tests: adds WSP functional test for local package

* Adds WSP functional test for local package path. * Refactored existing WSP functional test (add setup-teardown fixtures). * Refactored `utils.ftp_list_files` to `utils.list_files` for reusability. * Fixed WSP local package crawling mechanism. Closes inspirehep#106 Signed-off-by: Spiros Delviniotis <[email protected]>
spirosdelviniotis · May 9, 2017 · 76c79cc · 76c79cc
1 parent e74f9bb
commit 76c79cc
Show file tree

Hide file tree

Showing 4 changed files with 87 additions and 20 deletions.
diff --git a/docker-compose.test.yml b/docker-compose.test.yml
@@ -23,6 +23,7 @@ services:
       - ${DOCKER_DATA}/tmp/hepcrawl_venv:/hepcrawl_venv/
       - ${PWD}:/code/
       - ${PWD}/tests/functional/scrapyd_coverage_runner.conf:/etc/scrapyd/scrapyd.conf
+      - /tmp/WSP:/tmp/WSP  # the tmp folder that keeps the history of the crawled records
     links:
       - rabbitmq
       - celery

diff --git a/hepcrawl/spiders/wsp_spider.py b/hepcrawl/spiders/wsp_spider.py
@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of hepcrawl.
-# Copyright (C) 2015, 2016 CERN.
+# Copyright (C) 2015, 2016, 2017 CERN.
 #
 # hepcrawl is a free software; you can redistribute it and/or modify it
 # under the terms of the Revised BSD License; see LICENSE file for
@@ -23,6 +23,7 @@
 from ..utils import (
     ftp_list_files,
     ftp_connection_info,
+    local_list_files,
     get_license,
     unzip_xml_files,
 )
@@ -88,7 +89,13 @@ def __init__(self, package_path=None, ftp_folder="WSP", ftp_host=None, ftp_netrc
     def start_requests(self):
         """List selected folder on remote FTP and yield new zip files."""
         if self.package_path:
-            yield Request(self.package_path, callback=self.handle_package_file)
+            dummy, new_files = local_list_files(
+                self.package_path,
+                self.target_folder
+            )
+
+            for _file in new_files:
+                yield Request("file://{0}".format(_file), callback=self.handle_package_file)
         else:
             ftp_host, ftp_params = ftp_connection_info(self.ftp_host, self.ftp_netrc)
 
@@ -116,7 +123,7 @@ def start_requests(self):
 
     def handle_package_ftp(self, response):
         """Handle a zip package and yield every XML found."""
-        self.log("Visited %s" % response.url)
+        self.log("Visited url %s" % response.url)
         zip_filepath = response.body
         zip_target_folder, dummy = os.path.splitext(zip_filepath)
         xml_files = unzip_xml_files(zip_filepath, zip_target_folder)
@@ -128,6 +135,7 @@ def handle_package_ftp(self, response):
 
     def handle_package_file(self, response):
         """Handle a local zip package and yield every XML."""
+        self.log("Visited file %s" % response.url)
         zip_filepath = urlparse.urlsplit(response.url).path
         zip_target_folder, dummy = os.path.splitext(zip_filepath)
         xml_files = unzip_xml_files(zip_filepath, zip_target_folder)

diff --git a/hepcrawl/utils.py b/hepcrawl/utils.py
@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of hepcrawl.
-# Copyright (C) 2015, 2016 CERN.
+# Copyright (C) 2015, 2016, 2017 CERN.
 #
 # hepcrawl is a free software; you can redistribute it and/or modify it
 # under the terms of the Revised BSD License; see LICENSE file for
@@ -67,14 +67,25 @@ def ftp_list_files(server_folder, target_folder, server, user, password, passive
 
     with ftputil.FTPHost(server, user, password, session_factory=session_factory) as host:
         files = host.listdir(host.curdir + '/' + server_folder)
-        missing_files = []
-        all_files = []
-        for filename in files:
-            destination_file = os.path.join(target_folder, filename)
-            source_file = os.path.join(server_folder, filename)
-            if not os.path.exists(destination_file):
-                missing_files.append(source_file)
-            all_files.append(source_file)
+        return list_files(server_folder, target_folder, files)
+
+
+def local_list_files(local_folder, target_folder):
+    """List files from given package folder to target folder."""
+    files = os.listdir(local_folder)
+    return list_files(local_folder, target_folder, files)
+
+
+def list_files(remote_folder, target_folder, files):
+    missing_files = []
+    all_files = []
+    for filename in files:
+        destination_file = os.path.join(target_folder, filename)
+        source_file = os.path.join(remote_folder, filename)
+        if not os.path.exists(destination_file):
+            missing_files.append(source_file)
+        all_files.append(source_file)
+
     return all_files, missing_files
 
