Merge branch 'develop' into 'master'

Develop

See merge request !2

.gitlab-ci.yml

@@ -25,7 +25,7 @@ build:
   image: docker:18.09.7
   only:
     - master
-    - Init
+    - develop
   services:
     - 'docker:18.09.7-dind'
   script:

src/archive_extractor_worker.py

@@ -236,7 +236,7 @@ class ArchiveExtractor(nlp_ws.NLPWorker):
     def _safe_report(self, input_size_report: Dict, input_size_sum: int,
                      output_size_report: Dict, output_size_sum: int,
                      error_report_before: Dict, error_report_after: Dict,
-                     files_to_convert: List, dir_path: str):
+                     files_to_convert: List, output_path: str, dir_path: str):
         """Save size and error reports to json file.
 
         :param input_size_report: input size report
@@ -253,6 +253,8 @@ class ArchiveExtractor(nlp_ws.NLPWorker):
         :type error_report_after: Dict
         :param files_to_convert: converted files
         :type files_to_convert: List
+        :param output_path: Path to output location.
+        :type output_path: str
         :param dir_path: Path to directory where the
             worker will store result files.
         :type dir_path: str
@@ -267,10 +269,28 @@ class ArchiveExtractor(nlp_ws.NLPWorker):
             self.INPUT_SIZES_SUM: input_size_sum,
             self.OUTPUT_SIZES: output_size_report,
             self.OUTPUT_SIZES_SUM: output_size_sum}
-        filename = os.path.join(dir_path, 'report.json')
+        filename = os.path.join(output_path, 'report.json')
         with open(filename, 'w') as outfile:
             json.dump(data, outfile)
 
+    def _pack_result(self, output_path: str, dir_path: str):
+        """Pack cleaned files to zip, removes unpacked files.
+
+        :param output_path: Path to output location.
+        :type output_path: str
+        :param dir_path: Path to directory where the
+            worker store result files.
+        :type dir_path: str
+        """
+        zip_name = os.path.join(output_path, 'result.zip')
+        _log.info("Packing results to zip...")
+        with ZipFile(zip_name, 'w') as zip_obj:
+            for file_path, filename in self._file_generator(dir_path):
+                zip_obj.write(file_path, filename)
+
+        _log.info("Packing completed, removing unpacked files")
+        shutil.rmtree(dir_path)
+
     def process(self, input_path, task_options, output_path):
         """Called for each request made to the worker.
 
@@ -291,7 +311,8 @@ class ArchiveExtractor(nlp_ws.NLPWorker):
         if not is_zipfile(input_path):
             raise Exception("Input is not a zip archive")
 
-        dir_path = output_path
+        dir_path = os.path.join(output_path, 'result')
+        os.mkdir(dir_path)
         self._unpack_zip(input_path, dir_path)
         files_to_remove, files_to_convert, input_size_sum = [], [], 0
         input_size_report, error_report_before = {}, {member.value: []
@@ -345,6 +366,9 @@ class ArchiveExtractor(nlp_ws.NLPWorker):
                 f" {output_size_sum} is too large,"
                 f" max size is {self.output_max_size}")
 
+        self._pack_result(output_path, dir_path)
+
         self._safe_report(input_size_report, input_size_sum, output_size_report,
                           output_size_sum, error_report_before,
-                          error_report_after, files_to_convert, dir_path)
+                          error_report_after, files_to_convert, output_path,
+                          dir_path)