Merge branch 'main' into batch-embeddings

NEWS.org

@@ -1,5 +1,6 @@
 * Version 0.18.0
 - Add batch embeddings capability (currently for just Open AI and Ollama).
+- Add Microsoft Azure's Open AI
 - Remove testing and other development files from ELPA packaging.
 - Remove vendored =plz-event-source= and =plz-media-type=, and add requirements.
 - Update list of Ollama models for function calling.

README.org

@@ -33,6 +33,9 @@ You can set up with ~make-llm-openai~, with the following parameters:
 ** Open AI Compatible
 There are many Open AI compatible APIs and proxies of Open AI.  You can set up one with ~make-llm-openai-compatible~, with the following parameter:
 - ~:url~, the URL of leading up to the command ("embeddings" or "chat/completions").  So, for example, "https://api.openai.com/v1/" is the URL to use Open AI (although if you wanted to do that, just use ~make-llm-openai~ instead.
+** Azure's Open AI
+Microsoft Azure has an Open AI integration, although it doesn't support everything Open AI does, such as function calling.  You can set it up with ~make-llm-azure~, with the following parameter:
+- ~:url~, the endpoint URL, such as "https://docs-test-001.openai.azure.com/".
 ** Gemini (not via Google Cloud)
 This is Google's AI model.  You can get an API key via their [[https://makersuite.google.com/app/apikey][page on Google AI Studio]].
 Set this up with ~make-llm-gemini~, with the following parameters:

llm-azure.el

@@ -0,0 +1,58 @@
+;;; llm-azure.el --- llm module for integrating with Azure's Open AI service -*- lexical-binding: t; package-lint-main-file: "llm.el"; byte-compile-docstring-max-column: 200-*-
+
+;; Copyright (c) 2024  Free Software Foundation, Inc.
+
+;; Author: Andrew Hyatt <[email protected]>
+;; Homepage: https://github.com/ahyatt/llm
+;; 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 GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.
+
+;;; Commentary:
+;; This file implements the llm functionality defined in llm.el, for Azure's
+;; Open AI service.
+
+;;; Code:
+
+(require 'llm)
+(require 'llm-openai)
+(require 'cl-lib)
+
+(cl-defstruct (llm-azure (:include llm-openai-compatible)))
+
+(cl-defmethod llm-nonfree-message-info ((_ llm-azure))
+  "Return Azure's nonfree terms of service."
+  "https://azure.microsoft.com/en-us/support/legal/")
+
+(cl-defmethod llm-provider-chat-url ((provider llm-azure))
+  (format "%s/openai/deployments/%s/chat/completions?api-version=2024-08-01-preview"
+          (llm-azure-url provider)
+          (llm-azure-chat-model provider)))
+
+(cl-defmethod llm-provider-embedding-url ((provider llm-azure))
+  (format "%s/openai/deployments/%s/embeddings?api-version=2024-08-01-preview"
+          (llm-azure-url provider)
+          (llm-azure-embedding-model provider)))
+
+(cl-defmethod llm-provider-headers ((provider llm-azure))
+  `(("api-key" . ,(llm-azure-key provider))))
+
+(cl-defmethod llm-capabilities ((_ llm-azure))
+  (list 'streaming 'embedding))
+
+(cl-defmethod llm-name ((provider llm-azure))
+  (format "Azure OpenAI %s" (llm-azure-chat-model provider)))
+
+(provide 'llm-azure)
+;;; llm-azure.el ends here

llm-integration-test.el

@@ -31,8 +31,15 @@
 ;; - GEMINI_KEY: A Gemini API key.
 ;; - VERTEX_PROJECT: A Google Cloud Vertex project.
 ;; - OLLAMA_CHAT_MODELS: A list of Ollama models to test.
+;; - AZURE_URL: The URL of the Azure API.
+;; - AZURE_KEY: The key for the Azure API.
+;; - AZURE_CHAT_MODEL: The name of the chat model to test.
+;; - AZURE_EMBEDDING_MODEL: The name of the embedding model to test.
+;; - AZURE_SLEEP: The number of seconds to sleep between tests, to avoid rate
+;;   limiting.
 ;;
-;; If any of these are set, the corresponding provider will be tested.
+;; If any of these are set (except for Azure, which needs multiple), the
+;; corresponding provider will be tested.
 
 
 ;;; Code:
@@ -87,6 +94,11 @@
     ("capital_of_country" . "Italy"))
   "The correct answer to the function call prompt.")
 
