALLOC, "exclude source file regular expression", NULL, NULL, NULL);
 
 static int tokudb_backup_check_dir(THD *thd, struct st_mysql_sys_var *var, void *save, struct st_mysql_value *value);
 static void tokudb_backup_update_dir(THD *thd, struct st_mysql_sys_var *var, void *var_ptr, const void *save);
@@ -66,9 +69,30 @@ static struct st_mysql_sys_var *tokudb_backup_system_variables[] = {
     MYSQL_SYSVAR(dir),
     MYSQL_SYSVAR(last_error),
     MYSQL_SYSVAR(last_error_string),
+    MYSQL_SYSVAR(exclude),
     NULL,
 };
 
+struct tokudb_backup_exclude_copy_extra {
+    THD *_thd;
+    char *exclude_string;
+    regex_t *re;
+};
+
+static int tokudb_backup_exclude_copy_fun(const char *source_file, void *extra) {
+    tokudb_backup_exclude_copy_extra *exclude_extra = static_cast<tokudb_backup_exclude_copy_extra *>(extra);
+    int r = 0;
+    if (0) fprintf(stderr, "%s %s\n", __FUNCTION__, source_file);
+    if (exclude_extra->exclude_string) {
+        int regr = regexec(exclude_extra->re, source_file, 0, NULL, 0);
+        if (regr == 0) {
+            if (1) fprintf(stderr, "tokudb backup exclude %s\n", source_file);
+            r = 1;
+        }
+    }
+    return r;
+}
+
 struct tokudb_backup_progress_extra {
     THD *_thd;
     char *_the_string;
@@ -531,6 +555,19 @@ static void tokudb_backup_run(THD *thd, const char *dest_dir) {
         return;
     }
 
+    char *exclude_string = THDVAR(thd, exclude);
+    regex_t exclude_re;
+    if (exclude_string) {
+        int regr = regcomp(&exclude_re, exclude_string, REG_EXTENDED);
+        if (regr) {
+            error = EINVAL;
+            char reg_error[100+strlen(exclude_string)];
+            snprintf(reg_error, sizeof reg_error, "tokudb backup exclude %s regcomp %d", exclude_string, regr);
+            tokudb_backup_set_error(thd, error, reg_error);
+            return;
+        }
+    }
+
     const char *source_dirs[MYSQL_MAX_DIR_COUNT] = {};
     const char *dest_dirs[MYSQL_MAX_DIR_COUNT] = {};
     int count = sources.set_valid_dirs_and_get_count(source_dirs, MYSQL_MAX_DIR_COUNT);
@@ -544,7 +581,14 @@ static void tokudb_backup_run(THD *thd, const char *dest_dir) {
     // do the backup
     tokudb_backup_progress_extra progress_extra = { thd, NULL };
     tokudb_backup_error_extra error_extra = { thd };
-    error = tokubackup_create_backup(source_dirs, dest_dirs, count, tokudb_backup_progress_fun, &progress_extra, tokudb_backup_error_fun, &error_extra);
+    tokudb_backup_exclude_copy_extra exclude_copy_extra = { thd, exclude_string, &exclude_re };
+    error = tokubackup_create_backup(source_dirs, dest_dirs, count,
+                                     tokudb_backup_progress_fun, &progress_extra,
+                                     tokudb_backup_error_fun, &error_extra,
+                                     tokudb_backup_exclude_copy_fun, &exclude_copy_extra);
+
+    if (exclude_string)
+        regfree(&exclude_re);
 
     // cleanup
     thd_proc_info(thd, "tokudb backup done"); // must be a static string