bcc, ompi and crun are added

varats-core/varats/project/project_domain.py

@@ -31,6 +31,8 @@ class ProjectDomains(Enum):
     UNIX_TOOLS = "UNIX utils"
     VERSION_CONTROL = "Version control"
     WEB_TOOLS = "Web tools"
+    RUNTIME = "Runtime"
+    MPI = "Message Passing Interface"
 
     def __str__(self) -> str:
         return str(self.value)

varats/varats/plots/blame_interaction_graph_plots.py

@@ -1,5 +1,6 @@
 """Module for BlameInteractionGraph plots."""
-
+import configparser
+import subprocess
 import typing as tp
 from datetime import datetime
 from pathlib import Path
@@ -9,6 +10,7 @@
 import networkx as nx
 import pandas as pd
 import plotly.offline as offply
+import pygit2
 
 from varats.data.reports.blame_interaction_graph import (
     create_blame_interaction_graph,
@@ -18,7 +20,7 @@
     CAIGNodeAttrs,
 )
 from varats.experiments.vara.blame_report_experiment import (
-    BlameReportExperiment,
+    BlameReportExperiment, BlameReportExperimentRegion,
 )
 from varats.mapping.commit_map import get_commit_map
 from varats.paper_mgmt.case_study import (
@@ -64,7 +66,7 @@
         revision = commit_map.convert_to_full_or_warn(short_revision)
 
         cig = create_blame_interaction_graph(
-            project_name, revision, BlameReportExperiment
+            project_name, revision, BlameReportExperimentRegion
         ).commit_interaction_graph()
         nx.set_node_attributes(
             cig,
@@ -109,7 +111,7 @@
     NodeTy, NodeTy, EdgeInfoTy]]]:
     commit_lookup = create_commit_lookup_helper(project_name)
     cig = create_blame_interaction_graph(
-        project_name, revision, BlameReportExperiment
+        project_name, revision, BlameReportExperimentRegion
     ).commit_interaction_graph()
 
     def filter_nodes(node: CommitRepoPair) -> bool:
@@ -280,6 +282,138 @@
 )
 
