extract 解压缩文件

'''
Created on Feb 22, 2013

@author: changxue

@summary: extract all archive files in src_dir into target_dir

'''
import os, shutil

_list_postfix = ['.rar', '.zip', '.jar']

tool_path = os.path.join(r'C:\Program Files\7-Zip')

from common_util import _create_dirs, _execute_cmd, _printlog, _get_dirname_from_filename

def is_archive_valid(file_name):
    ''' to evaluate whether the file could be extracted(True) or not(False).'''
    for postfix in _list_postfix:
        if file_name.endswith(postfix):
            return True
    return False

def get_type_src(full_filename):
    ''' to check whether there are .class files or .java files.
        return bin if only there is one .class file.'''
    result = ''
    os.chdir(tool_path)
    command = '7z l %s > temp.txt' %(full_filename) # list the content of archive
    os.system(command)
    with open('temp.txt', 'r') as f:
        fhandler = f.read()
    
    _printlog(fhandler)
    if fhandler.find('.class') != -1:
        result = 'bin'
    elif fhandler.find('.java') != -1:
        result = 'src'
    
    os.remove('temp.txt')
    return result

def extract_archive(archive_file_path, target_dir):
    os.chdir(tool_path)
    command = '7z x %s -o%s' % (archive_file_path, target_dir)
    _execute_cmd(command)

def extract_single(file_name, src_path, target_dir):
    ''' extract non-jar archive files into target directory.'''
    dirname = _get_dirname_from_filename(file_name)
    if os.path.exists(os.path.join(target_dir, 'src', dirname)):
        raise Exception('WARNING: directory already exist: %s\n'%os.path.join(target_dir, 'src', dirname))
    elif os.path.exists(os.path.join(target_dir, 'bin', dirname)):
        raise Exception('WARNING: directory already exist: %s\n'%os.path.join(target_dir, 'bin', dirname))
    else:
        archive_file_path = os.path.join(src_path, file_name)
        file_type = get_type_src(archive_file_path)
        if not file_type:
            dst_dir = os.path.join(src_path, 'old', 'invalid')
            _create_dirs(dst_dir)
            shutil.move(archive_file_path, dst_dir)
            raise Exception('WARNING: no java or class file in %s, MOVE TO old/invalid.\n'%archive_file_path)
        dest = os.path.join(target_dir, file_type, dirname)
        os.makedirs(dest)
        extract_archive(archive_file_path, dest)

def extract_all(src_path, extract_dir):
    ''' Main Entry:
        process all archive files in src_path:
           .jar: copy it to jar_dir and extract into specified dir(bin)
           others: extract into specified dir(src or bin)
    '''
    _printlog('############## Extract starting ##############\n')
    total_count = 0
    extract_list =[]
    os.chdir(src_path)
    for f in os.listdir(src_path):
        archive_file_path = os.path.join(src_path, f)
        try:
            if not os.path.isfile(archive_file_path):
                raise Exception('%s is not a file.\n'%f)
            else:
                total_count += 1
                if is_archive_valid(archive_file_path):
                    _printlog('Extracting %s\n'% archive_file_path)
                    extract_single(f, src_path, extract_dir)
                    extract_list.append(f)
        except Exception, e:
            _printlog(e)
            continue
    _printlog('total_count=%s\n'%total_count)
    _printlog('success_count=%s\n'%len(extract_list))
    _printlog('success_list=\n%s'%'\n'.join(extract_list))
    _printlog('############## Extract end ##############\n')

if __name__ == '__main__':
    usage = '''Usage: extractAll.py archive_path target_dir'''
    import sys
    argv_len = len(sys.argv)
    if argv_len == 1:
        src_path = 'C:\\works\\workload_src'
        target_dir = "C:\\works\\workload_out"
        extract_all(src_path, target_dir)
    elif argv_len == 3:
        extract_all(sys.argv[1], sys.argv[2])
    else:
        print usage