Update checked in version of gsutil to version 4.6

gslib/commands/setmeta.py

Index: gslib/commands/setmeta.py
===================================================================
--- gslib/commands/setmeta.py	(revision 33376)
+++ gslib/commands/setmeta.py	(working copy)
@@ -1,5 +1,5 @@
+# -*- coding: utf-8 -*-
 # Copyright 2012 Google Inc. All Rights Reserved.
-#coding=utf8
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,44 +12,34 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+"""Implementation of setmeta command for setting cloud object metadata."""
 
-import boto
-import csv
-import random
-import StringIO
-import time
+from __future__ import absolute_import
 
-from boto.exception import GSResponseError
-from boto.s3.key import Key
-from gslib.command import COMMAND_NAME
-from gslib.command import COMMAND_NAME_ALIASES
+from gslib.cloud_api import AccessDeniedException
+from gslib.cloud_api import PreconditionException
+from gslib.cloud_api import Preconditions
 from gslib.command import Command
-from gslib.command import FILE_URIS_OK
-from gslib.command import MAX_ARGS
-from gslib.command import MIN_ARGS
-from gslib.command import PROVIDER_URIS_OK
-from gslib.command import SUPPORTED_SUB_ARGS
-from gslib.command import URIS_START_ARG
+from gslib.cs_api_map import ApiSelector
 from gslib.exception import CommandException
-from gslib.help_provider import HELP_NAME
-from gslib.help_provider import HELP_NAME_ALIASES
-from gslib.help_provider import HELP_ONE_LINE_SUMMARY
-from gslib.help_provider import HELP_TEXT
-from gslib.help_provider import HELP_TYPE
-from gslib.help_provider import HelpType
 from gslib.name_expansion import NameExpansionIterator
+from gslib.storage_url import StorageUrlFromString
+from gslib.translation_helper import CopyObjectMetadata
+from gslib.translation_helper import ObjectMetadataFromHeaders
+from gslib.util import GetCloudApiInstance
 from gslib.util import NO_MAX
 from gslib.util import Retry
 
-_detailed_help_text = ("""
+
+_DETAILED_HELP_TEXT = ("""
 <B>SYNOPSIS</B>
-    gsutil setmeta [-n] -h [header:value|header] ... uri...
+    gsutil setmeta [-n] -h [header:value|header] ... url...
 
 
 <B>DESCRIPTION</B>
   The gsutil setmeta command allows you to set or remove the metadata on one
   or more objects. It takes one or more header arguments followed by one or
-  more URIs, where each header argument is in one of two forms:
+  more URLs, where each header argument is in one of two forms:
 
   - if you specify header:value, it will set the given header on all
     named objects.
@@ -87,7 +77,7 @@
 
 
 <B>OPERATION COST</B>
-  This command uses four operations per URI (one to read the ACL, one to read
+  This command uses four operations per URL (one to read the ACL, one to read
   the current metadata, one to set the new metadata, and one to set the ACL).
 
   For cases where you want all objects to have the same ACL you can avoid half
@@ -98,7 +88,6 @@
 <B>OPTIONS</B>
   -h          Specifies a header:value to be added, or header to be removed,
               from each named object.
-
   -n          Causes the operations for reading and writing the ACL to be
               skipped. This halves the number of operations performed per
               request, improving the speed and reducing the cost of performing
@@ -106,77 +95,96 @@
               all objects to have the same ACL, for which you have set a default
               ACL on the bucket(s) containing the objects. See "help gsutil
               defacl".
-
-  -R, -r      Performs setmeta request recursively, to all objects under the
-              specified URI.
 """)
 
+# Setmeta assumes a header-like model which doesn't line up with the JSON way
+# of doing things. This list comes from functionality that was supported by
+# gsutil3 at the time gsutil4 was released.
+SETTABLE_FIELDS = ['cache-control', 'content-disposition',
+                   'content-encoding', 'content-language',
+                   'content-md5', 'content-type']
+
+
 def _SetMetadataExceptionHandler(cls, e):
   """Exception handler that maintains state about post-completion status."""
   cls.logger.error(e)
   cls.everything_set_okay = False
-  
-def _SetMetadataFuncWrapper(cls, name_expansion_result):
-  cls._SetMetadataFunc(name_expansion_result)
 
 
+def _SetMetadataFuncWrapper(cls, name_expansion_result, thread_state=None):
+  cls.SetMetadataFunc(name_expansion_result, thread_state=thread_state)
+
+
 class SetMetaCommand(Command):
   """Implementation of gsutil setmeta command."""
 
