New upstream version 1.1~beta1upstream/1.1_beta1debian-upstream

3 files changed, 238 insertions, 121 deletions

diff --git a/tests/helpers.py b/tests/helpers.py
new file mode 100644
index 0000000..df474b0
--- /dev/null
+++ b/tests/helpers.py
@@ -0,0 +1,51 @@
+# SPDX-License-Identifier: CC0-1.0
+
+# Copyright (C) 2022 Wojtek Kosior <koszko@koszko.org>
+#
+# Available under the terms of Creative Commons Zero v1.0 Universal.
+
+import re
+
+variable_word_re = re.compile(r'^<(.+)>$')
+
+def process_command(command, expected_command):
+    """Validate the command line and extract its variable parts (if any)."""
+    assert len(command) == len(expected_command)
+
+    extracted = {}
+    for word, expected_word in zip(command, expected_command):
+        match = variable_word_re.match(expected_word)
+        if match:
+            extracted[match.group(1)] = word
+        else:
+            assert word == expected_word
+
+    return extracted
+
+def run_missing_executable(command, **kwargs):
+    """
+    Instead of running a command, raise FileNotFoundError as if its executable
+    was missing.
+    """
+    raise FileNotFoundError('dummy')
+
+class MockedCompletedProcess:
+    """
+    Object with some fields similar to those of subprocess.CompletedProcess.
+    """
+    def __init__(self, args, returncode=0,
+                 stdout='some output', stderr='some error output',
+                 text_output=True):
+        """
+        Initialize MockedCompletedProcess. Convert strings to bytes if needed.
+        """
+        self.args       = args
+        self.returncode = returncode
+
+        if type(stdout) is str and not text_output:
+            stdout = stdout.encode()
+        if type(stderr) is str and not text_output:
+            stderr = stderr.encode()
+
+        self.stdout = stdout
+        self.stderr = stderr
diff --git a/tests/source-package-example/index.json b/tests/source-package-example/index.json
index 7162dd7..9aa6e70 100644
--- a/tests/source-package-example/index.json
+++ b/tests/source-package-example/index.json
@@ -3,9 +3,9 @@
 // Copyright (C) 2021, 2022 Wojtek Kosior <koszko@koszko.org>
 // Available under the terms of Creative Commons Zero v1.0 Universal.
 
-// This is an example index.json file describing Hydrilla site content. As you
-// can see, for storing site content information Hydrilla utilizes JSON with an
-// additional extension in the form of '//' comments support.
+// This is an example index.json file describing Hydrilla packages. As you can
+// see, for storing this information Hydrilla utilizes JSON with an additional
+// extension in the form of '//' comments support.
 
 // An index.json file conveys definitions of site resources and pattern->payload
 // mappings. The definitions may reference files under index.json's containing
@@ -62,18 +62,15 @@
     // will also belong here once they get implemented.
     "definitions": [
 	{
-	    // Value of "type" can currently be one of: "resource" and
-	    // "mapping". The one we have here, "resource", defines a list
-	    // of injectable scripts that can be used as a payload or as a
-	    // dependency of another "resource". In the future CSS style sheets
-	    // and WASM modules will also be composite parts of a "resource" as
-	    // scripts are now.
-	    "type": "resource",
+	    // Value of "type" can currently be one of: "resource", "mapping"
+	    // and "mapping_and_resource" for a combined definition. The one we
+	    // have here, "mapping", associates resources with pages on which
+	    // they are to be used.
+	    "type": "mapping",
 
-	    // Used when referring to this resource in "dependencies" list of
-	    // another resource or in "payload" field of a mapping. Should
-	    // be consize and can only use a restricted set of characters. It
-	    // has to match: [-0-9a-z]+
+	    // Used when referring to this mapping in "required_mappings" list
+	    // of another item. Should be consize and can only use a restricted
+	    // set of characters. It has to match: [-0-9a-z]+
 	    "identifier": "helloapple",
 
