gptel/gptel-transient.el

741 lines
29 KiB
EmacsLisp
Raw Normal View History

;;; gptel-transient.el --- Transient menu for GPTel -*- lexical-binding: t; -*-
;; Copyright (C) 2023 Karthik Chikmagalur
;; Author: Karthik Chikmagalur <karthikchikmagalur@gmail.com>
;; Keywords: convenience
;; SPDX-License-Identifier: GPL-3.0-or-later
;; This program is free software; you can redistribute it and/or modify
;; it under the terms of the GNU General Public License as published by
;; the Free Software Foundation, either version 3 of the License, or
;; (at your option) any later version.
;; This program is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
;; GNU General Public License for more details.
;; You should have received a copy of the GNU General Public License
;; along with this program. If not, see <https://www.gnu.org/licenses/>.
;;; Commentary:
;;
;;; Code:
(require 'cl-lib)
(require 'gptel)
(require 'transient)
(declare-function ediff-regions-internal "ediff")
(declare-function ediff-make-cloned-buffer "ediff-utils")
;; * Helper functions
(defun gptel--refactor-or-rewrite ()
"Rewrite should be refactored into refactor.
Or is it the other way around?"
(if (derived-mode-p 'prog-mode)
"Refactor" "Rewrite"))
(defvar-local gptel--rewrite-message nil)
(defun gptel--rewrite-message ()
"Set a generic refactor/rewrite message for the buffer."
(if (derived-mode-p 'prog-mode)
(format "You are a %s programmer. Refactor the following code. Generate only code, no explanation."
(substring (symbol-name major-mode) nil -5))
(format "You are a prose editor. Rewrite the following text to be more professional.")))
(defvar gptel--crowdsourced-prompts-url
"https://github.com/f/awesome-chatgpt-prompts/raw/main/prompts.csv"
"URL for crowdsourced LLM system prompts.")
(defvar gptel--crowdsourced-prompts
(make-hash-table :test #'equal)
"Crowdsourced LLM system prompts.")
(defun gptel--crowdsourced-prompts ()
"Acquire and read crowdsourced LLM system prompts.
These are stored in the variable `gptel--crowdsourced-prompts',
which see."
(when (hash-table-p gptel--crowdsourced-prompts)
(when (hash-table-empty-p gptel--crowdsourced-prompts)
(unless gptel-crowdsourced-prompts-file
(run-at-time 0 nil #'gptel-system-prompt)
(user-error "No crowdsourced prompts available"))
(unless (and (file-exists-p gptel-crowdsourced-prompts-file)
(time-less-p
(time-subtract (current-time) (days-to-time 14))
(file-attribute-modification-time
(file-attributes gptel-crowdsourced-prompts-file))))
(when (y-or-n-p
(concat
"Fetch crowdsourced system prompts from "
(propertize "https://github.com/f/awesome-chatgpt-prompts" 'face 'link)
"?"))
;; Fetch file
(message "Fetching prompts...")
(let ((dir (file-name-directory gptel-crowdsourced-prompts-file)))
(unless (file-exists-p dir) (mkdir dir 'create-parents))
(if (url-copy-file gptel--crowdsourced-prompts-url
gptel-crowdsourced-prompts-file
'ok-if-already-exists)
(message "Fetching prompts... done.")
(message "Could not retrieve new prompts.")))))
(if (not (file-readable-p gptel-crowdsourced-prompts-file))
(progn (message "No crowdsourced prompts available")
(call-interactively #'gptel-system-prompt))
(with-temp-buffer
(insert-file-contents gptel-crowdsourced-prompts-file)
(goto-char (point-min))
(forward-line 1)
(while (not (eobp))
(when-let ((act (read (current-buffer))))
(forward-char)
(save-excursion
(while (re-search-forward "\"\"" (line-end-position) t)
(replace-match "\\\\\"")))
(when-let ((prompt (read (current-buffer))))
(puthash act prompt gptel--crowdsourced-prompts)))
(forward-line 1)))))
gptel--crowdsourced-prompts))
;; * Transient Prefixes
(define-obsolete-function-alias 'gptel-send-menu 'gptel-menu "0.3.2")
;; BUG: The `:incompatible' spec doesn't work if there's a `:description' below it.