+(defun llm-integration-test-rate-limit (provider)
+  (cond ((eq (type-of provider) 'llm-azure)
+         ;; The free Azure tier has extremely restrictive rate limiting.
+         (sleep-for (string-to-number (or (getenv "AZURE_SLEEP") "60"))))))
+
 (defun llm-integration-test-providers ()
   "Return a list of providers to test."
   (let ((providers))
@@ -102,6 +114,14 @@
     (when (getenv "VERTEX_PROJECT")
       (require 'llm-vertex)
       (push (make-llm-vertex :project (getenv "VERTEX_PROJECT")) providers))
+    (when (and (getenv "AZURE_URL") (getenv "AZURE_KEY"))
+      (require 'llm-azure)
+      ;; Because Azure requires setting up each model, we can't just use any
+      ;; model it supports, but the model that the user running tests has set up.
+      (push (make-llm-azure :key (getenv "AZURE_KEY") :url (getenv "AZURE_URL")
+                            :chat-model (getenv "AZURE_CHAT_MODEL")
+                            :embedding-model (getenv "AZURE_EMBEDDING_MODEL"))
+            providers))
     (when (getenv "OLLAMA_CHAT_MODELS")
       (require 'llm-ollama)
       ;; This variable is a list of models to test.
@@ -113,9 +133,11 @@
   "Define an integration test."
   (declare (indent defun))
   `(ert-deftest ,name ()
-     (dolist (,(car arglist) (llm-integration-test-providers))
-       (ert-info ((format "Using provider %s" (llm-name provider)))
-         ,@body))))
+     (let ((llm-warn-on-nonfree nil))
+       (dolist (,(car arglist) (llm-integration-test-providers))
+         (llm-integration-test-rate-limit provider)
+         (ert-info ((format "Using provider %s" (llm-name provider)))
+           ,@body)))))
 
 (llm-def-integration-test llm-embedding (provider)
   (when (member 'embeddings (llm-capabilities provider))
@@ -179,17 +201,19 @@
 (llm-def-integration-test llm-chat-async (provider)
   (let ((result nil)
         (buf (current-buffer))
-        (llm-warn-on-nonfree nil))
+        (llm-warn-on-nonfree nil)
+        (err-result nil))
     (llm-chat-async
      provider
      (llm-make-chat-prompt llm-integration-test-chat-prompt)
      (lambda (response)
        (should (eq (current-buffer) buf))
        (setq result response))
-     (lambda (error)
-       (error "Error: %s" error)))
-    (while (null result)
+     (lambda (_ err)
+       (setq err-result err)))
+    (while (not (or result err-result))
       (sleep-for 0.1))
+    (if err-result (error err-result))
     (should (equal (string-trim result) llm-integration-test-chat-answer))))
 
 (llm-def-integration-test llm-chat-streaming (provider)
@@ -198,7 +222,8 @@
           (returned-result nil)
           (llm-warn-on-nonfree nil)
           (buf (current-buffer))
-          (start-time (current-time)))
+          (start-time (current-time))
+          (err-result nil))
       (llm-chat-streaming
        provider
        (llm-make-chat-prompt llm-integration-test-chat-prompt)
@@ -208,29 +233,33 @@
        (lambda (response)
          (should (eq (current-buffer) buf))
          (setq returned-result response))
-       (lambda (error)
-         (error "Error: %s" error)))
+       (lambda (_ err)
+         (setq err-result err)))
       (while (and (null returned-result)
+                  (null err-result)
                   (time-less-p (time-subtract (current-time) start-time) 10))
         (sleep-for 0.1))
+      (if err-result (error err-result))
       (should (equal (string-trim returned-result) llm-integration-test-chat-answer))
       (should (equal (string-trim streamed-result) llm-integration-test-chat-answer)))))
 
 (llm-def-integration-test llm-function-call (provider)
-  (let ((prompt (llm-integration-test-fc-prompt)))
-    (should (equal
-             (llm-chat provider prompt)
-             llm-integration-test-fc-answer))
-    ;; Test that we can send the function back to the provider without error.
-    (llm-chat provider prompt)))
+  (when (member 'function-calls (llm-capabilities provider))
+    (let ((prompt (llm-integration-test-fc-prompt)))
+      (should (equal
+               (llm-chat provider prompt)
+               llm-integration-test-fc-answer))
+      ;; Test that we can send the function back to the provider without error.
+      (llm-chat provider prompt))))
 
 (llm-def-integration-test llm-function-call-multiple (provider)
-  (let ((prompt (llm-integration-test-fc-multiple-prompt)))
-    ;; Sending back multiple answers often doesn't happen, so we can't reliably
-    ;; check for this yet.
-    (llm-chat provider prompt)
-    ;; Test that we can send the function back to the provider without error.
-    (llm-chat provider prompt)))
+  (when (member 'function-calls (llm-capabilities provider))
+    (let ((prompt (llm-integration-test-fc-multiple-prompt)))
+      ;; Sending back multiple answers often doesn't happen, so we can't reliably
+      ;; check for this yet.
+      (llm-chat provider prompt)
+      ;; Test that we can send the function back to the provider without error.
+      (llm-chat provider prompt))))
 
 (llm-def-integration-test llm-count-tokens (provider)
   (let ((result (llm-count-tokens provider "What is the capital of France?")))