annotate json_data_source_mod.py @ 8:4bb8859f919c draft

Uploaded 20170525
author fabio
date Thu, 25 May 2017 15:49:02 -0400
parents 7815152f70c6
children c0be9583df97
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
1 #!/usr/bin/env python
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
2 import json
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
3 import optparse
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
4 import urllib
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
5 import os.path
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
6 import os
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
7 from operator import itemgetter
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
8 import tarfile
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
9
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
10 __version__ = "1.0.0"
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
11 CHUNK_SIZE = 2**20 #1mb
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
12 VALID_CHARS = '.-()[]0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ '
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
13
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
14
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
15 def splitext(path):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
16 for ext in ['.tar.gz', '.tar.bz2']:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
17 if path.endswith(ext):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
18 path, ext = path[:-len(ext)], path[-len(ext):]
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
19 break
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
20 else:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
21 path, ext = os.path.splitext(path)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
22 return path, ext[1:]
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
23
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
24
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
25 def chunk_write( source_stream, target_stream, source_method = "read", target_method="write" ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
26 source_method = getattr( source_stream, source_method )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
27 target_method = getattr( target_stream, target_method )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
28 while True:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
29 chunk = source_method( CHUNK_SIZE )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
30 if chunk:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
31 target_method( chunk )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
32 else:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
33 break
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
34
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
35
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
36 def deconstruct_multi_filename( multi_filename ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
37 keys = [ 'primary', 'id', 'name', 'visible', 'file_type' ]
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
38 return ( dict( zip( keys, multi_filename.split('_') ) ) )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
39
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
40
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
41 def construct_multi_filename( id, name, file_type ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
42 """ Implementation of *Number of Output datasets cannot be determined until tool run* from documentation_.
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
43 .. _documentation: http://wiki.galaxyproject.org/Admin/Tools/Multiple%20Output%20Files
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
44 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
45 filename = "%s_%s_%s_%s_%s" % ( 'primary', id, name, 'visible', file_type )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
46 return filename
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
47
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
48
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
49 def download_from_query( query_data, target_output_filename ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
50 """ Download file from the json data and write it to target_output_filename.
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
51 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
52 query_url = query_data.get( 'url' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
53 query_file_type = query_data.get( 'extension' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
54 query_stream = urllib.urlopen( query_url )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
55 output_stream = open( target_output_filename, 'wb' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
56 chunk_write( query_stream, output_stream )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
57 query_stream.close()
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
58 output_stream.close()
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
59
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
60 def store_file_from_archive( file_object, target_output_filename ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
61 """ Store file after extracting from archive and organize them as a collection using the structure
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
62 (collection-name)_(file-name).ext as file name
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
63 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
64 output_stream = open( target_output_filename, 'wb' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
65 chunk_write( file_object.read(), output_stream )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
66 output_stream.close()
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
67
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
68
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
69 def download_extra_data( query_ext_data, base_path ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
70 """ Download any extra data defined in the JSON.
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
71 NOTE: the "path" value is a relative path to the file on our
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
72 file system. This is slightly dangerous and we should make every effort
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
73 to avoid a malicious absolute path to write the file elsewhere on the
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
74 filesystem.
