jqueue: Improve inotify error reporting
[ganeti-local] / lib / ovf.py
index 21816e8..dc75816 100644 (file)
@@ -1,7 +1,7 @@
 #!/usr/bin/python
 #
 
 #!/usr/bin/python
 #
 
-# Copyright (C) 2011 Google Inc.
+# Copyright (C) 2011, 2012 Google Inc.
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 
 """
 
 
 """
 
+# pylint: disable=F0401, E1101
+
+# F0401 because ElementTree is not default for python 2.4
+# E1101 makes no sense - pylint assumes that ElementTree object is a tuple
+
+
+import ConfigParser
+import errno
+import logging
+import os
 import os.path
 import os.path
+import re
 import shutil
 import shutil
+import tarfile
+import tempfile
+import xml.dom.minidom
+import xml.parsers.expat
+try:
+  import xml.etree.ElementTree as ET
+except ImportError:
+  import elementtree.ElementTree as ET
 
 
+try:
+  ParseError = ET.ParseError # pylint: disable=E1103
+except AttributeError:
+  ParseError = None
+
+from ganeti import constants
 from ganeti import errors
 from ganeti import utils
 from ganeti import errors
 from ganeti import utils
+from ganeti import pathutils
+
+
+# Schemas used in OVF format
+GANETI_SCHEMA = "http://ganeti"
+OVF_SCHEMA = "http://schemas.dmtf.org/ovf/envelope/1"
+RASD_SCHEMA = ("http://schemas.dmtf.org/wbem/wscim/1/cim-schema/2/"
+               "CIM_ResourceAllocationSettingData")
+VSSD_SCHEMA = ("http://schemas.dmtf.org/wbem/wscim/1/cim-schema/2/"
+               "CIM_VirtualSystemSettingData")
+XML_SCHEMA = "http://www.w3.org/2001/XMLSchema-instance"
+
+# File extensions in OVF package
+OVA_EXT = ".ova"
+OVF_EXT = ".ovf"
+MF_EXT = ".mf"
+CERT_EXT = ".cert"
+COMPRESSION_EXT = ".gz"
+FILE_EXTENSIONS = [
+  OVF_EXT,
+  MF_EXT,
+  CERT_EXT,
+]
+
+COMPRESSION_TYPE = "gzip"
+NO_COMPRESSION = [None, "identity"]
+COMPRESS = "compression"
+DECOMPRESS = "decompression"
+ALLOWED_ACTIONS = [COMPRESS, DECOMPRESS]
+
+VMDK = "vmdk"
+RAW = "raw"
+COW = "cow"
+ALLOWED_FORMATS = [RAW, COW, VMDK]
+
+# ResourceType values
+RASD_TYPE = {
+  "vcpus": "3",
+  "memory": "4",
+  "scsi-controller": "6",
+  "ethernet-adapter": "10",
+  "disk": "17",
+}
+
+SCSI_SUBTYPE = "lsilogic"
+VS_TYPE = {
+  "ganeti": "ganeti-ovf",
+  "external": "vmx-04",
+}
+
+# AllocationUnits values and conversion
+ALLOCATION_UNITS = {
+  "b": ["bytes", "b"],
+  "kb": ["kilobytes", "kb", "byte * 2^10", "kibibytes", "kib"],
+  "mb": ["megabytes", "mb", "byte * 2^20", "mebibytes", "mib"],
+  "gb": ["gigabytes", "gb", "byte * 2^30", "gibibytes", "gib"],
+}
+CONVERT_UNITS_TO_MB = {
+  "b": lambda x: x / (1024 * 1024),
+  "kb": lambda x: x / 1024,
+  "mb": lambda x: x,
+  "gb": lambda x: x * 1024,
+}
+
+# Names of the config fields
+NAME = "name"
+OS = "os"
+HYPERV = "hypervisor"
+VCPUS = "vcpus"
+MEMORY = "memory"
+AUTO_BALANCE = "auto_balance"
+DISK_TEMPLATE = "disk_template"
+TAGS = "tags"
+VERSION = "version"
+
+# Instance IDs of System and SCSI controller
+INSTANCE_ID = {
+  "system": 0,
+  "vcpus": 1,
+  "memory": 2,
+  "scsi": 3,
+}
+
+# Disk format descriptions
+DISK_FORMAT = {
+  RAW: "http://en.wikipedia.org/wiki/Byte",
+  VMDK: "http://www.vmware.com/interfaces/specifications/vmdk.html"
+          "#monolithicSparse",
+  COW: "http://www.gnome.org/~markmc/qcow-image-format.html",
+}
+
+
+def CheckQemuImg():
+  """ Make sure that qemu-img is present before performing operations.
+
+  @raise errors.OpPrereqError: when qemu-img was not found in the system
+
+  """
+  if not constants.QEMUIMG_PATH:
+    raise errors.OpPrereqError("qemu-img not found at build time, unable"
+                               " to continue", errors.ECODE_STATE)
+
+
+def LinkFile(old_path, prefix=None, suffix=None, directory=None):
+  """Create link with a given prefix and suffix.
+
+  This is a wrapper over os.link. It tries to create a hard link for given file,
+  but instead of rising error when file exists, the function changes the name
+  a little bit.
+
+  @type old_path:string
+  @param old_path: path to the file that is to be linked
+  @type prefix: string
+  @param prefix: prefix of filename for the link
+  @type suffix: string
+  @param suffix: suffix of the filename for the link
+  @type directory: string
+  @param directory: directory of the link
+
+  @raise errors.OpPrereqError: when error on linking is different than
+    "File exists"
+
+  """
+  assert(prefix is not None or suffix is not None)
+  if directory is None:
+    directory = os.getcwd()
+  new_path = utils.PathJoin(directory, "%s%s" % (prefix, suffix))
+  counter = 1
+  while True:
+    try:
+      os.link(old_path, new_path)
+      break
+    except OSError, err:
+      if err.errno == errno.EEXIST:
+        new_path = utils.PathJoin(directory,
+                                  "%s_%s%s" % (prefix, counter, suffix))
+        counter += 1
+      else:
+        raise errors.OpPrereqError("Error moving the file %s to %s location:"
+                                   " %s" % (old_path, new_path, err),
+                                   errors.ECODE_ENVIRON)
+  return new_path
+
+
+class OVFReader(object):
+  """Reader class for OVF files.
+
+  @type files_list: list
+  @ivar files_list: list of files in the OVF package
+  @type tree: ET.ElementTree
+  @ivar tree: XML tree of the .ovf file
+  @type schema_name: string
+  @ivar schema_name: name of the .ovf file
+  @type input_dir: string
+  @ivar input_dir: directory in which the .ovf file resides
+
+  """
+  def __init__(self, input_path):
+    """Initialiaze the reader - load the .ovf file to XML parser.
+
+    It is assumed that names of manifesto (.mf), certificate (.cert) and ovf
+    files are the same. In order to account any other files as part of the ovf
+    package, they have to be explicitly mentioned in the Resources section
+    of the .ovf file.
+
+    @type input_path: string
+    @param input_path: absolute path to the .ovf file
+
+    @raise errors.OpPrereqError: when .ovf file is not a proper XML file or some
+      of the files mentioned in Resources section do not exist
+
+    """
+    self.tree = ET.ElementTree()
+    try:
+      self.tree.parse(input_path)
+    except (ParseError, xml.parsers.expat.ExpatError), err:
+      raise errors.OpPrereqError("Error while reading %s file: %s" %
+                                 (OVF_EXT, err), errors.ECODE_ENVIRON)
+
+    # Create a list of all files in the OVF package
+    (input_dir, input_file) = os.path.split(input_path)
+    (input_name, _) = os.path.splitext(input_file)
+    files_directory = utils.ListVisibleFiles(input_dir)
+    files_list = []
+    for file_name in files_directory:
+      (name, extension) = os.path.splitext(file_name)
+      if extension in FILE_EXTENSIONS and name == input_name:
+        files_list.append(file_name)
+    files_list += self._GetAttributes("{%s}References/{%s}File" %
+                                      (OVF_SCHEMA, OVF_SCHEMA),
+                                      "{%s}href" % OVF_SCHEMA)
+    for file_name in files_list:
+      file_path = utils.PathJoin(input_dir, file_name)
+      if not os.path.exists(file_path):
+        raise errors.OpPrereqError("File does not exist: %s" % file_path,
+                                   errors.ECODE_ENVIRON)
+    logging.info("Files in the OVF package: %s", " ".join(files_list))
+    self.files_list = files_list
+    self.input_dir = input_dir
+    self.schema_name = input_name
+
+  def _GetAttributes(self, path, attribute):
+    """Get specified attribute from all nodes accessible using given path.
+
+    Function follows the path from root node to the desired tags using path,
+    then reads the apropriate attribute values.
+
+    @type path: string
+    @param path: path of nodes to visit
+    @type attribute: string
