Merge pull request #13 from microsoft/laserprec/bugfix/trailinglash

Bugfix img not generated due to missing trailling '/' in dist folder

genalog/pipeline.py

@@ -61,7 +61,7 @@ class AnalogDocumentGeneration(object):
 
         generator = self.doc_generator.create_generator(content, [template])
         # Generate the image
-        doc = next(generator)
+        doc = next(generator)  # TODO: this does not exhaust all of the style combinations in the generator
         src = doc.render_array(resolution=self.resolution, channel="GRAYSCALE")
         # Degrade the image
         dst = self.degrader.apply_effects(src)
@@ -73,7 +73,7 @@ class AnalogDocumentGeneration(object):
             # save it onto disk
             text_filename = os.path.basename(full_text_path)
             img_filename = text_filename.replace(".txt", ".png")
-            img_dst_path = target_folder + "img/" + img_filename
+            img_dst_path = os.path.join(target_folder, "img", img_filename)
             cv2.imwrite(img_dst_path, dst)
             return
 

tests/e2e/test_document_generation.py

@@ -18,7 +18,7 @@ MULTI_PAGE_TEMPLATE_NAME = "multipage.html.jinja"
 UNDEFINED_TEMPLATE_NAME = "not a valid template"
 
 TEST_OUTPUT_DIR = "test_out"
-FILE_DESTINATION = TEST_OUTPUT_DIR + "/save.png"
+FILE_DESTINATION = os.path.join(TEST_OUTPUT_DIR, "save.png")
 
 CUSTOM_STYLE = {
     "font_family": ["Calibri", "Times"],

tests/e2e/test_pipeline.py

@@ -1,10 +1,20 @@
+import os
 import glob
 
 import pytest
 
 from genalog import pipeline
+from genalog.generation.document import DocumentGenerator
 
 EXAMPLE_TEXT_FILE = "tests/unit/text/data/gt_1.txt"
+INPUT_TEXT_FILENAMES = glob.glob("tests/unit/text/data/gt_*.txt")
+
+STYLES = {"font_size": ["5px"]}
+STYLES_COMBINATION = {"font_size": ["5px", "6px"]}  # Multiple values per style are not supported right now
+DEGRATIONS = [
+    ("blur", {"radius": 3}),
+    ("morphology", {"operation": "close"})
+]
 
 
 @pytest.fixture
@@ -14,14 +24,13 @@ def default_analog_generator():
 
 @pytest.fixture
 def custom_analog_generator():
-    custom_styles = {"font_size": ["5px"]}
-    custom_degradation = [("blur", {"radius": 3})]
     return pipeline.AnalogDocumentGeneration(
-        styles=custom_styles, degradations=custom_degradation, resolution=300
+        styles=STYLES, degradations=DEGRATIONS, resolution=300
     )
 
 
 def test_default_generate_img(default_analog_generator):
+    assert len(default_analog_generator.list_templates()) > 0
     example_template = default_analog_generator.list_templates()[0]
     default_analog_generator.generate_img(
         EXAMPLE_TEXT_FILE, example_template, target_folder=None
@@ -29,15 +38,27 @@ def test_default_generate_img(default_analog_generator):
 
 
 def test_custom_generate_img(custom_analog_generator):
+    assert len(custom_analog_generator.list_templates()) > 0
     example_template = custom_analog_generator.list_templates()[0]
     custom_analog_generator.generate_img(
         EXAMPLE_TEXT_FILE, example_template, target_folder=None
     )
 
 
-def test_generate_dataset_multiprocess():
-    INPUT_TEXT_FILENAMES = glob.glob("tests/unit/text/data/gt_*.txt")
-    with pytest.deprecated_call():
-        pipeline.generate_dataset_multiprocess(
-            INPUT_TEXT_FILENAMES, "test_out", {}, [], "text_block.html.jinja"
+@pytest.mark.parametrize("styles", [
+    STYLES,
+    pytest.param(
+        STYLES_COMBINATION, marks=pytest.mark.xfail(
+            reason="Style combinations are not supported. Only one value per style", strict=True)
     )
+])
+@pytest.mark.parametrize("folder_name", ["result", "result/"])
+def test_generate_dataset_multiprocess(tmpdir, folder_name, styles):
+    assert len(INPUT_TEXT_FILENAMES) > 0
+    output_folder = os.path.join(tmpdir, folder_name)
+    pipeline.generate_dataset_multiprocess(
+        INPUT_TEXT_FILENAMES, output_folder, styles, DEGRATIONS, "text_block.html.jinja"
+    )
+    num_generated_img = glob.glob(os.path.join(output_folder, "**/*.png"))
+    assert len(num_generated_img) > 0
+    assert len(num_generated_img) == len(INPUT_TEXT_FILENAMES) * len(DocumentGenerator.expand_style_combinations(styles))