Return more informative error from deduplication

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: orderly
 Title: Lightweight Reproducible Reporting
-Version: 1.4.4
+Version: 1.4.5
 Description: Order, create and store reports from R.  By defining a
     lightweight interface around the inputs and outputs of an
     analysis, a lot of the repetitive work for reproducible research

R/deduplicate.R

@@ -68,6 +68,11 @@ orderly_deduplicate <- function(root = NULL, locate = TRUE, dry_run = TRUE,
 
 
 orderly_deduplicate_info <- function(config) {
+  report_metadata <- list.files(path_metadata(config$root))
+  if (length(report_metadata) > 0) {
+    stop(paste("Cannot deduplicate archive reports have been pulled from",
+               "remote with recursive = FALSE."))
+  }
   con <- orderly_db("destination", config)
   on.exit(DBI::dbDisconnect(con))
 
@@ -98,8 +103,13 @@ orderly_deduplicate_info <- function(config) {
     unname(tapply(files$inode, files$hash, function(x) x[[1L]]))[i]
 
   ## Quick check:
-  stopifnot(all(vlapply(split(files, files$hash), function(x)
-    all(x$inode_first == x$inode[[1]]))))
+  can_deduplicate <- all(vlapply(split(files, files$hash), function(x) {
+    isTRUE(all(x$inode_first == x$inode[[1]]))
+  }))
+  if (!can_deduplicate) {
+    stop(paste("Cannot deduplicate archive as database references files",
+                "which don't exist."))
+  }
 
   ## Classify the files into different states
   files$state <- rep("distinct", nrow(files))

docker/build_website

@@ -22,6 +22,7 @@ DB_NAME=orderly
 SCHEMASPY_NAME=vimc/orderly-schemaspy
 
 DOCS_DIR=$PACKAGE_ROOT/docs
+VIGNETTES_DIR=$PACKAGE_ROOT/vignettes
 
 docker build \
        --build-arg ORDERLY_BASE=$ORDERLY_BASE \
@@ -65,6 +66,7 @@ docker run --rm \
        --network=$DB_NW \
        -w /orderly \
        -v $DOCS_DIR:/orderly/docs \
+       -v $VIGNETTES_DIR:/orderly/vignettes \
        --user "$USER_STR" \
        $ORDERLY_DEV \
        Rscript -e 'pkgdown::build_site(document = FALSE)'

tests/testthat/test-deduplicate.R

@@ -172,3 +172,32 @@ test_that("relink error handling", {
   expect_error(relink(from, to), "Some error linking")
   expect_true(all(fs::file_info(c(from, to))$inode == info))
 })
+
+
+test_that("deduplicate fails if file is missing", {
+  skip_on_cran()
+  path <- orderly_example("demo")
+  id1 <- orderly_run("minimal", root = path, echo = FALSE)
+  id2 <- orderly_run("minimal", root = path, echo = FALSE)
+  orderly_commit(id1, root = path)
+  orderly_commit(id2, root = path)
+
+  unlink(file.path(path, "archive", "minimal", id1, "script.R"))
+
+  expect_error(orderly_deduplicate_info(orderly_config(path)), paste(
+    "Cannot deduplicate archive as database references files",
+    "which don't exist."))
+})
+
+test_that("deduplicate fails if report pulled from remote recursive FALSE", {
+  skip_on_cran()
+  dat <- prepare_orderly_remote_example()
+  id3 <- orderly_run("depend", root = dat$path_remote, echo = FALSE)
+  orderly_commit(id3, root = dat$path_remote)
+
+  orderly_pull_archive("depend", root = dat$config, remote = dat$remote,
+                       recursive = FALSE)
+  expect_error(orderly_deduplicate_info(dat$config), paste(
+    "Cannot deduplicate archive reports have been pulled from",
+    "remote with recursive = FALSE."))
+})