+    @param attribute: attribute for which we gather the information
+    @rtype: list
+    @return: for each accessible tag with the attribute value set, value of the
+      attribute
+
+    """
+    current_list = self.tree.findall(path)
+    results = [x.get(attribute) for x in current_list]
+    return filter(None, results)
+
+  def _GetElementMatchingAttr(self, path, match_attr):
+    """Searches for element on a path that matches certain attribute value.
+
+    Function follows the path from root node to the desired tags using path,
+    then searches for the first one matching the attribute value.
+
+    @type path: string
+    @param path: path of nodes to visit
+    @type match_attr: tuple
+    @param match_attr: pair (attribute, value) for which we search
+    @rtype: ET.ElementTree or None
+    @return: first element matching match_attr or None if nothing matches
+
+    """
+    potential_elements = self.tree.findall(path)
+    (attr, val) = match_attr
+    for elem in potential_elements:
+      if elem.get(attr) == val:
+        return elem
+    return None
+
+  def _GetElementMatchingText(self, path, match_text):
+    """Searches for element on a path that matches certain text value.
+
+    Function follows the path from root node to the desired tags using path,
+    then searches for the first one matching the text value.
+
+    @type path: string
+    @param path: path of nodes to visit
+    @type match_text: tuple
+    @param match_text: pair (node, text) for which we search
+    @rtype: ET.ElementTree or None
+    @return: first element matching match_text or None if nothing matches
+
+    """
+    potential_elements = self.tree.findall(path)
+    (node, text) = match_text
+    for elem in potential_elements:
+      if elem.findtext(node) == text:
+        return elem
+    return None
+
+  @staticmethod
+  def _GetDictParameters(root, schema):
+    """Reads text in all children and creates the dictionary from the contents.
+
+    @type root: ET.ElementTree or None
+    @param root: father of the nodes we want to collect data about
+    @type schema: string
+    @param schema: schema name to be removed from the tag
+    @rtype: dict
+    @return: dictionary containing tags and their text contents, tags have their
+      schema fragment removed or empty dictionary, when root is None
+
+    """
+    if not root:
+      return {}
+    results = {}
+    for element in list(root):
+      pref_len = len("{%s}" % schema)
+      assert(schema in element.tag)
+      tag = element.tag[pref_len:]
+      results[tag] = element.text
+    return results
+
+  def VerifyManifest(self):
+    """Verifies manifest for the OVF package, if one is given.
+
+    @raise errors.OpPrereqError: if SHA1 checksums do not match
+
+    """
+    if "%s%s" % (self.schema_name, MF_EXT) in self.files_list:
+      logging.warning("Verifying SHA1 checksums, this may take a while")
+      manifest_filename = "%s%s" % (self.schema_name, MF_EXT)
+      manifest_path = utils.PathJoin(self.input_dir, manifest_filename)
+      manifest_content = utils.ReadFile(manifest_path).splitlines()
+      manifest_files = {}
+      regexp = r"SHA1\((\S+)\)= (\S+)"
+      for line in manifest_content:
+        match = re.match(regexp, line)
+        if match:
+          file_name = match.group(1)
+          sha1_sum = match.group(2)
+          manifest_files[file_name] = sha1_sum
+      files_with_paths = [utils.PathJoin(self.input_dir, file_name)
+                          for file_name in self.files_list]
+      sha1_sums = utils.FingerprintFiles(files_with_paths)
+      for file_name, value in manifest_files.iteritems():
+        if sha1_sums.get(utils.PathJoin(self.input_dir, file_name)) != value:
+          raise errors.OpPrereqError("SHA1 checksum of %s does not match the"
+                                     " value in manifest file" % file_name,
+                                     errors.ECODE_ENVIRON)
+      logging.info("SHA1 checksums verified")
+
+  def GetInstanceName(self):
+    """Provides information about instance name.
+
+    @rtype: string
+    @return: instance name string
+
+    """
+    find_name = "{%s}VirtualSystem/{%s}Name" % (OVF_SCHEMA, OVF_SCHEMA)
+    return self.tree.findtext(find_name)
+
+  def GetDiskTemplate(self):
+    """Returns disk template from .ovf file
+
+    @rtype: string or None
+    @return: name of the template
+    """
+    find_template = ("{%s}GanetiSection/{%s}DiskTemplate" %
+                     (GANETI_SCHEMA, GANETI_SCHEMA))
+    return self.tree.findtext(find_template)
+
+  def GetHypervisorData(self):
+    """Provides hypervisor information - hypervisor name and options.
+
+    @rtype: dict
+    @return: dictionary containing name of the used hypervisor and all the
+      specified options
+
+    """
+    hypervisor_search = ("{%s}GanetiSection/{%s}Hypervisor" %
+                         (GANETI_SCHEMA, GANETI_SCHEMA))
+    hypervisor_data = self.tree.find(hypervisor_search)
+    if not hypervisor_data:
+      return {"hypervisor_name": constants.VALUE_AUTO}
+    results = {
+      "hypervisor_name": hypervisor_data.findtext("{%s}Name" % GANETI_SCHEMA,
+                                                  default=constants.VALUE_AUTO),
+    }
+    parameters = hypervisor_data.find("{%s}Parameters" % GANETI_SCHEMA)
+    results.update(self._GetDictParameters(parameters, GANETI_SCHEMA))
+    return results
+
+  def GetOSData(self):
+    """ Provides operating system information - os name and options.
+
+    @rtype: dict
+    @return: dictionary containing name and options for the chosen OS
+
+    """
+    results = {}
+    os_search = ("{%s}GanetiSection/{%s}OperatingSystem" %
+                 (GANETI_SCHEMA, GANETI_SCHEMA))
+    os_data = self.tree.find(os_search)
+    if os_data:
+      results["os_name"] = os_data.findtext("{%s}Name" % GANETI_SCHEMA)
+      parameters = os_data.find("{%s}Parameters" % GANETI_SCHEMA)
+      results.update(self._GetDictParameters(parameters, GANETI_SCHEMA))
+    return results
+
+  def GetBackendData(self):
+    """ Provides backend information - vcpus, memory, auto balancing options.
+
+    @rtype: dict
+    @return: dictionary containing options for vcpus, memory and auto balance
+      settings
+
+    """
+    results = {}
+
+    find_vcpus = ("{%s}VirtualSystem/{%s}VirtualHardwareSection/{%s}Item" %
+                   (OVF_SCHEMA, OVF_SCHEMA, OVF_SCHEMA))
+    match_vcpus = ("{%s}ResourceType" % RASD_SCHEMA, RASD_TYPE["vcpus"])
+    vcpus = self._GetElementMatchingText(find_vcpus, match_vcpus)
+    if vcpus:
+      vcpus_count = vcpus.findtext("{%s}VirtualQuantity" % RASD_SCHEMA,
+                                   default=constants.VALUE_AUTO)
+    else:
+      vcpus_count = constants.VALUE_AUTO
+    results["vcpus"] = str(vcpus_count)
+
+    find_memory = find_vcpus
+    match_memory = ("{%s}ResourceType" % RASD_SCHEMA, RASD_TYPE["memory"])
+    memory = self._GetElementMatchingText(find_memory, match_memory)
+    memory_raw = None
+    if memory:
+      alloc_units = memory.findtext("{%s}AllocationUnits" % RASD_SCHEMA)
+      matching_units = [units for units, variants in ALLOCATION_UNITS.items()
+                        if alloc_units.lower() in variants]
+      if matching_units == []:
+        raise errors.OpPrereqError("Unit %s for RAM memory unknown" %
+                                   alloc_units, errors.ECODE_INVAL)
+      units = matching_units[0]
+      memory_raw = int(memory.findtext("{%s}VirtualQuantity" % RASD_SCHEMA,
+                                       default=constants.VALUE_AUTO))
+      memory_count = CONVERT_UNITS_TO_MB[units](memory_raw)
+    else:
+      memory_count = constants.VALUE_AUTO
+    results["memory"] = str(memory_count)