 
+def get_gitmodules_info():
+    repo = pygit2.Repository(REPO_PATH)
+    # Get the .gitmodules blob object
+    commit = repo[repo.head.target]
+    tree = commit.tree
+    gitmodules_entry = tree[".gitmodules"]
+    gitmodules_blob = repo[gitmodules_entry.oid]
+
+    # Parse the .gitmodules file contents
+    gitmodules_data = gitmodules_blob.data
+    gitmodules_config = configparser.ConfigParser()
+    gitmodules_config.read_string(gitmodules_data.decode('utf-8'))
+
+    # Collect the submodule path and URL information
+    submodules = []
+    for section in gitmodules_config.sections():
+        if section.startswith('submodule '):
+            submodule_name = section[len('submodule '):]
+            submodule_path = gitmodules_config.get(section, 'path')
+            submodule_url = gitmodules_config.get(section, 'url')
+            submodules.append((submodule_name, submodule_path, submodule_url))
+
+    return submodules
+
+
+def pull_commit_history_information(submodules):
+
+    # Get the tree object for the specified commit
+    repo = pygit2.Repository(REPO_PATH)
+
+    commits = []
+    i = 0
+    for submodule in submodules:
+        walker = repo.walk(repo.head.target, pygit2.GIT_SORT_TOPOLOGICAL | pygit2.GIT_SORT_REVERSE)
+        subdirectory_path = submodule[1]
+        for commit in walker:
+            # Initialize a flag to check if the 'third_party' has changes in this commit
+            # has_changes = False
+            # Compare the tree of this commit with its parent
+            if len(commit.parents) == 0:
+                continue
+            diff = repo.diff(commit.parents[0], commit)
+            # Loop through all patches in the diff
+            files_changed = [patch.delta.new_file.path for patch in diff]
+            if any([file.startswith(subdirectory_path) for file in files_changed]):
+                i += 1
+                print(i)  # total 1864 times commits is for third party
+                commits.append({
+                    'author_name': commit.author.name,
+                    'author_email': commit.author.email,
+                    # 'message': commit.message,
+                    'commit_hash': commit.hex,
+                    'commit_time': datetime.datetime.fromtimestamp(commit.commit_time)
+                    # 'changed_folder': get_change_folder(commit)
+                })
+
+    df = pd.DataFrame(commits)
+    # df.to_csv(TASK_PATH + 'openssl_commit_history2.csv', index=False)
+
+    return df
+
+
+class SubCommitInteractionGraphPlot(Plot, plot_name='sub-cig-plot'):
+    submodule_info = get_gitmodules_info()
+    df = pull_commit_history_information(submodule_info)
+
+    def is_submodule_change(self, commit_hash):
+        for index, row in self.df.iterrows():
+            if row['commit_hash'] == commit_hash:
+                return True
+            else:
+                return False
+
+    def plot(self, view_mode: bool) -> None:
+        project_name = self.plot_kwargs["case_study"].project_name
+        commit_map = get_commit_map(project_name)
+        short_revision = ShortCommitHash(self.plot_kwargs["revision"])
+        revision = commit_map.convert_to_full_or_warn(short_revision)
+
+        cig = create_blame_interaction_graph(
+            project_name, revision, BlameReportExperimentRegion
+        ).commit_interaction_graph()
+        nx.set_node_attributes(
+            cig,
+            {node: cig.nodes[node]["commit"].commit_hash for node in cig.nodes},
+            "label"
+        )
+
+        # Find nodes representing commits that changed submodules
+        submodule_commit_nodes = [node for node in cig.nodes if
+                                  self.is_submodule_change(cig.nodes[node]["commit"].commit_hash)]
+
+        # Count the number of unique authors for main repository and submodule changes
+        main_repo_authors = set()
+        submodule_change_authors = set()
+
+        for node in cig.nodes:
+            commit_info = cig.nodes[node]["commit"]
+            author = commit_info.author_name  # Assuming the existence of the author_name attribute, modify as needed
+
+            main_repo_authors.add(author)
+
+            if node in submodule_commit_nodes:
+                submodule_change_authors.add(author)
+
+        main_repo_author_count = len(main_repo_authors)
+        submodule_change_author_count = len(submodule_change_authors)
+
+        print(f"Main Repository Authors: {main_repo_author_count}")
+        print(f"Authors with Submodule Changes: {submodule_change_author_count}")
+
+
+
+    def calc_missing_revisions(
+        self, boundary_gradient: float
+    ) -> tp.Set[FullCommitHash]:
+        raise UnsupportedOperation
+
+
+class SubCIGPlotGenerator(
+    PlotGenerator,
+    generator_name="sub-cig-plot",
+    options=[REQUIRE_CASE_STUDY, REQUIRE_REVISION]
+):
+    """Generates a plot for a submodule's commit interaction graph."""
+
+    def generate(self) -> tp.List[Plot]:
+        return [
+            SubCommitInteractionGraphPlot(self.plot_config, **self.plot_kwargs)
+        ]
+
+
 class CommitInteractionGraphNodeDegreePlot(Plot, plot_name='cig_node_degrees'):
     """
     Plot node degrees of a commit interaction graph.

