Merge pull request #21617 from mvdbeek/fix-toolshed-tests

Fix toolshed tests on Python 3.14

Author: mvdbeek

lib/galaxy/dependencies/pinned-requirements.txt

@@ -170,7 +170,8 @@ markdown-it-py==4.0.0 ; python_full_version >= '3.11'
 markupsafe==3.0.3
 mcp==1.25.0 ; python_full_version >= '3.10'
 mdurl==0.1.2
-mercurial==7.1.2
+mercurial==7.1.2 ; python_version < '3.14'
+mercurial==7.2rc0 ; python_version >= '3.14'
 mistralai==1.9.11
 mistune==3.1.4
 more-itertools==10.8.0

lib/galaxy/tool_shed/metadata/metadata_generator.py

@@ -392,7 +392,9 @@ def generate_metadata_for_changeset_revision(self):
         if invalid_tool_configs:
             metadata_dict["invalid_tools"] = invalid_tool_configs
         self.metadata_dict = metadata_dict
-        remove_dir(work_dir)
+        # Only remove work_dir if not resetting all metadata - in that case the caller handles cleanup
+        if not self.resetting_all_metadata_on_repository:
+            remove_dir(work_dir)
 
     def generate_package_dependency_metadata(self, elem, valid_tool_dependencies_dict, invalid_tool_dependencies_dict):
         """

lib/tool_shed/metadata/repository_metadata_manager.py

@@ -799,14 +799,24 @@ def reset_all_metadata_on_repository_in_tool_shed(self, repository_clone_url=Non
         metadata_dict = None
         ancestor_changeset_revision = None
         ancestor_metadata_dict = None
-        for changeset in self.repository.get_changesets_for_setting_metadata(self.app):
-            work_dir = tempfile.mkdtemp(prefix="tmp-toolshed-ramorits")
-            ctx = repo[changeset]
-            log.debug("Cloning repository changeset revision: %s", str(ctx.rev()))
-            assert self.repository_clone_url
-            repository_clone_url = repository_clone_url or self.repository_clone_url
-            cloned_ok, error_message = hg_util.clone_repository(repository_clone_url, work_dir, str(ctx.rev()))
-            if cloned_ok:
+
+        # Clone repository once, then update for each changeset revision.
+        work_dir = tempfile.mkdtemp(prefix="tmp-toolshed-ramorits")
+        assert self.repository_clone_url
+        repository_clone_url = repository_clone_url or self.repository_clone_url
+        log.debug("Cloning repository for metadata reset")
+        cloned_ok, error_message = hg_util.clone_repository(repository_clone_url, work_dir)
+        if not cloned_ok:
+            log.error(f"Failed to clone repository: {error_message}")
+            basic_util.remove_dir(work_dir)
+            self._clean_repository_metadata(changeset_revisions)
+            return
+
+        try:
+            for changeset in self.repository.get_changesets_for_setting_metadata(self.app):
+                ctx = repo[changeset]
+                log.debug("Updating to changeset revision: %s", str(ctx.rev()))
+                hg_util.update_repository(work_dir, str(ctx.rev()))
                 log.debug("Generating metadata for changeset revision: %s", str(ctx.rev()))
                 self.set_changeset_revision(str(ctx))
                 self.set_repository_files_dir(work_dir)
@@ -823,11 +833,11 @@ def reset_all_metadata_on_repository_in_tool_shed(self, repository_clone_url=Non
                         # self.SUBSET - ancestor metadata is a subset of current metadata, so continue from current
                         # self.NOT_EQUAL_AND_NOT_SUBSET - ancestor metadata is neither equal to nor a subset of current
                         # metadata, so persist ancestor metadata.
-                        log.info(f"amd {ancestor_metadata_dict}")
+                        log.debug(f"amd {ancestor_metadata_dict}")
                         comparison = self.compare_changeset_revisions(
                             ancestor_changeset_revision, ancestor_metadata_dict
                         )
-                        log.info(f"comparison {comparison}")
+                        log.debug(f"comparison {comparison}")
                         if comparison in [self.NO_METADATA, self.EQUAL, self.SUBSET]:
                             ancestor_changeset_revision = self.changeset_revision
                             ancestor_metadata_dict = self.metadata_dict
@@ -858,6 +868,7 @@ def reset_all_metadata_on_repository_in_tool_shed(self, repository_clone_url=Non
                         changeset_revisions.append(metadata_changeset_revision)
                         ancestor_changeset_revision = None
                         ancestor_metadata_dict = None
+        finally:
             basic_util.remove_dir(work_dir)
         # Delete all repository_metadata records for this repository that do not have a changeset_revision
         # value in changeset_revisions.

lib/tool_shed/webapp/controllers/repository.py

@@ -4,10 +4,9 @@
 import string
 import tempfile
 from datetime import date
+from typing import Any
 
 from mercurial import (
-    cmdutil,
-    commands,
     mdiff,
     patch,
 )
@@ -74,15 +73,15 @@
 malicious_error_can_push = "  Correct this changeset as soon as possible, it potentially produces malicious behavior or contains inappropriate content."
 
 
-def get_mercurial_default_options_dict(command):
-    """Borrowed from repoman - get default parameters for a mercurial command."""
-    possible = cmdutil.findpossible(command, commands.table)
-    # Mercurial >= 3.4 returns a tuple whose first element is the old return dict
-    if type(possible) is tuple:
-        possible = possible[0]
-    if len(possible) != 1:
-        raise Exception(f'unable to find mercurial command "{command}"')
-    return {r[1].replace(b"-", b"_"): r[2] for r in next(iter(possible.values()))[1][1]}
+def get_mercurial_default_options_dict(command: str) -> dict[str, Any]:
+    """Get default parameters for a mercurial command."""
+    # Use mdiff.diffopts defaults directly instead of introspecting command table
+    # (the old cmdutil.findpossible API was removed in Mercurial 7.2)
+    if command == "diff":
+        # Convert byte keys to strings but preserve value types (int, bool)
+        # as mdiff.diffopts expects properly typed values
+        return {(k.decode("utf-8") if isinstance(k, bytes) else k): v for k, v in mdiff.diffopts.defaults.items()}
+    raise Exception(f'unable to find mercurial command "{command}"')
 
 
 class RepositoryController(BaseUIController, ratings_util.ItemRatings):
@@ -2366,10 +2365,8 @@ def view_changeset(self, trans, id, ctx_str, **kwd):
         else:
             ctx_child = None
         diffs = []
-        options_dict = get_mercurial_default_options_dict(b"diff")
-        # Not quite sure if the following settings make any difference, but with a combination of them and the size check on each
-        # diff, we don't run out of memory when viewing the changelog of the cisortho2 repository on the test tool shed.
-        options_dict = {util.unicodify(k): util.unicodify(v) for k, v in options_dict.items()}
+        # Get default diff options with string keys and properly typed values
+        options_dict = get_mercurial_default_options_dict("diff")
         options_dict["maxfile"] = basic_util.MAXDIFFSIZE
         options_dict["maxtotal"] = basic_util.MAXDIFFSIZE
         diffopts = mdiff.diffopts(**options_dict)

pyproject.toml

@@ -57,7 +57,8 @@ dependencies = [
     "Mako",
     "Markdown",
     "MarkupSafe",
-    "mercurial>=6.8.2",  # Python 3.13 support
+    "mercurial>=6.8.2 ; python_version<'3.14'",  # Python 3.13 support
+    "mercurial>=7.2rc0 ; python_version>='3.14'",  # Python 3.14 support
     "mrcfile",
     "more-itertools",
     "msal",