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
75 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
76 for ext_data in query_ext_data:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
77 if not os.path.exists( base_path ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
78 os.mkdir( base_path )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
79 query_stream = urllib.urlopen( ext_data.get( 'url' ) )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
80 ext_path = ext_data.get( 'path' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
81 os.makedirs( os.path.normpath( '/'.join( [ base_path, os.path.dirname( ext_path ) ] ) ) )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
82 output_stream = open( os.path.normpath( '/'.join( [ base_path, ext_path ] ) ), 'wb' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
83 chunk_write( query_stream, output_stream )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
84 query_stream.close()
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
85 output_stream.close()
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
86
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
87
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
88 def metadata_to_json_for_archive_entry( dataset_id, extension, metaname, filename, ds_type='dataset', primary=False ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
89 """ Return line separated JSON """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
90 meta_dict = dict( type = ds_type,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
91 ext = extension,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
92 filename = filename,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
93 name = metaname,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
94 metadata = {} )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
95 if primary:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
96 meta_dict[ 'base_dataset_id' ] = dataset_id
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
97 else:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
98 meta_dict[ 'dataset_id' ] = dataset_id
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
99 return "%s\n" % json.dumps( meta_dict )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
100
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
101
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
102 def metadata_to_json( dataset_id, metadata, filename, ds_type='dataset', primary=False):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
103 """ Return line separated JSON """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
104 meta_dict = dict( type = ds_type,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
105 ext = metadata.get( 'extension' ),
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
106 filename = filename,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
107 name = metadata.get( 'name' ),
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
108 metadata = metadata.get( 'metadata', {} ) )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
109 if metadata.get( 'extra_data', None ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
110 meta_dict[ 'extra_files' ] = '_'.join( [ filename, 'files' ] )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
111 if primary:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
112 meta_dict[ 'base_dataset_id' ] = dataset_id
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
113 else:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
114 meta_dict[ 'dataset_id' ] = dataset_id
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
115 return "%s\n" % json.dumps( meta_dict )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
116
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
117
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
118 def download_files_and_write_metadata(query_item, json_params, output_base_path, metadata_parameter_file, primary, appdata_path):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
119 """ Main work function that operates on the JSON representation of
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
120 one dataset and its metadata. Returns True.
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
121 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
122 dataset_url, output_filename, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
123 extra_files_path, file_name, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
124 ext, out_data_name, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
125 hda_id, dataset_id = set_up_config_values(json_params)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
126 extension = query_item.get( 'extension' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
127 filename = query_item.get( 'url' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
128 extra_data = query_item.get( 'extra_data', None )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
129 if primary:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
130 filename = ''.join( c in VALID_CHARS and c or '-' for c in filename )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
131 name = construct_multi_filename( hda_id, filename, extension )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
132 target_output_filename = os.path.normpath( '/'.join( [ output_base_path, name ] ) )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
133 metadata_parameter_file.write( metadata_to_json( dataset_id, query_item,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
134 target_output_filename,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
135 ds_type='new_primary_dataset',
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
136 primary=primary) )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
137 else:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
138 target_output_filename = output_filename
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
139 metadata_parameter_file.write( metadata_to_json( dataset_id, query_item,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
140 target_output_filename,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
141 ds_type='dataset',
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
142 primary=primary) )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
143 download_from_query( query_item, target_output_filename )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
144 if extra_data:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
145 extra_files_path = ''.join( [ target_output_filename, 'files' ] )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
146 download_extra_data( extra_data, extra_files_path )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
147
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
148 check_ext = ""
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
149 if ( fname.endswith( "gz" ) ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
150 check_ext = "r:gz"
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
151 elif ( fname.endswith( "bz2" ) ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
152 check_ext = "r:bz2"
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
153 elif ( fname.endswith( "tar" ) ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
154 check_ext = "r:"
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
155 if ( bool( check_ext and check_ext.strip() ) ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
156 with tarfile.open( target_output_filename, check_ext ) as tf:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
157 for entry in tf:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
158 fileobj = tf.extractfile( entry )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
159 if entry.isfile():
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
160 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
161 dataset_url, output_filename, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
162 extra_files_path, file_name, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
163 ext, out_data_name, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
164 hda_id, dataset_id = set_up_config_values(json_params)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
165 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
166 filename = os.path.basename( entry.name )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
167 extension = splitext( filename )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
168 extra_data = None
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
169 #target_output_filename = output_filename
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
170 """ (?P<archive_name>.*)_(?P<file_name>.*)\..* """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
171 filename_with_collection_prefix = query_item.get( 'name' ) + "_" + filename
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
172 target_output_filename = os.path.join(appdata_path, filename_with_collection_prefix)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
173 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
174 metadata_parameter_file.write( metadata_to_json_for_archive_entry( dataset_id, extension,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
175 filename, target_output_filename,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
176 ds_type='dataset',
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
177 primary=primary) )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
178 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
179 store_file_from_archive( fileobj, target_output_filename )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
180
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
181 return True
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
182
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
183
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
184 def set_up_config_values():
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
185 extra_files_path, file_name, ext, out_data_name, hda_id, dataset_id = \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
186 itemgetter('extra_files_path', 'file_name', 'ext', 'out_data_name', 'hda_id', 'dataset_id')(output_data[0])
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
187
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
188 def set_up_config_values(json_params):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
189 """ Parse json_params file and return a tuple of necessary configuration
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
190 values.
