From bd0f25745296be6f23fa1203cf8f50991af9bb2f Mon Sep 17 00:00:00 2001
From: Wolfgang Maier <maierw@informatik.uni-freiburg.de>
Date: Thu, 8 Feb 2024 16:43:39 +0100
Subject: [PATCH] Switch data_source.py to modern tool-provided metadata

Allows moving all data source tools using the script to profile 20.09.
For more recent profiles data_source.py would have to work outside of
Galaxy's Python environment.
---
 tools/data_source/biomart.xml                   | 2 +-
 tools/data_source/biomart_test.xml              | 2 +-
 tools/data_source/cbi_rice_mart.xml             | 2 +-
 tools/data_source/data_source.py                | 5 +++--
 tools/data_source/ebi_sra.xml                   | 2 +-
 tools/data_source/eupathdb.xml                  | 2 +-
 tools/data_source/fly_modencode.xml             | 2 +-
 tools/data_source/flymine.xml                   | 2 +-
 tools/data_source/flymine_test.xml              | 2 +-
 tools/data_source/gramene_mart.xml              | 2 +-
 tools/data_source/hapmapmart.xml                | 2 +-
 tools/data_source/hbvar.xml                     | 2 +-
 tools/data_source/intermine.xml                 | 2 +-
 tools/data_source/metabolicmine.xml             | 2 +-
 tools/data_source/modmine.xml                   | 2 +-
 tools/data_source/mousemine.xml                 | 2 +-
 tools/data_source/ratmine.xml                   | 2 +-
 tools/data_source/sra.xml                       | 2 +-
 tools/data_source/ucsc_tablebrowser.xml         | 2 +-
 tools/data_source/ucsc_tablebrowser_archaea.xml | 2 +-
 tools/data_source/ucsc_tablebrowser_test.xml    | 2 +-
 tools/data_source/worm_modencode.xml            | 2 +-
 tools/data_source/wormbase.xml                  | 2 +-
 tools/data_source/wormbase_test.xml             | 2 +-
 tools/data_source/yeastmine.xml                 | 2 +-
 tools/data_source/zebrafishmine.xml             | 2 +-
 26 files changed, 28 insertions(+), 27 deletions(-)

diff --git a/tools/data_source/biomart.xml b/tools/data_source/biomart.xml
index b79a01f6048..b8b819e5580 100644
--- a/tools/data_source/biomart.xml
+++ b/tools/data_source/biomart.xml
@@ -7,7 +7,7 @@
     TODO: Hack to get biomart to work - the 'add_to_URL' param can be eliminated when the Biomart team encodes URL prior to sending, meanwhile
     everything including and beyond the first '&' is truncated from URL.  They said they'll let us know when this is fixed at their end.
 -->
-<tool name="BioMart" id="biomart" tool_type="data_source" version="1.0.1">
+<tool name="BioMart" id="biomart" tool_type="data_source" version="1.0.1" profile="20.09">
     <description>Ensembl server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/biomart_test.xml b/tools/data_source/biomart_test.xml
index 5334b94fbc2..b9932752058 100644
--- a/tools/data_source/biomart_test.xml
+++ b/tools/data_source/biomart_test.xml
@@ -7,7 +7,7 @@
     TODO: Hack to get biomart to work - the 'add_to_URL' param can be eliminated when the Biomart team encodes URL prior to sending, meanwhile
     everything including and beyond the first '&' is truncated from URL.  They said they'll let us know when this is fixed at their end.
 -->
-<tool name="BioMart" id="biomart_test" tool_type="data_source" version="1.0.1">
+<tool name="BioMart" id="biomart_test" tool_type="data_source" version="1.0.1" profile="20.09">
     <description>Test server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/cbi_rice_mart.xml b/tools/data_source/cbi_rice_mart.xml
index 99763f5f178..658df685448 100644
--- a/tools/data_source/cbi_rice_mart.xml
+++ b/tools/data_source/cbi_rice_mart.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="CBI Rice Mart" id="cbi_rice_mart" tool_type="data_source" version="1.0.1">
+<tool name="CBI Rice Mart" id="cbi_rice_mart" tool_type="data_source" version="1.0.1" profile="20.09">
     <description>rice mart</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/data_source.py b/tools/data_source/data_source.py
index f97538074bf..a2f25986276 100644
--- a/tools/data_source/data_source.py
+++ b/tools/data_source/data_source.py
@@ -81,10 +81,11 @@ def __main__():
             ext = sniff.handle_uploaded_dataset_file(cur_filename, datatypes_registry, ext=data_dict["ext"])
         except Exception as e:
             sys.exit(str(e))
-        info = dict(type="dataset", dataset_id=data_dict["dataset_id"], ext=ext)
+
+        tool_provided_metadata = {out_data_name: {"ext": ext}}
 
         with open(params["job_config"]["TOOL_PROVIDED_JOB_METADATA_FILE"], "w") as json_file:
-            json.dump(info, json_file)
+            json.dump(tool_provided_metadata, json_file)
 
 
 if __name__ == "__main__":
diff --git a/tools/data_source/ebi_sra.xml b/tools/data_source/ebi_sra.xml
index 11dbf6eef9b..5a1e946b815 100644
--- a/tools/data_source/ebi_sra.xml
+++ b/tools/data_source/ebi_sra.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="EBI SRA" id="ebi_sra_main" tool_type="data_source" version="1.0.1">
