* Adds spanish documentation
* Adds spanish documentation to CHANGES

…imeError (#791)

* fix: remove content-type from headers in _add_from_file to avoid RuntimeError

The recorder stores content_type as a top-level YAML key but may also
capture a 'content-type' entry inside the headers dict (since HTTP
servers return Content-Type as a response header).

When _add_from_file() calls add() with both content_type= and
headers={'content-type': ...}, add() raises:

    RuntimeError: You cannot define both content_type and
    headers[Content-Type].

Fix: strip any 'content-type' key from headers before calling add()
when content_type is also present in the loaded response dict.  The
content_type kwarg takes precedence, which is already documented as
the recommended approach.

Fixes #741

* test: add regression test for Content-Type conflict in _add_from_file

Fixes the case where a YAML fixture has Content-Type in both headers
and content_type fields — the recorder captures both, causing RuntimeError
when _add_from_file calls add(). Added test_add_from_file_content_type_in_headers
to verify the fix works end-to-end with a YAML fixture.

* test: use mismatched Content-Type values to verify content_type precedence

Per markstory's review: using different values for headers['Content-Type']
('text/html') and content_type ('application/json'/'text/plain') makes the
assertion non-trivial — it now explicitly verifies that content_type wins
over the conflicting header, not just that both happen to carry the same value.

* fix(test): use .yaml extension so _add_from_file selects YAML loader

As suggested by @markstory — the fixture file needs an explicit extension
so that _add_from_file can detect the correct loader. Changed out_file
to use a .yaml suffix for the content_type_in_headers test.

* fix(test): clean up .yaml/.toml fixture variants in teardown_method

The content_type test creates response_record.yaml but teardown_method
only deleted response_record (no extension). Add cleanup for .yaml and
.toml variants to avoid stale artifacts across test runs.

* fix(test): restore _parse_response_file in teardown + reformat for black

Root cause of the CI failure: test_add_from_file[tomli_w] monkey-patches
responses.mock._parse_response_file to a TOML loader and never restores
it, so test_add_from_file_content_type_in_headers inherits the TOML
parser and chokes on the YAML fixture (tomllib.TOMLDecodeError).

The prior fixes (.yaml extension, .yaml/.toml teardown cleanup) don't
help because _parse_response_file always calls yaml.safe_load regardless
of file extension — the bug is in test state leakage, not the loader
dispatch.

- teardown_method: always reset _parse_response_file back to the class
  default so any per-test monkey-patch is contained.
- __init__.py: black wrap for the Content-Type-stripping dict comp
  (was 91 chars, over the 88-char limit).

* fix(test): unwind mock inside test_add_from_file instead of teardown

per markstory's review: capture the original `_parse_response_file`
before monkey-patching it for the tomli_w parametrize variant, then
restore it in a finally block so the mock is contained to the test
that introduces it. removes the class-wide reset from teardown_method
that was leaking responsibility upward.

---------

Co-authored-by: Julio César Suástegui <juliosuas@users.noreply.github.com>

* chore: Fix lint build and update changes
* Fix mypy better

* Add strict_match parameter to urlencoded_params_matcher
* Add to CHANGES the new strict_match param for urlencoded_params_matcher
* Prevent non-strict matching of empty params with non-empty body
* Test for successful strict matching
* Check for falsy/truthy params and request_body instead of None
  With this change, "params={}" has the same effect as "params=None".
* Show note when match fails and strict match is ON

_remove_default_headers matched header names by exact case, so the
default headers (Content-Type, Date, Server, and the others) were only
stripped when servers used the canonical casing. HTTP header names are
case-insensitive and HTTP/2 servers send them lowercase, so responses
recorded from such servers kept these redundant headers in the file.
Match the names without regard to case.

…801)

Numeric values (int/float) in the params mapping were being converted to
strings in-place, silently modifying the original dict passed by the
caller. Copy the mapping into a local dict before performing the
int/float→str normalisation.

---------

Co-authored-by: gaoflow <claude@vinzzy.uk>

Right now the License uses `Apache 2.0` instead of the SPDX compliant `Apache-2.0`.

This causes problems with some license scanners like https://github.com/anchore/grant

Move recently merged changes under a new heading instead of under 0.26.1
which has already been released.