;;;###autoload (autoload 'gptel-menu "gptel-transient" nil t)
(transient-define-prefix gptel-menu ()
"Change parameters of prompt to send to the LLM."
;; :incompatible '(("-m" "-n" "-k" "-e"))
[:description
(lambda () (format "Directive: %s"
(truncate-string-to-width
gptel--system-message (max (- (window-width) 14) 20) nil nil t)))
("h" "Set directives for chat" gptel-system-prompt :transient t)]
[["Session Parameters"
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(gptel--infix-provider)
;; (gptel--infix-model)
(gptel--infix-max-tokens)
(gptel--infix-num-messages-to-send)
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(gptel--infix-temperature)]
["Prompt from"
("p" "Minibuffer instead" "p")
("y" "Kill-ring instead" "y")
""
("i" "Replace/Delete prompt" "i")]
["Response to"
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
("m" "Minibuffer instead" "m")
("g" "gptel session" "g"
:class transient-option
:prompt "Existing or new gptel session: "
:reader
(lambda (prompt _ _history)
(read-buffer
prompt (generate-new-buffer-name
(concat "*" (gptel-backend-name gptel-backend) "*"))
nil (lambda (buf-name)
(if (consp buf-name) (setq buf-name (car buf-name)))
(let ((buf (get-buffer buf-name)))
(and (buffer-local-value 'gptel-mode buf)
(not (eq (current-buffer) buf))))))))
("b" "Any buffer" "b"
:class transient-option
:prompt "Output to buffer: "
:reader
(lambda (prompt _ _history)
(read-buffer prompt (buffer-name (other-buffer)) nil)))
("k" "Kill-ring" "k")]]
[["Send"
(gptel--suffix-send)
("M-RET" "Regenerate" gptel--regenerate :if gptel--in-response-p)]
[:description gptel--refactor-or-rewrite
:if use-region-p
("r"
;;FIXME: Transient complains if I use `gptel--refactor-or-rewrite' here. It
;;reads this function as a suffix instead of a function that returns the
;;description.
(lambda () (if (derived-mode-p 'prog-mode)
"Refactor" "Rewrite"))
gptel-rewrite-menu)]
["Tweak Response" :if gptel--in-response-p :pad-keys t
("SPC" "Mark" gptel--mark-response)
("P" "Previous variant" gptel--previous-variant
:if gptel--at-response-history-p
:transient t)
("N" "Next variant" gptel--previous-variant
:if gptel--at-response-history-p
:transient t)
("E" "Ediff previous" gptel--ediff
:if gptel--at-response-history-p)]]
(interactive)
(gptel--sanitize-model)
(transient-setup 'gptel-menu))
;; ** Prefix for setting the system prompt.
(defun gptel-system-prompt--setup (_)
"Set up suffixes for system prompt."
(transient-parse-suffixes
'gptel-system-prompt
(cl-loop for (type . prompt) in gptel-directives
with taken
with width = (window-width)
for name = (symbol-name type)
for key =
(let ((idx 0) pos)
(while (or (not pos) (member pos taken))
(setq pos (substring name idx (1+ idx)))
(cl-incf idx))
(push pos taken)
pos)
;; The explicit declaration ":transient transient--do-return" here
;; appears to be required for Transient v0.5 and up. Without it, these
;; are treated as suffixes when invoking `gptel-system-prompt' directly,
;; and infixes when going through `gptel-menu'.
;; TODO: Raise an issue with Transient.
collect (list (key-description key)
(concat (capitalize name) " "
(propertize " " 'display '(space :align-to 20))
(propertize
(concat
"("
(string-replace
"\n" " "
(truncate-string-to-width prompt (- width 30) nil nil t))
")")
'face 'shadow))
`(lambda () (interactive)
(message "Directive: %s" ,prompt)
(setq gptel--system-message ,prompt))
:transient 'transient--do-return)
into prompt-suffixes
finally return
(nconc
(list (list 'gptel--suffix-system-message))
prompt-suffixes
(list (list "SPC" "Pick crowdsourced prompt"
'gptel--read-crowdsourced-prompt
;; NOTE: Quitting the completing read when picking a
;; crowdsourced prompt will cause the transient to exit
;; instead of returning to the system prompt menu.