-  # Command specification (processed by parent class).
-  command_spec = {
-    # Name of command.
-    COMMAND_NAME : 'setmeta',
-    # List of command name aliases.
-    COMMAND_NAME_ALIASES : ['setheader'],
-    # Min number of args required by this command.
-    MIN_ARGS : 1,
-    # Max number of args required by this command, or NO_MAX.
-    MAX_ARGS : NO_MAX,
-    # Getopt-style string specifying acceptable sub args.
-    SUPPORTED_SUB_ARGS : 'h:nrR',
-    # True if file URIs acceptable for this command.
-    FILE_URIS_OK : False,
-    # True if provider-only URIs acceptable for this command.
-    PROVIDER_URIS_OK : False,
-    # Index in args of first URI arg.
-    URIS_START_ARG : 1,
-  }
-  help_spec = {
-    # Name of command or auxiliary help info for which this help applies.
-    HELP_NAME : 'setmeta',
-    # List of help name aliases.
-    HELP_NAME_ALIASES : ['setheader'],
-    # Type of help:
-    HELP_TYPE : HelpType.COMMAND_HELP,
-    # One line summary of this help.
-    HELP_ONE_LINE_SUMMARY : 'Set metadata on already uploaded objects',
-    # The full help text.
-    HELP_TEXT : _detailed_help_text,
-  }
+  # Command specification. See base class for documentation.
+  command_spec = Command.CreateCommandSpec(
+      'setmeta',
+      command_name_aliases=['setheader'],
+      min_args=1,
+      max_args=NO_MAX,
+      supported_sub_args='h:nrR',
+      file_url_ok=False,
+      provider_url_ok=False,
+      urls_start_arg=1,
+      gs_api_support=[ApiSelector.XML, ApiSelector.JSON],
+      gs_default_api=ApiSelector.JSON,
+  )
+  # Help specification. See help_provider.py for documentation.
+  help_spec = Command.HelpSpec(
+      help_name='setmeta',
+      help_name_aliases=['setheader'],
+      help_type='command_help',
+      help_one_line_summary='Set metadata on already uploaded objects',
+      help_text=_DETAILED_HELP_TEXT,
+      subcommand_help_text={},
+  )
 
-  # Command entry point.
   def RunCommand(self):
-    if (len(self.args) == 1 and not self.recursion_requested
-        and not self.suri_builder.StorageUri(self.args[0]).names_object()):
-      raise CommandException('URI (%s) must name an object' % self.args[0])
+    """Command entry point for the setmeta command."""
+    headers = []
+    if self.sub_opts:
+      for o, a in self.sub_opts:
+        if o == '-n':
+          self.logger.warning(
+              'Warning: gsutil setmeta -n is now on by default, and will be '
+              'removed in the future.\nPlease use gsutil acl set ... to set '
+              'canned ACLs.')
+        elif o == '-h':
+          if 'x-goog-acl' in a or 'x-amz-acl' in a:
+            raise CommandException(
+                'gsutil setmeta no longer allows canned ACLs. Use gsutil acl '
+                'set ... to set canned ACLs.')
+          headers.append(a)
 
+    (metadata_minus, metadata_plus) = self._ParseMetadataHeaders(headers)
+
+    self.metadata_change = metadata_plus
+    for header in metadata_minus:
+      self.metadata_change[header] = ''
+
+    if len(self.args) == 1 and not self.recursion_requested:
+      url = StorageUrlFromString(self.args[0])
+      if not (url.IsCloudUrl() and url.IsObject()):
+        raise CommandException('URL (%s) must name an object' % self.args[0])
+
     # Used to track if any objects' metadata failed to be set.
     self.everything_set_okay = True
 
     name_expansion_iterator = NameExpansionIterator(
-        self.command_name, self.proj_id_handler, self.headers, self.debug,
-        self.logger, self.bucket_storage_uri_class, self.args,
-        self.recursion_requested, flat=self.recursion_requested)
+        self.command_name, self.debug, self.logger, self.gsutil_api,
+        self.args, self.recursion_requested, all_versions=self.all_versions,
+        continue_on_error=self.parallel_operations)
+
     try:
       # Perform requests in parallel (-m) mode, if requested, using
       # configured number of parallel processes and threads. Otherwise,
       # perform requests with sequential function calls in current process.
       self.Apply(_SetMetadataFuncWrapper, name_expansion_iterator,
                  _SetMetadataExceptionHandler, fail_on_error=True)
-    except GSResponseError as e:
-      if e.code == 'AccessDenied' and e.reason == 'Forbidden' \
-          and e.status == 403:
+    except AccessDeniedException as e:
+      if e.status == 403:
         self._WarnServiceAccounts()
       raise
 
@@ -184,40 +192,58 @@
       raise CommandException('Metadata for some objects could not be set.')
 
     return 0
-  
-  @Retry(GSResponseError, tries=3, timeout_secs=1)
-  def _SetMetadataFunc(self, name_expansion_result):
-    headers = []
-    preserve_acl = True
-    if self.sub_opts:
-      for o, a in self.sub_opts:
-        if o == '-n':
-          preserve_acl = False
-        elif o == '-h':
-          headers.append(a)
 
-    (metadata_minus, metadata_plus) = self._ParseMetadataHeaders(headers)
+  @Retry(PreconditionException, tries=3, timeout_secs=1)
+  def SetMetadataFunc(self, name_expansion_result, thread_state=None):
+    """Sets metadata on an object.
 