varats/varats/projects/c_projects/bcc.py

@@ -0,0 +1,101 @@
+"""Project file for the bcc."""
+import typing as tp
+
+import benchbuild as bb
+from benchbuild.utils.cmd import make, mkdir, cmake, sudo
+from benchbuild.utils.settings import get_number_of_jobs
+from plumbum import local
+
+from varats.containers.containers import get_base_image, ImageBase
+from varats.paper.paper_config import (
+    project_filter_generator,
+    PaperConfigSpecificGit,
+)
+from varats.project.project_domain import ProjectDomains
+from varats.project.project_util import (
+    ProjectBinaryWrapper,
+    BinaryType,
+    verify_binaries,
+    get_local_project_git_path,
+)
+from varats.project.varats_project import VProject
+from varats.utils.git_util import ShortCommitHash, RevisionBinaryMap
+from varats.utils.settings import bb_cfg
+
+
+class Bcc(VProject):
+
+    NAME = 'bcc'
+    GROUP = 'c_projects'
+    DOMAIN = ProjectDomains.UNIX_TOOLS
+
+    SOURCE = [
+        PaperConfigSpecificGit(
+            project_name="bcc",
+            remote="https://github.com/iovisor/bcc",
+            local="bcc",
+            refspec="origin/HEAD",
+            limit=None,
+            shallow=False
+        ),
+        bb.source.GitSubmodule(
+            remote="https://github.com/libbpf/libbpf.git",
+            local="bcc/src/cc/libbpf",
+            refspec="origin/HEAD",
+            limit=None,
+            shallow=False,
+            version_filter=project_filter_generator("bcc")
+        ),
+        bb.source.GitSubmodule(
+            remote="https://github.com/libbpf/bpftool",
+            local="bcc/libbpf-tools/bpftool",
+            refspec="origin/HEAD",
+            limit=None,
+            shallow=False,
+            version_filter=project_filter_generator("bcc")
+        ),
+        bb.source.GitSubmodule(
+            remote="https://github.com/libbpf/blazesym",
+            local="bcc/ibbpf-tools/blazesym",
+            refspec="origin/HEAD",
+            limit=None,
+            shallow=False,
+            version_filter=project_filter_generator("bcc")
+        )
+    ]
+
+    @staticmethod
+    def binaries_for_revision(
+        revision: ShortCommitHash
+    ) -> tp.List[ProjectBinaryWrapper]:
+        binary_map = RevisionBinaryMap(get_local_project_git_path(Bcc.NAME))
+
+        binary_map.specify_binary("build/src/cc/libbcc.so", BinaryType.SHARED_LIBRARY)
+
+        return binary_map[revision]
+
+    def compile(self) -> None:
+        """Compile the BCC project."""
+        bcc_source = local.path(self.source_of_primary)
+        mkdir("-p", bcc_source / "build")
+
+        cc_compiler = bb.compiler.cc(self)
+        cxx_compiler = bb.compiler.cxx(self)
+        with local.cwd(bcc_source / 'build'):
+            with local.env(CC=str(cc_compiler), CXX=cxx_compiler):
+                # llvm_path = "/lib/llvm-14/lib/cmake/llvm/"
+                bb.watch(cmake)("..",
+                                # "-DCMAKE_BUILD_TYPE=Release",
+                                # f"-DLLVM_DIR={llvm_path}",
+                                # "-DCMAKE_POSITION_INDEPENDENT_CODE=ON",
+                                "-DCMAKE_C_FLAGS=-fPIE", "-DCMAKE_CXX_FLAGS=-fPIE",
+                                # "-DCMAKE_EXE_LINKER_FLAGS=-pie"
+                                )  # Configuring the project
+                # Compiling the project with make, specifying the number of jobs
+                bb.watch(make)("-j", get_number_of_jobs(bb_cfg()))
+
+        with local.cwd(bcc_source):
+            verify_binaries(self)
+
+
+

varats/varats/projects/c_projects/crun.py

@@ -0,0 +1,72 @@
+"""Project file for the Crun."""
+import typing as tp
+
+import benchbuild as bb
+from benchbuild.utils.cmd import make
+from benchbuild.utils.settings import get_number_of_jobs
+from plumbum import local
+
+from varats.containers.containers import get_base_image, ImageBase
+from varats.paper.paper_config import (
+    project_filter_generator,
+    PaperConfigSpecificGit,
+)
+from varats.project.project_domain import ProjectDomains
+from varats.project.project_util import (
+    ProjectBinaryWrapper,
+    BinaryType,
+    verify_binaries,
+    get_local_project_git_path,
+)
+from varats.project.varats_project import VProject
+from varats.utils.git_util import ShortCommitHash, RevisionBinaryMap
+from varats.utils.settings import bb_cfg
+
+
+class Crun(VProject):
+
+    NAME = 'crun'
+    GROUP = 'c_projects'
+    DOMAIN = ProjectDomains.RUNTIME
+
+    SOURCE = [
+        PaperConfigSpecificGit(
+            project_name="crun",
+            remote="https://github.com/containers/crun",
+            local="crun",
+            refspec="origin/HEAD",
+            limit=None,
+            shallow=False
+        ),
+        bb.source.GitSubmodule(
+            remote="https://github.com/containers/libocispec.git",
+            local="crun/libocispec",
+            refspec="origin/HEAD",
+            limit=None,
+            shallow=False,
+            version_filter=project_filter_generator("crun")
+        )
+    ]
+
+    @staticmethod
+    def binaries_for_revision(
+        revision: ShortCommitHash
+    ) -> tp.List[ProjectBinaryWrapper]:
+        binary_map = RevisionBinaryMap(get_local_project_git_path(Crun.NAME))
+
+        binary_map.specify_binary("crun", BinaryType.EXECUTABLE)
+
+        return binary_map[revision]
+
+    def compile(self) -> None:
+        crun_source = local.path(self.source_of_primary)
+        compiler = bb.compiler.cc(self)
+        with local.cwd(crun_source):
+            with local.env(CC=str(compiler)):
+                bb.watch(local["./autogen.sh"])()
+                bb.watch(local["./configure"])("--disable-gcc-warnings")
+
+            bb.watch(make)("-j", get_number_of_jobs(bb_cfg()))
+
+            verify_binaries(self)
+