+
+    find_balance = ("{%s}GanetiSection/{%s}AutoBalance" %
+                    (GANETI_SCHEMA, GANETI_SCHEMA))
+    balance = self.tree.findtext(find_balance, default=constants.VALUE_AUTO)
+    results["auto_balance"] = balance
+
+    return results
+
+  def GetTagsData(self):
+    """Provides tags information for instance.
+
+    @rtype: string or None
+    @return: string of comma-separated tags for the instance
+
+    """
+    find_tags = "{%s}GanetiSection/{%s}Tags" % (GANETI_SCHEMA, GANETI_SCHEMA)
+    results = self.tree.findtext(find_tags)
+    if results:
+      return results
+    else:
+      return None
+
+  def GetVersionData(self):
+    """Provides version number read from .ovf file
+
+    @rtype: string
+    @return: string containing the version number
+
+    """
+    find_version = ("{%s}GanetiSection/{%s}Version" %
+                    (GANETI_SCHEMA, GANETI_SCHEMA))
+    return self.tree.findtext(find_version)
+
+  def GetNetworkData(self):
+    """Provides data about the network in the OVF instance.
+
+    The method gathers the data about networks used by OVF instance. It assumes
+    that 'name' tag means something - in essence, if it contains one of the
+    words 'bridged' or 'routed' then that will be the mode of this network in
+    Ganeti. The information about the network can be either in GanetiSection or
+    VirtualHardwareSection.
+
+    @rtype: dict
+    @return: dictionary containing all the network information
+
+    """
+    results = {}
+    networks_search = ("{%s}NetworkSection/{%s}Network" %
+                       (OVF_SCHEMA, OVF_SCHEMA))
+    network_names = self._GetAttributes(networks_search,
+                                        "{%s}name" % OVF_SCHEMA)
+    required = ["ip", "mac", "link", "mode", "network"]
+    for (counter, network_name) in enumerate(network_names):
+      network_search = ("{%s}VirtualSystem/{%s}VirtualHardwareSection/{%s}Item"
+                        % (OVF_SCHEMA, OVF_SCHEMA, OVF_SCHEMA))
+      ganeti_search = ("{%s}GanetiSection/{%s}Network/{%s}Nic" %
+                       (GANETI_SCHEMA, GANETI_SCHEMA, GANETI_SCHEMA))
+      network_match = ("{%s}Connection" % RASD_SCHEMA, network_name)
+      ganeti_match = ("{%s}name" % OVF_SCHEMA, network_name)
+      network_data = self._GetElementMatchingText(network_search, network_match)
+      network_ganeti_data = self._GetElementMatchingAttr(ganeti_search,
+                                                         ganeti_match)
+
+      ganeti_data = {}
+      if network_ganeti_data:
+        ganeti_data["mode"] = network_ganeti_data.findtext("{%s}Mode" %
+                                                           GANETI_SCHEMA)
+        ganeti_data["mac"] = network_ganeti_data.findtext("{%s}MACAddress" %
+                                                          GANETI_SCHEMA)
+        ganeti_data["ip"] = network_ganeti_data.findtext("{%s}IPAddress" %
+                                                         GANETI_SCHEMA)
+        ganeti_data["link"] = network_ganeti_data.findtext("{%s}Link" %
+                                                           GANETI_SCHEMA)
+        ganeti_data["network"] = network_ganeti_data.findtext("{%s}Net" %
+                                                              GANETI_SCHEMA)
+      mac_data = None
+      if network_data:
+        mac_data = network_data.findtext("{%s}Address" % RASD_SCHEMA)
+
+      network_name = network_name.lower()
+
+      # First, some not Ganeti-specific information is collected
+      if constants.NIC_MODE_BRIDGED in network_name:
+        results["nic%s_mode" % counter] = "bridged"
+      elif constants.NIC_MODE_ROUTED in network_name:
+        results["nic%s_mode" % counter] = "routed"
+      results["nic%s_mac" % counter] = mac_data
+
+      # GanetiSection data overrides 'manually' collected data
+      for name, value in ganeti_data.iteritems():
+        results["nic%s_%s" % (counter, name)] = value
+
+      # Bridged network has no IP - unless specifically stated otherwise
+      if (results.get("nic%s_mode" % counter) == "bridged" and
+          not results.get("nic%s_ip" % counter)):
+        results["nic%s_ip" % counter] = constants.VALUE_NONE
+
+      for option in required:
+        if not results.get("nic%s_%s" % (counter, option)):
+          results["nic%s_%s" % (counter, option)] = constants.VALUE_AUTO
+
+    if network_names:
+      results["nic_count"] = str(len(network_names))
+    return results
+
+  def GetDisksNames(self):
+    """Provides list of file names for the disks used by the instance.
+
+    @rtype: list
+    @return: list of file names, as referenced in .ovf file
+
+    """
+    results = []
+    disks_search = "{%s}DiskSection/{%s}Disk" % (OVF_SCHEMA, OVF_SCHEMA)
+    disk_ids = self._GetAttributes(disks_search, "{%s}fileRef" % OVF_SCHEMA)
+    for disk in disk_ids:
+      disk_search = "{%s}References/{%s}File" % (OVF_SCHEMA, OVF_SCHEMA)
+      disk_match = ("{%s}id" % OVF_SCHEMA, disk)
+      disk_elem = self._GetElementMatchingAttr(disk_search, disk_match)
+      if disk_elem is None:
+        raise errors.OpPrereqError("%s file corrupted - disk %s not found in"
+                                   " references" % (OVF_EXT, disk),
+                                   errors.ECODE_ENVIRON)
+      disk_name = disk_elem.get("{%s}href" % OVF_SCHEMA)
+      disk_compression = disk_elem.get("{%s}compression" % OVF_SCHEMA)
+      results.append((disk_name, disk_compression))
+    return results
+
+
+def SubElementText(parent, tag, text, attrib={}, **extra):
+# pylint: disable=W0102
+  """This is just a wrapper on ET.SubElement that always has text content.
+
+  """
+  if text is None:
+    return None
+  elem = ET.SubElement(parent, tag, attrib=attrib, **extra)
+  elem.text = str(text)
+  return elem
+
+
+class OVFWriter(object):
+  """Writer class for OVF files.
+
+  @type tree: ET.ElementTree
+  @ivar tree: XML tree that we are constructing
+  @type virtual_system_type: string
+  @ivar virtual_system_type: value of vssd:VirtualSystemType, for external usage
+    in VMWare this requires to be vmx
+  @type hardware_list: list
+  @ivar hardware_list: list of items prepared for VirtualHardwareSection
+  @type next_instance_id: int
+  @ivar next_instance_id: next instance id to be used when creating elements on
+    hardware_list
+
+  """
+  def __init__(self, has_gnt_section):
+    """Initialize the writer - set the top element.
+
+    @type has_gnt_section: bool
+    @param has_gnt_section: if the Ganeti schema should be added - i.e. this
+      means that Ganeti section will be present
+
+    """
+    env_attribs = {
+      "xmlns:xsi": XML_SCHEMA,
+      "xmlns:vssd": VSSD_SCHEMA,
+      "xmlns:rasd": RASD_SCHEMA,
+      "xmlns:ovf": OVF_SCHEMA,
+      "xmlns": OVF_SCHEMA,
+      "xml:lang": "en-US",
+    }
+    if has_gnt_section:
+      env_attribs["xmlns:gnt"] = GANETI_SCHEMA
+      self.virtual_system_type = VS_TYPE["ganeti"]
+    else:
+      self.virtual_system_type = VS_TYPE["external"]
+    self.tree = ET.Element("Envelope", attrib=env_attribs)
+    self.hardware_list = []
+    # INSTANCE_ID contains statically assigned IDs, starting from 0
+    self.next_instance_id = len(INSTANCE_ID) # FIXME: hackish
+
+  def SaveDisksData(self, disks):
+    """Convert disk information to certain OVF sections.
+
+    @type disks: list
+    @param disks: list of dictionaries of disk options from config.ini
+
+    """
+    references = ET.SubElement(self.tree, "References")
+    disk_section = ET.SubElement(self.tree, "DiskSection")
+    SubElementText(disk_section, "Info", "Virtual disk information")
+    for counter, disk in enumerate(disks):
