refactor: bincraftR -> bincraft

local/detect-duplicates.R

@@ -1,27 +1,63 @@
+library(future.apply)
+
+access_key <- Sys.getenv("HETZNER_S3_ACCESS_KEY_K3S")
+secret_key <- Sys.getenv("HETZNER_S3_SECRET_KEY_K3S")
+
+# Set up S3 connection
 s3fs::s3_file_system(
-  aws_access_key_id = Sys.getenv("HETZNER_S3_ACCESS_KEY_K3S"),
-  aws_secret_access_key = Sys.getenv("HETZNER_S3_SECRET_KEY_K3S"),
+  aws_access_key_id = access_key,
+  aws_secret_access_key = secret_key,
   endpoint = "https://hel1.your-objectstorage.com",
-  region_name = "hel1",
+  region_name = "hel1"
 )
 
-### Scope: Shows if there are duplicated packages at the S3 repository root (e.g. due to missed/aborted archiving)
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/amd64/jammy/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/amd64/noble/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/amd64/rhel8/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/amd64/rhel9/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/amd64/alpine320/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/amd64/alpine321/latest/src/contrib"))
+# Define architectures and codenames
+arches <- c("amd64", "arm64")
+codenames <- c("jammy", "noble", "rhel8", "rhel9", "alpine320", "alpine321")
 
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/arm64/jammy/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/arm64/noble/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/arm64/rhel8/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/arm64/rhel9/latest/src/contrib"))
-files <- s3fs::s3_dir_ls(sprintf("devxy-r-package-binaries-hel1/arm64/alpine320/latest/src/contrib"))
+# Create all combinations as a data frame
+combos <- expand.grid(arch = arches, codename = codenames)
 
-files_b=basename(files)
-files_b_split = sapply(strsplit(basename(files_b), "_"), function(x) x[1])
-files_b_split[duplicated(files_b_split)]
-
-bincraftR::archive_package("Keng", codename = "alpine320", arch = "arm64")
+# Set up parallel plan
+future::plan(multisession) # or multicore
 
+# Parallel processing
+future.apply::future_lapply(
+  seq_len(nrow(combos)),
+  function(i) {
+    s3fs::s3_file_system(
+      aws_access_key_id = access_key,
+      aws_secret_access_key = secret_key,
+      endpoint = "https://hel1.your-objectstorage.com",
+      region_name = "hel1",
+      refresh = TRUE
+    )
+    arch <- combos$arch[i]
+    codename <- combos$codename[i]
+    files <- s3fs::s3_dir_ls(
+      sprintf(
+        "devxy-r-package-binaries-hel1/%s/%s/latest/src/contrib",
+        arch, codename
+      )
+    )
+    if (length(files) == 0) return(NULL)
+    files_b <- basename(files)
+    pkg_names <- sapply(strsplit(files_b, "_"), `[`, 1)
+    dupes <- unique(pkg_names[duplicated(pkg_names)])
+    for (pkg in dupes) {
+      bincraft::archive_package(
+        pkg,
+        codename = codename,
+        arch = arch,
+        s3_region = "hel1",
+        s3_endpoint = "https://hel1.your-objectstorage.com",
+        s3_bucket = "devxy-r-package-binaries-hel1",
+        s3_access_key_id = access_key,
+        s3_secret_access_key = secret_key
+      )
+    }
+    NULL
+  },
+  future.globals = TRUE,
+  future.seed = TRUE
+)