 	    // "long_name" should be used to specify a user-friendly alternative
@@ -96,26 +93,80 @@
 	    // Different versions (e.g. 1.0 and 1.3) of the same resource can be
 	    // defined in separate index.json files. This makes it easy to
 	    // accidently cause an identifier clash. To help detect it, we allow
-	    // each resource to have a UUID associated with it. Attempt to
-	    // define multiple resources with the same identifier and different
-	    // UUIDs will result in an error being reported. Defining multiple
-	    // resources with different identifiers and the same UUID is
+	    // each item to have a UUID associated with it. Attempt to define
+	    // multiple mapping with the same identifier and different UUIDs
+	    // will result in an error being reported. Defining multiple
+	    // mappings with different identifiers and the same UUID is
 	    // disallowed for now (it may be later permitted if we consider it
 	    // good for some use-case).
-	    // As of package source schema version 1.0, UUIDs are optional and
+	    // As of package source schema version 2.0, UUIDs are optional and
 	    // can be omitted.
+	    "uuid": "54d23bba-472e-42f5-9194-eaa24c0e3ee7",
+
+	    // Thanks to the "version" field (and "revision" field in case of
+	    // "resource" or "mapping_and_resource", clients will know they have
+	    // to update certain item after a new version has appeared in the
+	    // repository. If multiple definitions of the same version of given
+	    // item are provided to Hydrilla server, an error is generated.
+	    // "version" differs from its counterpart in resource in that it has
+	    // no accompanying revision number. For combined definitions with
+	    // "mapping_and_resource" as type, the value of "revision" is
+	    // appended as the last component of the resulting mapping's
+	    // version. If type is simply "mapping", revision number is ignored.
+	    "version": [2021, 11, 10],
+
+	    // A short, meaningful description of what the mapping does.
+	    "description": "causes apple to get greeted on Hydrillabugs issue tracker",
+
+	    // If needed, a "comment" field can be added to provide some
+	    // additional information.
+	    // "comment": "this resource something something",
+
+	    // The "payloads" object specifies which payloads are to be applied
+	    // to which URLs.
+	    "payloads": {
+		// Each key should be a valid Haketilo URL pattern.
+		"https://hydrillabugs.koszko.org/***": {
+		    // Should be the name of an existing resource. The resource
+		    // may, but doesn't have to, be defined in the same
+		    // index.json file.
+		    "identifier": "helloapple"
+		},
+		// More associations may follow.
+		"https://hachettebugs.koszko.org/***": {
+		    "identifier": "helloapple"
+		}
+	    }
+	}, {
+	    // A "resource" item defines a list of injectable scripts that can
+	    // be used as a payload or as a dependency of another "resource". In
+	    // the future CSS style sheets and WASM modules will also be
+	    // composite parts of a "resource" as scripts are now.
+	    "type": "resource",
+
+	    // Has similar function to mapping's identifier. Used when
+	    // referring to this resource in "dependencies" list of another
+	    // resource or in "payload" field of a mapping. Should be consize
+	    // and can only use a restricted set of characters. It has to match:
+	    // [-0-9a-z]+
+	    // It can be the same as some mapping identifier (those are
+	    // different entities and are treated separately).
+	    "identifier": "helloapple",
+
+	    // "long name" and "uuid" have the same meaning as in the case of
+	    // resources and "uuid" is also optional. UUIDs of a resource and a
+	    // mapping can technically be the same but it is recommended to
+	    // avoid even this kind of repetition.
+	    "long_name": "Hello Apple",
 	    "uuid": "a6754dcb-58d8-4b7a-a245-24fd7ad4cd68",
 
 	    // Version should match the upstream version of the resource (e.g. a
-	    // version of javascript library). Revision number starts as 1 for
+	    // version of JavaScript library). Revision number starts as 1 for
 	    // each new resource version and gets incremented by 1 each time a
-	    // modification to the packaging of this version is done. Hydrilla
-	    // will allow multiple definitions of the same resource to load, as
-	    // long as their versions differ. Thanks to the "version" and
-	    // "revision" fields, clients will know they have to update certain
-	    // resource after it has been updated. If multiple definitions of
-	    // the same version of given resource are provided, an error is
-	    // generated (even if those definitions differ by revision number).
+	    // modification to the packaging of this version is done.
+	    // If multiple definitions of the same version of given resource are
+	    // provided to Hydrilla server, an error is generated (even if those
+	    // definitions differ by revision number).
 	    "version": [2021, 11, 10],
 	    "revision": 1,
 
@@ -123,16 +174,15 @@
 	    // what it does.
 	    "description": "greets an apple",
 
-	    // If needed, a "comment" field can be added to provide some
-	    // additional information.
-	    // "comment": "this resource something something",
+	    // A comment, if necessary.
+	    // "comment": "blah blah because bleh"
 
 	    // Resource's "dependencies" array shall contain names of other
 	    // resources that (in case of scripts at least) should get evaluated
 	    // on a page before this resource's own scripts.
 	    "dependencies": [{"identifier": "hello-message"}],
 