:transient 'transient--do-exit))))))
(transient-define-prefix gptel-system-prompt ()
"Change the LLM system prompt.
The \"system\" prompt establishes directives for the chat
session. Some examples of system prompts are:
You are a helpful assistant. Answer as concisely as possible.
Reply only with shell commands and no prose.
You are a poet. Reply only in verse.
Customize `gptel-directives' for task-specific prompts."
[:description
(lambda () (format "Current directive: %s"
(truncate-string-to-width
gptel--system-message 100 nil nil t)))
:class transient-column
:setup-children gptel-system-prompt--setup
:pad-keys t])
;; ** Prefix for rewriting/refactoring
(transient-define-prefix gptel-rewrite-menu ()
"Rewrite or refactor text region using an LLM."
[:description
(lambda ()
(format "Directive: %s"
(truncate-string-to-width
(or gptel--rewrite-message (gptel--rewrite-message))
(max (- (window-width) 14) 20) nil nil t)))
(gptel--infix-rewrite-prompt)]
[[:description "Diff Options"
("-w" "Wordwise diff" "-w")]
[:description
(lambda () (if (derived-mode-p 'prog-mode)
"Refactor" "Rewrite"))
(gptel--suffix-rewrite)
(gptel--suffix-rewrite-and-replace)
(gptel--suffix-rewrite-and-ediff)]]
(interactive)
(unless gptel--rewrite-message
(setq gptel--rewrite-message (gptel--rewrite-message)))
(transient-setup 'gptel-rewrite-menu))
;; * Transient Infixes
;; ** Infixes for model parameters
(defun gptel--transient-read-variable (prompt initial-input history)
"Read value from minibuffer and interpret the result as a Lisp object.
PROMPT, INITIAL-INPUT and HISTORY are as in the Transient reader
documention."
(ignore-errors
(read-from-minibuffer prompt initial-input read-expression-map t history)))
(transient-define-infix gptel--infix-num-messages-to-send ()
"Number of recent messages to send with each exchange.
By default, the full conversation history is sent with every new
prompt. This retains the full context of the conversation, but
can be expensive in token size. Set how many recent messages to
include."
:description "Number of past messages to send"
:class 'transient-lisp-variable
:variable 'gptel--num-messages-to-send
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
:key "-n"
:prompt "Number of past messages to include for context (leave empty for all): "
:reader 'gptel--transient-read-variable)
(transient-define-infix gptel--infix-max-tokens ()
"Max tokens per response.
This is roughly the number of words in the response. 100-300 is a
reasonable range for short answers, 400 or more for longer
responses."
:description "Response length (tokens)"
:class 'transient-lisp-variable
:variable 'gptel-max-tokens
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
:key "-c"
:prompt "Response length in tokens (leave empty: default, 80-200: short, 200-500: long): "
:reader 'gptel--transient-read-variable)
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(defclass gptel-provider-variable (transient-lisp-variable)
((model :initarg :model)
(model-value :initarg :model-value)
(always-read :initform t)
(set-value :initarg :set-value :initform #'set))
"Class used for gptel-backends.")
(cl-defmethod transient-format-value ((obj gptel-provider-variable))
(propertize (concat (gptel-backend-name (oref obj value)) ":"
(buffer-local-value (oref obj model) transient--original-buffer))
'face 'transient-value))
(cl-defmethod transient-infix-set ((obj gptel-provider-variable) value)
(pcase-let ((`(,backend-value ,model-value) value))
(funcall (oref obj set-value)
(oref obj variable)
(oset obj value backend-value))
(funcall (oref obj set-value)
(oref obj model)
(oset obj model-value model-value))))
(transient-define-infix gptel--infix-provider ()
"AI Provider for Chat."
