Allow for moving snapshot tag to another snapshot

dataworkspaces/commands/snapshot.py

@@ -1,16 +1,19 @@
 # Copyright 2018,2019 by MPI-SWS and Data-ken Research. Licensed under Apache 2.0. See LICENSE.txt.
 
 import os
-from os.path import join, exists, isdir
+from os.path import join, exists, isdir, basename, dirname
 import re
 import json
 import datetime
 import getpass
+from copy import copy
 
 import click
 
+from dataworkspaces.utils.subprocess_utils import call_subprocess
 from dataworkspaces.utils.git_utils import \
-    is_a_git_hash, is_a_shortened_git_hash, validate_git_fat_in_path_if_needed
+    is_a_git_hash, is_a_shortened_git_hash, \
+    validate_git_fat_in_path_if_needed, GIT_EXE_PATH
 from dataworkspaces.resources.resource import CurrentResources
 from dataworkspaces.resources.snapshot_utils import \
     expand_dir_template, validate_template, make_re_pattern_for_dir_template
@@ -22,7 +25,7 @@
 from dataworkspaces.utils.lineage_utils import \
     get_current_lineage_dir, get_snapshot_lineage_dir,\
     LineageStoreCurrent
-from dataworkspaces.errors import ConfigurationError
+from dataworkspaces.errors import ConfigurationError, UserAbort
 
 def find_metadata_for_tag(tag, workspace_dir):
     """Return the snapshot metadata for the specified
@@ -202,6 +205,36 @@ def run(self):
     def __str__(self):
         return "Write snapshot metadata file to .dataworkspace/snapshot_metadata/HASH_md.json"
 
+
+class RemoveTagFromOldSnapshot(actions.Action):
+    @actions.requires_from_ns('snapshot_hash', str)
+    def __init__(self, ns, verbose, workspace_dir, existing_tag_md, remove_tag):
+        super().__init__(ns, verbose)
+        self.remove_tag = remove_tag
+        self.old_hash_file = join(get_snapshot_metadata_dir_path(workspace_dir),
+                                  existing_tag_md['hash'] + '_md.json')
+        self.snapshot_data = copy(existing_tag_md)
+        old_len = len(self.snapshot_data['tags'])
+        self.snapshot_data['tags'] = [tag for tag in self.snapshot_data['tags']
+                                      if tag!=remove_tag]
+        assert old_len!=len(self.snapshot_data['tags']), \
+            "did not remove tag %s from %s"%(remove_tag, self.snapshot_data['hash'])
+        self.snapshot_data['updated_timestamp'] = datetime.datetime.now().isoformat()
+
+    def run(self):
+        if self.ns.snapshot_hash==self.snapshot_data['hash']:
+            print("No need to remove tag %s from snapshot %s, as that is also the new snapshot we are taking"%
+                  (self.remove_tag, self.snapshot_data['hash']))
+            return
+        with open(self.old_hash_file, 'w') as f:
+            json.dump(self.snapshot_data, f, indent=2)
+        call_subprocess([GIT_EXE_PATH, 'add', basename(self.old_hash_file)],
+                        cwd=dirname(self.old_hash_file), verbose=self.verbose)
+
+    def __str__(self):
+        return "Remove tag %s from old commit %s" % (self.remove_tag, self.old_hash)
+
+
 class SaveLineageData(actions.Action):
     @actions.requires_from_ns('snapshot_hash', str)
     @actions.requires_from_ns('map_of_hashes', dict)
@@ -267,9 +300,15 @@ def snapshot_command(workspace_dir, batch, verbose, tag=None, message=''):
     if tag is not None:
         existing_tag_md = find_metadata_for_tag(tag, workspace_dir)
         if existing_tag_md:
-            raise ConfigurationError("Tag '%s' already exists for snapshot %s taken %s"%
-                                     (tag, existing_tag_md['hash'],
-                                      existing_tag_md['timestamp']))
+            msg = "Tag '%s' already exists for snapshot %s taken %s"%\
+                                         (tag, existing_tag_md['hash'],
+                                          existing_tag_md['timestamp'])
+            if batch:
+                raise ConfigurationError(msg)
+            elif not click.confirm(msg + ". Remove this tag so we an add it to the new snapshot?"):
+                raise UserAbort()
+        else:
+            existing_tag_md = None
     current_resources = CurrentResources.read_current_resources(workspace_dir, batch, verbose)
     resource_names = current_resources.by_name.keys()
     plan = []
@@ -308,6 +347,9 @@ def snapshot_command(workspace_dir, batch, verbose, tag=None, message=''):
     plan.append(actions.GitAdd(ns, verbose, workspace_dir,
                                lambda:[ns.snapshot_filename,
                                        ns.snapshot_metadata_file]))
+    if existing_tag_md is not None:
+        plan.append(RemoveTagFromOldSnapshot(ns, verbose, workspace_dir,
+                                             existing_tag_md, tag))
     # see if we need to add lineage files
     save_lineage = SaveLineageData(ns, verbose, workspace_dir, resource_names,
                                    results_resources, rel_dest_root)