jolt 0.9.123__py3-none-any.whl → 0.9.435__py3-none-any.whl

jolt/plugins/git.py

@@ -1,11 +1,13 @@
 import os
 import pygit2
 import re
+from threading import RLock
+import urllib.parse
 
-from jolt.tasks import BooleanParameter, Export, Parameter, TaskRegistry, WorkspaceResource
+from jolt.tasks import BooleanParameter, Export, IntParameter, Parameter, TaskRegistry, WorkspaceResource
 from jolt.influence import FileInfluence, HashInfluenceRegistry
 from jolt.tools import Tools
-from jolt.loader import JoltLoader
+from jolt.loader import JoltLoader, workspace_locked
 from jolt import config
 from jolt import filesystem as fs
 from jolt import log
@@ -19,6 +21,13 @@ from jolt.error import raise_task_error_if
 log.verbose("[Git] Loaded")
 
 
+def locked(func):
+    def _f(self, *args, **kwargs):
+        with self._lock:
+            return func(self, *args, **kwargs)
+    return _f
+
+
 class GitRepository(object):
     def __init__(self, url, path, relpath, refspecs=None):
         self.path = path
@@ -28,11 +37,12 @@ class GitRepository(object):
         self.url = url
         self.default_refspecs = [
             '+refs/heads/*:refs/remotes/origin/*',
-            '+refs/tags/*:refs/remotes/origin/*',
+            '+refs/tags/*:refs/tags/*',
         ]
         self.refspecs = refspecs or []
         self._tree_hash = {}
         self._original_head = True
+        self._last_rev = None
         self._init_repo()
 
     def _init_repo(self):
@@ -66,18 +76,64 @@ class GitRepository(object):
     def is_indexed(self):
         return self.is_cloned() and fs.path.exists(self._git_index())
 
-    def clone(self):
+    def clone(self, submodules=False):
         log.info("Cloning into {0}", self.path)
+
+        # Get reference repository path
+        refroot = config.get("git", "reference", None)
+        refpath = None
+        if refroot:
+            # Append <host>/<path> to the reference root
+            url = urllib.parse.urlparse(str(self.url))
+            refpath = fs.path.join(refroot, url.hostname, url.path.lstrip("/"))
+            refpath = fs.path.abspath(refpath)
+
+        # If the directory exists, initialize the repository instead of cloning
         if fs.path.exists(self.path):
             with self.tools.cwd(self.path):
-                self.tools.run("git init && git remote add origin {} && git fetch",
-                               self.url, output_on_error=True)
+                self.tools.run("git init", output_on_error=True)
+
+                # Set the reference repository if available
+                if refpath:
+                    # Check if the reference repository is a git repository
+                    objpath = os.path.join(refpath, ".git", "objects")
+                    objpath_bare = os.path.join(refpath, "objects")
+                    if os.path.isdir(objpath):
+                        refpath = objpath
+                    elif os.path.isdir(objpath_bare):
+                        refpath = objpath_bare
+                    else:
+                        refpath = None
+
+                if refpath:
+                    self.tools.mkdir(".git/objects/info")
+                    self.tools.write_file(".git/objects/info/alternates", refpath)
+
+                utils.call_and_catch(self.tools.run, "git remote remove origin", output=False)
+                self.tools.run("git remote add origin {}", self.url, output_on_error=True)
+                self._fetch_origin(submodules=submodules)
+                self.tools.run("git checkout -f FETCH_HEAD", output_on_error=True)
         else:
-            self.tools.run("git clone {0} {1}", self.url, self.path, output_on_error=True)
+            # Get configurable extra clone options
+            extra_clone_options = config.get("git", "clone_options", "")
+
+            if refpath and os.path.isdir(refpath):
+                self.tools.run("git clone --reference-if-able  {0} {1} {2} {3}", refpath, extra_clone_options, self.url, self.path, output_on_error=True)
+            else:
+                self.tools.run("git clone {0} {1} {2}", extra_clone_options, self.url, self.path, output_on_error=True)
+
         self._init_repo()
         raise_error_if(
             self.repository is None,
-            "git: failed to clone repository '{0}'", self.relpath)
+            "Failed to clone repository '{0}'", self.relpath)
+
+    @utils.retried.on_exception(JoltCommandError, pattern="Command failed: git fetch", count=6, backoff=[2, 5, 10, 15, 20, 30])
+    def _fetch_origin(self, submodules=False):
+        # Get configurable extra clone options
+        extra_fetch_options = config.get("git", "fetch_options", "")
+
+        with self.tools.cwd(self.path):
+            self.tools.run("git fetch {0} {1} origin", "", extra_fetch_options, output_on_error=True)
 