:description "GPT Model: "
:class 'gptel-provider-variable
:prompt "Model provider: "
:variable 'gptel-backend
:model 'gptel-model
:key "-m"
:reader (lambda (prompt &rest _)
(let* ((backend-name
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(if (<= (length gptel--known-backends) 1)
(caar gptel--known-backends)
(completing-read
prompt
(mapcar #'car gptel--known-backends))))
(backend (alist-get backend-name gptel--known-backends
nil nil #'equal))
(backend-models (gptel-backend-models backend))
(model-name (if (= (length backend-models) 1)
(car backend-models)
(completing-read
"Model: " backend-models))))
(list backend model-name))))
(transient-define-infix gptel--infix-model ()
"AI Model for Chat."
:description "GPT Model: "
:class 'transient-lisp-variable
:variable 'gptel-model
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
:key "-m"
:choices '("gpt-3.5-turbo" "gpt-3.5-turbo-16k" "gpt-4" "gpt-4-1106-preview")
:reader (lambda (prompt &rest _)
(completing-read
prompt
'("gpt-3.5-turbo" "gpt-3.5-turbo-16k" "gpt-4" "gpt-4-1106-preview"))))
(transient-define-infix gptel--infix-temperature ()
"Temperature of request."
:description "Randomness (0 - 2.0)"
:class 'transient-lisp-variable
:variable 'gptel-temperature
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
:key "-t"
:prompt "Set temperature (0.0-2.0, leave empty for default): "
:reader 'gptel--transient-read-variable)
;; ** Infix for the refactor/rewrite system message
(transient-define-infix gptel--infix-rewrite-prompt ()
"Chat directive (system message) to use for rewriting or refactoring."
:description (lambda () (if (derived-mode-p 'prog-mode)
"Set directives for refactor"
"Set directives for rewrite"))
:format "%k %d"
:class 'transient-lisp-variable
:variable 'gptel--rewrite-message
:key "h"
:prompt "Set directive for rewrite: "
:reader (lambda (prompt _ history)
(read-string
prompt (gptel--rewrite-message) history)))
;; * Transient Suffixes
;; ** Suffix to send prompt
(transient-define-suffix gptel--suffix-send (args)
"Send ARGS."
:key "RET"
:description "Send prompt"
(interactive (list (transient-args transient-current-command)))
(let ((stream gptel-stream)
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(in-place (and (member "i" args) t))
(output-to-other-buffer-p)
(backend gptel-backend)
(model gptel-model)
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(backend-name (gptel-backend-name gptel-backend))
(buffer) (position)
(callback) (gptel-buffer-name)
;; Input redirection: grab prompt from elsewhere?
(prompt
(cond
((member "p" args)
(read-string
(format "Ask %s: " (gptel-backend-name gptel-backend))
(apply #'buffer-substring-no-properties
(if (use-region-p)
(list (region-beginning) (region-end))
(list (line-beginning-position) (line-end-position))))))
((member "y" args)
(unless (car-safe kill-ring)
(user-error "`kill-ring' is empty! Nothing to send"))
(if current-prefix-arg
(read-from-kill-ring "Prompt from kill-ring: ")
(current-kill 0))))))
;; Output redirection: Send response elsewhere?
(cond
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
((member "m" args)
(setq stream nil)
(setq callback
(lambda (resp info)
(if resp
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(message "%s response: %s" backend-name resp)
(message "%s response error: %s" backend-name (plist-get info :status))))))
((member "k" args)
(setq stream nil)
(setq callback
(lambda (resp info)
(if (not resp)
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(message "%s response error: %s" backend-name (plist-get info :status))
(kill-new resp)
(message "%s response: \"%s\" copied to kill-ring."
