changeset 51:55b12ec24a9f draft

"planemo upload commit 59b014e9f6e2d668cbd7c4844b10db3d59baefd8-dirty"
author proteore
date Fri, 05 Jun 2020 13:24:48 +0000
parents 7a6d8aafb269
children 93bb1170280d
files data_manager/resource_building.py data_manager/resource_building.xml
diffstat 2 files changed, 10 insertions(+), 8 deletions(-) [+]
line wrap: on
line diff
--- a/data_manager/resource_building.py	Mon Feb 03 14:38:16 2020 +0000
+++ b/data_manager/resource_building.py	Fri Jun 05 13:24:48 2020 +0000
@@ -132,7 +132,8 @@
 #######################################################################################################
 # 3. ID mapping file
 #######################################################################################################
-import ftplib, gzip
+import ftplib,  gzip
+from io import StringIO
 csv.field_size_limit(sys.maxsize) # to handle big files
 
 def id_mapping_sources (data_manager_dict, species, target_directory, tool_data_path) :
@@ -280,15 +281,17 @@
     return (path)
 
 def id_list_from_nextprot_ftp(file,target_directory) :
-    ftp_dir = "pub/current_release/ac_lists/"
     path = os.path.join(target_directory, file)
     ftp = ftplib.FTP("ftp.nextprot.org")
     ftp.login("anonymous", "anonymous") 
-    ftp.cwd(ftp_dir)
-    ftp.retrbinary("RETR " + file, open(path, 'wb').write)
+    ftp.cwd("pub/current_release/ac_lists/")
+    r = StringIO()
+    ftp.retrlines("RETR " + file, lambda line: r.write(line + '\n'))
     ftp.quit()
-
-    return (path)
+    r.seek(0)
+    ids = r.readlines()
+    ids = [id.strip('\n') for id in ids if id != '']
+    return (ids)
 
 #return '' if there's no value in a dictionary, avoid error
 def access_dictionary (dico,key1,key2) :
@@ -549,7 +552,6 @@
     writer.writerows(nextprot_file)
     
     for id in ids :
-        #print (id)
         query="https://api.nextprot.org/entry/"+id+".json"
         resp = requests.get(url=query)
         data = resp.json()
--- a/data_manager/resource_building.xml	Mon Feb 03 14:38:16 2020 +0000
+++ b/data_manager/resource_building.xml	Fri Jun 05 13:24:48 2020 +0000
@@ -1,4 +1,4 @@
-<tool id="data_manager_proteore" name="Get source files for proteore tools" version="2020.02.03.1" tool_type="manage_data">
+<tool id="data_manager_proteore" name="Get source files for proteore tools" version="2020.06.05" tool_type="manage_data">
 <description>
 to create or update reference files for proteore tools
 </description>