+      file_id = "file%s" % counter
+      disk_id = "disk%s" % counter
+      file_attribs = {
+        "ovf:href": disk["path"],
+        "ovf:size": str(disk["real-size"]),
+        "ovf:id": file_id,
+      }
+      disk_attribs = {
+        "ovf:capacity": str(disk["virt-size"]),
+        "ovf:diskId": disk_id,
+        "ovf:fileRef": file_id,
+        "ovf:format": DISK_FORMAT.get(disk["format"], disk["format"]),
+      }
+      if "compression" in disk:
+        file_attribs["ovf:compression"] = disk["compression"]
+      ET.SubElement(references, "File", attrib=file_attribs)
+      ET.SubElement(disk_section, "Disk", attrib=disk_attribs)
+
+      # Item in VirtualHardwareSection creation
+      disk_item = ET.Element("Item")
+      SubElementText(disk_item, "rasd:ElementName", disk_id)
+      SubElementText(disk_item, "rasd:HostResource", "ovf:/disk/%s" % disk_id)
+      SubElementText(disk_item, "rasd:InstanceID", self.next_instance_id)
+      SubElementText(disk_item, "rasd:Parent", INSTANCE_ID["scsi"])
+      SubElementText(disk_item, "rasd:ResourceType", RASD_TYPE["disk"])
+      self.hardware_list.append(disk_item)
+      self.next_instance_id += 1
+
+  def SaveNetworksData(self, networks):
+    """Convert network information to NetworkSection.
+
+    @type networks: list
+    @param networks: list of dictionaries of network options form config.ini
+
+    """
+    network_section = ET.SubElement(self.tree, "NetworkSection")
+    SubElementText(network_section, "Info", "List of logical networks")
+    for counter, network in enumerate(networks):
+      network_name = "%s%s" % (network["mode"], counter)
+      network_attrib = {"ovf:name": network_name}
+      ET.SubElement(network_section, "Network", attrib=network_attrib)
+
+      # Item in VirtualHardwareSection creation
+      network_item = ET.Element("Item")
+      SubElementText(network_item, "rasd:Address", network["mac"])
+      SubElementText(network_item, "rasd:Connection", network_name)
+      SubElementText(network_item, "rasd:ElementName", network_name)
+      SubElementText(network_item, "rasd:InstanceID", self.next_instance_id)
+      SubElementText(network_item, "rasd:ResourceType",
+                     RASD_TYPE["ethernet-adapter"])
+      self.hardware_list.append(network_item)
+      self.next_instance_id += 1
+
+  @staticmethod
+  def _SaveNameAndParams(root, data):
+    """Save name and parameters information under root using data.
+
+    @type root: ET.Element
+    @param root: root element for the Name and Parameters
+    @type data: dict
+    @param data: data from which we gather the values
+
+    """
+    assert(data.get("name"))
+    name = SubElementText(root, "gnt:Name", data["name"])
+    params = ET.SubElement(root, "gnt:Parameters")
+    for name, value in data.iteritems():
+      if name != "name":
+        SubElementText(params, "gnt:%s" % name, value)
+
+  def SaveGanetiData(self, ganeti, networks):
+    """Convert Ganeti-specific information to GanetiSection.
+
+    @type ganeti: dict
+    @param ganeti: dictionary of Ganeti-specific options from config.ini
+    @type networks: list
+    @param networks: list of dictionaries of network options form config.ini
+
+    """
+    ganeti_section = ET.SubElement(self.tree, "gnt:GanetiSection")
+
+    SubElementText(ganeti_section, "gnt:Version", ganeti.get("version"))
+    SubElementText(ganeti_section, "gnt:DiskTemplate",
+                   ganeti.get("disk_template"))
+    SubElementText(ganeti_section, "gnt:AutoBalance",
+                   ganeti.get("auto_balance"))
+    SubElementText(ganeti_section, "gnt:Tags", ganeti.get("tags"))
+
+    osys = ET.SubElement(ganeti_section, "gnt:OperatingSystem")
+    self._SaveNameAndParams(osys, ganeti["os"])
+
+    hypervisor = ET.SubElement(ganeti_section, "gnt:Hypervisor")
+    self._SaveNameAndParams(hypervisor, ganeti["hypervisor"])
+
+    network_section = ET.SubElement(ganeti_section, "gnt:Network")
+    for counter, network in enumerate(networks):
+      network_name = "%s%s" % (network["mode"], counter)
+      nic_attrib = {"ovf:name": network_name}
+      nic = ET.SubElement(network_section, "gnt:Nic", attrib=nic_attrib)
+      SubElementText(nic, "gnt:Mode", network["mode"])
+      SubElementText(nic, "gnt:MACAddress", network["mac"])
+      SubElementText(nic, "gnt:IPAddress", network["ip"])
+      SubElementText(nic, "gnt:Link", network["link"])
+      SubElementText(nic, "gnt:Net", network["network"])
+
+  def SaveVirtualSystemData(self, name, vcpus, memory):
+    """Convert virtual system information to OVF sections.
+
+    @type name: string
+    @param name: name of the instance
+    @type vcpus: int
+    @param vcpus: number of VCPUs
+    @type memory: int
+    @param memory: RAM memory in MB
+
+    """
+    assert(vcpus > 0)
+    assert(memory > 0)
+    vs_attrib = {"ovf:id": name}
+    virtual_system = ET.SubElement(self.tree, "VirtualSystem", attrib=vs_attrib)
+    SubElementText(virtual_system, "Info", "A virtual machine")
+
+    name_section = ET.SubElement(virtual_system, "Name")
+    name_section.text = name
+    os_attrib = {"ovf:id": "0"}
+    os_section = ET.SubElement(virtual_system, "OperatingSystemSection",
+                               attrib=os_attrib)
+    SubElementText(os_section, "Info", "Installed guest operating system")
+    hardware_section = ET.SubElement(virtual_system, "VirtualHardwareSection")
+    SubElementText(hardware_section, "Info", "Virtual hardware requirements")
+
+    # System description
+    system = ET.SubElement(hardware_section, "System")
+    SubElementText(system, "vssd:ElementName", "Virtual Hardware Family")
+    SubElementText(system, "vssd:InstanceID", INSTANCE_ID["system"])
+    SubElementText(system, "vssd:VirtualSystemIdentifier", name)
+    SubElementText(system, "vssd:VirtualSystemType", self.virtual_system_type)
+
+    # Item for vcpus
+    vcpus_item = ET.SubElement(hardware_section, "Item")
+    SubElementText(vcpus_item, "rasd:ElementName",
+                   "%s virtual CPU(s)" % vcpus)
+    SubElementText(vcpus_item, "rasd:InstanceID", INSTANCE_ID["vcpus"])
+    SubElementText(vcpus_item, "rasd:ResourceType", RASD_TYPE["vcpus"])
+    SubElementText(vcpus_item, "rasd:VirtualQuantity", vcpus)
+
+    # Item for memory
+    memory_item = ET.SubElement(hardware_section, "Item")
+    SubElementText(memory_item, "rasd:AllocationUnits", "byte * 2^20")
+    SubElementText(memory_item, "rasd:ElementName", "%sMB of memory" % memory)
+    SubElementText(memory_item, "rasd:InstanceID", INSTANCE_ID["memory"])
+    SubElementText(memory_item, "rasd:ResourceType", RASD_TYPE["memory"])
+    SubElementText(memory_item, "rasd:VirtualQuantity", memory)
+
+    # Item for scsi controller
+    scsi_item = ET.SubElement(hardware_section, "Item")
+    SubElementText(scsi_item, "rasd:Address", INSTANCE_ID["system"])
+    SubElementText(scsi_item, "rasd:ElementName", "scsi_controller0")
+    SubElementText(scsi_item, "rasd:InstanceID", INSTANCE_ID["scsi"])
+    SubElementText(scsi_item, "rasd:ResourceSubType", SCSI_SUBTYPE)
+    SubElementText(scsi_item, "rasd:ResourceType", RASD_TYPE["scsi-controller"])
+
+    # Other items - from self.hardware_list
+    for item in self.hardware_list:
+      hardware_section.append(item)
+
+  def PrettyXmlDump(self):
+    """Formatter of the XML file.
+
+    @rtype: string
+    @return: XML tree in the form of nicely-formatted string
+
+    """
+    raw_string = ET.tostring(self.tree)
+    parsed_xml = xml.dom.minidom.parseString(raw_string)
+    xml_string = parsed_xml.toprettyxml(indent="  ")
+    text_re = re.compile(">\n\s+([^<>\s].*?)\n\s+</", re.DOTALL)
+    return text_re.sub(">\g<1></", xml_string)
 
