From 18ba4b59b0bebf1282a6b47bd2ac93e0ef6fefb3 Mon Sep 17 00:00:00 2001
From: Emmanuel Raviart <emmanuel@raviart.com>
Date: Tue, 28 Nov 2023 21:45:07 +0100
Subject: [PATCH] Handle symlinks when unzipping & updating datasets
 repositories

---
 src/lib/server/file_systems.ts       |  2 +-
 src/scripts/download_dila_dataset.ts | 25 ++++++++++++++++++++++++-
 2 files changed, 25 insertions(+), 2 deletions(-)

diff --git a/src/lib/server/file_systems.ts b/src/lib/server/file_systems.ts
index 9a3fc59..212fc66 100644
--- a/src/lib/server/file_systems.ts
+++ b/src/lib/server/file_systems.ts
@@ -12,7 +12,7 @@ export function* walkDir(
     }
     const filePath = path.join(dir, filename)
     const relativeSplitPath = [...relativeSplitDir, filename]
-    if (fs.statSync(filePath).isDirectory()) {
+    if (fs.lstatSync(filePath).isDirectory()) {
       yield* walkDir(rootDir, relativeSplitPath)
     } else {
       yield relativeSplitPath
diff --git a/src/scripts/download_dila_dataset.ts b/src/scripts/download_dila_dataset.ts
index 77aa3e2..81139de 100644
--- a/src/scripts/download_dila_dataset.ts
+++ b/src/scripts/download_dila_dataset.ts
@@ -5,6 +5,8 @@ import path from "path"
 import sade from "sade"
 import { $, cd } from "zx"
 
+import { walkDir } from "$lib/server/file_systems"
+
 async function downloadDataset(
   datasetName: string,
   dilaDir: string,
@@ -99,7 +101,28 @@ async function downloadDataset(
             ", ",
           )}`,
         )
-        await $`cp -r ${date}/${datasetName}/* ${datasetName}/`
+
+        // When archive contains a symbolic link and dataset contains at the same path
+        // something that is not a symbolic link, remove it before copy, otherwise
+        // copy of symbolic link will fail.
+        const archiveDatasetDir = path.join(date, datasetName)
+        for (const relativeSplitPath of walkDir(archiveDatasetDir)) {
+          const archiveFilePath = path.join(
+            archiveDatasetDir,
+            ...relativeSplitPath,
+          )
+          if (fs.lstatSync(archiveFilePath).isSymbolicLink()) {
+            const datasetFilePath = path.join(datasetName, ...relativeSplitPath)
+            if (
+              fs.pathExistsSync(datasetFilePath) &&
+              !fs.lstatSync(datasetFilePath).isSymbolicLink()
+            ) {
+              fs.removeSync(datasetFilePath)
+            }
+          }
+        }
+
+        await $`cp -r ${archiveDatasetDir}/* ${datasetName}/`
         const removalListFilePath = path.join(
           date,
           `liste_suppression_${datasetName}.dat`,