+<tool name="EBI SRA" id="ebi_sra_main" tool_type="data_source" version="1.0.1" profile="20.09">
     <description>ENA SRA</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/eupathdb.xml b/tools/data_source/eupathdb.xml
index 9cb410ccaa9..7fe82b76564 100644
--- a/tools/data_source/eupathdb.xml
+++ b/tools/data_source/eupathdb.xml
@@ -1,4 +1,4 @@
-<tool name="EuPathDB" id="eupathdb" tool_type="data_source" url_method="post" version="1.0.0">
+<tool name="EuPathDB" id="eupathdb" tool_type="data_source" url_method="post" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/fly_modencode.xml b/tools/data_source/fly_modencode.xml
index f5e776ab775..2bcd60015a4 100644
--- a/tools/data_source/fly_modencode.xml
+++ b/tools/data_source/fly_modencode.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="modENCODE fly" id="modENCODEfly" tool_type="data_source" version="1.0.1">
+<tool name="modENCODE fly" id="modENCODEfly" tool_type="data_source" version="1.0.1"  profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/flymine.xml b/tools/data_source/flymine.xml
index 6928cc1fb04..f7b7a2abd0c 100644
--- a/tools/data_source/flymine.xml
+++ b/tools/data_source/flymine.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="Flymine" id="flymine" tool_type="data_source" version="1.0.0">
+<tool name="Flymine" id="flymine" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/flymine_test.xml b/tools/data_source/flymine_test.xml
index 899b73969c7..ef0524499b8 100644
--- a/tools/data_source/flymine_test.xml
+++ b/tools/data_source/flymine_test.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="Flymine test" id="flymine_test" tool_type="data_source" version="1.0.0">
+<tool name="Flymine test" id="flymine_test" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/gramene_mart.xml b/tools/data_source/gramene_mart.xml
index 64feef7dbec..7abf0b06d1f 100644
--- a/tools/data_source/gramene_mart.xml
+++ b/tools/data_source/gramene_mart.xml
@@ -7,7 +7,7 @@
     TODO: Hack to get biomart to work - the 'add_to_URL' param can be eliminated when the Biomart team encodes URL prior to sending, meanwhile
     everything including and beyond the first '&' is truncated from URL.  They said they'll let us know when this is fixed at their end.
 -->
-<tool name="GrameneMart" id="gramenemart" tool_type="data_source" version="1.0.1">
+<tool name="GrameneMart" id="gramenemart" tool_type="data_source" version="1.0.1" profile="20.09">
     <description> Central server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/hapmapmart.xml b/tools/data_source/hapmapmart.xml
index c43fa5761e0..5ad6e0ac795 100644
--- a/tools/data_source/hapmapmart.xml
+++ b/tools/data_source/hapmapmart.xml
@@ -11,7 +11,7 @@
     TODO: Hack to get biomart to work - the 'add_to_URL' param can be eliminated when the Biomart team encodes URL prior to sending, meanwhile
     everything including and beyond the first '&' is truncated from URL.  They said they'll let us know when this is fixed at their end.
 -->
-<tool name="HapMapMart" id="hapmapmart" tool_type="data_source" version="0.0.01">
+<tool name="HapMapMart" id="hapmapmart" tool_type="data_source" version="0.0.01" profile="20.09">
     <description>HapMap Biomart</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/hbvar.xml b/tools/data_source/hbvar.xml
index 74f926e8c13..b6867850995 100644
--- a/tools/data_source/hbvar.xml
+++ b/tools/data_source/hbvar.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="HbVar" id="hbvar" tool_type="data_source" version="2.0.0">
+<tool name="HbVar" id="hbvar" tool_type="data_source" version="2.0.0" profile="20.09">
     <description>Human Hemoglobin Variants and Thalassemias</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/intermine.xml b/tools/data_source/intermine.xml
index 6f574b9b598..89c8b95e730 100644
--- a/tools/data_source/intermine.xml
+++ b/tools/data_source/intermine.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="InterMine" id="intermine" tool_type="data_source" version="1.0.0">
+<tool name="InterMine" id="intermine" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/metabolicmine.xml b/tools/data_source/metabolicmine.xml
index 17e2c7a0535..be4f1c9dd4d 100644
--- a/tools/data_source/metabolicmine.xml
+++ b/tools/data_source/metabolicmine.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="metabolicMine" id="metabolicmine" tool_type="data_source" version="1.0.0">
+<tool name="metabolicMine" id="metabolicmine" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/modmine.xml b/tools/data_source/modmine.xml
index 3fd3bc9595f..b7c7ee6ab4f 100644
--- a/tools/data_source/modmine.xml
+++ b/tools/data_source/modmine.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="modENCODE modMine" id="modmine" tool_type="data_source" version="1.0.0">
+<tool name="modENCODE modMine" id="modmine" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/mousemine.xml b/tools/data_source/mousemine.xml
index 508811f9aac..bbda4c1aef8 100644
--- a/tools/data_source/mousemine.xml