-	    // Array of javascript files that belong to this resource.
+	    // Array of JavaScript files that belong to this resource.
 	    "scripts": [
 		{"file": "hello.js"},
 		{"file":   "bye.js"}
@@ -148,48 +198,6 @@
 	    // If "dependencies" is empty, it can also be omitted.
 	    // "dependencies": [],
 	    "scripts": [{"file": "message.js"}]
-	}, {
-	    "type": "mapping",
-
-	    // Has similar function to resource's identifier. Should be consize
-	    // and can only use a restricted set of characters. It has to match:
-	    // [-0-9a-z]+
-	    // It can be the same as some resource identifier (those are
-	    // different entities and are treated separately).
-	    "identifier": "helloapple",
-
-	    // "long name" and "uuid" have the same meaning as in the case of
-	    // resources and "uuid" is also optional. UUIDs of a resource and a
-	    // mapping can technically be the same but it is recommended to
-	    // avoid even this kind of repetition.
-	    "long_name": "Hello Apple",
-	    "uuid": "54d23bba-472e-42f5-9194-eaa24c0e3ee7",
-
-	    // "version" differs from its counterpart in resource in that it has
-	    // no accompanying revision number.
-	    "version": [2021, 11, 10],
-
-	    // A short, meaningful description of what the mapping does.
-	    "description": "causes apple to get greeted on Hydrillabugs issue tracker",
-
-	    // A comment, if necessary.
-	    // "comment": "blah blah because bleh"
-
-	    // The "payloads" object specifies which payloads are to be applied
-	    // to which URLs.
-	    "payloads": {
-		// Each key should be a valid Haketilo URL pattern.
-		"https://hydrillabugs.koszko.org/***": {
-		    // Should be the name of an existing resource. The resource
-		    // may, but doesn't have to, be defined in the same
-		    // index.json file.
-		    "identifier": "helloapple"
-		},
-		// More associations may follow.
-		"https://hachettebugs.koszko.org/***": {
-		    "identifier": "helloapple"
-		}
-	    }
 	}
     ],
     // We can also list additional files to include in the produced source
diff --git a/tests/test_server.py b/tests/test_server.py
index 0820d5c..65ae8ce 100644
--- a/tests/test_server.py
+++ b/tests/test_server.py
@@ -31,6 +31,7 @@ import pytest
 import sys
 import shutil
 import json
+import functools as ft
 
 from pathlib import Path
 from hashlib import sha256
@@ -42,9 +43,10 @@ from markupsafe import escape
 from werkzeug import Response
 
 from hydrilla import util as hydrilla_util
-from hydrilla.builder import Build
-from hydrilla.server import config, _version
-from hydrilla.server.serve import HydrillaApp
+from hydrilla.builder import build
+from hydrilla.server import config, _version, serve
+
+from .helpers import *
 
 here        = Path(__file__).resolve().parent
 config_path = here / 'config.json'
@@ -55,13 +57,41 @@ expected_generated_by = {
     'version': _version.version
 }
 
-SetupMod = Optional[Callable['Setup', None]]
+SetupMod = Optional[Callable[['Setup'], None]]
 
 source_files = (
     'index.json', 'hello.js', 'bye.js', 'message.js', 'README.txt',
     'README.txt.license', '.reuse/dep5', 'LICENSES/CC0-1.0.txt'
 )
 
+def run_reuse(command, **kwargs):
+    """
+    Instead of running a 'reuse' command, check if 'mock_reuse_missing' file
+    exists under root directory. If yes, raise FileNotFoundError as if 'reuse'
+    command was missing. If not, check if 'README.txt.license' file exists
+    in the requested directory and return zero if it does.
+    """
+    expected = ['reuse', '--root', '<root>',
+                'lint' if 'lint' in command else 'spdx']
+
+    root_path = Path(process_command(command, expected)['root'])
+
+    if (root_path / 'mock_reuse_missing').exists():
+        raise FileNotFoundError('dummy')
+
+    is_reuse_compliant = (root_path / 'README.txt.license').exists()
+
+    return MockedCompletedProcess(command, 1 - is_reuse_compliant,
+                                  stdout=f'dummy {expected[-1]} output',
+                                  text_output=kwargs.get('text'))
+
+@pytest.fixture
+def mock_reuse(mock_subprocess_run):
+    """
+    Mock the REUSE command when executed through subprocess.run() from serve.py.
+    """
+    mock_subprocess_run(build, run_reuse)
+
 class Setup:
     """
     Facilitate preparing test malcontent directory, Hydrilla config file and the
@@ -95,8 +125,8 @@ class Setup:
         if self._modify_before_build:
             self._modify_before_build(self)
 
