* gptel-transient.el (gptel-menu, gptel-lisp-variable,
gptel--infix-num-messages-to-send, gptel--infix-max-tokens,
gptel--infix-provider): Improve the wording and the display of the
default settings. Define a new class `gptel-lisp-variable` to
faciliate this.
* gptel.el (gptel--inspect-query): Add a function to inspect the
query that will be sent. This is useful to ensure that (only and
all of) what you expect to be sent is being sent.
* gptel-transient.el (gptel-menu): Allow the query in progress to
be inspected from `gptel-menu` as an elisp or JSON object.
* gptel.el (gptel-request, gptel-send, gptel--url-get-response):
Consolidate the HTTP query construction into `gptel-request`,
and use it as the single point for sending gptel queries. This
simplifies the code, makes it easier to debug and (later) advise
or otherwise modify. To this end, `gptel-send` reuses
`gptel-request` and no longer does its own thing. Adjust other
HTTP request-related functions accordingly.
* gptel-curl.el (gptel-curl--get-args, gptel-curl--get-response):
Receive the full request data instead of constructing it partly in
`gptel-curl--get-args`.
* gptel.el (gptel--url-get-response): Log request headers as JSON objects.
* gptel-curl.el (gptel-curl--get-args, gptel-curl--get-response):
Log request headers as JSON objects, Curl command as a shell
command you can copy and paste into the terminal.
Using the libjansson JSON parser gives us a modest boost in speed.
It's not as significant a speedup as it is for LSP clients since
our jSON payloads are smaller and less frequent -- but we might as
well use it.
* gptel.el (gptel--json-read, gptel--json-encode,
gptel--url-get-response, gptel--parse-response): Define macros to
use the libjansson-supported `json-parse-buffer` and
`json-serialize`. Replace use of `json-encode` and `json-read`
appropriately.
* gptel-openai.el: (gptel-curl--parse-stream) : Use
`gptel--json-read` instead of `json-read`.
* gptel-ollama.el (gptel-curl--parse-stream): Use
`gptel--json-read` instead of `json-read`.
* gptel-gemini.el (gptel-curl--parse-stream): Use
`gptel--json-read` instead of `json-read`.
* gptel-curl.el (gptel-curl--get-args, gptel-curl--get-response,
gptel-curl--log-response, gptel-curl--stream-cleanup,
gptel-curl--parse-response): Use `gptel--json-read` and
`gptel--json-encode` in place of the json.el versions.
* gptel-anthropic.el (gptel-curl--parse-stream): Use
`gptel--json-read` instead of `json-read`.
* test/gptel-org-test.el: Use `gptel--json-read`.
* gptel.el (gptel--sanitize-model, gptel): Check for mismatches
between the default values of gptel-backend and gptel-model when
starting a new gptel chat. Previously the default value of
gptel-backend was compared against the buffer-local value of
gptel-model.
* gptel.el (gptel--insert-response, gptel--restore-state): Don't
set the rear-nonsticky property on the gptel response text.
Instead, the gptel text-property is globally declared to be
nonsticky via `text-property-default-nonsticky`.
This change is since markdown-mode makes rear-nonsticky a
font-lock-managed property and we can't set it manually. Setting
this property explicitly also makes all text properties in the
range nonsticky, which can have unintended side effects.
* gptel-curl.el (gptel-curl--stream-insert-response): Ditto.
gptel.el (gptel--convert-markdown->org,
gptel--stream-convert-markdown->org): Handle the case of Markdown
bullet lists that begin with a "*", which the Gemini LLMs produce
often. Address #238.
gptel-anthropic.el (gptel-curl--parse-stream): Reset point
explicitly when parsing streaming responses returned by the
Anthropic API. Try to address #233.
gptel.el (gptel--url-parse-response): Handle HTTP 100 followed by
200. Note: this fix is brittle, it will break if 100 is followed
by an error code.
gptel-curl.el (gptel-curl--stream-filter,
gptel-curl--parse-stream): Ditto. Address #194.
gptel-curl.el (gptel-curl--stream-cleanup): The LLM response can
be empty with HTTP status 200, for example when the API responds
with an error description instead. Handle this case gracefully.
When `gptel-mode` is enabled, also inform the user that the
response was empty. Fix#234.
gptel.el (gptel-post-response-functions): Documentation. Explain
the arguments passed to each hook function in this hook when the
response fails or is empty.
gptel-curl.el (gptel-curl--stream-cleanup,
gptel-curl--stream-insert-response): Don't consider
`gptel-response-prefix-string` part of the response for the
purpose of running `gptel-post-response-functions`.
* gptel.el: Mention Anthropic in the package description.
* gptel-anthropic.el (gptel-anthropic, gptel-make-anthropic,
gptel--parse-response, gptel--request-data, gptel--parse-buffer,
gptel-curl--parse-stream): Add support for Anthropic AI's Claude 3
models.
