content/file.py

# -*- coding: utf-8 -*-

# Server agnostic file operation similar to local files
# Currently supported - local, ftp, http

import os, sys, re
import urllib2, ftplib

def parse_url(url):
    #logger.info("Url: %s" % url)
    url = url.strip()
    import re
    m = re.search(r"^(?P<protocol>\w+)://(?:(?P<domain>[^;]+);)?(?:(?P<user>[^:@]+)[:|@])?(?:(?P<password>[^@]+)@)?(?P<host>[^/^:]+)?(?::(?P<port>[^/]+))?(?P<path>[/]?.*?)$", url, re.DOTALL | re.MULTILINE)
    if m:
        res = m.groupdict()
    else:
        res = {"domain": None, "host": None, "password": None, "path": url, "port": None, "protocol": "file", "user": None}
    return res


class FileObject(object):

    def __init__(self, url, mode="r", encoding=""):
        self.url = url
        self.mode = mode
        self.encoding = encoding
        self.file = None
        p = parse_url(url)
        self.p = p
        if p["protocol"] == "file":
            self.file = __builtins__["open"](p["path"], mode)
        elif p["protocol"] == "ftp":
            import ftplib, urllib2
            if "w" in mode or "a" in mode:
                self.file = ftplib.FTP(p["host"], p["user"], p["password"])
            elif "r" in mode:
                self.file = urllib2.urlopen(url)
        elif p["protocol"] in ["http", "https"]:
            self.file = urllib2.urlopen(url)
        else:
            raise Exception("Not valid protocol")
        self.first = True

    def read(self, bytes=None):
        if self.p["protocol"] == "file":
            return self.file.read(bytes)
        elif self.p["protocol"] == "ftp":
            return self.file.read(bytes)
        elif self.p["protocol"] in ("http", "https"):
            return self.file.read(bytes)

    def write(self, data, bytes=None):
        if self.p["protocol"] == "file":
            return self.file.write(data)
        elif self.p["protocol"] == "ftp":
            from StringIO import StringIO
            if "w" in self.mode and self.first:
                cmd = "STOR "
            elif "a" in self.mode or "w" in self.mode:
                cmd = "APPE "
            else:
                raise Exception("Can not write in r mode")
            if "b" in self.mode:
                self.file.storbinary(cmd + self.p["path"], StringIO(data))
            else:
                self.file.storlines(cmd + self.p["path"], StringIO(data))
            self.first = False
        elif self.p["protocol"] in ("http", "https"):
            raise Exception("Write not allowed for http")
        else:
            raise Exception("Not valid protocol")

    def close(self):
        if self.p["protocol"] == "file":
            self.file.close()
        elif self.p["protocol"] == "ftp":
            if "w" in self.mode or "a" in self.mode:
                self.file.close()
        elif self.p["protocol"] in ("http", "https"):
            pass
        else:
            raise Exception("Not valid protocol")

def open(url, mode="r", encoding=""):
    return FileObject(url, mode, encoding)

def exists(url):
    p = parse_url(url)
    if p["protocol"] == "file":
        if isinstance(url, str):
            url = url.decode("utf8")
        return os.path.exists(url)
    elif p["protocol"] in ("ftp", "http", "https"):
        try:
            f = urllib2.urlopen(url)
            del f
            return True
        except:
            return False
    else:
        raise Exception("Not valid protocol")

def isdir(url):
    p = parse_url(url)
    if p["protocol"] == "file":
        if isinstance(url, str):
            url = url.decode("utf8")
        return os.path.isdir(url)
    elif p["protocol"] in ("ftp"):
        file = ftplib.FTP(p["host"], p["user"], p["password"])
        cmd = "MLSD "+p["path"]
        lines = []
        file.retrbinary(cmd, lines.append)
        file.close()
        return False if lines and "No such file or directory" in lines[0] else True
    else:
        raise Exception("Not valid protocol")

def mkdir(url):
    p = parse_url(url)
    if p["protocol"] == "file":
        os.mkdir(url)
    elif p["protocol"] in ("ftp"):
        file = ftplib.FTP(p["host"], p["user"], p["password"])
        file.mkd(p["path"])
        file.close()
    else:
        raise Exception("Not valid protocol")

