[wip] audio input option

MichelNivard · Aug 23, 2024 · f302d96 · f302d96
1 parent 78c3078
commit f302d96
Show file tree

Hide file tree

Showing 7 changed files with 65 additions and 30 deletions.
diff --git a/R/app_config.R b/R/app_config.R
@@ -6,7 +6,8 @@ save_user_config <- function(code_style,
                              model,
                              custom_prompt,
                              stream,
-                             read_docs) {
+                             read_docs,
+                             audio_input) {
   if (is.null(custom_prompt)) custom_prompt <- ""
   config <-
     data.frame(
@@ -18,7 +19,8 @@ save_user_config <- function(code_style,
       model,
       custom_prompt,
       stream,
-      read_docs
+      read_docs,
+      audio_input
     )
   user_config_path <- tools::R_user_dir("gptstudio", which = "config")
   user_config <- file.path(user_config_path, "config.yml")
@@ -40,7 +42,8 @@ set_user_options <- function(config) {
     gptstudio.custom_prompt = config$custom_prompt,
     gptstudio.stream        = config$stream,
     # added in v.3.1+ dev version
-    gptstudio.read_docs     = config$read_docs
+    gptstudio.read_docs     = config$read_docs,
+    gptstudio.audio_input   = config$audio_input
   )
   options(op_gptstudio)
   invisible()

diff --git a/R/mod_chat.R b/R/mod_chat.R
@@ -29,27 +29,7 @@ mod_chat_ui <- function(id, translator = create_translator()) {
             style = css(
               "width" = "100%"
             ),
-            div(
-              text_area_input_wrapper(
-                inputId = ns("chat_input"),
-                label = NULL,
-                width = "100%",
-                placeholder = translator$t("Write your prompt here"),
-                value = "",
-                resize = "none",
-                textarea_class = "chat-prompt"
-              )
-            ),
-            div(
-              class = "position-absolute top-50 end-0 translate-middle",
-              bslib::input_task_button(
-                id = ns("chat"),
-                label = icon("fas fa-paper-plane"),
-                label_busy = NULL,
-                class = "w-100 btn-primary p-1 chat-send-btn"
-              ) %>%
-                bslib::tooltip("Send (click or Enter)")
-            )
+            uiOutput(ns("chat_input"))
           )
         )
       )
@@ -144,5 +124,46 @@ mod_chat_server <- function(id,
       updateTextAreaInput(session, "chat_input", value = "")
     }) %>%
       bindEvent(input$chat)
+
+    output$chat_input <- renderUI({
+      tagList(
+        fluidRow(
+          column(
+            width = 3,
+            div(
+              style = "display: flex; align-items: center; height: 100%;",
+              input_audio_clip("clip", show_mic_settings = FALSE)
+            )
+          ),
+          column(
+            width = 9,
+            div(
+              style = "display: flex; align-items: center; position: relative; height: 100%;",
+              div(
+                style = "flex-grow: 1;",
+                text_area_input_wrapper(
+                  inputId = "chat_input",
+                  label = NULL,
+                  width = "100%",
+                  value = "",
+                  resize = "none",
+                  textarea_class = "chat-prompt"
+                )
+              ),
+              div(
+                style = "position: absolute; right: 10px;",
+                bslib::input_task_button(
+                  id = "chat",
+                  label = icon("fas fa-paper-plane"),
+                  label_busy = NULL,
+                  class = "w-100 btn-primary p-1 chat-send-btn"
+                ) %>%
+                  bslib::tooltip("Send (click or Enter)")
+              )
+            )
+          )
+        )
+      )
+    })
   })
 }
diff --git a/R/mod_settings.R b/R/mod_settings.R
@@ -77,6 +77,12 @@ mod_settings_ui <- function(id, translator = create_translator()) {
         label = "Stream Response",
         value = as.logical(getOption("gptstudio.stream")),
         width = "100%"
+      ),
+      bslib::input_switch(
+        id = ns("audio_input"),
+        label = "Audio as Input",
+        value = as.logical(getOption("gptstudio.audio_input")),
+        width = "100%"
       )
     ),
     bslib::accordion_panel(
@@ -130,6 +136,7 @@ mod_settings_server <- function(id) {
     rv$selected_history <- 0L
     rv$modify_session_settings <- 0L
     rv$create_new_chat <- 0L
+    rv$record_input <- 0L
 
     observe({
       msg <- glue::glue("Fetching models for {input$service} service...")
@@ -217,7 +224,8 @@ mod_settings_server <- function(id) {
         model = input$model,
         custom_prompt = input$custom_prompt,
         stream = input$stream,
-        read_docs = input$read_docs
+        read_docs = input$read_docs,
+        audio_input = input$audio_input
       )
 
       rv$modify_session_settings <- rv$modify_session_settings + 1L
@@ -259,6 +267,7 @@ mod_settings_server <- function(id) {
       rv$service <- input$service %||% getOption("gptstudio.service")
       rv$stream <- as.logical(input$stream %||% getOption("gptstudio.stream"))
       rv$custom_prompt <- input$custom_prompt %||% getOption("gptstudio.custom_prompt")
+      rv$audio_input <- input$audio_input %||% getOption("gptstudio.audio_input")
 
       rv$create_new_chat <- rv$create_new_chat + 1L
     }) %>%

diff --git a/R/record-audio.R b/R/record-audio.R
@@ -117,7 +117,7 @@ multimodal_dep <- function() {
     name = "multimodal",
     version = "0.0.1",
     package = "shinymedia",
-    src = "dist",
+    src = system.file("dist", package = "gptstudio"),
     script = "index.js",
     stylesheet = "index.css"
   )

diff --git a/R/zzz.R b/R/zzz.R
@@ -27,7 +27,8 @@
     gptstudio.stream        = config$stream,
     # options added after v3.0 will need a safe check because the user's
     # config file might not have values for new features
-    gptstudio.read_docs     = config$read_docs %||% FALSE
+    gptstudio.read_docs     = config$read_docs %||% FALSE,
+    gptstudio.audio_input   = config$audio_input %||% FALSE
   )
 
   toset <- !(names(op_gptstudio) %in% names(op))

diff --git a/inst/rstudio/config.yml b/inst/rstudio/config.yml
@@ -7,3 +7,4 @@ model: "gpt-4-turbo-preview"
 custom_prompt: null
 stream: true
 read_docs: false
+audio_input: false
diff --git a/tests/testthat/_snaps/models.md b/tests/testthat/_snaps/models.md
@@ -15,9 +15,9 @@
     Code
       models
     Output
-      [1] "command-r"             "command-nightly"       "command-r-plus"       
-      [4] "c4ai-aya-23-35b"       "command-light-nightly" "c4ai-aya-23-8b"       
-      [7] "command"               "command-light"        
+      [1] "c4ai-aya-23-35b"       "command-r"             "command-nightly"      
+      [4] "command-r-plus"        "command-light-nightly" "command"              
+      [7] "command-r-08-2024"     "command-light"         "c4ai-aya-23-8b"       
 
 # get_available_models works for google