-        build = Build(self.source_dir, self.index_json)
-        build.write_package_files(self.malcontent_dir)
+        build_obj = build.Build(self.source_dir, self.index_json)
+        build_obj.write_package_files(self.malcontent_dir)
 
         if self._modify_after_build:
             self._modify_after_build(self)
@@ -114,48 +144,64 @@ class Setup:
         Provide app client that serves the objects from built sample package.
         """
         if self._client is None:
-            app = HydrillaApp(self.config(), flask_config={'TESTING': True})
+            app = serve.HydrillaApp(self.config(),
+                                    flask_config={'TESTING': True})
             self._client = app.test_client()
 
         return self._client
 
-def remove_all_uuids(setup: Setup) -> None:
-    """Modify sample packages before build to contain no (optional) UUIDs"""
-    index_json = (setup.source_dir / 'index.json').read_text()
-    index_json = json.loads(hydrilla_util.strip_json_comments(index_json))
+def index_json_modification(modify_index_json):
+    """Decorator for function that modifies index.json before build."""
+    def handle_index_json(setup):
+        """Modify index.json before build."""
+        index_path = setup.source_dir / 'index.json'
+        index_json, _ = hydrilla_util.load_instance_from_file(index_path)
+
+        index_json = modify_index_json(index_json) or index_json
+
+        index_json = f'''
+        // SPDX-License-Identifier: CC0-1.0
+        // Copyright (C) 2021, 2022 Wojtek Kosior
+        {json.dumps(index_json)}
+        '''
+
+        index_path.write_text(index_json)
 
+    return handle_index_json
+
+@index_json_modification
+def remove_all_uuids(index_json):
+    """Modify sample packages to contain no (optional) UUIDs"""
     for definition in index_json['definitions']:
         del definition['uuid']
 
-    index_json = ("// SPDX-License-Identifier: CC0-1.0\n" +
-                  "// Copyright (C) 2021, 2022 Wojtek Kosior\n" +
-                  json.dumps(index_json))
+@index_json_modification
+def bump_schema_v2(index_json) -> None:
+    """Modify sample packages to use version 2 of Hydrilla JSON schemas."""
+    for definition in index_json['definitions']:
+        definition['min_haketilo_version'] = [1, 1]
 
-    (setup.source_dir / 'index.json').write_text(index_json)
+        if definition['identifier'] == 'helloapple' and \
+           definition['type'] == 'resource':
+            definition['required_mappings'] = {'identifier': 'helloapple'}
 
-default_setup = Setup()
-uuidless_setup = Setup(modify_before_build=remove_all_uuids)
+default_setup = lambda: Setup()
+uuidless_setup = lambda: Setup(modify_before_build=remove_all_uuids)
+schema_v2_setup = lambda: Setup(modify_before_build=bump_schema_v2)
 
-def def_get(url: str) -> Response:
-    """Convenience wrapper for def_get()"""
-    return default_setup.client().get(url)
+setup_makers = [default_setup, uuidless_setup, schema_v2_setup]
 
-def test_project_url() -> None:
-    """Fetch index.html and verify project URL from config is present there."""
-    response = def_get('/')
-    assert b'html' in response.data
-    project_url = default_setup.config()['hydrilla_project_url']
-    assert escape(project_url).encode() in response.data
-
-@pytest.mark.parametrize('setup', [default_setup, uuidless_setup])
+@pytest.mark.usefixtures('mock_reuse')
+@pytest.mark.parametrize('setup_maker', setup_makers)
 @pytest.mark.parametrize('item_type', ['resource', 'mapping'])
-def test_get_newest(setup: Setup, item_type: str) -> None:
+def test_get_newest(setup_maker, item_type: str) -> None:
     """
     Verify that
         GET '/{item_type}/{item_identifier}.json'
     returns proper definition that is also served at:
         GET '/{item_type}/{item_identifier}/{item_version}'
     """
+    setup = setup_maker()
     response = setup.client().get(f'/{item_type}/helloapple.json')
     assert response.status_code == 200
     definition = json.loads(response.data.decode())
@@ -166,46 +212,58 @@ def test_get_newest(setup: Setup, item_type: str) -> None:
     assert response.status_code == 200
     assert definition == json.loads(response.data.decode())
 
-    assert ('uuid' in definition) == (setup is not uuidless_setup)
+    assert ('uuid' in definition) == (setup_maker is not uuidless_setup)
 
     hydrilla_util.validator_for(f'api_{item_type}_description-1.0.1.schema.json')\
                  .validate(definition)
 