     @utils.cached.instance
     def diff_unchecked(self):
@@ -95,7 +151,7 @@ class GitRepository(object):
 
     def diff(self):
         diff = self.diff_unchecked()
-        dlim = config.getsize("git", "maxdiffsize", "1M")
+        dlim = config.getsize("git", "maxdiffsize", "1 MiB")
         raise_error_if(
             len(diff) > dlim,
             "Repository '{}' has uncommitted changes. Size of patch exceeds configured transfer limit ({} > {} bytes)."
@@ -105,8 +161,8 @@ class GitRepository(object):
     def patch(self, patch):
         if not patch:
             return
-        with self.tools.cwd(self.path), self.tools.tmpdir("git") as t:
-            patchfile = fs.path.join(t.path, "jolt.diff")
+        with self.tools.cwd(self.path), self.tools.tmpdir("git") as tmp:
+            patchfile = fs.path.join(tmp, "jolt.diff")
             with open(patchfile, "wb") as f:
                 f.write(patch.encode())
             log.info("Applying patch to {0}", self.path)
@@ -133,17 +189,18 @@ class GitRepository(object):
             try:
                 commit = self.repository.revparse_single(rev)
             except KeyError:
-                self.fetch()
+                self.fetch(commit=rev)
                 try:
                     commit = self.repository.revparse_single(rev)
                 except Exception:
-                    raise_error("invalid git reference: {}", rev)
+                    raise_error("Invalid git reference: {}", rev)
             try:
                 return str(commit.id)
             except Exception:
                 return str(commit)
 
     @utils.cached.instance
+    @workspace_locked
     def write_tree(self):
         tools = Tools()
         with tools.cwd(self._git_folder()):
@@ -154,30 +211,30 @@ class GitRepository(object):
                 output_on_error=True)
         return tree
 
-    def tree_hash(self, sha=None, path="/"):
-        # When sha is None, the caller want the tree hash of the repository's
+    def tree_hash(self, rev=None, path="/"):
+        # When rev is None, the caller want the tree hash of the repository's
         # current workspace state. If no checkout has been made, that would be the
         # tree that was written upon initialization of the repository as it
         # includes any uncommitted changes. If a checkout has been made since
         # the repo was initialized, make this an explicit request for the current
         # head - there can be no local changes.
-        if sha is None:
+        if rev is None:
             if self.is_original_head():
                 tree = self.repository.get(self.write_tree())
             else:
-                sha = self.head()
+                rev = self.head()
 
         path = fs.path.normpath(path)
         full_path = fs.path.join(self.path, path) if path != "/" else self.path
 
         # Lookup tree hash value in cache
-        value = self._tree_hash.get((full_path, sha))
+        value = self._tree_hash.get((full_path, rev))
         if value is not None:
             return value
 
-        # Translate explicit sha to tree
-        if sha is not None:
-            commit = self.rev_parse(sha)
+        # Translate explicit rev to tree
+        if rev is not None:
+            commit = self.rev_parse(rev)
             obj = self.repository.get(commit)
             try:
                 tree = obj.tree
@@ -189,7 +246,7 @@ class GitRepository(object):
             tree = tree[fs.as_posix(path)]
 
         # Update tree hash cache
-        self._tree_hash[(full_path, sha)] = value = tree.id
+        self._tree_hash[(full_path, rev)] = value = tree.id
 
         return value
 
@@ -201,24 +258,49 @@ class GitRepository(object):
         with self.tools.cwd(self.path):
             return self.tools.run("git reset --hard", output_on_error=True)
 