 
 class Converter(object):
 
 
 class Converter(object):
@@ -61,7 +868,8 @@ class Converter(object):
     """
     input_path = os.path.abspath(input_path)
     if not os.path.isfile(input_path):
     """
     input_path = os.path.abspath(input_path)
     if not os.path.isfile(input_path):
-      raise errors.OpPrereqError("File does not exist: %s" % input_path)
+      raise errors.OpPrereqError("File does not exist: %s" % input_path,
+                                 errors.ECODE_ENVIRON)
     self.options = options
     self.temp_file_manager = utils.TemporaryFileManager()
     self.temp_dir = None
     self.options = options
     self.temp_file_manager = utils.TemporaryFileManager()
     self.temp_dir = None
@@ -77,6 +885,115 @@ class Converter(object):
     """
     raise NotImplementedError()
 
     """
     raise NotImplementedError()
 
+  def _CompressDisk(self, disk_path, compression, action):
+    """Performs (de)compression on the disk and returns the new path
+
+    @type disk_path: string
+    @param disk_path: path to the disk
+    @type compression: string
+    @param compression: compression type
+    @type action: string
+    @param action: whether the action is compression or decompression
+    @rtype: string
+    @return: new disk path after (de)compression
+
+    @raise errors.OpPrereqError: disk (de)compression failed or "compression"
+      is not supported
+
+    """
+    assert(action in ALLOWED_ACTIONS)
+    # For now we only support gzip, as it is used in ovftool
+    if compression != COMPRESSION_TYPE:
+      raise errors.OpPrereqError("Unsupported compression type: %s"
+                                 % compression, errors.ECODE_INVAL)
+    disk_file = os.path.basename(disk_path)
+    if action == DECOMPRESS:
+      (disk_name, _) = os.path.splitext(disk_file)
+      prefix = disk_name
+    elif action == COMPRESS:
+      prefix = disk_file
+    new_path = utils.GetClosedTempfile(suffix=COMPRESSION_EXT, prefix=prefix,
+                                       dir=self.output_dir)
+    self.temp_file_manager.Add(new_path)
+    args = ["gzip", "-c", disk_path]
+    run_result = utils.RunCmd(args, output=new_path)
+    if run_result.failed:
+      raise errors.OpPrereqError("Disk %s failed with output: %s"
+                                 % (action, run_result.stderr),
+                                 errors.ECODE_ENVIRON)
+    logging.info("The %s of the disk is completed", action)
+    return (COMPRESSION_EXT, new_path)
+
+  def _ConvertDisk(self, disk_format, disk_path):
+    """Performes conversion to specified format.
+
+    @type disk_format: string
+    @param disk_format: format to which the disk should be converted
+    @type disk_path: string
+    @param disk_path: path to the disk that should be converted
+    @rtype: string
+    @return path to the output disk
+
+    @raise errors.OpPrereqError: convertion of the disk failed
+
+    """
+    CheckQemuImg()
+    disk_file = os.path.basename(disk_path)
+    (disk_name, disk_extension) = os.path.splitext(disk_file)
+    if disk_extension != disk_format:
+      logging.warning("Conversion of disk image to %s format, this may take"
+                      " a while", disk_format)
+
+    new_disk_path = utils.GetClosedTempfile(
+      suffix=".%s" % disk_format, prefix=disk_name, dir=self.output_dir)
+    self.temp_file_manager.Add(new_disk_path)
+    args = [
+      constants.QEMUIMG_PATH,
+      "convert",
+      "-O",
+      disk_format,
+      disk_path,
+      new_disk_path,
+    ]
+    run_result = utils.RunCmd(args, cwd=os.getcwd())
+    if run_result.failed:
+      raise errors.OpPrereqError("Convertion to %s failed, qemu-img output was"
+                                 ": %s" % (disk_format, run_result.stderr),
+                                 errors.ECODE_ENVIRON)
+    return (".%s" % disk_format, new_disk_path)
+
+  @staticmethod
+  def _GetDiskQemuInfo(disk_path, regexp):
+    """Figures out some information of the disk using qemu-img.
+
+    @type disk_path: string
+    @param disk_path: path to the disk we want to know the format of
+    @type regexp: string
+    @param regexp: string that has to be matched, it has to contain one group
+    @rtype: string
+    @return: disk format
+
+    @raise errors.OpPrereqError: format information cannot be retrieved
+
+    """
+    CheckQemuImg()
+    args = [constants.QEMUIMG_PATH, "info", disk_path]
+    run_result = utils.RunCmd(args, cwd=os.getcwd())
+    if run_result.failed:
+      raise errors.OpPrereqError("Gathering info about the disk using qemu-img"
+                                 " failed, output was: %s" % run_result.stderr,
+                                 errors.ECODE_ENVIRON)
+    result = run_result.output
+    regexp = r"%s" % regexp
+    match = re.search(regexp, result)
+    if match:
+      disk_format = match.group(1)
+    else:
+      raise errors.OpPrereqError("No file information matching %s found in:"
+                                 " %s" % (regexp, result),
+                                 errors.ECODE_ENVIRON)
+    return disk_format
+
   def Parse(self):
     """Parses the data and creates a structure containing all required info.
 
   def Parse(self):
     """Parses the data and creates a structure containing all required info.
 
@@ -100,22 +1017,836 @@ class Converter(object):
 
 
 class OVFImporter(Converter):
 
 
 class OVFImporter(Converter):
+  """Converter from OVF to Ganeti config file.
+
+  @type input_dir: string
+  @ivar input_dir: directory in which the .ovf file resides
+  @type output_dir: string
+  @ivar output_dir: directory to which the results of conversion shall be
+    written
+  @type input_path: string
+  @ivar input_path: complete path to the .ovf file
+  @type ovf_reader: L{OVFReader}
+  @ivar ovf_reader: OVF reader instance collects data from .ovf file
+  @type results_name: string
+  @ivar results_name: name of imported instance
+  @type results_template: string
+  @ivar results_template: disk template read from .ovf file or command line
+    arguments
+  @type results_hypervisor: dict
+  @ivar results_hypervisor: hypervisor information gathered from .ovf file or
+    command line arguments
+  @type results_os: dict
+  @ivar results_os: operating system information gathered from .ovf file or
+    command line arguments
+  @type results_backend: dict
+  @ivar results_backend: backend information gathered from .ovf file or
+    command line arguments
+  @type results_tags: string
+  @ivar results_tags: string containing instance-specific tags
+  @type results_version: string
+  @ivar results_version: version as required by Ganeti import
+  @type results_network: dict
+  @ivar results_network: network information gathered from .ovf file or command
+    line arguments
+  @type results_disk: dict
+  @ivar results_disk: disk information gathered from .ovf file or command line
+    arguments
+
+  """
   def _ReadInputData(self, input_path):
   def _ReadInputData(self, input_path):
-    pass
+    """Reads the data on which the conversion will take place.
+
+    @type input_path: string
+    @param input_path: absolute path to the .ovf or .ova input file
+
+    @raise errors.OpPrereqError: if input file is neither .ovf nor .ova
+
+    """
+    (input_dir, input_file) = os.path.split(input_path)
+    (_, input_extension) = os.path.splitext(input_file)
+
+    if input_extension == OVF_EXT:
+      logging.info("%s file extension found, no unpacking necessary", OVF_EXT)
+      self.input_dir = input_dir
+      self.input_path = input_path
+      self.temp_dir = None
+    elif input_extension == OVA_EXT:
+      logging.info("%s file extension found, proceeding to unpacking", OVA_EXT)
+      self._UnpackOVA(input_path)
+    else:
+      raise errors.OpPrereqError("Unknown file extension; expected %s or %s"
+                                 " file" % (OVA_EXT, OVF_EXT),
+                                 errors.ECODE_INVAL)
+    assert ((input_extension == OVA_EXT and self.temp_dir) or
+            (input_extension == OVF_EXT and not self.temp_dir))
+    assert self.input_dir in self.input_path
+
+    if self.options.output_dir:
+      self.output_dir = os.path.abspath(self.options.output_dir)
+      if (os.path.commonprefix([pathutils.EXPORT_DIR, self.output_dir]) !=
+          pathutils.EXPORT_DIR):
+        logging.warning("Export path is not under %s directory, import to"
+                        " Ganeti using gnt-backup may fail",
+                        pathutils.EXPORT_DIR)
+    else:
+      self.output_dir = pathutils.EXPORT_DIR
+
+    self.ovf_reader = OVFReader(self.input_path)
+    self.ovf_reader.VerifyManifest()
+
+  def _UnpackOVA(self, input_path):
+    """Unpacks the .ova package into temporary directory.
+
+    @type input_path: string
+    @param input_path: path to the .ova package file
+
+    @raise errors.OpPrereqError: if file is not a proper tarball, one of the
+        files in the archive seem malicious (e.g. path starts with '../') or
+        .ova package does not contain .ovf file
+
+    """
+    input_name = None
+    if not tarfile.is_tarfile(input_path):
+      raise errors.OpPrereqError("The provided %s file is not a proper tar"
+                                 " archive" % OVA_EXT, errors.ECODE_ENVIRON)
+    ova_content = tarfile.open(input_path)
+    temp_dir = tempfile.mkdtemp()
+    self.temp_dir = temp_dir
+    for file_name in ova_content.getnames():
+      file_normname = os.path.normpath(file_name)
+      try:
+        utils.PathJoin(temp_dir, file_normname)
+      except ValueError, err:
+        raise errors.OpPrereqError("File %s inside %s package is not safe" %
+                                   (file_name, OVA_EXT), errors.ECODE_ENVIRON)
+      if file_name.endswith(OVF_EXT):
+        input_name = file_name
+    if not input_name:
+      raise errors.OpPrereqError("No %s file in %s package found" %
+                                 (OVF_EXT, OVA_EXT), errors.ECODE_ENVIRON)
+    logging.warning("Unpacking the %s archive, this may take a while",
+                    input_path)
+    self.input_dir = temp_dir
+    self.input_path = utils.PathJoin(self.temp_dir, input_name)
+    try:
+      try:
+        extract = ova_content.extractall
+      except AttributeError:
+        # This is a prehistorical case of using python < 2.5
+        for member in ova_content.getmembers():
+          ova_content.extract(member, path=self.temp_dir)
+      else:
+        extract(self.temp_dir)
+    except tarfile.TarError, err:
+      raise errors.OpPrereqError("Error while extracting %s archive: %s" %
+                                 (OVA_EXT, err), errors.ECODE_ENVIRON)
+    logging.info("OVA package extracted to %s directory", self.temp_dir)
 