 

diff --git a/tests/functional/WSP/test_wsp.py b/tests/functional/WSP/test_wsp.py
@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 #
 # This file is part of hepcrawl.
-# Copyright (C) 2015, 2016, 2017 CERN.
+# Copyright (C) 2017 CERN.
 #
 # hepcrawl is a free software; you can redistribute it and/or modify it
 # under the terms of the Revised BSD License; see LICENSE file for
@@ -14,6 +14,7 @@
 import pytest
 import json
 import os
+import shutil
 
 from scrapyd_api import ScrapydAPI
 from time import sleep
@@ -50,14 +51,14 @@ def expected_results():
     return expected_data
 
 
-@pytest.fixture(scope="module")
-def set_up_environment():
+@pytest.fixture(scope="function")
+def set_up_ftp_environment():
     netrc_location = os.path.join(os.path.dirname(
         os.path.realpath(__file__)),
         'fixtures/ftp_server/.netrc'
     )
 
-    return {
+    yield {
         'CRAWLER_HOST_URL': 'http://scrapyd:6800',
         'CRAWLER_PROJECT': 'hepcrawl',
         'CRAWLER_ARGUMENTS': {
@@ -66,9 +67,55 @@ def set_up_environment():
         }
     }
 
+    cleaner()
+
+
+@pytest.fixture(scope="function")
+def set_up_local_environment():
+    package_location = os.path.join(os.path.dirname(
+        os.path.realpath(__file__)),
+        'fixtures/ftp_server/WSP/'
+    )
+
+    yield {
+        'CRAWLER_HOST_URL': 'http://scrapyd:6800',
+        'CRAWLER_PROJECT': 'hepcrawl',
+        'CRAWLER_ARGUMENTS': {
+            'package_path': package_location,
+        }
+    }
+
+    cleaner()
+    cleaner(package_location + 'IDAQPv20i01-03160015-1510863')
+
+
+def cleaner(path='/tmp/WSP/'):
+    shutil.rmtree(path, ignore_errors=True)
+
+
+def test_wsp_ftp(set_up_ftp_environment, expected_results):
+    crawler = get_crawler_instance(set_up_ftp_environment.get('CRAWLER_HOST_URL'))
+
+    # The test must wait until the docker environment is up (takes about 5 seconds).
+    sleep(5)
+
+    results = CeleryMonitor.do_crawl(
+        app=app,
+        monitor_timeout=5,
+        monitor_iter_limit=100,
+        crawler_instance=crawler,
+        project=set_up_ftp_environment.get('CRAWLER_PROJECT'),
+        spider='WSP',
+        settings={},
+        **set_up_ftp_environment.get('CRAWLER_ARGUMENTS')
+    )
+
+    assert [override_generated_fields(result) for result in results] == \
+           [override_generated_fields(expected) for expected in expected_results]
+
 
-def test_wsp_normal_set_of_records(set_up_environment, expected_results):
-    crawler = get_crawler_instance(set_up_environment.get('CRAWLER_HOST_URL'))
+def test_wsp_local_package_path(set_up_local_environment, expected_results):
+    crawler = get_crawler_instance(set_up_local_environment.get('CRAWLER_HOST_URL'))
 
     # The test must wait until the docker environment is up (takes about 5 seconds).
     sleep(5)
@@ -78,10 +125,10 @@ def test_wsp_normal_set_of_records(set_up_environment, expected_results):
         monitor_timeout=5,
         monitor_iter_limit=100,
         crawler_instance=crawler,
-        project=set_up_environment.get('CRAWLER_PROJECT'),
+        project=set_up_local_environment.get('CRAWLER_PROJECT'),
         spider='WSP',
         settings={},
-        **set_up_environment.get('CRAWLER_ARGUMENTS')
+        **set_up_local_environment.get('CRAWLER_ARGUMENTS')
     )
 
     assert [override_generated_fields(result) for result in results] == \