schemathesis 3.25.6__py3-none-any.whl → 3.39.7__py3-none-any.whl

schemathesis/specs/openapi/formats.py

@@ -1,5 +1,8 @@
 from __future__ import annotations
 
+import string
+from base64 import b64encode
+from functools import lru_cache
 from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:
@@ -33,5 +36,46 @@ def unregister_string_format(name: str) -> None:
         raise ValueError(f"Unknown Open API format: {name}") from exc
 
 
+def header_values(blacklist_characters: str = "\n\r") -> st.SearchStrategy[str]:
+    from hypothesis import strategies as st
+
+    return st.text(
+        alphabet=st.characters(min_codepoint=0, max_codepoint=255, blacklist_characters=blacklist_characters)
+        # Header values with leading non-visible chars can't be sent with `requests`
+    ).map(str.lstrip)
+
+
+HEADER_FORMAT = "_header_value"
+
+
+@lru_cache
+def get_default_format_strategies() -> dict[str, st.SearchStrategy]:
+    """Get all default "format" strategies."""
+    from hypothesis import strategies as st
+    from requests.auth import _basic_auth_str
+
+    from ...serializers import Binary
+
+    def make_basic_auth_str(item: tuple[str, str]) -> str:
+        return _basic_auth_str(*item)
+
+    latin1_text = st.text(alphabet=st.characters(min_codepoint=0, max_codepoint=255))
+
+    # Define valid characters here to avoid filtering them out in `is_valid_header` later
+    header_value = header_values()
+
+    return {
+        "binary": st.binary().map(Binary),
+        "byte": st.binary().map(lambda x: b64encode(x).decode()),
+        # RFC 7230, Section 3.2.6
+        "_header_name": st.text(
+            min_size=1, alphabet=st.sampled_from("!#$%&'*+-.^_`|~" + string.digits + string.ascii_letters)
+        ),
+        HEADER_FORMAT: header_value,
+        "_basic_auth": st.tuples(latin1_text, latin1_text).map(make_basic_auth_str),
+        "_bearer_auth": header_value.map("Bearer {}".format),
+    }
+
+
 register = register_string_format
 unregister = unregister_string_format

schemathesis/specs/openapi/links.py

@@ -2,26 +2,31 @@
 
 Based on https://swagger.io/docs/specification/links/
 """
+
 from __future__ import annotations
+
 from dataclasses import dataclass, field
 from difflib import get_close_matches
-from typing import Any, Generator, NoReturn, Sequence, Union, TYPE_CHECKING
-
-from ...models import APIOperation, Case
-from ...parameters import ParameterSet
-from ...stateful import ParsedData, StatefulTest
-from ...stateful.state_machine import Direction
-from ...types import NotSet
+from types import SimpleNamespace
+from typing import TYPE_CHECKING, Any, Generator, Literal, NoReturn, Sequence, TypedDict, Union, cast
 
 from ...constants import NOT_SET
 from ...internal.copy import fast_deepcopy
+from ...models import APIOperation, Case, TransitionId
+from ...stateful import ParsedData, StatefulTest, UnresolvableLink
+from ...stateful.state_machine import Direction
 from . import expressions
 from .constants import LOCATION_TO_CONTAINER
 from .parameters import OpenAPI20Body, OpenAPI30Body, OpenAPIParameter
-
+from .references import RECURSION_DEPTH_LIMIT, Unresolvable
 
 if TYPE_CHECKING:
+    from hypothesis.vendor.pretty import RepresentationPrinter
+    from jsonschema import RefResolver
+
+    from ...parameters import ParameterSet
     from ...transports.responses import GenericResponse
+    from ...types import NotSet
 