backend-name
(truncate-string-to-width resp 30))))))
((setq gptel-buffer-name
(cl-some (lambda (s) (and (string-prefix-p "g" s)
gptel: Add multi-llm support README.org: Update README with new information and a multi-llm demo. gptel.el (gptel-host, gptel--known-backends, gptel--api-key, gptel--create-prompt, gptel--request-data, gptel--parse-buffer, gptel-request, gptel--parse-response, gptel--openai, gptel--debug, gptel--restore-state, gptel, gptel-backend): Integrate multiple LLMs through the introcution of gptel-backends. Each backend is composed of two pieces: 1. An instance of a cl-struct, containing connection, authentication and model information. See the cl-struct `gptel-backend` for details. A separate cl-struct type is defined for each supported backend (OpenAI, Azure, GPT4All and Ollama) that inherits from the generic gptel-backend type. 2. cl-generic implementations of specific tasks, like gathering up and formatting context (previous user queries and LLM responses), parsing responses or responses streams etc. The four tasks currently specialized this way are carried out by `gptel--parse-buffer` and `gptel--request-data` (for constructing the query) and `gptel--parse-response` and `gptel-curl--parse-stream` (for parsing the response). See their implementations for details. Some effort has been made to limit the number of times dispatching is done when reading streaming responses. When a backend is created, it is registered in the collection `gptel--known-backends` and can be accessed by name later, such as from the transient menu. Only one of these backends is active at any time in a buffer, stored in the buffer-local variable `gptel-backend`. Most messaging, authentication etc accounts for the active backend, although there might be some leftovers. When using `gptel-request` or `gptel-send`, the active backend can be changed or let-bound. - Obsolete `gptel-host` - Fix the rear-sticky property when restoring sessions from files. - Document some variables (not user options), like `gptel--debug` gptel-openai.el (gptel-backend, gptel-make-openai, gptel-make-azure, gptel-make-gpt4all): This file (currently always loaded) sets up the generic backend struct and includes constructors for creating OpenAI, GPT4All and Azure backends. They all use the same API so a single set of defgeneric implemenations suffices for all of them. gptel-ollama.el (gptel-make-ollama): This file includes the cl-struct, constructor and requisite defgeneric implementations for Ollama support. gptel-transient.el (gptel-menu, gptel-provider-variable, gptel--infix-provider, gptel-suffix-send): - Provide access to all available LLM backends and models from `gptel-menu`. - Adjust keybindings in gptel-menu: setting the model and query parameters is now bound to two char keybinds, while redirecting input and output is bound to single keys.
2023-10-23 23:40:52 -07:00
(substring s 1)))
args))
(setq output-to-other-buffer-p t)
(let ((reduced-prompt ;For inserting into the gptel buffer as
;context, not the prompt used for the
;request itself
(or prompt
(if (use-region-p)
(buffer-substring-no-properties (region-beginning)
(region-end))
(buffer-substring-no-properties
(save-excursion
(text-property-search-backward
'gptel 'response
(when (get-char-property (max (point-min) (1- (point)))
'gptel)
t))
(point))
(gptel--at-word-end (point)))))))
(cond
((buffer-live-p (get-buffer gptel-buffer-name))
;; Insert into existing gptel session
(progn
(setq buffer (get-buffer gptel-buffer-name))
(with-current-buffer buffer
(goto-char (point-max))
(unless (or buffer-read-only
(get-char-property (point) 'read-only))
(insert reduced-prompt))
(setq position (point))
(when gptel-mode
(gptel--update-status " Waiting..." 'warning)))))
;; Insert into new gptel session
(t (setq buffer
(gptel gptel-buffer-name
(condition-case nil
(gptel--get-api-key)
((error user-error)
(setq gptel-api-key
(read-passwd
(format "%s API key: "
(gptel-backend-name
gptel-backend))))))
reduced-prompt))
;; Set backend and model in new session from current buffer
(with-current-buffer buffer
(setq gptel-backend backend)
(setq gptel-model model)
(gptel--update-status " Waiting..." 'warning)
(setq position (point)))))))
((setq gptel-buffer-name
(cl-some (lambda (s) (and (string-prefix-p "b" s)
(substring s 1)))
args))
(setq output-to-other-buffer-p t)
(setq buffer (get-buffer-create gptel-buffer-name))
(with-current-buffer buffer (setq position (point)))))
;; Create prompt, unless doing input-redirection above
(unless prompt
(setq prompt (gptel--create-prompt (gptel--at-word-end (point)))))
(when in-place
;; Kill the latest prompt
(let ((beg
(if (use-region-p)
(region-beginning)
(save-excursion
(text-property-search-backward
'gptel 'response
(when (get-char-property (max (point-min) (1- (point)))
'gptel)
t))
(point))))
(end (if (use-region-p) (region-end) (point))))
(unless output-to-other-buffer-p
;; store the killed text in gptel-history
(gptel--attach-response-history
(list (buffer-substring-no-properties beg end))))
(kill-region beg end)))
(gptel-request
prompt
:buffer (or buffer (current-buffer))
:position position
:in-place (and in-place (not output-to-other-buffer-p))
:stream stream
:callback callback)
(when output-to-other-buffer-p
(message (concat "Prompt sent to buffer: "
(propertize gptel-buffer-name 'face 'help-key-binding)))
(display-buffer
buffer '((display-buffer-reuse-window
display-buffer-pop-up-window)
(reusable-frames . visible))))))
;; ** Suffix to regenerate response
(defun gptel--regenerate ()
"Regenerate gptel response at point."
