From d7d58dda4afe105c50879fa179652c4b73d7aa98 Mon Sep 17 00:00:00 2001
From: b97pla <pontus.larsson@medsci.uu.se>
Date: Wed, 11 Oct 2023 17:56:34 +0200
Subject: [PATCH] fix tar commands

---
 archive_upload/handlers/dsmc_handlers.py | 113 ++++++++++++++---------
 1 file changed, 69 insertions(+), 44 deletions(-)

diff --git a/archive_upload/handlers/dsmc_handlers.py b/archive_upload/handlers/dsmc_handlers.py
index a1a2725..0cdace9 100644
--- a/archive_upload/handlers/dsmc_handlers.py
+++ b/archive_upload/handlers/dsmc_handlers.py
@@ -39,7 +39,7 @@ def initialize(self, config, runner_service):
         :param config: configuration used by the service
         :param runner_service: runner service to use. Must fulfill `archive_upload.lib.jobrunner.JobRunnerAdapter` interface
         """
-        self.config = config
+        self.config = config.get_app_config()
         self.runner_service = runner_service
 
     @staticmethod
@@ -402,7 +402,7 @@ def post(self, runfolder_archive):
 
         path_to_archive = os.path.join(monitored_dir, runfolder_archive)
         dsmc_log_root_dir = self.config["log_directory"]
-        dsmc_extra_args = self.config.get("dsmc_extra_args", "")
+        dsmc_extra_args = self.config.get("dsmc_extra_args", {})
 
         if not self._is_valid_log_dir(dsmc_log_root_dir):
             msg = "Error when validating log dir. {} is not a directory.".format(dsmc_log_root_dir)
@@ -491,7 +491,7 @@ def post(self, runfolder_archive):
 
         path_to_archive = os.path.join(monitored_dir, runfolder_archive)
         dsmc_log_root_dir = self.config["log_directory"]
-        dsmc_extra_args = self.config.get("dsmc_extra_args", "")
+        dsmc_extra_args = self.config.get("dsmc_extra_args", {})
         uniq_id = str(uuid.uuid4())
 
         if not self._is_valid_log_dir(dsmc_log_root_dir):
@@ -781,6 +781,58 @@ class CompressArchiveHandler(BaseDsmcHandler):
     Handler for compressing certain files in the archive before uploading.
     """
 
+    @staticmethod
+    def _create_tarball_cmd(tarball_name, path_to_archive, exclude_from_tarball):
+        exclude_patterns = " ".join(
+            [
+                "--exclude={}".format(p)
+                for p in exclude_from_tarball + [tarball_name]
+            ]
+        )
+        return "cd {} && " \
+               "touch {} && " \
+               "tar " \
+               "--create " \
+               "--gzip " \
+               "--dereference " \
+               "--hard-dereference " \
+               "--file={} " \
+               "{} " \
+               ".".format(
+            path_to_archive,
+            tarball_name,
+            tarball_name,
+            exclude_patterns
+        )
+
+    @staticmethod
+    def _remove_tarballed_files_cmd(path_to_archive, tarball_name):
+        return "cd {} && " \
+               "tar " \
+               "--list " \
+               "--file={} |" \
+               "xargs " \
+               "-n1 " \
+               "rm -f".format(
+            path_to_archive,
+            tarball_name
+        )
+
+    @staticmethod
+    def _remove_empty_dirs_cmd(path_to_archive):
+        return "cd {} && " \
+               "find " \
+               ". " \
+               "-depth " \
+               "-mindepth 1 " \
+               "-type d |" \
+               "xargs " \
+               "-n1 " \
+               "rmdir " \
+               "--ignore-fail-on-non-empty".format(
+            path_to_archive
+        )
+
     def post(self, archive):
         """
         Create a gziped tarball of most files in the archive, with the exception of
@@ -799,7 +851,8 @@ def post(self, archive):
                 archive, path_to_archive_root)
             raise ArchiveException(reason=msg, status_code=400)
 
-        tarball_path = "{}.tar.gz".format(os.path.join(path_to_archive, archive))
+        tarball_name = "{}.tar.gz".format(archive)
+        tarball_path = os.path.join(path_to_archive, tarball_name)
 
         log.debug("Checking to see if {} exists".format(tarball_path))
 
@@ -807,48 +860,20 @@ def post(self, archive):
             msg = "Error when creating archive tarball. {} already exists.".format(tarball_path)
             raise ArchiveException(reason=msg, status_code=400)
 
-        exclude_from_tarball = self.config.get("exclude_from_tarball", [])
-        exclude_patterns = " ".join(
-            [
-                "--exclude={}".format(p)
-                for p in exclude_from_tarball
-            ]
+        exclude_from_tarball = self.config["exclude_from_tarball"]
+        cmd = " ( {} ) && ( {} ) ; ( {} )".format(
+            self._create_tarball_cmd(
+                tarball_name,
+                path_to_archive,
+                exclude_from_tarball),
+            self._remove_tarballed_files_cmd(
+                path_to_archive,
+                tarball_name),
+            self._remove_empty_dirs_cmd(
+                path_to_archive)
         )
 
-        cmd = "cd {} && " \
-              "tar " \
-              "--create " \
-              "--gzip " \
-              "--dereference " \
-              "--hard-dereference " \
-              "--verify " \
-              "--file={} " \
-              "{} " \
-              ".".format(
-            path_to_archive,
-            tarball_path,
-            exclude_patterns
-        )
-        cmd = "{} && " \
-              "tar " \
-              "--list " \
-              "--file={} |" \
-              "xargs " \
-              "-n1 " \
-              "rm".format(
-            cmd,
-            tarball_path
-        )
-        cmd = "{} && " \
-              "find . " \
-              "-depth " \
-              "-type d |" \
-              "xargs " \
-              "-n1 " \
-              "rmdir".format(
-            cmd,
-            path_to_archive
-        )
+        log.info("run command: {}".format(cmd))
         log.info(
             "Creating tarball {}, then removing files from {} that were added to tarball".format(
                 tarball_path,