added jacoco injections

handlers.py

@@ -104,10 +104,28 @@ class BuildHandler(ABC):
         finally:
             signal.alarm(0)  # Cancel the alarm
 
-    def generate_coverage_report(self):
+    def generate_coverage_report(self, already_injected_manually: bool = False):
         result = self.container.exec_run(self.generate_coverage_report_cmd())
         if result.exit_code != 0:
-            raise CantExecJacoco(clean_output(result.output))
+            if already_injected_manually:
+                raise CantExecJacoco(clean_output(result.output))
+
+            build_file_path = os.path.join(self.path, self.build_file)
+            if not os.path.exists(build_file_path):
+                raise CantInjectJacoco("pom.xml not found")
+            with open(build_file_path, "r") as f:
+                og_content = f.read()
+            try:
+                self._try_to_inject_jacoco(build_file_path)
+                self.generate_coverage_report(already_injected_manually=True)
+            except (CantInjectJacoco, CantExecJacoco) as e:
+                with open(build_file_path, "w") as f:
+                    f.write(og_content)
+                    raise e
+
+    @abstractmethod
+    def _try_to_inject_jacoco(self, build_file_path: str) -> None:
+        pass
 
     def check_coverage(self, filename: str) -> Iterator[Tuple[str, float]]:
         """
@@ -261,6 +279,54 @@ class MavenHandler(BuildHandler):
         if not found_at_least_one:
             raise NoCoverageReportFound(f"Couldn't find any 'jacoco.xml' in {self.path}")
 
+    def _try_to_inject_jacoco(self, build_file_path: str) -> None:
+        with open(build_file_path, "r", encoding="utf-8") as f:
+            content = f.read()
+
+        if "<artifactId>jacoco-maven-plugin</artifactId>" in content:
+            return   # already present
+
+        jacoco_plugin = """
+    <plugin>
+        <groupId>org.jacoco</groupId>
+        <artifactId>jacoco-maven-plugin</artifactId>
+        <version>0.8.8</version>
+        <executions>
+            <execution>
+                <goals>
+                    <goal>prepare-agent</goal>
+                </goals>
+            </execution>
+            <execution>
+                <id>report</id>
+                <phase>test</phase>
+                <goals>
+                    <goal>report</goal>
+                </goals>
+            </execution>
+        </executions>
+    </plugin>
+"""
+
+        if "<plugins>" in content:
+            # just insert inside existing plugins
+            content = content.replace("<plugins>", f"<plugins>\n{jacoco_plugin}")
+        elif "</project>" in content:
+            # plugins section doesn't exist, create full <build> section
+            build_block = f"""
+        <build>
+            <plugins>
+    {jacoco_plugin}
+            </plugins>
+        </build>
+    """
+            content = content.replace("</project>", f"{build_block}\n</project>")
+        else:
+            raise CantInjectJacoco("Could not find insertion point for plugins in pom.xml")
+
+        with open(build_file_path, "w", encoding="utf-8") as f:
+            f.write(content)
+
 
 class GradleHandler(BuildHandler):
     def __init__(self, repo_path: str, build_file: str, updates: dict) -> None:
@@ -341,6 +407,39 @@ class GradleHandler(BuildHandler):
                 f"Couldn't find any 'index.html' inside any 'reports/jacoco' in {self.path}"
             )
 
+    def _try_to_inject_jacoco(self, build_file_path: str) -> None:
+        with open(build_file_path, "r", encoding="utf-8") as f:
+            content = f.read()
+
+        if "id 'jacoco'" in content or "apply plugin: 'jacoco'" in content:
+            return  # already present
+
+        jacoco_snippet = """
+plugins {
+    id 'jacoco'
+}
+
+jacoco {
+    toolVersion = "0.8.8"
+}
+
+test {
+    finalizedBy jacocoTestReport
+}
+
+jacocoTestReport {
+    dependsOn test
+    reports {
+        xml.required = true
+        html.required = true
+    }
+}"""
+
+        content = jacoco_snippet + "\n\n" + content
+
+        with open(build_file_path, "w", encoding="utf-8") as f:
+            f.write(content)
+
 
 class HandlerException(Exception, ABC):
     reason_for_failure = "Generic handler expection (this shouldn't appear)"
@@ -366,6 +465,10 @@ class CantExecJacoco(HandlerException):
     reason_for_failure = "Couldn't execute jacoco"
 