 
 @dataclass(repr=False)
@@ -29,6 +34,7 @@ class Link(StatefulTest):
     operation: APIOperation
     parameters: dict[str, Any]
     request_body: Any = NOT_SET
+    merge_body: bool = True
 
     def __post_init__(self) -> None:
         if self.request_body is not NOT_SET and not self.operation.body:
@@ -48,6 +54,7 @@ class Link(StatefulTest):
             operation = source_operation.schema.get_operation_by_id(definition["operationId"])  # type: ignore
         else:
             operation = source_operation.schema.get_operation_by_reference(definition["operationRef"])  # type: ignore
+        extension = definition.get(SCHEMATHESIS_LINK_EXTENSION)
         return cls(
             # Pylint can't detect that the API operation is always defined at this point
             # E.g. if there is no matching operation or no operations at all, then a ValueError will be risen
@@ -55,21 +62,25 @@ class Link(StatefulTest):
             operation=operation,
             parameters=definition.get("parameters", {}),
             request_body=definition.get("requestBody", NOT_SET),  # `None` might be a valid value - `null`
+            merge_body=extension.get("merge_body", True) if extension is not None else True,
         )
 
     def parse(self, case: Case, response: GenericResponse) -> ParsedData:
         """Parse data into a structure expected by links definition."""
         context = expressions.ExpressionContext(case=case, response=response)
-        parameters = {
-            parameter: expressions.evaluate(expression, context) for parameter, expression in self.parameters.items()
-        }
-        return ParsedData(
-            parameters=parameters,
-            # https://github.com/OAI/OpenAPI-Specification/blob/master/versions/3.0.3.md#link-object
-            # > A literal value or {expression} to use as a request body when calling the target operation.
-            # In this case all literals will be passed as is, and expressions will be evaluated
-            body=expressions.evaluate(self.request_body, context),
-        )
+        parameters = {}
+        for parameter, expression in self.parameters.items():
+            evaluated = expressions.evaluate(expression, context)
+            if isinstance(evaluated, Unresolvable):
+                raise UnresolvableLink(f"Unresolvable reference in the link: {expression}")
+            parameters[parameter] = evaluated
+        body = expressions.evaluate(self.request_body, context, evaluate_nested=True)
+        if self.merge_body:
+            body = merge_body(case.body, body)
+        return ParsedData(parameters=parameters, body=body)
+
+    def is_match(self) -> bool:
+        return self.operation.schema.filter_set.match(SimpleNamespace(operation=self.operation))
 
     def make_operation(self, collected: list[ParsedData]) -> APIOperation:
         """Create a modified version of the original API operation with additional data merged in."""
@@ -152,17 +163,27 @@ class Link(StatefulTest):
 
 def get_links(response: GenericResponse, operation: APIOperation, field: str) -> Sequence[Link]:
     """Get `x-links` / `links` definitions from the schema."""
-    responses = operation.definition.resolved["responses"]
+    responses = operation.definition.raw["responses"]
     if str(response.status_code) in responses:
-        response_definition = responses[str(response.status_code)]
+        definition = responses[str(response.status_code)]
     elif response.status_code in responses:
-        response_definition = responses[response.status_code]
+        definition = responses[response.status_code]
     else:
-        response_definition = responses.get("default", {})
-    links = response_definition.get(field, {})
+        definition = responses.get("default", {})
+    if not definition:
+        return []
+    _, definition = operation.schema.resolver.resolve_in_scope(definition, operation.definition.scope)  # type: ignore[attr-defined]
+    links = definition.get(field, {})
     return [Link.from_definition(name, definition, operation) for name, definition in links.items()]
 