(interactive)
(when (gptel--in-response-p)
(pcase-let* ((`(,beg . ,end) (gptel--get-bounds))
(history (get-char-property (point) 'gptel-history))
(prev-responses (cons (buffer-substring-no-properties beg end)
history)))
(when gptel-mode ;Remove prefix/suffix
(save-excursion
(goto-char beg)
(when (looking-back (concat "\n+" (regexp-quote (gptel-response-prefix-string)))
(point-min) 'greedy)
(setq beg (match-beginning 0)))
(goto-char end)
(when (looking-at
(concat "\n+" (regexp-quote (gptel-prompt-prefix-string))))
(setq end (match-end 0)))))
(delete-region beg end)
(gptel--attach-response-history prev-responses)
(call-interactively #'gptel--suffix-send))))
;; ** Set system message
(defun gptel--read-crowdsourced-prompt ()
"Pick a crowdsourced system prompt for gptel.
This uses the prompts in the variable
`gptel--crowdsourced-prompts', which see."
(interactive)
(if (not (hash-table-empty-p (gptel--crowdsourced-prompts)))
(let ((choice
(completing-read
"Pick and edit prompt: "
(lambda (str pred action)
(if (eq action 'metadata)
`(metadata
(affixation-function .
(lambda (cands)
(mapcar
(lambda (c)
(list c ""
(concat (propertize " " 'display '(space :align-to 22))
" " (propertize (gethash c gptel--crowdsourced-prompts)
'face 'completions-annotations))))
cands))))
(complete-with-action action gptel--crowdsourced-prompts str pred)))
nil t)))
(when-let ((prompt (gethash choice gptel--crowdsourced-prompts)))
(setq gptel--system-message prompt)
(call-interactively #'gptel--suffix-system-message)))
(message "No prompts available.")))
(transient-define-suffix gptel--suffix-system-message ()
"Edit LLM directives."
:transient 'transient--do-exit
:description "Set custom directives"
:key "h"
(interactive)
(let ((orig-buf (current-buffer))
(msg-start (make-marker)))
(with-current-buffer (get-buffer-create "*gptel-system*")
(let ((inhibit-read-only t))
(erase-buffer)
(text-mode)
(setq header-line-format
(concat
"Edit your system message below and press "
(propertize "C-c C-c" 'face 'help-key-binding)
" when ready, or "
(propertize "C-c C-k" 'face 'help-key-binding)
" to abort."))
(insert
"# Example: You are a helpful assistant. Answer as concisely as possible.\n"
"# Example: Reply only with shell commands and no prose.\n"
"# Example: You are a poet. Reply only in verse.\n\n")
(add-text-properties
(point-min) (1- (point))
(list 'read-only t 'face 'font-lock-comment-face))
;; TODO: make-separator-line requires Emacs 28.1+.
