I do have a set of old lucene index files, probably 15-20 years old. Every index directory contains four files, something like these...(size is different)
_0.cfx 47, 942 KB _s.cfs 1,78,687 KB segments.gen 1 KB segments_2 1 KB When I try to open an index using PyLucene 10.0.0 or java 10.0.0 searcher = IndexSearcher(DirectoryReader.open(directory)) DirectoryReader throws an error org.apache.lucene.index.IndexFormatTooOldException: Format version is not supported (resource BufferedChecksumIndexInput(NIOFSIndexInput(path="D:\database\segments_2"))): -9 (needs to be between 1071082519 and 1071082519). This version of Lucene only supports indexes created with release 9.0 and later. I started opening the index using older versions of Lucene (java based not PyLucene) one by one and finally the index was opened using Lucene 2.9.4. The slight difference in the code is File oldIndexDir = new File(INDEX_PATH); Directory directory = FSDirectory.open(oldIndexDir); IndexSearcher searcher = new IndexSearcher(directory, true); // read-only=true What I need is to upgrade the indexes so that they can be opened using the latest 10.0.0 version. One logical solution I can think of is to export the data to sqlite database first and later create a new index by reading the data back using the latest version. I am new to Lucene and recently started diving in. Is there any solution offered by API for this problem? I would prefer a solution based on PyLucene but if it's not possible then a java based solution would also be fine. Prashant
