Merge branch 'devel'

[qcg-portal.git] / filex / ftp.py

diff --git a/filex/ftp.py b/filex/ftp.py
index 2c92f12..85161a4 100644 (file)
--- a/filex/ftp.py
+++ b/filex/ftp.py
@@ -4,18 +4,23 @@ from itertools import chain
 import os
 import re
 from threading import Event
-from django.utils.http import urlunquote
+from urlparse import urlparse, urlunparse
 
+from django.utils.http import urlquote
 from django.utils.timezone import localtime, UTC
-from gridftp import FTPClient, Buffer, HandleAttr, OperationAttr
 
 
 class FTPError(Exception):
-    pass
+    def __init__(self, message, verbose=None, *args, **kwargs):
+        super(FTPError, self).__init__(message, *args, **kwargs)
+
+        self.verbose = verbose
 
 
 class FTPOperation:
     def __init__(self, proxy=None, buffer_size=4096):
+        from gridftp import FTPClient, Buffer, HandleAttr, OperationAttr
+
         self._end = Event()
         self._error = None
         self._buffer = Buffer(buffer_size)
@@ -67,14 +72,11 @@ class FTPOperation:
         self._end.clear()
 
         if self._error is not None:
-            # TODO logging
-            print 'GridFTP ERROR:', self._error
-
             match = re.search(r'A system call failed: (.*)$', self._error.replace('\r\n', '\n'), re.MULTILINE)
 
             msg = match.groups()[0] if match else "Unknown error"
 
-            raise FTPError(msg)
+            raise FTPError(msg, self._error)
 
     def listing(self, url):
         self.cli.verbose_list(url, self._done, None, self.op_attr)
@@ -99,13 +101,14 @@ class FTPOperation:
                 for attr in attrs.split(';'):
                     try:
                         key, value = attr.split('=', 1)
-                        attrs_dict[key] = value
                     except ValueError:
-                        pass
+                        key, value = attr, ''
+
+                    attrs_dict[key] = value
 
                 yield {
                     'name': name,
-                    'type': 'directory' if attrs_dict['Type'] == 'dir' else 'file',
+                    'type': 'directory' if attrs_dict['Type'].endswith('dir') else 'file',
                     'size': int(attrs_dict['Size']),
                     'date': localtime(datetime.strptime(attrs_dict['Modify'], "%Y%m%d%H%M%S").replace(tzinfo=UTC())),
                 }
@@ -140,6 +143,18 @@ class FTPOperation:
 
         return data
 
+    def exists(self, url):
+        self.cli.exists(url, self._done, None, self.op_attr)
+
+        try:
+            self.wait()
+        except FTPError as e:
+            if 'No such file or directory' in e.message:
+                return False
+            raise
+        else:
+            return True
+
     def delete(self, url):
         self.cli.delete(url, self._done, None, self.op_attr)
 
@@ -150,7 +165,16 @@ class FTPOperation:
 
         self.wait()
 
-    def mkdir(self, url):
+    def mkdir(self, url, parents=False):
+        if parents:
+            if self.exists(url):
+                return
+
+            u = urlparse(url)
+            parent_url = urlunparse((u.scheme, u.netloc, os.path.dirname(os.path.normpath(u.path)), '', '', ''))
+
+            self.mkdir(parent_url, parents=True)
+
         self.cli.mkdir(url, self._done, None, self.op_attr)
 
         self.wait()
@@ -166,15 +190,22 @@ class FTPOperation:
         self._check_disk_stack_args(*([path, archive] + files))
 
         if self.match_ext(archive, '.tar.gz', '.tgz'):
-            cmd, args = 'tar', ['cvzf', archive, '-C', path] + files
+            cmd, args = 'tar', ['czf', archive, '-C', path] + files
         elif self.match_ext(archive, '.tar.bz2', '.tbz'):
-            cmd, args = 'tar', ['cvjf', archive, '-C', path] + files
+            cmd, args = 'tar', ['cjf', archive, '-C', path] + files
         elif self.match_ext(archive, '.zip'):
-            cmd, args = 'jar', (['cvMf', archive] + list(chain.from_iterable(('-C', path, f) for f in files)))
+            # zip doesn't support unicode file names
+            for arg in files:
+                try:
+                    arg.encode('ascii')
+                except UnicodeEncodeError as e:
+                    raise ValueError(u'Unsupported character `{}` in `{}`!'.format(arg[e.start:e.start + 1], arg))
+
+            cmd, args = 'jar', (['cMf', archive] + list(chain.from_iterable(('-C', path, f) for f in files)))
         else:
             raise ValueError('Unknown archive type: {}'.format(archive))
 
-        self.op_attr.set_disk_stack('#'.join(["popen:argv=", cmd] + args))
+        self.op_attr.set_disk_stack('#'.join(["popen:argv=", cmd] + map(urlquote, args)))
 
         return self.get(server)
 
@@ -182,21 +213,21 @@ class FTPOperation:
         self._check_disk_stack_args(*[archive, dst])
 
         if self.match_ext(archive, '.tar.gz', '.tgz'):
-            cmd, args = 'tar', ('xvzf', archive, '-C', dst)
+            cmd, args = 'tar', ('xzf', archive, '-C', dst)
         elif self.match_ext(archive, '.tar.bz2', '.tbz'):
-            cmd, args = 'tar', ('xvjf', archive, '-C', dst)
+            cmd, args = 'tar', ('xjf', archive, '-C', dst)
         elif self.match_ext(archive, '.zip'):
-            cmd, args = 'unzip', (archive, '-d', dst)
+            cmd, args = 'unzip', ('-qo', archive, '-d', dst)
         else:
             raise ValueError('Unknown archive type: {}'.format(archive))
 
-        self.op_attr.set_disk_stack('#'.join(("popen:argv=", cmd) + args))
+        self.op_attr.set_disk_stack('#'.join(["popen:argv=", cmd] + map(urlquote, args)))
 
         return self.get(server)
 
     @staticmethod
     def _check_disk_stack_args(*args):
-        for char in ['#', ',', ';', '%23', '%3B']:
+        for char in ['#', ';']:
             for arg in args:
                 if char in arg:
-                    raise ValueError('Unsupported character `{}` in `{}`!'.format(urlunquote(char), urlunquote(arg)))
+                    raise ValueError(u'Unsupported character `{}` in `{}`!'.format(char, arg))