enable bashlib test in test_modules_process_helpers

tests/cli/test_bashlib.py

@@ -1,6 +1,8 @@
+from contextlib import contextmanager
 from tests.base import CapturingTestCase as TestCase, main, assets, copy_of_directory
 
 import os, sys
+from os import environ
 import traceback
 import subprocess
 import tempfile
@@ -110,34 +112,38 @@ def test_bashlib_minversion(self):
         assert "ERROR: ocrd/core is too old" in err
 
     def test_bashlib_cp_processor(self):
-        script = (Path(__file__).parent.parent / 'data/bashlib_cp_processor.sh').read_text()
-        ocrd_tool = json.loads((Path(__file__).parent.parent / 'data/bashlib_cp_processor.ocrd-tool.json').read_text())
-        with copy_of_directory(assets.path_to('kant_aufklaerung_1784/data')) as wsdir:
-            with pushd_popd(wsdir):
-                with open('ocrd-tool.json', 'w') as f:
-                    f.write(json.dumps(ocrd_tool))
-                # run on 1 input
-                exit_code, out, err = self.invoke_bash(
-                    script, '-I', 'OCR-D-GT-PAGE', '-O', 'OCR-D-GT-PAGE2', '-P', 'message', 'hello world',
-                    executable='ocrd-cp')
-                print({'exit_code': exit_code, 'out': out, 'err': err})
-                assert 'single input fileGrp' in err
-                assert 'processing PAGE-XML' in err
-                assert exit_code == 0
-                assert 'hello world' in out
-                path = pathlib.Path('OCR-D-GT-PAGE2')
-                assert path.is_dir()
-                assert next(path.glob('*.xml'), None)
-                # run on 2 inputs
-                exit_code, out, err = self.invoke_bash(
-                    script, '-I', 'OCR-D-IMG,OCR-D-GT-PAGE', '-O', 'OCR-D-IMG2',
-                    executable='ocrd-cp')
-                assert 'multiple input fileGrps' in err
-                assert exit_code == 0
-                assert 'ignoring application/vnd.prima.page+xml' in err
-                path = pathlib.Path('OCR-D-IMG2')
-                assert path.is_dir()
-                assert next(path.glob('*.tif'), None)
+        # script = (Path(__file__).parent.parent / 'data/bashlib_cp_processor.sh').read_text()
+        # ocrd_tool = json.loads((Path(__file__).parent.parent / 'data/bashlib_cp_processor.ocrd-tool.json').read_text())
+        scriptdir = Path(__file__).parent.parent / 'data'
+
+        with copy_of_directory(assets.path_to('kant_aufklaerung_1784/data')) as wsdir, pushd_popd(wsdir):
+            with open(f'{scriptdir}/ocrd-cp', 'r', encoding='utf-8') as script_f:
+                script = script_f.read()
+            with open(f'{scriptdir}/ocrd-cp.ocrd-tool.json', 'r', encoding='utf-8') as tool_in, \
+                open(f'{wsdir}/ocrd-tool.json', 'w', encoding='utf-8') as tool_out:
+                tool_out.write(tool_in.read())
+            # run on 1 input
+            exit_code, out, err = self.invoke_bash(
+                script, '-I', 'OCR-D-GT-PAGE', '-O', 'OCR-D-GT-PAGE2', '-P', 'message', 'hello world',
+                executable='ocrd-cp')
+            print({'exit_code': exit_code, 'out': out, 'err': err})
+            assert 'single input fileGrp' in err
+            assert 'processing PAGE-XML' in err
+            assert exit_code == 0
+            assert 'hello world' in out
+            path = pathlib.Path('OCR-D-GT-PAGE2')
+            assert path.is_dir()
+            assert next(path.glob('*.xml'), None)
+            # run on 2 inputs
+            exit_code, out, err = self.invoke_bash(
+                script, '-I', 'OCR-D-IMG,OCR-D-GT-PAGE', '-O', 'OCR-D-IMG2',
+                executable='ocrd-cp')
+            assert 'multiple input fileGrps' in err
+            assert exit_code == 0
+            assert 'ignoring application/vnd.prima.page+xml' in err
+            path = pathlib.Path('OCR-D-IMG2')
+            assert path.is_dir()
+            assert next(path.glob('*.tif'), None)
 