-    def fetch(self):
+    @utils.retried.on_exception(JoltCommandError, pattern="Command failed: git fetch", count=6, backoff=[2, 5, 10, 15, 20, 30])
+    def fetch(self, commit=None):
+        if commit and not self.is_valid_sha(commit):
+            commit = None
+
+        # Get configurable extra clone options
+        extra_fetch_options = config.get("git", "fetch_options", "")
+
         refspec = " ".join(self.default_refspecs + self.refspecs)
         with self.tools.cwd(self.path):
-            log.info("Fetching {0} from {1}", refspec or 'commits', self.url)
-            self.tools.run("git fetch {url} {refspec}",
-                           url=self.url,
-                           refspec=refspec or '',
-                           output_on_error=True)
-
-    def checkout(self, rev):
-        log.info("Checking out {0} in {1}", rev, self.path)
+            log.info("Fetching {0} from {1}", commit or refspec or 'commits', self.url)
+            self.tools.run(
+                "git fetch --force --prune {extra_fetch_options} {url} {what}",
+                extra_fetch_options=extra_fetch_options,
+                url=self.url,
+                what=commit or refspec or '',
+                output_on_error=True)
+
+    def checkout(self, rev, commit=None, submodules=False):
+        if rev == self._last_rev:
+            log.debug("Checkout skipped, already @ {}", rev)
+            return False
+        log.verbose("Checking out {0} in {1}", rev, self.path)
         with self.tools.cwd(self.path):
             try:
-                return self.tools.run("git checkout -f {rev}", rev=rev, output=False)
+                self.tools.run("git checkout -f {rev}", rev=rev, output=False)
+                if submodules:
+                    self._update_submodules()
             except Exception:
-                self.fetch()
-                return self.tools.run("git checkout -f {rev}", rev=rev, output_on_error=True)
+                self.fetch(commit=commit)
+                try:
+                    self.tools.run("git checkout -f {rev}", rev=rev, output_on_error=True)
+                    if submodules:
+                        self._update_submodules()
+                except Exception:
+                    raise_error("Commit does not exist in remote for '{}': {}", self.relpath, rev)
         self._original_head = False
+        self._last_rev = rev
+        return True
+
+    def _update_submodules(self):
+        with self.tools.cwd(self.path):
+            self.tools.run("git submodule update --init --recursive", output_on_error=True)
 
 
 _gits = {}
@@ -228,9 +310,9 @@ def new_git(url, path, relpath, refspecs=None):
     refspecs = utils.as_list(refspecs or [])
     try:
         git = _gits[path]
-        raise_error_if(git.url != url, "multiple git repositories required at {}", relpath)
+        raise_error_if(git.url != url, "Multiple git repositories required at {}", relpath)
         raise_error_if(git.refspecs != refspecs,
-                       "conflicting refspecs detected for git repository at  {}", relpath)
+                       "Conflicting refspecs detected for git repository at  {}", relpath)
         return git
     except Exception:
         git = _gits[path] = GitRepository(url, path, relpath, refspecs)
@@ -241,7 +323,7 @@ class GitInfluenceProvider(FileInfluence):
     name = "Git"
 
     def __init__(self, path):
-        super(GitInfluenceProvider, self).__init__(path)
+        super().__init__(path)
         self.path = path.rstrip(fs.sep)
         self.name = GitInfluenceProvider.name
 
@@ -256,7 +338,7 @@ class GitInfluenceProvider(FileInfluence):
                 return path
             ppath = path
             path = fs.path.dirname(path)
-        raise_error("no git repository found at '{}'", self.path)
+        raise_error("No git repository found at '{}'", self.path)
 
     @utils.cached.instance
     def get_influence(self, task):
@@ -304,23 +386,101 @@ def influence(path, git_cls=GitInfluenceProvider):
     return _decorate
 
 
-class GitSrc(WorkspaceResource, FileInfluence):
-    """ Clones a Git repo.
+class ErrorDict(dict):
+    """ A dict that raises an error if the value of a key is None. """
+
+    def __init__(self, repo):
+        self.repo = repo
+
+    def __getitem__(self, key):
+        value = super().__getitem__(key)
+        raise_task_error_if(value is None, self.repo, "Git repository '{0}' referenced in influence collection before being cloned/checked out. Assign hash=true to the git requirement.", key)
+        return value
+
+
+class Git(WorkspaceResource, FileInfluence):
     """
+    Resource that clones and monitors a Git repo.
+
+    By default, the repo is cloned into a build directory named after
+    the resource. The 'path' parameter can be used to specify a different
+    location relative to the workspace root.
+
+    The path of the cloned repo is made available to consuming tasks
+    through their 'git' attribute. The 'git' attribute is a dictionary
+    where the key is the name of the git repository and the value is
+    the relative path to the repository from the consuming task's
+    workspace.
+
+    The resource influences the hash of consuming tasks, causing tasks
+    to be re-executed if the cloned repo is modified.
+
+    The plugin must be loaded before it can be used. This is done by
+    importing the module, or by adding the following line to the
+    configuration file:
+
+    .. code-block:: ini
+
+            [git]
+
+    Example:
+
+    .. code-block:: python
+
+            from jolt.plugins import git
+
+            class Example(Task):
+                requires = ["git:url=https://github.com/user/repo.git"]
+
+                def run(self, deps, tools):
+                    self.info("The git repo is located at: {git[repo]}")
+                    with tools.cwd(self.git["repo"]):
+                        tools.run("make")
+
+    """
+    name = "git"
 