   def Parse(self):
 
   def Parse(self):
-    pass
+    """Parses the data and creates a structure containing all required info.
+
+    The method reads the information given either as a command line option or as
+    a part of the OVF description.
+
+    @raise errors.OpPrereqError: if some required part of the description of
+      virtual instance is missing or unable to create output directory
+
+    """
+    self.results_name = self._GetInfo("instance name", self.options.name,
+                                      self._ParseNameOptions,
+                                      self.ovf_reader.GetInstanceName)
+    if not self.results_name:
+      raise errors.OpPrereqError("Name of instance not provided",
+                                 errors.ECODE_INVAL)
+
+    self.output_dir = utils.PathJoin(self.output_dir, self.results_name)
+    try:
+      utils.Makedirs(self.output_dir)
+    except OSError, err:
+      raise errors.OpPrereqError("Failed to create directory %s: %s" %
+                                 (self.output_dir, err), errors.ECODE_ENVIRON)
+
+    self.results_template = self._GetInfo(
+      "disk template", self.options.disk_template, self._ParseTemplateOptions,
+      self.ovf_reader.GetDiskTemplate)
+    if not self.results_template:
+      logging.info("Disk template not given")
+
+    self.results_hypervisor = self._GetInfo(
+      "hypervisor", self.options.hypervisor, self._ParseHypervisorOptions,
+      self.ovf_reader.GetHypervisorData)
+    assert self.results_hypervisor["hypervisor_name"]
+    if self.results_hypervisor["hypervisor_name"] == constants.VALUE_AUTO:
+      logging.debug("Default hypervisor settings from the cluster will be used")
+
+    self.results_os = self._GetInfo(
+      "OS", self.options.os, self._ParseOSOptions, self.ovf_reader.GetOSData)
+    if not self.results_os.get("os_name"):
+      raise errors.OpPrereqError("OS name must be provided",
+                                 errors.ECODE_INVAL)
+
+    self.results_backend = self._GetInfo(
+      "backend", self.options.beparams,
+      self._ParseBackendOptions, self.ovf_reader.GetBackendData)
+    assert self.results_backend.get("vcpus")
+    assert self.results_backend.get("memory")
+    assert self.results_backend.get("auto_balance") is not None
+
+    self.results_tags = self._GetInfo(
+      "tags", self.options.tags, self._ParseTags, self.ovf_reader.GetTagsData)
+
+    ovf_version = self.ovf_reader.GetVersionData()
+    if ovf_version:
+      self.results_version = ovf_version
+    else:
+      self.results_version = constants.EXPORT_VERSION
+
+    self.results_network = self._GetInfo(
+      "network", self.options.nics, self._ParseNicOptions,
+      self.ovf_reader.GetNetworkData, ignore_test=self.options.no_nics)
+
+    self.results_disk = self._GetInfo(
+      "disk", self.options.disks, self._ParseDiskOptions, self._GetDiskInfo,
+      ignore_test=self.results_template == constants.DT_DISKLESS)
+
+    if not self.results_disk and not self.results_network:
+      raise errors.OpPrereqError("Either disk specification or network"
+                                 " description must be present",
+                                 errors.ECODE_STATE)
+
+  @staticmethod
+  def _GetInfo(name, cmd_arg, cmd_function, nocmd_function,
+               ignore_test=False):
+    """Get information about some section - e.g. disk, network, hypervisor.
+
+    @type name: string
+    @param name: name of the section
+    @type cmd_arg: dict
+    @param cmd_arg: command line argument specific for section 'name'
+    @type cmd_function: callable
+    @param cmd_function: function to call if 'cmd_args' exists
+    @type nocmd_function: callable
+    @param nocmd_function: function to call if 'cmd_args' is not there
+
+    """
+    if ignore_test:
+      logging.info("Information for %s will be ignored", name)
+      return {}
+    if cmd_arg:
+      logging.info("Information for %s will be parsed from command line", name)
+      results = cmd_function()
+    else:
+      logging.info("Information for %s will be parsed from %s file",
+                   name, OVF_EXT)
+      results = nocmd_function()
+    logging.info("Options for %s were succesfully read", name)
+    return results
+
+  def _ParseNameOptions(self):
+    """Returns name if one was given in command line.
+
+    @rtype: string
+    @return: name of an instance
+
+    """
+    return self.options.name
+
+  def _ParseTemplateOptions(self):
+    """Returns disk template if one was given in command line.
+
+    @rtype: string
+    @return: disk template name
+
+    """
+    return self.options.disk_template
+
+  def _ParseHypervisorOptions(self):
+    """Parses hypervisor options given in a command line.
+
+    @rtype: dict
+    @return: dictionary containing name of the chosen hypervisor and all the
+      options
+
+    """
+    assert type(self.options.hypervisor) is tuple
+    assert len(self.options.hypervisor) == 2
+    results = {}
+    if self.options.hypervisor[0]:
+      results["hypervisor_name"] = self.options.hypervisor[0]
+    else:
+      results["hypervisor_name"] = constants.VALUE_AUTO
+    results.update(self.options.hypervisor[1])
+    return results
+
+  def _ParseOSOptions(self):
+    """Parses OS options given in command line.
+
+    @rtype: dict
+    @return: dictionary containing name of chosen OS and all its options
+
+    """
+    assert self.options.os
+    results = {}
+    results["os_name"] = self.options.os
+    results.update(self.options.osparams)
+    return results
+
+  def _ParseBackendOptions(self):
+    """Parses backend options given in command line.
+
+    @rtype: dict
+    @return: dictionary containing vcpus, memory and auto-balance options
+
+    """
+    assert self.options.beparams
+    backend = {}
+    backend.update(self.options.beparams)
+    must_contain = ["vcpus", "memory", "auto_balance"]
+    for element in must_contain:
+      if backend.get(element) is None:
+        backend[element] = constants.VALUE_AUTO
+    return backend
+
+  def _ParseTags(self):
+    """Returns tags list given in command line.
+
+    @rtype: string
+    @return: string containing comma-separated tags
+
+    """
+    return self.options.tags
+
+  def _ParseNicOptions(self):
+    """Parses network options given in a command line or as a dictionary.
+
+    @rtype: dict
+    @return: dictionary of network-related options
+
+    """
+    assert self.options.nics
+    results = {}
+    for (nic_id, nic_desc) in self.options.nics:
+      results["nic%s_mode" % nic_id] = \
+        nic_desc.get("mode", constants.VALUE_AUTO)
+      results["nic%s_mac" % nic_id] = nic_desc.get("mac", constants.VALUE_AUTO)
+      results["nic%s_link" % nic_id] = \
+        nic_desc.get("link", constants.VALUE_AUTO)
+      results["nic%s_network" % nic_id] = \
+        nic_desc.get("network", constants.VALUE_AUTO)
+      if nic_desc.get("mode") == "bridged":
+        results["nic%s_ip" % nic_id] = constants.VALUE_NONE
+      else:
+        results["nic%s_ip" % nic_id] = constants.VALUE_AUTO
+    results["nic_count"] = str(len(self.options.nics))
+    return results
+
+  def _ParseDiskOptions(self):
+    """Parses disk options given in a command line.
+
+    @rtype: dict
+    @return: dictionary of disk-related options
+
+    @raise errors.OpPrereqError: disk description does not contain size
+      information or size information is invalid or creation failed
+
+    """
+    CheckQemuImg()
+    assert self.options.disks
+    results = {}
+    for (disk_id, disk_desc) in self.options.disks:
+      results["disk%s_ivname" % disk_id] = "disk/%s" % disk_id
+      if disk_desc.get("size"):
+        try:
+          disk_size = utils.ParseUnit(disk_desc["size"])
+        except ValueError:
+          raise errors.OpPrereqError("Invalid disk size for disk %s: %s" %
+                                     (disk_id, disk_desc["size"]),
+                                     errors.ECODE_INVAL)
+        new_path = utils.PathJoin(self.output_dir, str(disk_id))
+        args = [
+          constants.QEMUIMG_PATH,
+          "create",
+          "-f",
+          "raw",
+          new_path,
+          disk_size,
+        ]
+        run_result = utils.RunCmd(args)
+        if run_result.failed:
+          raise errors.OpPrereqError("Creation of disk %s failed, output was:"
+                                     " %s" % (new_path, run_result.stderr),
+                                     errors.ECODE_ENVIRON)
+        results["disk%s_size" % disk_id] = str(disk_size)
+        results["disk%s_dump" % disk_id] = "disk%s.raw" % disk_id
+      else:
+        raise errors.OpPrereqError("Disks created for import must have their"
+                                   " size specified",
+                                   errors.ECODE_INVAL)
+    results["disk_count"] = str(len(self.options.disks))
+    return results
+
+  def _GetDiskInfo(self):
+    """Gathers information about disks used by instance, perfomes conversion.
+
+    @rtype: dict
+    @return: dictionary of disk-related options
+
+    @raise errors.OpPrereqError: disk is not in the same directory as .ovf file
+
+    """
+    results = {}
+    disks_list = self.ovf_reader.GetDisksNames()
+    for (counter, (disk_name, disk_compression)) in enumerate(disks_list):
+      if os.path.dirname(disk_name):
+        raise errors.OpPrereqError("Disks are not allowed to have absolute"
+                                   " paths or paths outside main OVF"
+                                   " directory", errors.ECODE_ENVIRON)
+      disk, _ = os.path.splitext(disk_name)
+      disk_path = utils.PathJoin(self.input_dir, disk_name)
+      if disk_compression not in NO_COMPRESSION:
+        _, disk_path = self._CompressDisk(disk_path, disk_compression,
+                                          DECOMPRESS)
+        disk, _ = os.path.splitext(disk)
+      if self._GetDiskQemuInfo(disk_path, "file format: (\S+)") != "raw":
+        logging.info("Conversion to raw format is required")
+      ext, new_disk_path = self._ConvertDisk("raw", disk_path)
+
+      final_disk_path = LinkFile(new_disk_path, prefix=disk, suffix=ext,
+                                 directory=self.output_dir)
+      final_name = os.path.basename(final_disk_path)
+      disk_size = os.path.getsize(final_disk_path) / (1024 * 1024)
+      results["disk%s_dump" % counter] = final_name
+      results["disk%s_size" % counter] = str(disk_size)
+      results["disk%s_ivname" % counter] = "disk/%s" % str(counter)
+    if disks_list:
+      results["disk_count"] = str(len(disks_list))
+    return results
 