 if __name__ == "__main__":
     main(__file__)

tests/model/test_ocrd_mets.py

@@ -1,5 +1,4 @@
 # -*- coding: utf-8 -*-
-
 from datetime import datetime
 
 from os.path import join

tests/network/test_modules_process_helpers.py

@@ -1,38 +1,54 @@
+from contextlib import contextmanager
+from os import environ
 from pathlib import Path
-from src.ocrd.processor.builtin.dummy_processor import DummyProcessor
-from src.ocrd_network.constants import NetworkLoggingDirs
-from src.ocrd_network.logging_utils import get_root_logging_dir
-from src.ocrd_network.process_helpers import invoke_processor
-from src.ocrd_network.utils import generate_id
+
+from ocrd.processor.builtin.dummy_processor import DummyProcessor
+from ocrd_network.constants import NetworkLoggingDirs
+from ocrd_network.logging_utils import get_root_logging_dir
+from ocrd_network.process_helpers import invoke_processor
+from ocrd_network.utils import generate_id
+
 from tests.base import assets
 
+@contextmanager
+def temp_env_var(k, v):
+    v_before = environ.get(k, None)
+    environ[k] = v
+    yield
+    if v_before is not None:
+        environ[k] = v_before
+    else:
+        del environ[k]
 
-# TODO: Fix this
-def _test_invoke_processor_bash():
-    bash_lib_cp_processor = "tests/data/bashlib_cp_processor.sh"
-    assert Path(bash_lib_cp_processor).exists(), f"Bash lib test processor not found in: {bash_lib_cp_processor}"
 
-    workspace_root = "kant_aufklaerung_1784/data"
-    path_to_mets = assets.path_to(f"{workspace_root}/mets.xml")
-    assert Path(path_to_mets).exists()
-    log_dir_root = get_root_logging_dir(module_name=NetworkLoggingDirs.PROCESSING_JOBS)
-    job_id = generate_id()
-    path_to_log_file = Path(log_dir_root, job_id)
-    input_file_grp = "OCR-D-IMG"
-    output_file_grp = f"OCR-D-BASH-TEST-{job_id}"
-    invoke_processor(
-        processor_class=None,  # required only for pythonic processors
-        executable=bash_lib_cp_processor,
-        abs_path_to_mets=path_to_mets,
-        input_file_grps=[input_file_grp],
-        output_file_grps=[output_file_grp],
-        page_id="PHYS_0017,PHYS_0020",
-        parameters={},
-        log_filename=path_to_log_file,
-        log_level="DEBUG"
-    )
-    assert Path(assets.path_to(f"{workspace_root}/{output_file_grp}")).exists()
-    assert Path(path_to_log_file).exists()
+def test_invoke_processor_bash():
+    scriptdir = Path(__file__).parent.parent / 'data'
+    with temp_env_var('PATH', f'{scriptdir}:{environ["PATH"]}'):
+        workspace_root = "kant_aufklaerung_1784/data"
+        path_to_mets = assets.path_to(f"{workspace_root}/mets.xml")
+        assert Path(path_to_mets).exists()
+        log_dir_root = get_root_logging_dir(module_name=NetworkLoggingDirs.PROCESSING_JOBS)
+        job_id = generate_id()
+        path_to_log_file = Path(log_dir_root, job_id)
+        input_file_grp = "OCR-D-IMG"
+        output_file_grp = f"OCR-D-BASH-TEST-{job_id}"
+        try:
+            invoke_processor(
+                processor_class=None,  # required only for pythonic processors
+                executable='ocrd-cp',
+                abs_path_to_mets=path_to_mets,
+                input_file_grps=[input_file_grp],
+                output_file_grps=[output_file_grp],
+                page_id="PHYS_0017,PHYS_0020",
+                parameters={},
+                log_filename=path_to_log_file,
+                log_level="DEBUG"
+            )
+        except:
+            with open(path_to_log_file, 'r', encoding='utf-8') as f:
+                print(f.read())
+        assert Path(assets.path_to(f"{workspace_root}/{output_file_grp}")).exists()
+        assert Path(path_to_log_file).exists()
 
 
 def test_invoke_processor_pythonic():