Merge branch 'main' of github.com:galaxyproject/galaxy_tool_metadata_…

…extractor into merge_keep_exclude_list

CODE_OF_CONDUCT.md

@@ -0,0 +1,8 @@
+Code of Conduct
+===============
+
+As part of the Galaxy Community, this project is committed to providing a
+welcoming and harassment-free experience for everyone. We therefore expect
+participants to abide by our Code of Conduct, which can be found at:
+
+https://galaxyproject.org/community/coc/

bin/extract_all_tools.sh

@@ -5,7 +5,7 @@ mkdir -p 'results/'
 python bin/extract_galaxy_tools.py \
         extractools \
         --api $GITHUB_API_KEY \
-        --all_tools 'results/all_tools.tsv'
+        --all-tools 'results/all_tools.tsv'
 
 python bin/create_interactive_table.py \
         --table "results/all_tools.tsv" \

bin/extract_all_tools_stepwise.sh

@@ -4,9 +4,20 @@ mkdir -p 'results/'
 
 output="results/${1}_tools.tsv"
 
-python bin/extract_galaxy_tools.py \
+if [[ $1 =~ "01" ]]; then
+   python bin/extract_galaxy_tools.py \
         extractools \
         --api $GITHUB_API_KEY \
-        --all_tools $output \
-        --planemorepository $1
+        --all-tools $output \
+        --planemo-repository-list $1
+else
+   python bin/extract_galaxy_tools.py \
+        extractools \
+        --api $GITHUB_API_KEY \
+        --all-tools $output \
+        --planemo-repository-list $1 \
+        --avoid-extra-repositories
+fi
+
+
 

bin/extract_all_tools_test.sh

@@ -7,7 +7,7 @@ output="results/${1}_tools.tsv"
 python bin/extract_galaxy_tools.py \
         extractools \
         --api $GITHUB_API_KEY \
-        --all_tools $output \
-        --planemorepository $1 \
+        --all-tools $output \
+        --planemo-repository-list $1 \
         --test
 

bin/extract_galaxy_tools.py

@@ -123,14 +123,14 @@ def get_string_content(cf: ContentFile) -> str:
 
 
 def get_tool_github_repositories(
-    g: Github, RepoSelection: Optional[str], run_test: bool, add_extra_repositories: bool = True
+    g: Github, repository_list: Optional[str], run_test: bool, add_extra_repositories: bool = True
 ) -> List[str]:
     """
     Get list of tool GitHub repositories to parse
 
     :param g: GitHub instance
-    :param RepoSelection: The selection to use from the repository (needed to split the process for CI jobs)
-    :run_test: for testing only parse the repository
+    :param repository_list: The selection to use from the repository (needed to split the process for CI jobs)
+    :param run_test: for testing only parse the repository
     """
 
     if run_test:
@@ -140,8 +140,8 @@ def get_tool_github_repositories(
     repo_list: List[str] = []
     for i in range(1, 5):
         repo_selection = f"repositories0{i}.list"
-        if RepoSelection:  # only get these repositories
-            if RepoSelection == repo_selection:
+        if repository_list:  # only get these repositories
+            if repository_list == repo_selection:
                 repo_f = repo.get_contents(repo_selection)
                 repo_l = get_string_content(repo_f).rstrip()
                 repo_list.extend(repo_l.split("\n"))
@@ -616,11 +616,21 @@ def filter_tools(
     # Extract tools
     extractools = subparser.add_parser("extractools", help="Extract tools")
     extractools.add_argument("--api", "-a", required=True, help="GitHub access token")
-    extractools.add_argument("--all_tools", "-o", required=True, help="Filepath to TSV with all extracted tools")
+    extractools.add_argument("--all-tools", "-o", required=True, help="Filepath to TSV with all extracted tools")
     extractools.add_argument(
-        "--planemorepository", "-pr", required=False, help="Repository list to use from the planemo-monitor repository"
+        "--planemo-repository-list",
+        "-pr",
+        required=False,
+        help="Repository list to use from the planemo-monitor repository",
+    )
+    extractools.add_argument(
+        "--avoid-extra-repositories",
+        "-e",
+        action="store_true",
+        default=False,
+        required=False,
+        help="Do not parse extra repositories in conf file",
     )
-
     extractools.add_argument(
         "--test",
         "-t",
@@ -639,13 +649,13 @@ def filter_tools(
         help="Filepath to TSV with all extracted tools, generated by extractools command",
     )
     filtertools.add_argument(
-        "--ts_filtered_tools",
+        "--ts-filtered-tools",
         "-t",
         required=True,
         help="Filepath to TSV with tools filtered based on ToolShed category",
     )
     filtertools.add_argument(
-        "--filtered_tools",
+        "--filtered-tools",
         "-f",
         required=True,
         help="Filepath to TSV with tools filtered based on ToolShed category and manual curation",
@@ -666,7 +676,12 @@ def filter_tools(
         # connect to GitHub
         g = Github(args.api)
         # get list of GitHub repositories to parse
-        repo_list = get_tool_github_repositories(g, args.planemorepository, args.test)
+        repo_list = get_tool_github_repositories(
+            g=g,
+            repository_list=args.planemo_repository_list,
+            run_test=args.test,
+            add_extra_repositories=not args.avoid_extra_repositories,
+        )
         # parse tools in GitHub repositories to extract metada, filter by TS categories and export to output file
         tools: List[Dict] = []
         for r in repo_list:

bin/get_community_tools.sh

@@ -12,8 +12,8 @@ for com_data_fp in data/communities/* ; do
                 python bin/extract_galaxy_tools.py \
                         filtertools \
                         --tools "results/all_tools.tsv" \
-                        --ts_filtered_tools "results/$community/tools_filtered_by_ts_categories.tsv" \
-                        --filtered_tools "results/$community/tools.tsv" \
+                        --ts-filtered-tools "results/$community/tools_filtered_by_ts_categories.tsv" \
+                        --filtered-tools "results/$community/tools.tsv" \
                         --categories "data/communities/$community/categories" \
                         --status "data/communities/$community/tool_status.tsv"
 

data/conf.yml

@@ -1,3 +1,2 @@
 extra-repositories:
   - https://github.com/qiime2/galaxy-tools
-  - https://github.com/geraldinepascal/FROGS-wrappers