-    exp_src_uri = self.suri_builder.StorageUri(
-        name_expansion_result.GetExpandedUriStr())
-    self.logger.info('Setting metadata on %s...', exp_src_uri)
+    Args:
+      name_expansion_result: NameExpansionResult describing target object.
+      thread_state: gsutil Cloud API instance to use for the operation.
+    """
+    gsutil_api = GetCloudApiInstance(self, thread_state=thread_state)
 
-    key = exp_src_uri.get_key()
-    metageneration = getattr(key, 'metageneration', None)
-    generation = getattr(key, 'generation', None)
+    exp_src_url = name_expansion_result.expanded_storage_url
+    self.logger.info('Setting metadata on %s...', exp_src_url)
 
-    headers = {}
-    if generation:
-      headers['x-goog-if-generation-match'] = generation
-    if metageneration:
-      headers['x-goog-if-metageneration-match'] = metageneration
+    fields = ['generation', 'metadata', 'metageneration']
+    cloud_obj_metadata = gsutil_api.GetObjectMetadata(
+        exp_src_url.bucket_name, exp_src_url.object_name,
+        generation=exp_src_url.generation, provider=exp_src_url.scheme,
+        fields=fields)
 
-    # If this fails because of a precondition, it will raise a
-    # GSResponseError for @Retry to handle.
-    exp_src_uri.set_metadata(metadata_plus, metadata_minus, preserve_acl,
-                               headers=headers)
+    preconditions = Preconditions(
+        gen_match=cloud_obj_metadata.generation,
+        meta_gen_match=cloud_obj_metadata.metageneration)
 
+    # Patch handles the patch semantics for most metadata, but we need to
+    # merge the custom metadata field manually.
+    patch_obj_metadata = ObjectMetadataFromHeaders(self.metadata_change)
+
+    api = gsutil_api.GetApiSelector(provider=exp_src_url.scheme)
+    # For XML we only want to patch through custom metadata that has
+    # changed.  For JSON we need to build the complete set.
+    if api == ApiSelector.XML:
+      pass
+    elif api == ApiSelector.JSON:
+      CopyObjectMetadata(patch_obj_metadata, cloud_obj_metadata,
+                         override=True)
+      patch_obj_metadata = cloud_obj_metadata
+
+    gsutil_api.PatchObjectMetadata(
+        exp_src_url.bucket_name, exp_src_url.object_name, patch_obj_metadata,
+        generation=exp_src_url.generation, preconditions=preconditions,
+        provider=exp_src_url.scheme)
+
   def _ParseMetadataHeaders(self, headers):
+    """Validates and parses metadata changes from the headers argument.
+
+    Args:
+      headers: Header dict to validate and parse.
+
+    Returns:
+      (metadata_plus, metadata_minus): Tuple of header sets to add and remove.
+    """
     metadata_minus = set()
     cust_metadata_minus = set()
     metadata_plus = {}
@@ -271,21 +297,21 @@
         or metadata_minus.intersection(set(metadata_plus.keys()))):
       raise CommandException('Each header must appear at most once.')
     other_than_base_fields = (set(metadata_plus.keys())
-        .difference(Key.base_user_settable_fields))
+                              .difference(SETTABLE_FIELDS))
     other_than_base_fields.update(
-        metadata_minus.difference(Key.base_user_settable_fields))
+        metadata_minus.difference(SETTABLE_FIELDS))
     for f in other_than_base_fields:
       # This check is overly simple; it would be stronger to check, for each
-      # URI argument, whether f.startswith the
-      # uri.get_provider().metadata_prefix, but here we just parse the spec
-      # once, before processing any of the URIs. This means we will not
+      # URL argument, whether f.startswith the
+      # provider metadata_prefix, but here we just parse the spec
+      # once, before processing any of the URLs. This means we will not
       # detect if the user tries to set an x-goog-meta- field on an another
       # provider's object, for example.
       if not _IsCustomMeta(f):
-        raise CommandException('Invalid or disallowed header (%s).\n'
-                               'Only these fields (plus x-goog-meta-* fields)'
-                               ' can be set or unset:\n%s' % (f,
-                               sorted(list(Key.base_user_settable_fields))))
+        raise CommandException(
+            'Invalid or disallowed header (%s).\nOnly these fields (plus '
+            'x-goog-meta-* fields) can be set or unset:\n%s' % (
+                f, sorted(list(SETTABLE_FIELDS))))
     metadata_plus.update(cust_metadata_plus)
     metadata_minus.update(cust_metadata_minus)
     return (metadata_minus, metadata_plus)