fix: automatically remove invalid characters for aliases generated from index names

src/storage/elasticsearch/Elasticsearch.ts

@@ -226,6 +226,43 @@ export class Elasticsearch implements Database {
     return 'stapps_*_*_*';
   }
 
+  /**
+   * Checks for invalid character in alias names and removes them
+   * @param alias The alias name
+   * @param uid   The UID of the current bulk (for debugging purposes)
+   */
+  private removeAliasChars(alias: string, uid: string | undefined): string {
+    // spaces are included in some types, so throwing an error in this case would clutter up the log unnecessarily
+    alias = alias.replace(' ', '');
+    // List of invalid characters: https://www.elastic.co/guide/en/elasticsearch/reference/6.6/indices-create-index.html
+    ['\\', '/', '*', '?', '"', '<', '>', '|', ',', '#'].forEach((value) => {
+      if (alias.includes(value)) {
+        alias = alias.replace(value, '');
+        logger.warn(`Type of the bulk ${uid} contains an invalid character '${value}'. This can lead to two bulks `
+          + `having the same alias despite having different types, as invalid characters are removed automatically. ` +
+        `New alias name is "${alias}."`);
+      }
+    });
+    ['-', '_', '+'].forEach((value) => {
+      if (alias.charAt(0) === value) {
+        alias = alias.substring(1);
+        logger.warn(`Type of the bulk ${uid} begins with '${value}'. This can lead to two bulks `
+          + `having the same alias despite having different types, as invalid characters are removed automatically. ` +
+          `New alias name is "${alias}."`);
+      }
+    });
+    if (alias === '.' || alias === '..') {
+      logger.warn(`Type of the bulk ${uid} is ${alias}. This is an invalid name, please consider using another ` +
+      `one, as it will be replaced with 'alias_placeholder', which can lead to strange errors.`);
+      return 'alias_placeholder';
+    }
+    if (alias.includes(':')) {
+      logger.warn(`Type of the bulk ${uid} contains a ':'. This isn't an issue now, but will be in future ` +
+      `Elasticsearch versions!`);
+    }
+    return alias;
+  }
+
   /**
    * Should be called, when a new bulk was created. Creates a new index and applies a the mapping to the index
    * @param bulk the bulk process that was created
@@ -241,7 +278,7 @@ export class Elasticsearch implements Database {
 
     // there already is an index with this type and source. We will index the new one and switch the alias to it
     // the old one is deleted
-    const alias = bulk.type;
+    const alias = this.removeAliasChars(bulk.type, bulk.uid);
 
     if (typeof this.aliasMap[alias] === 'undefined') {
       this.aliasMap[alias] = {};
@@ -295,7 +332,7 @@ export class Elasticsearch implements Database {
     const index: string = this.getIndex(bulk.type, bulk.source, bulk);
 
     // alias for the indices
-    const alias = bulk.type;
+    const alias = this.removeAliasChars(bulk.type, bulk.uid);
 
     if (typeof this.aliasMap[alias] === 'undefined') {
       this.aliasMap[alias] = {};