def rmdir(url):
    p = parse_url(url)
    if p["protocol"] == "file":
        os.rmdir(url)
    elif p["protocol"] in ("ftp"):
        file = ftplib.FTP(p["host"], p["user"], p["password"])
        file.rmd(p["path"])
        file.close()
    else:
        raise Exception("Not valid protocol")

def remove(url):
    p = parse_url(url)
    if p["protocol"] == "file":
        os.remove(url)
    elif p["protocol"] in ("ftp"):
        file = ftplib.FTP(p["host"], p["user"], p["password"])
        file.delete(p["path"])
        file.close()
    else:
        raise Exception("Not valid protocol")

def rename(url, url2):
    p = parse_url(url)
    p2 = parse_url(url2)
    if p["protocol"] == "file":
        os.rename(url, url2)
    elif p["protocol"] in ("ftp"):
        file = ftplib.FTP(p["host"], p["user"], p["password"])
        file.rename(p["path"], p2["path"])
        file.close()
    else:
        raise Exception("Not valid protocol")

def listdir(url):
    p = parse_url(url)
    if p["protocol"] == "file":
        return os.listdir(url)
    elif p["protocol"] in ("ftp"):
        cmd = "MLSD "+p["path"]
        file = ftplib.FTP(p["host"], p["user"], p["password"])
        lines = []
        file.retrlines(cmd, lines.append)
        file.close()
        return lines
    else:
        raise Exception("Not valid protocol")

def join(*arg):
    n = len(arg)
    if n == 0:
        return ""
    elif n == 1:
        return arg[0]
    p = parse_url(arg[0])
    sep = os.path.sep if p["protocol"] == "file" else "/"
    if "/" in arg[0]:
        sep = "/"
    elif "\\" in arg[0]:
        sep = "\\"
    path = arg[0]
    for i in xrange(1, n):
        if arg[i-1][-1] == sep:
            path = path + arg[i]
        else:
            path = path + sep + arg[i]
    return path

def encoding(url):
    p = parse_url(url)
    if p["protocol"] == "file":
        enc = sys.getfilesystemencoding()
        #enc = "utf8"  # TODO
    else:
        enc = "utf8"
    return enc

def make_fname(name):
    #"\[.+?\]" "[/\n\r\t,:]"
    name = re.sub(r"[/\n\r\t,:/\\]", " ", name)
    name = re.sub("['""\?\*<>]","", name)
    name = re.sub(r"(\[[^\[]*\])", "", name)
    name = name.replace("&quot;","")
    name = name.strip()
    return name


def encode(url):
    p = parse_url(url)
    if p["protocol"] == "file":  # unicode
        if not isinstance(url, unicode):
            url = url.decode("utf8")
    else:
        if isinstance(url, unicode):
            url = url.encode("utf8")
    return url

if __name__ == "__main__":
    res = join("aaa", "bbb")
    #url = "smb://user:xxxx@192.168.77.197/hdd/test"
    url = "ftp://user:xxxx@home.blue.lv:21/hdd/test"
    url = "ftp://user:xxxx@192.168.77.197:21/hdd/test"

    mkdir("ftp://user:xxxx@192.168.77.197:21/hdd/movie/AAA")
    res = listdir("ftp://user:xxxx@home.blue.lv/hdd/movie/")
    print res
    rmdir("ftp://user:xxxx@192.168.77.197:21/hdd/movie/AAA")
    rename("ftp://user:xxxx@192.168.77.197:21/hdd/test", "ftp://user:xxxx@192.168.77.197:21/hdd/test2")
    remove("ftp://user:xxxx@192.168.77.197:21/hdd/test2")
    e = isdir("ftp://user:xxxx@192.168.77.197:21/hdd/movies/TV2")
    print e

    f = open(url, "wb")
    f.write("6.rinda\n")
    f.write("7.rinda\n")
    f.close()
    f = open(url, "rb")
    res = f.read()
    print res
    f.close()
    e = exists(url)
    print e
    e = exists(url+"2")
    print e