;; (insert (propertize (make-separator-line) 'rear-nonsticky t))
(set-marker msg-start (point))
(save-excursion
(insert (buffer-local-value 'gptel--system-message orig-buf))
(push-mark nil 'nomsg))
(activate-mark))
(display-buffer (current-buffer)
`((display-buffer-below-selected)
(body-function . ,#'select-window)
(window-height . ,#'fit-window-to-buffer)))
(let ((quit-to-menu
(lambda ()
(interactive)
(local-unset-key (kbd "C-c C-c"))
(local-unset-key (kbd "C-c C-k"))
(quit-window)
(display-buffer
orig-buf
`((display-buffer-reuse-window
display-buffer-use-some-window)
(body-function . ,#'select-window)))
(call-interactively #'gptel-menu))))
(local-set-key (kbd "C-c C-c")
(lambda ()
(interactive)
(let ((system-message
(buffer-substring msg-start (point-max))))
(with-current-buffer orig-buf
(setq gptel--system-message system-message)))
(funcall quit-to-menu)))
(local-set-key (kbd "C-c C-k") quit-to-menu)))))
;; ** Suffixes for rewriting/refactoring
(transient-define-suffix gptel--suffix-rewrite ()
"Rewrite or refactor region contents."
:key "r"
:description #'gptel--refactor-or-rewrite
(interactive)
(let* ((prompt (buffer-substring-no-properties
(region-beginning) (region-end)))
(stream gptel-stream)
(gptel--system-message gptel--rewrite-message))
(gptel-request prompt :stream stream)))
(transient-define-suffix gptel--suffix-rewrite-and-replace ()
"Refactor region contents and replace it."
:key "R"
:description (lambda () (concat (gptel--refactor-or-rewrite) " in place"))
(interactive)
(let* ((prompt (buffer-substring-no-properties
(region-beginning) (region-end)))
(stream gptel-stream)
(gptel--system-message gptel--rewrite-message))
(kill-region (region-beginning) (region-end))
(message "Original text saved to kill-ring.")
(gptel-request prompt :stream stream :in-place t)))
(transient-define-suffix gptel--suffix-rewrite-and-ediff (args)
"Refactoring or rewrite region contents and run Ediff."
:key "E"
:description (lambda () (concat (gptel--refactor-or-rewrite) " and Ediff"))
(interactive (list (transient-args transient-current-command)))
(letrec ((prompt (buffer-substring-no-properties
(region-beginning) (region-end)))
(gptel--system-message gptel--rewrite-message)
;; TODO: Technically we should save the window config at the time
;; `ediff-setup-hook' runs, but this will do for now.
(cwc (current-window-configuration))
(gptel--ediff-restore
(lambda ()
(when (window-configuration-p cwc)
(set-window-configuration cwc))
(remove-hook 'ediff-quit-hook gptel--ediff-restore))))
(message "Waiting for response... ")
(gptel-request
prompt
:context (cons (region-beginning) (region-end))
:callback
(lambda (response info)
(if (not response)
(message "ChatGPT response error: %s" (plist-get info :status))
(let* ((gptel-buffer (plist-get info :buffer))
(gptel-bounds (plist-get info :context))
(buffer-mode
(buffer-local-value 'major-mode gptel-buffer)))
(pcase-let ((`(,new-buf ,new-beg ,new-end)
(with-current-buffer (get-buffer-create "*gptel-rewrite-Region.B-*")
(let ((inhibit-read-only t))
(erase-buffer)
(funcall buffer-mode)
(insert response)
(goto-char (point-min))
(list (current-buffer) (point-min) (point-max))))))
(require 'ediff)
(add-hook 'ediff-quit-hook gptel--ediff-restore)
(apply
#'ediff-regions-internal
(get-buffer (ediff-make-cloned-buffer gptel-buffer "-Region.A-"))
(car gptel-bounds) (cdr gptel-bounds)
new-buf new-beg new-end
nil
(if (transient-arg-value "-w" args)
(list 'ediff-regions-wordwise 'word-wise nil)
(list 'ediff-regions-linewise nil nil))))))))))
(provide 'gptel-transient)
;;; gptel-transient.el ends here
;; Local Variables:
;; outline-regexp: "^;; \\*+"
;; eval: (outline-minor-mode 1)
;; End: