different-issue-to-markdown aiders attempty

.Rbuildignore

@@ -1,3 +1,6 @@
+aider-instructions.md
+issue.md
+combined-instructions.md
 coverage.rds
 coverage-report.html
 cobertura.xml

.github/workflows/aider.yml

@@ -48,13 +48,23 @@ jobs:
       - name: Install Package
         run: make install
 
-      - name: Get Issue Content
-        id: get-issue
-        uses: actions-cool/issues-helper@v3
+      - name: Convert Issue to Markdown
+        uses: eunjae-lee/issue-to-markdown@v1
         with:
-          actions: 'get-issue'
           token: ${{ secrets.GITHUB_TOKEN }}
-          issue-number: ${{ github.event.inputs.issue_number }}
+          dest: 'tmp'
+          extension: '.md'
+          slug_as_folder_name: false
+          label: ''
+          inject_title: true
+          inject_created_at: false
+          inject_updated_at: false
+          authors: zachmayer
+
+      - name: Prepare Instructions
+        run: |
+          cat aider-instructions.md "tmp/${{ github.event.inputs.issue_number }}/index.md" > combined_instructions.md
+          rm -rf tmp/
 
       - name: Setup Python
         uses: actions/setup-python@v5
@@ -64,11 +74,6 @@ jobs:
       - name: Install Aider
         run: pip install -U aider-chat
 
-      - name: Prepare Instructions
-        run: |
-          printf '%s' "${{ steps.get-issue.outputs.issue-body }}" > issue.md
-          cat aider-instructions.md issue.md > combined_instructions.md
-
       - name: Run Aider
         run: |
           aider \
@@ -91,5 +96,3 @@ jobs:
           title: "fix: Automated fixes for issue #${{ github.event.inputs.issue_number }}"
           body: |
             This PR contains automated fixes for issue #${{ github.event.inputs.issue_number }} using aider.
-
-            Original issue: ${{ steps.get-issue.outputs.issue-title }}

R/caretList.R

@@ -37,7 +37,8 @@ caretList <- function(
     tuneList = NULL,
     metric = NULL,
     continue_on_fail = FALSE,
-    trim = TRUE) {
+    trim = TRUE,
+    sort_preds = TRUE) {
   # Checks
   if (is.null(tuneList) && is.null(methodList)) {
     stop("Please either define a methodList or tuneList", call. = FALSE)
@@ -79,7 +80,10 @@ caretList <- function(
   global_args[["metric"]] <- metric
 
   # Loop through the tuneLists and fit caret models with those specs
-  modelList <- lapply(tuneList, caretTrain, global_args = global_args, continue_on_fail = continue_on_fail, trim = trim)
+  modelList <- lapply(tuneList, caretTrain,
+    global_args = global_args,
+    continue_on_fail = continue_on_fail, trim = trim, sort_preds = sort_preds
+  )
   names(modelList) <- names(tuneList)
   nulls <- vapply(modelList, is.null, logical(1L))
   modelList <- modelList[!nulls]

R/caretPredict.R

@@ -82,7 +82,7 @@ caretPredict <- function(object, newdata = NULL, excluded_class_id = 1L, ...) {
 #' If `TRUE`, the function will remove some elements that are not needed from the output model.
 #' @return The output of the `train` function.
 #' @keywords internal
-caretTrain <- function(local_args, global_args, continue_on_fail = FALSE, trim = TRUE) {
+caretTrain <- function(local_args, global_args, continue_on_fail = FALSE, trim = TRUE, sort_preds = TRUE) {
   # Combine args
   # I think my handling here is correct (update globals with locals, which allows locals be partial)
   # but it would be nice to have some tests
@@ -100,7 +100,7 @@ caretTrain <- function(local_args, global_args, continue_on_fail = FALSE, trim =
 
   # Only save stacked predictions for the best model
   if ("pred" %in% names(model)) {
-    model[["pred"]] <- extractBestPreds(model)
+    model[["pred"]] <- extractBestPreds(model, sort_preds = sort_preds)
   }
 
   if (trim) {
@@ -147,9 +147,10 @@ aggregate_mean_or_first <- function(x) {
 #' @title Extract the best predictions from a train object
 #' @description Extract the best predictions from a train object.
 #' @param x a train object
+#' @param sort_preds logical, should predictions be sorted by rowIndex. Default TRUE.
 #' @return a data.table::data.table with predictions
 #' @keywords internal
-extractBestPreds <- function(x) {
+extractBestPreds <- function(x, sort_preds = TRUE) {
   stopifnot(methods::is(x, "train"))
   if (is.null(x$pred)) {
     stop("No predictions saved during training. Please set savePredictions = 'final' in trainControl", call. = FALSE)
@@ -173,8 +174,10 @@ extractBestPreds <- function(x) {
   data.table::setkeyv(pred, keys)
   pred <- pred[, lapply(.SD, aggregate_mean_or_first), by = keys]
 
-  # Order results consistently
-  data.table::setorderv(pred, keys)
+  # Order results consistently if requested
+  if (sort_preds) {
+    data.table::setorderv(pred, keys)
+  }
 
   # Return
   pred