 
+class CantInjectJacoco(HandlerException):
+    reason_for_failure = "Couldn't inject jacoco in the build file"
+
+
 class NoCoverageReportFound(HandlerException):
     reason_for_failure = "No coverage report was found"
 

pull_requests.py

@@ -28,10 +28,18 @@ def get_good_projects(csv_file: str) -> pd.DataFrame:
     return df.loc[(df['good_repo_for_crab'] == True) & (df['n_tests'] > 0)]
 
 
-def is_pull_good(pull: PullRequest, verbose: bool = False):
+def is_pull_good(pull: PullRequest, verbose: bool = False) -> bool:
-    return pull.user.type != "Bot" and has_only_1_comment(
+    comments = pull.get_review_comments()
-        pull.get_commits(), pull.get_review_comments(), verbose=verbose
+    if pull.user.type != "Bot" or comments.totalCount > 2:
-    )
+        return False
+
+    if comments.totalCount == 2:
+        comment_list = list(comments)
+        second_comment = comment_list[1]
+        if second_comment.user.login != pull.user.login:
+            return False
+
+    return has_only_1_comment(pull.get_commits(), pull.get_review_comments(), verbose=verbose)
 
 
 def run_git_cmd(cmd: list[str], repo_path: str) -> subprocess.CompletedProcess:
@@ -88,8 +96,7 @@ def process_pull(
 
     try:
         diffs_after = {
-            file.filename: file.patch
+            file.filename: file.patch for file in repo.compare(first_commit.sha, last_commit.sha).files
-            for file in repo.compare(first_commit.sha, last_commit.sha).files
         }
     except GithubException as e:
         return
@@ -254,6 +261,48 @@ def process_repos(
             pbar.update(1)
 
 
+def only_inject_jacoco(
+    dataset: Dataset,
+    repos_dir: str,
+    cache: dict[str, dict[int, DatasetEntry]] = {},
+):
+    n_successfull_injections = 0
+    n_tried_injections = 0
+    with tqdm(cache, desc="Processing repos (only for injection") as top_bar:
+        for repo_name in top_bar:
+            top_bar.set_postfix(
+                {
+                    "# successfull injections": f"{n_successfull_injections}/{n_tried_injections} ({n_successfull_injections/n_tried_injections if n_tried_injections > 0 else 0:.2%})"
+                }
+            )
+            with tqdm(total=len(cache[repo_name]), desc=f"Processing prs", leave=False) as pbar:
+                # extracting keys so that it doesn't get messy as I pop elements from the dict
+                pr_numbers = list(cache[repo_name].keys())
+                for pr_number in pr_numbers:
+                    pbar.set_postfix({"repo": repo_name, "pr": pr_number})
+
+                    entry = cache[repo_name].pop(pr_number)
+                    if entry.metadata.reason_for_failure != "Couldn't execute jacoco":
+                        dataset.entries.append(entry)
+                        dataset.to_json(args.output)
+                        pbar.update(1)
+                        continue
+
+                    n_tried_injections += 1
+                    repo = g.get_repo(repo_name)
+                    pull = repo.get_pull(pr_number)
+                    process_pull(repo, pull, dataset, repos_dir, cache)
+                    pbar.update(1)
+                    last_addition = dataset.entries[-1]
+                    last_metadata = last_addition.metadata
+                    if (
+                        last_metadata.repo == repo_name
+                        and last_metadata.pr_number == pr_number
+                        and last_metadata.successful
+                    ):
+                        n_successfull_injections += 1
+
+
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description='Creates the triplets for the CRAB dataset.')
     parser.add_argument(
@@ -287,6 +336,11 @@ if __name__ == "__main__":
         type=str,
         help="If this argument is not provided, all the repos in the '--repos' csv will be processed. If instead you want to run the script on a single repo (for testing purposes mainly) provide a string of form 'XXX/YYY' to this argument, where XXX is the owner of the repo and YYY is the name of the repo",
     )
+    parser.add_argument(
+        "--only-inject-jacoco",
+        action="store_true",
+        help="You must provide a cache with --cache. It will take that cache and go through all the entries that failed because they couldn't execute jacoco and process them again, trying to inject jacoco manually",
+    )
 
     args = parser.parse_args()
     g = Github(os.environ["GITHUB_AUTH_TOKEN_CRAB"])
@@ -306,7 +360,9 @@ if __name__ == "__main__":
 
     dataset = Dataset()
     try:
-        # try and finally to save, regardless of an error occuring or the program finished correctly
+        if args.only_inject_jacoco:
-        process_repos(df, dataset, args.repos, cache)
+            only_inject_jacoco(dataset, args.repos, cache)
+        else:
+            process_repos(df, dataset, args.repos, cache)
     finally:
         dataset.to_json(args.output)