+++ b/tools/data_source/mousemine.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="MouseMine" id="mousemine" tool_type="data_source" version="1.0.0">
+<tool name="MouseMine" id="mousemine" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/ratmine.xml b/tools/data_source/ratmine.xml
index 69aca47743c..4a33f08cc75 100644
--- a/tools/data_source/ratmine.xml
+++ b/tools/data_source/ratmine.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="Ratmine" id="ratmine" tool_type="data_source" version="1.0.0">
+<tool name="Ratmine" id="ratmine" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/sra.xml b/tools/data_source/sra.xml
index 31b014e01cf..07f01e5fed5 100644
--- a/tools/data_source/sra.xml
+++ b/tools/data_source/sra.xml
@@ -1,4 +1,4 @@
-<tool name="SRA" id="sra_source" tool_type="data_source" version="1.0.1" profile="16.04">
+<tool name="SRA" id="sra_source" tool_type="data_source" version="1.0.1" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/ucsc_tablebrowser.xml b/tools/data_source/ucsc_tablebrowser.xml
index 928ff6498fe..19852ecf6fc 100644
--- a/tools/data_source/ucsc_tablebrowser.xml
+++ b/tools/data_source/ucsc_tablebrowser.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="UCSC Main" id="ucsc_table_direct1" tool_type="data_source" version="1.0.0">
+<tool name="UCSC Main" id="ucsc_table_direct1" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>table browser</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/ucsc_tablebrowser_archaea.xml b/tools/data_source/ucsc_tablebrowser_archaea.xml
index 62258145892..a1707e441a1 100644
--- a/tools/data_source/ucsc_tablebrowser_archaea.xml
+++ b/tools/data_source/ucsc_tablebrowser_archaea.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="UCSC Archaea" id="ucsc_table_direct_archaea1" tool_type="data_source" version="1.0.0">
+<tool name="UCSC Archaea" id="ucsc_table_direct_archaea1" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>table browser</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/ucsc_tablebrowser_test.xml b/tools/data_source/ucsc_tablebrowser_test.xml
index e90bc6d14cd..9782f8d2b41 100644
--- a/tools/data_source/ucsc_tablebrowser_test.xml
+++ b/tools/data_source/ucsc_tablebrowser_test.xml
@@ -4,7 +4,7 @@
     the initial response.  If value of 'URL_method' is 'post', any additional params coming back in the
     initial response ( in addition to 'URL' ) will be encoded and appended to URL and a post will be performed.
 -->
-<tool name="UCSC Test" id="ucsc_table_direct_test1" tool_type="data_source" version="1.0.1">
+<tool name="UCSC Test" id="ucsc_table_direct_test1" tool_type="data_source" version="1.0.1" profile="20.09">
     <description>table browser</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/worm_modencode.xml b/tools/data_source/worm_modencode.xml
index 1c6f4be9c74..519430bbb40 100644
--- a/tools/data_source/worm_modencode.xml
+++ b/tools/data_source/worm_modencode.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="modENCODE worm" id="modENCODEworm" tool_type="data_source" version="1.0.1">
+<tool name="modENCODE worm" id="modENCODEworm" tool_type="data_source" version="1.0.1" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/wormbase.xml b/tools/data_source/wormbase.xml
index d279bdf7f76..a1657018ebe 100644
--- a/tools/data_source/wormbase.xml
+++ b/tools/data_source/wormbase.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="WormBase" id="wormbase" tool_type="data_source" version="1.0.1">
+<tool name="WormBase" id="wormbase" tool_type="data_source" version="1.0.1" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/wormbase_test.xml b/tools/data_source/wormbase_test.xml
index 9e2e9609471..a9fb4e05d70 100644
--- a/tools/data_source/wormbase_test.xml
+++ b/tools/data_source/wormbase_test.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="Wormbase" id="wormbase_test" tool_type="data_source" version="1.0.0">
+<tool name="Wormbase" id="wormbase_test" tool_type="data_source" version="1.0.0" profile="20.09">
 	<description>test server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/yeastmine.xml b/tools/data_source/yeastmine.xml
index 87ff7fdf73e..6a669cd33f6 100644
--- a/tools/data_source/yeastmine.xml
+++ b/tools/data_source/yeastmine.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="YeastMine" id="yeastmine" tool_type="data_source" version="1.0.0">
+<tool name="YeastMine" id="yeastmine" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>
diff --git a/tools/data_source/zebrafishmine.xml b/tools/data_source/zebrafishmine.xml
index 18c51fce213..c3296167776 100644
--- a/tools/data_source/zebrafishmine.xml
+++ b/tools/data_source/zebrafishmine.xml
@@ -1,5 +1,5 @@
 <?xml version="1.0"?>
-<tool name="ZebrafishMine" id="zebrafishmine" tool_type="data_source" version="1.0.0">
+<tool name="ZebrafishMine" id="zebrafishmine" tool_type="data_source" version="1.0.0" profile="20.09">
     <description>server</description>
     <edam_operations>
         <edam_operation>operation_0224</edam_operation>