   def Save(self):
 
   def Save(self):
-    pass
+    """Saves all the gathered information in a constant.EXPORT_CONF_FILE file.
+
+    @raise errors.OpPrereqError: when saving to config file failed
+
+    """
+    logging.info("Conversion was succesfull, saving %s in %s directory",
+                 constants.EXPORT_CONF_FILE, self.output_dir)
+    results = {
+      constants.INISECT_INS: {},
+      constants.INISECT_BEP: {},
+      constants.INISECT_EXP: {},
+      constants.INISECT_OSP: {},
+      constants.INISECT_HYP: {},
+    }
+
+    results[constants.INISECT_INS].update(self.results_disk)
+    results[constants.INISECT_INS].update(self.results_network)
+    results[constants.INISECT_INS]["hypervisor"] = \
+      self.results_hypervisor["hypervisor_name"]
+    results[constants.INISECT_INS]["name"] = self.results_name
+    if self.results_template:
+      results[constants.INISECT_INS]["disk_template"] = self.results_template
+    if self.results_tags:
+      results[constants.INISECT_INS]["tags"] = self.results_tags
+
+    results[constants.INISECT_BEP].update(self.results_backend)
+
+    results[constants.INISECT_EXP]["os"] = self.results_os["os_name"]
+    results[constants.INISECT_EXP]["version"] = self.results_version
+
+    del self.results_os["os_name"]
+    results[constants.INISECT_OSP].update(self.results_os)
+
+    del self.results_hypervisor["hypervisor_name"]
+    results[constants.INISECT_HYP].update(self.results_hypervisor)
+
+    output_file_name = utils.PathJoin(self.output_dir,
+                                      constants.EXPORT_CONF_FILE)
+
+    output = []
+    for section, options in results.iteritems():
+      output.append("[%s]" % section)
+      for name, value in options.iteritems():
+        if value is None:
+          value = ""
+        output.append("%s = %s" % (name, value))
+      output.append("")
+    output_contents = "\n".join(output)
+
+    try:
+      utils.WriteFile(output_file_name, data=output_contents)
+    except errors.ProgrammerError, err:
+      raise errors.OpPrereqError("Saving the config file failed: %s" % err,
+                                 errors.ECODE_ENVIRON)
+
+    self.Cleanup()
+
+
+class ConfigParserWithDefaults(ConfigParser.SafeConfigParser):
+  """This is just a wrapper on SafeConfigParser, that uses default values
+
+  """
+  def get(self, section, options, raw=None, vars=None): # pylint: disable=W0622
+    try:
+      result = ConfigParser.SafeConfigParser.get(self, section, options,
+                                                 raw=raw, vars=vars)
+    except ConfigParser.NoOptionError:
+      result = None
+    return result
+
+  def getint(self, section, options):
+    try:
+      result = ConfigParser.SafeConfigParser.get(self, section, options)
+    except ConfigParser.NoOptionError:
+      result = 0
+    return int(result)
 
 
 class OVFExporter(Converter):
 