-    name = "git-src"
     url = Parameter(help="URL to the git repo to be cloned. Required.")
-    sha = Parameter(required=False, help="Specific commit or tag to be checked out. Optional.")
+    """ URL to the git repo to be cloned. Required. """
+
+    rev = Parameter(required=False, help="Specific commit or tag to be checked out. Optional.")
+    """ Specific commit or tag to be checked out. Optional. """
+
+    hash = BooleanParameter(required=False, help="Let repo content influence the hash of consuming tasks.")
+    """ Let repo content influence the hash of consuming tasks. Default ``True``. Optional. """
+
     path = Parameter(required=False, help="Local path where the repository should be cloned.")
-    defer = BooleanParameter(False, help="Defer cloning until a consumer task must be built.")
+    """ Alternative path where the repository should be cloned. Relative to ``joltdir``. Optional. """
+
+    submodules = BooleanParameter(default=False, help="Initialize and update git submodules after cloning.")
+    """ Initialize and update git submodules after cloning. Default ``False``. Optional. """
+
+    clean = BooleanParameter(default=False, help="Clean repository after it has been used")
+    """ Removes untracked files from the repository. """
+
     _revision = Export(value=lambda t: t._export_revision())
+    """ To worker exported value of the revision to be checked out. """
+
     _diff = Export(value=lambda t: t.git.diff(), encoded=True)
+    """ To worker exported value of the diff of the repo. """
 
     def __init__(self, *args, **kwargs):
-        super(GitSrc, self).__init__(*args, **kwargs)
+        super().__init__(*args, **kwargs)
+        self._lock = RLock()
         self.joltdir = JoltLoader.get().joltdir
-        self.relpath = str(self.path) or self._get_name()
-        self.abspath = fs.path.join(self.joltdir, self.relpath)
+
+        # Set the path to the repo
+        if self.path.is_unset():
+            self.abspath = self.tools.builddir(utils.canonical(self.short_qualified_name), incremental="always", unique=False)
+            self.relpath = fs.path.relpath(self.abspath, self.tools.wsroot)
+        else:
+            self.abspath = fs.path.join(self.joltdir, str(self.path) or self._get_name())
+            self.relpath = fs.path.relpath(self.abspath, self.tools.wsroot)
+
+        self.abspath = fs.path.normpath(self.abspath)
+        self.relpath = fs.path.normpath(self.relpath)
+
+        # Create the git repository
         self.refspecs = kwargs.get("refspecs", [])
         self.git = new_git(self.url, self.abspath, self.relpath, self.refspecs)
 
@@ -331,72 +491,113 @@ class GitSrc(WorkspaceResource, FileInfluence):
         return name
 
     def _export_revision(self):
-        return self.sha.value or self.git.head()
+        return self.rev.value or self.git.head()
 
     def _get_revision(self):
         if self._revision.is_imported:
             return self._revision.value
-        if not self.sha.is_unset():
-            return self.sha.get_value()
+        if not self.rev.is_unset():
+            return self.rev.get_value()
         return None
 
-    def acquire(self, **kwargs):
+    def _assign_git(self, task, none=False):
+        if not hasattr(task, "git"):
+            task.git = ErrorDict(self)
+        if none:
+            # None means the git repo is not cloned or checked out
+            # and should not be included in the git dictionary
+            # of the consuming task yet. If the consuming task
+            # requires the git repo for its influence collection,
+            # the dict will raise an error. The solution is to
+            # assign hash=true to the git requirement which
+            # will cause the git repo to be cloned and checked out
+            # before the influence collection is performed.
+            task.git[self._get_name()] = None
+        else:
+            # Assign the git repo to the consuming task.
+            # The git repo is cloned and checked out before
+            # any influence collection is performed.
+            task.git[self._get_name()] = fs.path.relpath(self.abspath, task.joltdir)
+
+    def acquire(self, artifact, deps, tools, owner):
         self._acquire_ws()
+        self._assign_git(owner)
+        artifact.worktree = fs.path.relpath(self.abspath, owner.joltdir)
+
+    def release(self, artifact, deps, tools, owner):
+        if self.clean:
+            self.git.clean()
+            self.git.reset()
+
+    def prepare_ws_for(self, task):
+        """ Prepare the workspace for the task.
+
+        :param task: The task to prepare the workspace for.
+        """
+        if not self._must_influence():
+            # The content of the git repo is not required to influence the hash of the
+            # consumer task. The repo is therefore not cloned or checked out
+            # until the consumer is executed. Raise an error if the git repo
+            # is required for the influence collection of the consumer task.
+            self._assign_git(task, none=True)
+            return
+        # The content of the git repo is required to influence the hash of the consumer task.
+        self._assign_git(task)
 