 
+SCHEMATHESIS_LINK_EXTENSION = "x-schemathesis"
+
+
+class SchemathesisLink(TypedDict):
+    merge_body: bool
+
+
 @dataclass(repr=False)
 class OpenAPILink(Direction):
     """Alternative approach to link processing.
@@ -174,41 +195,87 @@ class OpenAPILink(Direction):
     status_code: str
     definition: dict[str, Any]
     operation: APIOperation
-    parameters: list[tuple[str | None, str, str]] = field(init=False)
+    parameters: list[tuple[Literal["path", "query", "header", "cookie", "body"] | None, str, str]] = field(init=False)
     body: dict[str, Any] | NotSet = field(init=False)
+    merge_body: bool = True
+
+    def __repr__(self) -> str:
+        path = self.operation.path
+        method = self.operation.method
+        return f"state.schema['{path}']['{method}'].links['{self.status_code}']['{self.name}']"
+
+    def _repr_pretty_(self, printer: RepresentationPrinter, cycle: bool) -> None:
+        return printer.text(repr(self))
 
     def __post_init__(self) -> None:
+        extension = self.definition.get(SCHEMATHESIS_LINK_EXTENSION)
         self.parameters = [
             normalize_parameter(parameter, expression)
             for parameter, expression in self.definition.get("parameters", {}).items()
         ]
         self.body = self.definition.get("requestBody", NOT_SET)
+        if extension is not None:
+            self.merge_body = extension.get("merge_body", True)
 
     def set_data(self, case: Case, elapsed: float, **kwargs: Any) -> None:
         """Assign all linked definitions to the new case instance."""
         context = kwargs["context"]
-        self.set_parameters(case, context)
-        self.set_body(case, context)
-        case.set_source(context.response, context.case, elapsed)
+        overrides = self.set_parameters(case, context)
+        self.set_body(case, context, overrides)
+        overrides_all_parameters = True
+        if case.operation.body and "body" not in overrides.get("body", []):
+            overrides_all_parameters = False
+        if overrides_all_parameters:
+            for parameter in case.operation.iter_parameters():
+                if parameter.name not in overrides.get(parameter.location, []):
+                    overrides_all_parameters = False
+                    break
+        case.set_source(
+            context.response,
+            context.case,
+            elapsed,
+            overrides_all_parameters,
+            transition_id=TransitionId(
+                name=self.name,
+                status_code=self.status_code,
+            ),
+        )
 
-    def set_parameters(self, case: Case, context: expressions.ExpressionContext) -> None:
+    def set_parameters(
+        self, case: Case, context: expressions.ExpressionContext
+    ) -> dict[Literal["path", "query", "header", "cookie", "body"], list[str]]:
+        overrides: dict[Literal["path", "query", "header", "cookie", "body"], list[str]] = {}
         for location, name, expression in self.parameters:
-            container = get_container(case, location, name)
+            location, container = get_container(case, location, name)
             # Might happen if there is directly specified container,
             # but the schema has no parameters of such type at all.
             # Therefore the container is empty, otherwise it will be at least an empty object
             if container is None:
                 message = f"No such parameter in `{case.operation.method.upper()} {case.operation.path}`: `{name}`."
-                possibilities = [param.name for param in case.operation.definition.parameters]
+                possibilities = [param.name for param in case.operation.iter_parameters()]
                 matches = get_close_matches(name, possibilities)
                 if matches:
                     message += f" Did you mean `{matches[0]}`?"
                 raise ValueError(message)
-            container[name] = expressions.evaluate(expression, context)
-
-    def set_body(self, case: Case, context: expressions.ExpressionContext) -> None:
+            value = expressions.evaluate(expression, context)
+            if value is not None:
+                container[name] = value
+                overrides.setdefault(location, []).append(name)
+        return overrides
+
+    def set_body(
+        self,
+        case: Case,
+        context: expressions.ExpressionContext,
+        overrides: dict[Literal["path", "query", "header", "cookie", "body"], list[str]],
+    ) -> None:
         if self.body is not NOT_SET:
-            case.body = expressions.evaluate(self.body, context)
+            evaluated = expressions.evaluate(self.body, context, evaluate_nested=True)
+            overrides["body"] = ["body"]
+            if self.merge_body:
+                case.body = merge_body(case.body, evaluated)
+            else:
+                case.body = evaluated
 
     def get_target_operation(self) -> APIOperation:
         if "operationId" in self.definition:
@@ -216,21 +283,32 @@ class OpenAPILink(Direction):
         return self.operation.schema.get_operation_by_reference(self.definition["operationRef"])  # type: ignore
 
