fix up test_filter should now runnable from same directory as maven

Author: mspruc

python/src/pywy/tests/test_filter.py

@@ -15,21 +15,50 @@
 #  limitations under the License.
 #
 
+from pathlib import Path
+import subprocess
+import time
 from pywy.dataquanta import WayangContext
 from pywy.platforms.java import JavaPlugin
 from pywy.platforms.spark import SparkPlugin
+from pywy.tests import resources as resources_folder
+
+from importlib import resources
+from pathlib import Path
 
 def test_filter_to_json():
-    ctx = WayangContext() \
-        .register({JavaPlugin, SparkPlugin})
-    left = ctx.textfile("file:///var/www/html/README.md") \
-        .filter(lambda w: "Apache" in w, str) \
-        .flatmap(lambda w: w.split(), str, str) \
-        .map(lambda w: (len(w), w), str, (int, str))
-    right = ctx.textfile("file:///var/www/html/README.md") \
-        .filter(lambda w: "Wayang" in w, str) \
-        .map(lambda w: (len(w), w), str, (int, str))
-    join = left.join(lambda w: w[0], right, lambda w: w[0], (int, str)) \
-        .store_textfile("file:///var/www/html/data/wordcount-out-python.txt")
-    
-    assert join is not None, "Could not construct join."
+    wayang_runner_dir = Path.cwd() / 'wayang-assembly' / 'target' / 'wayang-1.1.0' / 'bin'
+
+    print("Opening subprocess")
+    with resources.path(resources_folder, "sample_data.txt") as resource_path, \
+         resources.path(resources_folder, "wordcount_out_python.txt") as output_path, \
+         resources.path(resources_folder, "wayang.properties") as configuration_file_path:
+        print(f"Using resource path: {resource_path}")
+        print(f"Using output path: {output_path}")
+        print(f"Using configuration path: {configuration_file_path}")
+        proc = subprocess.Popen([
+            f"{wayang_runner_dir}/wayang-submit",
+            f"-Dwayang.configuration=file://{configuration_file_path}",
+            f"org.apache.wayang.api.json.Main", 
+            f"8080"], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+        time.sleep(5)
+        try:
+            print(f"Running process: {proc.pid} with args: {proc.args}")
+            ctx = WayangContext() \
+                .register({JavaPlugin, SparkPlugin})
+            left = ctx.textfile(f"file://{resource_path}") \
+                .filter(lambda w: "Apache" in w, str) \
+                .flatmap(lambda w: w.split(), str, str) \
+                .map(lambda w: (str(len(w)), w), str, (int, str))
+            right = ctx.textfile(f"file://{resource_path}") \
+                .filter(lambda w: "Wayang" in w, str) \
+                .map(lambda w: (str(len(w)), w), str, (int, str))
+            join = left.join(lambda w: w[0], right, lambda w: w[0], (int, str)) \
+                .store_textfile(f"file://{output_path}")
+            time.sleep(3)
+
+            for _ in range(1):
+                print(proc.stdout.readline())
+            assert join is not None, "Could not construct join."
+        finally:    
+            proc.kill()