* README.org: Add instructions for using Anthropic AI's Claude 3
models.
gptel-transient (gptel--system-prompt-setup): When assigning keys
to directives, avoid clashes with the system-message edit key
(`h`), and use an arbitrary unused key if no mnemonic key
assignments are possible.
gptel-openai.el (gptel-backend, gptel-make-openai,
gptel-make-azure): Add a curl-args slot to the backend struct for
additional Curl arguments.
Usage example: This can be used to set the `--cert` and `--key`
options in a custom backend that uses mutal TLS to communicate
with an OpenAI proxy/gateway.
gptel-curl.el (gptel-curl--get-args): Add backend-specific
curl-args when creating HTTP requests.
gptel-gemini.el (gptel-make-gemini): Add a curl-args slot to the
constructor.
gptel-kagi.el (gptel-make-kagi): Ditto.
gptel-ollama.el (gptel-make-ollama): Ditto.
* gptel.el (gptel-display-buffer-action, gptel): Add
`gptel-display-buffer-action` to customize the display of the
gptel buffer.
* README: Mention new option.
* gptel-gemini.el (gptel-make-gemini, gptel--request-data): The
Gemini API misclassifies harmless questions (like "What's 2+2",
see #208) as harmful. Use the most permissive safety settings the
API offers.
Also respect the value of `:stream` used when defining Gemini
backends.
* gptel.el (gptel--attach-response-history, gptel--ediff,
gptel--next-variant, gptel--previous-variant,
gptel--mark-response):
Add `gptel--attach-response-history` -- this can be used to add
text properties to the next gptel response in the buffer. This
is (currently) useful for tracking changes when the response
overwrites existing text.
The next three commands -- `gptel--ediff`,
`gptel--previous-variant`, `gptel--next-variant` -- provide
facilities for manipulating a gptel response at point when there
is history. `gptel--mark-response` marks the response at point.
These are considered internal functions for now and can be
accessed from the transient menu, where they work together with
`gptel--regenerate`.
The input arguments to these commands are expected to change to
support copilot-style functionality in the near future.
* gptel-transient.el (gptel-menu, gptel--suffix-send,
gptel--regenerate):
Change the transient menu layout to be more compact (with a newly
added column.) When overwriting the prompt with a response, save
the prompt to the gptel response's history. Add
`gptel--regenerate` to regenerate a response. This is accessible
from the transient menu when the point is inside response text.
* gptel.el (gptel--debug, gptel-log-level, gptel--log-buffer-name,
gptel--url-get-response, gptel--parse-response): Optionally log
all request and response data to `gptel--log-buffer-name`, with
the log level governed by `gptel-log-level`. Obsolete
`gptel--debug`.
* gptel-curl.el (gptel-curl--log-response, gptel-curl--get-args,
gptel-curl--get-response, gptel-curl--stream-cleanup,
gptel-curl--sentinel): Add support for logging.
* gptel-transient.el (gptel--suffix-system-message): Mark the
whole directive when editing it in the `*gptel-system*` buffer.
Previously only the first line of the directive was being marked.
* gptel.el (gptel--convert-markdown->org,
gptel--stream-convert-markdown->org): Don't touch underscores in
the source markdown. This will turn some emphasis blocks into
underlines in Org, but we can live with that. Fix#40.
* gptel.el (gptel--restore-state, gptel--save-state): Try to
support writing/reading multi-line directives in Org and other
modes correctly when saving the buffer. The support is
preliminary and works as follows:
- org-mode: Replace newlines with "\n" before writing the relevant
org property.
- other modes: escape newlines with print-escape-newlines when
updating local vars.
Neither of these is a good fix, but brings support for multi-line
directives up from completely broken to works-via-hack. They are
subject to change in the future and might break some chat files. :(
* gptel.el (gptel-request): Update docstring to clarify what
BUFFER and POSITION do. Addresses #191.
* gptel-transient.el (gptel-menu, gptel--suffix-send): Replace
"new session" and "existing session" redirection options with
"gptel session" and "any buffer", allowing for more flexibility
when redirecting. "gptel session" can be an existing or new
session. Fix bug where the prompt was generated from the contents
of the destination buffer instead of the current buffer when
redirecting to a gptel session. Add comments demarcating blocks
in `gptel--suffix-send`.
* gptel.el (gptel--openai): Don't specify header.
* gptel-openai.el (gptel-make-openai): Use a key-aware lambda for
the header argument. This should make it easier to define new
OpenAI-style API backends (see #177, #184)
* README.org: Update with instructions for together.ai and
Anyscale, both of which provide OpenAI-style APIs. Clean up the
config blocks for the other backends.
* gptel-transient.el (gptel-menu): Sanitize model if it's not in
the backend.
* gptel.el (gptel--sanitize-model): New helper.
* gptel.el (gptel-send): Also sanitize model in non-prefixed gptel-send.
* gptel-openai.el (cl-lib): Require it.
* gptel.el (compat): Leniently require compat so gptel.el can be
compiled standalone. This will expose other compiler errors that
are easily visible with M-x flymake.