-    def acquire_ws(self):
-        if self.defer is None or self.defer.is_false:
+    def acquire_ws(self, force=False):
+        """ Clone and/or checkout the git repo if required """
+        if force or self._must_influence() or self._revision.is_imported:
             self._acquire_ws()
 
+    @locked
     def _acquire_ws(self):
+        commit = None
         if not self.git.is_cloned():
-            self.git.clone()
+            self.git.clone(submodules=bool(self.submodules))
         if not self._revision.is_imported:
             self.git.diff_unchecked()
+        else:
+            commit = self._revision.value
         rev = self._get_revision()
         if rev is not None:
             raise_task_error_if(
-                not self._revision.is_imported and not self.sha.is_unset() and self.git.diff(), self,
-                "explicit sha requested but git repo '{0}' has local changes", self.git.relpath)
+                not self._revision.is_imported and not self.rev.is_unset() and self.git.diff(), self,
+                "Explicit revision requested but git repo '{0}' has local changes, refusing checkout", self.git.relpath)
             # Should be safe to do this now
             rev = self.git.rev_parse(rev)
             if not self.git.is_head(rev) or self._revision.is_imported:
-                self.git.checkout(rev)
-                self.git.clean()
-                self.git.patch(self._diff.value)
+                if self.git.checkout(rev, commit=commit, submodules=bool(self.submodules)):
+                    self.git.clean()
+                    self.git.patch(self._diff.value)
 
-    def get_influence(self, task):
-        return None
-
-    def is_influenced_by(self, task, path):
-        return fs.is_relative_to(path, self.abspath) and self.sha.is_set()
+    def _must_influence(self):
+        """ Check if the git repo must influence the hash of the consumer task."""
 
+        # If the hash parameter is set, honor it
+        if self.hash.is_set():
+            return self.hash
 
-TaskRegistry.get().add_task_class(GitSrc)
+        # If the revision parameter is not set, the git repo must influence the hash
+        if self.rev.is_unset():
+            return True
 
+        # If the revision parameter is set, no influence is needed since the
+        # revision is fixed and repository content will not change.
+        return False
 
-class Git(GitSrc):
-    """ Clones a Git repo.
-
-    Also influences the hash of consuming tasks, causing tasks to
-    be re-executed if the cloned repo is modified.
-
-    """
-    name = "git"
-    url = Parameter(help="URL to the git repo to be cloned. Required.")
-    sha = Parameter(required=False, help="Specific commit or tag to be checked out. Optional.")
-    path = Parameter(required=False, help="Local path where the repository should be cloned.")
-    defer = None
-    _revision = Export(value=lambda t: t._export_revision())
-    _diff = Export(value=lambda t: t.git.diff(), encoded=True)
+    def is_influenced_by(self, task, path):
+        influencing = self._must_influence() or self.rev.is_set()
+        return influencing and fs.is_relative_to(path, self.abspath)
 
-    def __init__(self, *args, **kwargs):
-        super(Git, self).__init__(*args, **kwargs)
-        self.influence.append(self)
+    def _influence(self):
+        influence = super()._influence()
+        return influence + [self] if self._must_influence() else influence
 
+    @locked
     @utils.cached.instance
     def get_influence(self, task):
         if not self.git.is_cloned():
-            self.git.clone()
+            self.git.clone(submodules=bool(self.submodules))
         if not self._revision.is_imported:
             self.git.diff_unchecked()
         rev = self._get_revision()
@@ -411,8 +612,5 @@ class Git(GitSrc):
 
         return "{0}: {1}".format(self.git.relpath, th)
 
-    def is_influenced_by(self, task, path):
-        return path.startswith(self.abspath + fs.sep)
-
 
 TaskRegistry.get().add_task_class(Git)

jolt/plugins/googletest.py

@@ -212,7 +212,7 @@ def filter(default: str = "*", param: bool = True, attr: str = "filter"):
     Task class decorator controlling the GTEST_FILTER environment variable.
 
     The variable instructs test applications to only run test-cases that matches
-    a wildcard filter string (default: *).
+    a wildcard filter string (default: ``*``).
 
     The value is taken from the parameter ``filter`` which is created
     by default. If no parameter is created, the value is instead taken from the class
@@ -473,6 +473,7 @@ def shuffle(default: bool = False, param: bool = True, attr: str = "shuffle"):
 @brief()
 @disabled()
 @fail_fast()
+@filter()
 @junit_report()
 @repeat()
 @seed()