#!/usr/bin/env python
# -*- coding: utf-8 -*-

"""
splitdump splits a mysqldump file in several files (one per table)
"""

__title__ = "splitdump"
__version__ = "0.5"
__author__= "Carles Amigo"
__email__= "fr3nd at fr3nd dot net"

import sys
import re
import gzip
from optparse import OptionParser

def split_file(file, included, excluded, included_databases, excluded_databases, directory, use_gzip, store_databases):
        database = None
        
        if use_gzip:
            extension = ".sql.gz"
        else:
            extension = ".sql"
        
        header_file = directory + "/00_header" + extension
        
        if use_gzip:
            output = gzip.open(header_file,"w")
        else:
            output = open(header_file,"w")
        
        pat = re.compile('^DROP TABLE IF EXISTS `?(.*)`;')
        pat2 = re.compile('^CREATE DATABASE .+`(.*)` .+;')
        
        previous_database = None
        database = None
        
        for line in file:
                matchdb = pat2.search(line)
                
                if matchdb:
                    previous_database = database
                    database = matchdb.group(1)
                
                if store_databases:
                    if (database not in excluded_databases) and ((database == None) or (len(included_databases) == 0) or (database in included_databases)):
                        if previous_database != database:
                            previous_database = database
                            if database:
                                filename = directory + "/" + database + extension
                                output.close()
                                if use_gzip:
                                    output = gzip.open(filename, "w")
                                else:
                                    output = open(filename, "w")
                        output.write(line)
                            
                                
                        
                else:
                    if (database not in excluded_databases) and ((database == None) or (len(included_databases) == 0) or (database in included_databases)):
                        match = pat.search(line)
                    
                        if match:
                            table_name = match.groups(1)[0]
                            
                            if (table_name not in excluded) and ((len(included) == 0) or (table_name in included)):
                                filename = directory + "/"
                                if database:
                                    filename = filename + database + "."
                                filename = filename + table_name + extension
                            else:
                                filename = "/dev/null"
                            output.close()
                            if use_gzip:
                                output = gzip.open(filename, "w")
                            else:
                                output = open(filename, "w")
                        output.write(line)
        output.close()

def main():
        usage = "usage: %prog [options]\n Split a mysqldumpfile in multiple files (one per database).\n Use --help to view options"
        parser = OptionParser(usage, version=__version__)
        parser.add_option("-f", "--file", action="store", dest="file", type="string", default=None, help="Input file. Defaults to stdin")
        parser.add_option("-i", "--include_table", action="append", dest="included_tables", default=[], help="Tables to be included. By default, all tables are dumped. Can be called more than once.")
        parser.add_option("-e", "--exclude_table", action="append", dest="excluded_tables", default=[], help="Tables to be excluded. By default, no table is excluded. Can be called more than once.")
        
        parser.add_option("-I", "--include_database", action="append", dest="included_databases", default=[], help="Databases to be included. By default, all databases are dumped. Can be called more than once.")
        parser.add_option("-E", "--exclude_database", action="append", dest="excluded_databases", default=[], help="Databases to be excluded. By default, no database is excluded. Can be called more than once.")
        
        parser.add_option("-D", "--databases", action="store_true", dest="store_databases", default=False, help="Store databases instead of tables in individual files.")
        
        parser.add_option("-d", "--directory", action="store", dest="directory", type="string", default=".", help="Output directory where to store dump files.")
        parser.add_option("-g", "--gzip", action="store_true", dest="gzip", default=False, help="Compress (.gz) output files.")

        (options, args) = parser.parse_args()

        if not options.file:
                if not sys.stdin.isatty():
                    try:
                        file = sys.stdin
                    except:
                        parser.error("Error opening stdin")
                else:
                        parser.error("No input file specified")
        else:
                try:
                        file = open(options.file,"r")
                except:
                        parser.error("Error opening file")

        split_file(file, options.included_tables, options.excluded_tables, options.included_databases, options.excluded_databases, options.directory, options.gzip, options.store_databases)

if __name__ == "__main__":
        main()