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
191 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
192 datasource_params = json_params.get( 'param_dict' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
193 dataset_url = datasource_params.get( 'URL' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
194 output_filename = datasource_params.get( 'output1', None )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
195 output_data = json_params.get( 'output_data' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
196 extra_files_path, file_name, ext, out_data_name, hda_id, dataset_id = \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
197 itemgetter('extra_files_path', 'file_name', 'ext', 'out_data_name', 'hda_id', 'dataset_id')(output_data[0])
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
198 return (dataset_url, output_filename,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
199 extra_files_path, file_name,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
200 ext, out_data_name,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
201 hda_id, dataset_id)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
202
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
203
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
204 def download_from_json_data( options, args ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
205 """ Parse the returned JSON data and download files. Write metadata
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
206 to flat JSON file.
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
207 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
208 output_base_path = options.path
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
209 appdata_path = options.appdata
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
210 if not os.path.exists(appdata_path):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
211 os.makedirs(appdata_path)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
212
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
213 # read tool job configuration file and parse parameters we need
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
214 json_params = json.loads( open( options.json_param_file, 'r' ).read() )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
215 dataset_url, output_filename, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
216 extra_files_path, file_name, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
217 ext, out_data_name, \
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
218 hda_id, dataset_id = set_up_config_values(json_params)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
219 # line separated JSON file to contain all dataset metadata
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
220 metadata_parameter_file = open( json_params['job_config']['TOOL_PROVIDED_JOB_METADATA_FILE'], 'wb' )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
221
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
222 # get JSON response from data source
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
223 # TODO: make sure response is not enormous
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
224 query_params = json.loads(urllib.urlopen( dataset_url ).read())
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
225 # download and write files
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
226 primary = False
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
227 # query_item, hda_id, output_base_path, dataset_id
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
228 for query_item in query_params:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
229 if isinstance( query_item, list ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
230 # TODO: do something with the nested list as a collection
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
231 for query_subitem in query_item:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
232 primary = download_files_and_write_metadata(query_subitem, json_params, output_base_path,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
233 metadata_parameter_file, primary, appdata_path)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
234
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
235 elif isinstance( query_item, dict ):
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
236 primary = download_files_and_write_metadata(query_item, json_params, output_base_path,
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
237 metadata_parameter_file, primary, appdata_path)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
238 metadata_parameter_file.close()
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
239
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
240 def __main__():
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
241 """ Read the JSON return from a data source. Parse each line and request
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
242 the data, download to "newfilepath", and write metadata.
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
243
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
244 Schema
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
245 ------
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
246
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
247 [ {"url":"http://url_of_file",
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
248 "name":"encode WigData",
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
249 "extension":"wig",
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
250 "metadata":{"db_key":"hg19"},
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
251 "extra_data":[ {"url":"http://url_of_ext_file",
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
252 "path":"rel/path/to/ext_file"}
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
253 ]
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
254 }
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
255 ]
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
256
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
257 """
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
258 # Parse the command line options
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
259 usage = "Usage: json_data_source_mod.py max_size --json_param_file filename [options]"
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
260 parser = optparse.OptionParser(usage = usage)
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
261 parser.add_option("-j", "--json_param_file", type="string",
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
262 action="store", dest="json_param_file", help="json schema return data")
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
263 parser.add_option("-p", "--path", type="string",
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
264 action="store", dest="path", help="new file path")
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
265 parser.add_option("-a", "--appdata", type="string",
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
266 action="store", dest="appdata", help="appdata folder name")
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
267 parser.add_option("-v", "--version", action="store_true", dest="version",
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
268 default=False, help="display version and exit")
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
269
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
270 (options, args) = parser.parse_args()
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
271 if options.version:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
272 print __version__
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
273 else:
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
274 download_from_json_data( options, args )
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
275
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
276
7815152f70c6 Uploaded 20170525
fabio
parents:
diff changeset
277 if __name__ == "__main__": __main__()