fix children directories not being added to parents, keep working on obscure exception, improve build script

README.md

@@ -10,8 +10,8 @@ I needed to serve a very large dataset full of small files publicly over the int
 existing solutions were subpar and I found myself having to create confusing Openresty scripts and complex CDN caching
 to keep things responsive and server load low. I gave up and decided to create my own solution.
 
-You will likely need to store your data on an SSD for this. With an SSD, my server was able to crawl over 6 million
-files stored in a very complicated directory tree in just 5 minutes.
+You absolutely need an SSD for this. With an SSD, my server was able to crawl over 6 million files stored in a very
+complicated directory tree in just 5 minutes.
 
 ## Features
 

build.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+
+SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
+
+if [ -z ${1+x} ]; then
+  VERSION="0.0.0"
+else
+  VERSION="$1"
+fi
+
+mkdir -p "$SCRIPT_DIR/dist"
+
+cd "$SCRIPT_DIR/src" || exit 1
+go build -v -trimpath -ldflags "-s -w -X main.VersionDate=$(date -u --iso-8601=minutes) -X main.Version=v$VERSION" -o "$SCRIPT_DIR/dist/crazyfs"
+
+chmod +x "$SCRIPT_DIR/dist/crazyfs"
+
+echo "Finished building ->  $SCRIPT_DIR/dist/crazyfs"

src/CacheItem/NewItem.go

@@ -90,8 +90,8 @@ func PathOutsideRoot(fullPath string) bool {
 	return !strings.HasPrefix(fullPath, config.GetConfig().RootDir)
 }
 
+// RetardCheck makes sure we never do anything outside the root dir.
 func RetardCheck(fullPath string) {
-	// Make sure we never do anything outside the root dir.
 	if PathOutsideRoot(fullPath) {
 		panic(fmt.Sprintf("NewItem was not passed an absolute path. The path must start with the RootDir (%s). Failing path: %s", config.GetConfig().RootDir, fullPath))
 	}

src/DirectoryCrawler/process.go

@@ -40,12 +40,11 @@ func (dc *DirectoryCrawler) processPath(fullPath string, info os.FileInfo) error
 			parentItem, found := SharedCache.Cache.Get(strippedParentDir)
 			if found {
 				// Remove the old version of the directory from the parent's Children field
-				newChildren, foundOldDir := removeOldDir(parentItem.Children, relPath)
+				newChildren, _ := removeOldDir(parentItem.Children, relPath)
+
+				// Always add the new version of the directory to the parent's Children field
+				parentItem.Children = append(newChildren, relPath)
 
-				// Add the new version of the directory to the parent's Children field only if it wasn't found
-				if !foundOldDir {
-					parentItem.Children = append(newChildren, relPath)
-				}
 				// Update the parent directory in the cache
 				SharedCache.Cache.Add(strippedParentDir, parentItem)
 			}

src/ResponseItem/ResponseItem.go

@@ -77,9 +77,10 @@ func NewResponseItem(cacheItem *CacheItem.Item) *ResponseItem {
 			// This also be triggered if we encounter a broken symlink. We don't check for broken symlinks when scanning
 			// because that would be an extra os.Lstat() call in processPath().
 			if !found {
-				log.Debugf("CRAWLER - %s not in cache, crawling", child)
+				crawlRelPath := filepath.Join(config.GetConfig().RootDir, child)
+				log.Debugf(`CRAWLER - "%s" ("%s") not in cache, crawling`, child, crawlRelPath)
 				dc := DirectoryCrawler.NewDirectoryCrawler()
-				item, err := dc.CrawlNoRecursion(filepath.Join(config.GetConfig().RootDir, child))
+				item, err := dc.CrawlNoRecursion(crawlRelPath)
 				if err != nil {
 					log.Errorf("NewResponseItem - CrawlNoRecursion - %s", err)
 					continue // skip this child
@@ -90,20 +91,24 @@ func NewResponseItem(cacheItem *CacheItem.Item) *ResponseItem {
 				}
 			}
 
-			copiedChildItem := &CacheItem.Item{
-				Path:      childItem.Path,
-				Name:      childItem.Name,
-				Size:      childItem.Size,
-				Extension: childItem.Extension,
-				Modified:  childItem.Modified,
-				Mode:      childItem.Mode,
-				IsDir:     childItem.IsDir,
-				IsSymlink: childItem.IsSymlink,
-				Cached:    childItem.Cached,
-				Children:  nil,
-				MimeType:  childItem.MimeType,
+			if childItem != nil {
+				copiedChildItem := &CacheItem.Item{
+					Path:      childItem.Path,
+					Name:      childItem.Name,
+					Size:      childItem.Size,
+					Extension: childItem.Extension,
+					Modified:  childItem.Modified,
+					Mode:      childItem.Mode,
+					IsDir:     childItem.IsDir,
+					IsSymlink: childItem.IsSymlink,
+					Cached:    childItem.Cached,
+					Children:  nil,
+					MimeType:  childItem.MimeType,
+				}
+				children = append(children, copiedChildItem)
+			} else {
+				log.Errorf(`NewResponseItem - copiedChildItem for "%s" was null! - %+v`, child, cacheItem)
 			}
-			children = append(children, copiedChildItem)
 		}
 		item.Children = children
 	}

src/build.sh

@@ -1,14 +0,0 @@
-#!/bin/bash
-
-
-if [ -z ${1+x} ]; then
-  VERSION="0.0.0"
-else
-  VERSION="$1"
-fi
-
-mkdir -p ../dist
-
-go build -v -trimpath -ldflags "-s -w -X main.VersionDate=$(date -u --iso-8601=minutes) -X main.Version=v$VERSION" -o ../dist/crazyfs
-
-chmod +x ../dist/crazyfs

src/elastic/add.go

@@ -26,7 +26,8 @@ func addToElasticsearch(fullPath string, info os.FileInfo, incomingErr error) er
 	if !shouldExclude(relPath, config.GetConfig().ElasticsearchExcludePatterns) {
 		cacheItem, found := SharedCache.Cache.Get(relPath)
 		if !found {
-			log.Debugf(`ELASTICSEARCH - path "%s" exists on disk, but not in the LRU cache. Deleting from Elastic.`, relPath)
+			// I don't think this should ever happen
+			log.Errorf(`ELASTICSEARCH - path "%s" exists on disk, but not in the LRU cache. Deleting from Elastic.`, relPath)
 			// Delete this item from Elastic in order to avoid any strange inconsistencies.
 			err := deleteFromElasticsearch(encodeToBase64(relPath))
 			if err != nil {