Merge pull request #1177 from strictdoc-project/stanislaw/1157-includ…

…e-exclude-filters

 helpers/path_filter: basic validation of user input

strictdoc/__init__.py

@@ -1,6 +1,6 @@
 from strictdoc.core.environment import SDocRuntimeEnvironment
 
-__version__ = "0.0.42a3"
+__version__ = "0.0.42a5"
 
 
 environment = SDocRuntimeEnvironment(__file__)

strictdoc/core/project_config.py

@@ -6,6 +6,7 @@
 import toml
 
 from strictdoc.helpers.auto_described import auto_described
+from strictdoc.helpers.path_filter import validate_mask
 
 
 class ProjectFeature(str, Enum):
@@ -167,18 +168,57 @@ def _load_from_dictionary(config_dict: dict) -> ProjectConfig:
                 "include_doc_paths", include_doc_paths
             )
             assert isinstance(include_doc_paths, list)
+            for include_doc_path in include_doc_paths:
+                try:
+                    validate_mask(include_doc_path)
+                except SyntaxError as exception:
+                    print(  # noqa: T201
+                        f"error: strictdoc.toml: 'include_doc_paths': "
+                        f"{exception} Provided string: '{include_doc_path}'."
+                    )
+                    sys.exit(1)
+
             exclude_doc_paths = project_content.get(
                 "exclude_doc_paths", exclude_doc_paths
             )
             assert isinstance(exclude_doc_paths, list)
+            for exclude_doc_path in exclude_doc_paths:
+                try:
+                    validate_mask(exclude_doc_path)
+                except SyntaxError as exception:
+                    print(  # noqa: T201
+                        f"error: strictdoc.toml: 'exclude_doc_paths': "
+                        f"{exception} Provided string: '{exclude_doc_path}'."
+                    )
+                    sys.exit(1)
+
             include_source_paths = project_content.get(
                 "include_source_paths", include_source_paths
             )
             assert isinstance(include_source_paths, list)
+            for include_source_path in include_source_paths:
+                try:
+                    validate_mask(include_source_path)
+                except SyntaxError as exception:
+                    print(  # noqa: T201
+                        f"error: strictdoc.toml: 'include_source_paths': "
+                        f"{exception} Provided string: '{include_source_path}'."
+                    )
+                    sys.exit(1)
+
             exclude_source_paths = project_content.get(
                 "exclude_source_paths", exclude_source_paths
             )
             assert isinstance(exclude_source_paths, list)
+            for exclude_source_path in exclude_source_paths:
+                try:
+                    validate_mask(exclude_source_path)
+                except SyntaxError as exception:
+                    print(  # noqa: T201
+                        f"error: strictdoc.toml: 'exclude_source_paths': "
+                        f"{exception} Provided string: '{exclude_source_path}'."
+                    )
+                    sys.exit(1)
 
         if "server" in config_dict:
             # FIXME: Introduce at least a basic validation for the host/port.

strictdoc/helpers/path_filter.py

@@ -11,18 +11,33 @@
     rf"(\\|\/)?({REGEX_NAME_PART})?((\\|\/){REGEX_NAME_PART})*(\\|\/)?"
 )
 
+REGEX_MASK_VALIDATION = rf"[^(\\|\/)]{REGEX_DOUBLE_WILDCARD}"
+
+MOST_COMMON_CHARACTERS = ("[", "]", "(", ")", "{", "}", "?", "+", "!")
+
 
 def validate_mask(mask: str):
-    assert not mask.startswith("/")
-    assert "***" not in mask
-    assert "//" not in mask
-    assert "\\" not in mask
-    assert "?" not in mask
-    assert "[" not in mask
-    assert "]" not in mask
-    assert "(" not in mask
-    assert ")" not in mask
-    assert "+" not in mask
+    if mask == "":
+        raise SyntaxError("Path mask must not be empty.")
+
+    if not mask[0].isalnum() and mask[0] != "*":
+        raise SyntaxError(
+            "Path mask must start with an alphanumeric character or "
+            "a wildcard symbol '*'."
+        )
+
+    if "//" in mask or "\\\\" in mask:
+        raise SyntaxError("Path mask must not contain double slashes.")
+
+    if "***" in mask:
+        raise SyntaxError("Invalid wildcard: '***'.")
+
+    for regex_symbol in MOST_COMMON_CHARACTERS:
+        if regex_symbol in mask:
+            raise SyntaxError(
+                f"Path mask must not contain any of the special characters: "
+                f"{MOST_COMMON_CHARACTERS}."
+            )
 
 
 def compile_regex_mask(path_mask: str) -> str:

tests/integration/options/options_per_project/include_doc_paths/02_invalid_option/docs1/document1.sdoc

@@ -0,0 +1,6 @@
+[DOCUMENT]
+TITLE: Document 1
+
+[FREETEXT]
+**Hello world**
+[/FREETEXT]

tests/integration/options/options_per_project/include_doc_paths/02_invalid_option/strictdoc.toml

@@ -0,0 +1,7 @@
+[project]
+
+include_doc_paths = [
+  " "
+]
+
+features = []

tests/integration/options/options_per_project/include_doc_paths/02_invalid_option/test.itest

@@ -0,0 +1,3 @@
+RUN: %expect_exit 1 %strictdoc export %S --output-dir Output/ | filecheck %s --dump-input=fail
+
+CHECK: error: strictdoc.toml: 'include_doc_paths': Path mask must start with an alphanumeric character or a wildcard symbol '*'. Provided string: ' '.

tests/unit/strictdoc/helpers/test_path_filter_mask_validation.py

@@ -0,0 +1,43 @@
+import pytest
+
+from strictdoc.helpers.path_filter import validate_mask
+
+
+def test_case_01_empty_mask_allows_everything():
+    # Does not raise is success.
+    validate_mask("docs/")
+    validate_mask("Docs/")
+
+    with pytest.raises(SyntaxError) as exc_info:
+        validate_mask("")
+    assert "Path mask must not be empty." in exc_info.value.args[0]
+
+    with pytest.raises(SyntaxError) as exc_info:
+        validate_mask(".")
+    assert (
+        "Path mask must start with an alphanumeric character or "
+        "a wildcard symbol '*'."
+    ) in exc_info.value.args[0]
+
+    with pytest.raises(SyntaxError) as exc_info:
+        validate_mask("foo//")
+    assert (
+        "Path mask must not contain double slashes." in exc_info.value.args[0]
+    )
+
+    with pytest.raises(SyntaxError) as exc_info:
+        validate_mask("bar\\\\")
+    assert (
+        "Path mask must not contain double slashes." in exc_info.value.args[0]
+    )
+
+    with pytest.raises(SyntaxError) as exc_info:
+        validate_mask("***")
+    assert "Invalid wildcard: '***'." in exc_info.value.args[0]
+
+    with pytest.raises(SyntaxError) as exc_info:
+        validate_mask("a[")
+    assert (
+        "Path mask must not contain any of the special characters: "
+        "('[', ']', '(', ')', '{', '}', '?', '+', '!')."
+    ) in exc_info.value.args[0]