 
 class OVFExporter(Converter):
+  """Converter from Ganeti config file to OVF
+
+  @type input_dir: string
+  @ivar input_dir: directory in which the config.ini file resides
+  @type output_dir: string
+  @ivar output_dir: directory to which the results of conversion shall be
+    written
+  @type packed_dir: string
+  @ivar packed_dir: if we want OVA package, this points to the real (i.e. not
+    temp) output directory
+  @type input_path: string
+  @ivar input_path: complete path to the config.ini file
+  @type output_path: string
+  @ivar output_path: complete path to .ovf file
+  @type config_parser: L{ConfigParserWithDefaults}
+  @ivar config_parser: parser for the config.ini file
+  @type reference_files: list
+  @ivar reference_files: files referenced in the ovf file
+  @type results_disk: list
+  @ivar results_disk: list of dictionaries of disk options from config.ini
+  @type results_network: list
+  @ivar results_network: list of dictionaries of network options form config.ini
+  @type results_name: string
+  @ivar results_name: name of the instance
+  @type results_vcpus: string
+  @ivar results_vcpus: number of VCPUs
+  @type results_memory: string
+  @ivar results_memory: RAM memory in MB
+  @type results_ganeti: dict
+  @ivar results_ganeti: dictionary of Ganeti-specific options from config.ini
+
+  """
   def _ReadInputData(self, input_path):
   def _ReadInputData(self, input_path):
-    pass
+    """Reads the data on which the conversion will take place.
+
+    @type input_path: string
+    @param input_path: absolute path to the config.ini input file
+
+    @raise errors.OpPrereqError: error when reading the config file
+
+    """
+    input_dir = os.path.dirname(input_path)
+    self.input_path = input_path
+    self.input_dir = input_dir
+    if self.options.output_dir:
+      self.output_dir = os.path.abspath(self.options.output_dir)
+    else:
+      self.output_dir = input_dir
+    self.config_parser = ConfigParserWithDefaults()
+    logging.info("Reading configuration from %s file", input_path)
+    try:
+      self.config_parser.read(input_path)
+    except ConfigParser.MissingSectionHeaderError, err:
+      raise errors.OpPrereqError("Error when trying to read %s: %s" %
+                                 (input_path, err), errors.ECODE_ENVIRON)
+    if self.options.ova_package:
+      self.temp_dir = tempfile.mkdtemp()
+      self.packed_dir = self.output_dir
+      self.output_dir = self.temp_dir
+
+    self.ovf_writer = OVFWriter(not self.options.ext_usage)
+
+  def _ParseName(self):
+    """Parses name from command line options or config file.
+
+    @rtype: string
+    @return: name of Ganeti instance
+
+    @raise errors.OpPrereqError: if name of the instance is not provided
+
+    """
+    if self.options.name:
+      name = self.options.name
+    else:
+      name = self.config_parser.get(constants.INISECT_INS, NAME)
+    if name is None:
+      raise errors.OpPrereqError("No instance name found",
+                                 errors.ECODE_ENVIRON)
+    return name
+
+  def _ParseVCPUs(self):
+    """Parses vcpus number from config file.
+
+    @rtype: int
+    @return: number of virtual CPUs
+
+    @raise errors.OpPrereqError: if number of VCPUs equals 0
+
+    """
+    vcpus = self.config_parser.getint(constants.INISECT_BEP, VCPUS)
+    if vcpus == 0:
+      raise errors.OpPrereqError("No CPU information found",
+                                 errors.ECODE_ENVIRON)
+    return vcpus
+
+  def _ParseMemory(self):
+    """Parses vcpus number from config file.
+
+    @rtype: int
+    @return: amount of memory in MB
+
+    @raise errors.OpPrereqError: if amount of memory equals 0
+
+    """
+    memory = self.config_parser.getint(constants.INISECT_BEP, MEMORY)
+    if memory == 0:
+      raise errors.OpPrereqError("No memory information found",
+                                 errors.ECODE_ENVIRON)
+    return memory
+
+  def _ParseGaneti(self):
+    """Parses Ganeti data from config file.
+
+    @rtype: dictionary
+    @return: dictionary of Ganeti-specific options
+
+    """
+    results = {}
+    # hypervisor
+    results["hypervisor"] = {}
+    hyp_name = self.config_parser.get(constants.INISECT_INS, HYPERV)
+    if hyp_name is None:
+      raise errors.OpPrereqError("No hypervisor information found",
+                                 errors.ECODE_ENVIRON)
+    results["hypervisor"]["name"] = hyp_name
+    pairs = self.config_parser.items(constants.INISECT_HYP)
+    for (name, value) in pairs:
+      results["hypervisor"][name] = value
+    # os
+    results["os"] = {}
+    os_name = self.config_parser.get(constants.INISECT_EXP, OS)
+    if os_name is None:
+      raise errors.OpPrereqError("No operating system information found",
+                                 errors.ECODE_ENVIRON)
+    results["os"]["name"] = os_name
+    pairs = self.config_parser.items(constants.INISECT_OSP)
+    for (name, value) in pairs:
+      results["os"][name] = value
+    # other
+    others = [
+      (constants.INISECT_INS, DISK_TEMPLATE, "disk_template"),
+      (constants.INISECT_BEP, AUTO_BALANCE, "auto_balance"),
+      (constants.INISECT_INS, TAGS, "tags"),
+      (constants.INISECT_EXP, VERSION, "version"),
+    ]
+    for (section, element, name) in others:
+      results[name] = self.config_parser.get(section, element)
+    return results
+
+  def _ParseNetworks(self):
+    """Parses network data from config file.
+
+    @rtype: list
+    @return: list of dictionaries of network options
+
+    @raise errors.OpPrereqError: then network mode is not recognized
+
+    """
+    results = []
+    counter = 0
+    while True:
+      data_link = \
+        self.config_parser.get(constants.INISECT_INS,
+                               "nic%s_link" % counter)
+      if data_link is None:
+        break
+      results.append({
+        "mode": self.config_parser.get(constants.INISECT_INS,
+                                       "nic%s_mode" % counter),
+        "mac": self.config_parser.get(constants.INISECT_INS,
+                                      "nic%s_mac" % counter),
+        "ip": self.config_parser.get(constants.INISECT_INS,
+                                     "nic%s_ip" % counter),
+        "network": self.config_parser.get(constants.INISECT_INS,
+                                          "nic%s_network" % counter),
+        "link": data_link,
+      })
+      if results[counter]["mode"] not in constants.NIC_VALID_MODES:
+        raise errors.OpPrereqError("Network mode %s not recognized"
+                                   % results[counter]["mode"],
+                                   errors.ECODE_INVAL)
+      counter += 1
+    return results
+
+  def _GetDiskOptions(self, disk_file, compression):
+    """Convert the disk and gather disk info for .ovf file.
+
+    @type disk_file: string
+    @param disk_file: name of the disk (without the full path)
+    @type compression: bool
+    @param compression: whether the disk should be compressed or not
+
+    @raise errors.OpPrereqError: when disk image does not exist
+
+    """
+    disk_path = utils.PathJoin(self.input_dir, disk_file)
+    results = {}
+    if not os.path.isfile(disk_path):
+      raise errors.OpPrereqError("Disk image does not exist: %s" % disk_path,
+                                 errors.ECODE_ENVIRON)
+    if os.path.dirname(disk_file):
+      raise errors.OpPrereqError("Path for the disk: %s contains a directory"
+                                 " name" % disk_path, errors.ECODE_ENVIRON)
+    disk_name, _ = os.path.splitext(disk_file)
+    ext, new_disk_path = self._ConvertDisk(self.options.disk_format, disk_path)
+    results["format"] = self.options.disk_format
+    results["virt-size"] = self._GetDiskQemuInfo(
+      new_disk_path, "virtual size: \S+ \((\d+) bytes\)")
+    if compression:
+      ext2, new_disk_path = self._CompressDisk(new_disk_path, "gzip",
+                                               COMPRESS)
+      disk_name, _ = os.path.splitext(disk_name)
+      results["compression"] = "gzip"
+      ext += ext2
+    final_disk_path = LinkFile(new_disk_path, prefix=disk_name, suffix=ext,
+                               directory=self.output_dir)
+    final_disk_name = os.path.basename(final_disk_path)
+    results["real-size"] = os.path.getsize(final_disk_path)
+    results["path"] = final_disk_name
+    self.references_files.append(final_disk_path)
+    return results
+
+  def _ParseDisks(self):
+    """Parses disk data from config file.
+
+    @rtype: list
+    @return: list of dictionaries of disk options
+
+    """
+    results = []
+    counter = 0
+    while True:
+      disk_file = \
+        self.config_parser.get(constants.INISECT_INS, "disk%s_dump" % counter)
+      if disk_file is None:
+        break
+      results.append(self._GetDiskOptions(disk_file, self.options.compression))
+      counter += 1
+    return results
 
   def Parse(self):
 
   def Parse(self):
-    pass
+    """Parses the data and creates a structure containing all required info.
+
+    """
+    try:
+      utils.Makedirs(self.output_dir)
+    except OSError, err:
+      raise errors.OpPrereqError("Failed to create directory %s: %s" %
+                                 (self.output_dir, err), errors.ECODE_ENVIRON)
+
+    self.references_files = []
+    self.results_name = self._ParseName()
+    self.results_vcpus = self._ParseVCPUs()
+    self.results_memory = self._ParseMemory()
+    if not self.options.ext_usage:
+      self.results_ganeti = self._ParseGaneti()
+    self.results_network = self._ParseNetworks()
+    self.results_disk = self._ParseDisks()
+
+  def _PrepareManifest(self, path):
+    """Creates manifest for all the files in OVF package.
+
+    @type path: string
+    @param path: path to manifesto file
+
+    @raise errors.OpPrereqError: if error occurs when writing file
+
+    """
+    logging.info("Preparing manifest for the OVF package")
+    lines = []
+    files_list = [self.output_path]
+    files_list.extend(self.references_files)
+    logging.warning("Calculating SHA1 checksums, this may take a while")
+    sha1_sums = utils.FingerprintFiles(files_list)
+    for file_path, value in sha1_sums.iteritems():
+      file_name = os.path.basename(file_path)
+      lines.append("SHA1(%s)= %s" % (file_name, value))
+    lines.append("")
+    data = "\n".join(lines)
+    try:
+      utils.WriteFile(path, data=data)
+    except errors.ProgrammerError, err:
+      raise errors.OpPrereqError("Saving the manifest file failed: %s" % err,
+                                 errors.ECODE_ENVIRON)
+
+  @staticmethod
+  def _PrepareTarFile(tar_path, files_list):
+    """Creates tarfile from the files in OVF package.
+
+    @type tar_path: string
+    @param tar_path: path to the resulting file
+    @type files_list: list
+    @param files_list: list of files in the OVF package
+
+    """
+    logging.info("Preparing tarball for the OVF package")
+    open(tar_path, mode="w").close()
+    ova_package = tarfile.open(name=tar_path, mode="w")
+    for file_path in files_list:
+      file_name = os.path.basename(file_path)
+      ova_package.add(file_path, arcname=file_name)
+    ova_package.close()
 
   def Save(self):
 
   def Save(self):
-    pass
+    """Saves the gathered configuration in an apropriate format.
+
+    @raise errors.OpPrereqError: if unable to create output directory
+
+    """
+    output_file = "%s%s" % (self.results_name, OVF_EXT)
+    output_path = utils.PathJoin(self.output_dir, output_file)
+    self.ovf_writer = OVFWriter(not self.options.ext_usage)
+    logging.info("Saving read data to %s", output_path)
+
+    self.output_path = utils.PathJoin(self.output_dir, output_file)
+    files_list = [self.output_path]
+
+    self.ovf_writer.SaveDisksData(self.results_disk)
+    self.ovf_writer.SaveNetworksData(self.results_network)
+    if not self.options.ext_usage:
+      self.ovf_writer.SaveGanetiData(self.results_ganeti, self.results_network)
+
+    self.ovf_writer.SaveVirtualSystemData(self.results_name, self.results_vcpus,
+                                          self.results_memory)
+
+    data = self.ovf_writer.PrettyXmlDump()
+    utils.WriteFile(self.output_path, data=data)
+
+    manifest_file = "%s%s" % (self.results_name, MF_EXT)
+    manifest_path = utils.PathJoin(self.output_dir, manifest_file)
+    self._PrepareManifest(manifest_path)
+    files_list.append(manifest_path)
+
+    files_list.extend(self.references_files)
+
+    if self.options.ova_package:
+      ova_file = "%s%s" % (self.results_name, OVA_EXT)
+      packed_path = utils.PathJoin(self.packed_dir, ova_file)
+      try:
+        utils.Makedirs(self.packed_dir)
+      except OSError, err:
+        raise errors.OpPrereqError("Failed to create directory %s: %s" %
+                                   (self.packed_dir, err),
+                                   errors.ECODE_ENVIRON)
+      self._PrepareTarFile(packed_path, files_list)
+    logging.info("Creation of the OVF package was successfull")
+    self.Cleanup()