+@pytest.fixture
+def setup(mock_reuse):
+    """Prepare server test environment in the default way."""
+    return default_setup()
+
+def test_project_url(setup) -> None:
+    """Fetch index.html and verify project URL from config is present there."""
+    response = setup.client().get('/')
+    assert b'html' in response.data
+    project_url = setup.config()['hydrilla_project_url']
+    assert escape(project_url).encode() in response.data
+
 @pytest.mark.parametrize('item_type', ['resource', 'mapping'])
-def test_get_nonexistent(item_type: str) -> None:
+def test_get_nonexistent(setup, item_type: str) -> None:
     """
     Verify that attempts to GET a JSON definition of a nonexistent item or item
     version result in 404.
     """
-    response = def_get(f'/{item_type}/nonexistentapple.json')
+    response = setup.client().get(f'/{item_type}/nonexistentapple.json')
     assert response.status_code == 404
-    response = def_get(f'/{item_type}/helloapple/1.2.3.999')
+    response = setup.client().get(f'/{item_type}/helloapple/1.2.3.999')
     assert response.status_code == 404
 
 @pytest.mark.parametrize('item_type', ['resource', 'mapping'])
-def test_file_refs(item_type: str) -> None:
+def test_file_refs(setup, item_type: str) -> None:
     """
     Verify that files referenced by definitions are accessible under their
     proper URLs and that their hashes match.
     """
-    response = def_get(f'/{item_type}/helloapple/2021.11.10')
+    response = setup.client().get(f'/{item_type}/helloapple/2021.11.10')
     assert response.status_code == 200
     definition = json.loads(response.data.decode())
 
     for file_ref in [*definition.get('scripts', []),
                      *definition['source_copyright']]:
         hash_sum = file_ref["sha256"]
-        response = def_get(f'/file/sha256/{hash_sum}')
+        response = setup.client().get(f'/file/sha256/{hash_sum}')
 
         assert response.status_code == 200
         assert sha256(response.data).digest().hex() == hash_sum
 
-def test_empty_query() -> None:
+def test_empty_query(setup) -> None:
     """
     Verify that querying mappings for URL gives an empty list when there're no
     mathes.
     """
-    response = def_get(f'/query?url=https://nonexiste.nt/example')
+    response = setup.client().get(f'/query?url=https://nonexiste.nt/example')
     assert response.status_code == 200
 
     response_object = json.loads(response.data.decode())
@@ -219,12 +277,12 @@ def test_empty_query() -> None:
     hydrilla_util.validator_for('api_query_result-1.0.1.schema.json')\
                  .validate(response_object)
 
-def test_query() -> None:
+def test_query(setup) -> None:
     """
     Verify that querying mappings for URL gives a list with reference(s) the the
     matching mapping(s).
     """
-    response = def_get(f'/query?url=https://hydrillabugs.koszko.org/')
+    response = setup.client().get(f'/query?url=https://hydrillabugs.koszko.org/')
     assert response.status_code == 200
 
     response_object = json.loads(response.data.decode())
@@ -239,12 +297,12 @@ def test_query() -> None:
         'generated_by': expected_generated_by
     }
 
-    hydrilla_util.validator_for('api_query_result-1.0.1.schema.json')\
+    hydrilla_util.validator_for('api_query_result-1.schema.json')\
                  .validate(response_object)
 
-def test_source() -> None:
+def test_source(setup) -> None:
     """Verify source descriptions are properly served."""
-    response = def_get(f'/source/hello.json')
+    response = setup.client().get(f'/source/hello.json')
     assert response.status_code == 200
 
     description = json.loads(response.data.decode())
@@ -254,18 +312,18 @@ def test_source() -> None:
         ['hello-message', 'helloapple', 'helloapple']
 
     zipfile_hash = description['source_archives']['zip']['sha256']
-    response = def_get(f'/source/hello.zip')
+    response = setup.client().get(f'/source/hello.zip')
     assert sha256(response.data).digest().hex() == zipfile_hash
 
-    hydrilla_util.validator_for('api_source_description-1.0.1.schema.json')\
+    hydrilla_util.validator_for('api_source_description-1.schema.json')\
                  .validate(description)
 
-def test_missing_source() -> None:
+def test_missing_source(setup) -> None:
     """Verify requests for nonexistent sources result in 404."""
-    response = def_get(f'/source/nonexistent.json')
+    response = setup.client().get(f'/source/nonexistent.json')
     assert response.status_code == 404
 
-    response = def_get(f'/source/nonexistent.zip')
+    response = setup.client().get(f'/source/nonexistent.zip')
     assert response.status_code == 404
 
 def test_normalize_version():