 
-def get_container(case: Case, location: str | None, name: str) -> dict[str, Any] | None:
+def merge_body(old: Any, new: Any) -> Any:
+    if isinstance(old, dict) and isinstance(new, dict):
+        return {**old, **new}
+    return new
+
+
+def get_container(
+    case: Case, location: Literal["path", "query", "header", "cookie", "body"] | None, name: str
+) -> tuple[Literal["path", "query", "header", "cookie", "body"], dict[str, Any] | None]:
     """Get a container that suppose to store the given parameter."""
     if location:
         container_name = LOCATION_TO_CONTAINER[location]
     else:
-        for param in case.operation.definition.parameters:
+        for param in case.operation.iter_parameters():
             if param.name == name:
+                location = param.location
                 container_name = LOCATION_TO_CONTAINER[param.location]
                 break
         else:
             raise ValueError(f"Parameter `{name}` is not defined in API operation `{case.operation.verbose_name}`")
-    return getattr(case, container_name)
+    return location, getattr(case, container_name)
 
 
-def normalize_parameter(parameter: str, expression: str) -> tuple[str | None, str, str]:
+def normalize_parameter(
+    parameter: str, expression: str
+) -> tuple[Literal["path", "query", "header", "cookie", "body"] | None, str, str]:
     """Normalize runtime expressions.
 
     Runtime expressions may have parameter names prefixed with their location - `path.id`.
@@ -240,13 +318,15 @@ def normalize_parameter(parameter: str, expression: str) -> tuple[str | None, st
     try:
         # The parameter name is prefixed with its location. Example: `path.id`
         location, name = tuple(parameter.split("."))
-        return location, name, expression
+        _location = cast(Literal["path", "query", "header", "cookie", "body"], location)
+        return _location, name, expression
     except ValueError:
         return None, parameter, expression
 
 
 def get_all_links(operation: APIOperation) -> Generator[tuple[str, OpenAPILink], None, None]:
-    for status_code, definition in operation.definition.resolved["responses"].items():
+    for status_code, definition in operation.definition.raw["responses"].items():
+        definition = operation.schema.resolver.resolve_all(definition, RECURSION_DEPTH_LIMIT - 8)  # type: ignore[attr-defined]
         for name, link_definition in definition.get(operation.schema.links_field, {}).items():  # type: ignore
             yield status_code, OpenAPILink(name, status_code, link_definition, operation)
 
@@ -273,6 +353,7 @@ def _get_response_by_status_code(responses: dict[StatusCode, dict[str, Any]], st
 
 
 def add_link(
+    resolver: RefResolver,
     responses: dict[StatusCode, dict[str, Any]],
     links_field: str,
     parameters: dict[str, str] | None,
@@ -282,6 +363,8 @@ def add_link(
     name: str | None = None,
 ) -> None:
     response = _get_response_by_status_code(responses, status_code)
+    if "$ref" in response:
+        _, response = resolver.resolve(response["$ref"])
     links_definition = response.setdefault(links_field, {})
     new_link: dict[str, str | dict[str, str]] = {}
     if parameters is not None:
@@ -295,8 +378,8 @@ def add_link(
         name = name or f"{target.method.upper()} {target.path}"
         # operationId is a dict lookup which is more efficient than using `operationRef`, since it
         # doesn't involve reference resolving when we will look up for this target during testing.
-        if "operationId" in target.definition.resolved:
-            new_link["operationId"] = target.definition.resolved["operationId"]
+        if "operationId" in target.definition.raw:
+            new_link["operationId"] = target.definition.raw["operationId"]
         else:
             new_link["operationRef"] = target.operation_reference
     # The name is arbitrary, so we don't really case what it is,