DocFetcher Server 1.4 -- 2025-09-14

• Features:
  • Added two word segmentation options in the preferences, designed for indexing Chinese and Japanese text, respectively.
  • The preview pane now shows line indentations. (E.g., if a line starts with four spaces or a tab, that indentation is preserved in the preview.)
• Bugfixes:
  • Fixed SSL connection errors ("Invalid SNI") that occurred when users tried to access the server through localhost URLs (e.g., https://localhost:31191) or direct IP addresses while using the default self-signed certificate.
  • Indexing crash caused by antivirus software blocking the application's access to its own temporary files.
  • Indexing crash when trying to index 7z archives, caused by locked DLL files.
  • Indexing crash on some extremely large files.
  • Crash when trying to index files while running the application from a read-only file system.
  • Fixed a "Too many open files" indexing crash that occurred with certain Outlook PST and OST files.
  • Fixed an indexing crash that occurred with emails where the combined filename length of the email's attachments was extremely large.
  • Fixed an AssertionError indexing crash that occurred with certain RAR archives.
  • Text extraction on MS Excel 2007 files failed completely if any cells with invalid content were encountered. Now such cells are silently ignored.
  • Text extraction on JPEG files failed completely if there were invalid metadata tags.
  • If the folder tree in the Search Scope pane is too large to be fully stored in the browser's local storage, the web UI now just suppresses the error instead of breaking completely.
  • Previously, changing the word segmentation in the preferences did not take effect until all indexes were rebuilt and the server was restarted. Now the server restart is not needed anymore.
  • On macOS, searches containing full-width katakana, umlauts, and other decomposable characters failed to match file/folder names containing those characters due to a Unicode normalization issue.
• Changes:
  • On Windows, all .exe and .dll files are now digitally signed.
  • Slightly more intuitive indexing of filenames. Now filenames are indexed the same way regardless of whether they contain whitespace or not.
  • The application now fully supports generic URLs as file paths, e.g., "smb://myserver/path/to/file.txt".
  • Upgraded bundled Java runtime from Java 11 to Java 21.
  • The oldest supported macOS version has been bumped from 10.10 to 10.12.

DocFetcher Server 1.3 -- 2023-08-06

• Features:
  • You can now open the web interface with a "?q=" URL parameter to immediately run a search with the query specified via the URL parameter. An example of what the full URL might look like: https://example.com/search/?q=dog cat. One use case for the new URL parameter is launching DocFetcher Server searches programmatically or from a terminal. Another use case is that you can now select some text on a website, and then with the help of a browser extension quickly initiate a DocFetcher Server search with the selected text as the query.
  • Added a path mapping feature that allows clients to see different file paths in their search results than the file paths the server sees. For example, the server can now see a file as being located at /path/to/documents/file.txt, but present it to the clients as being located at X:\docs\file.txt.
  • In the result table, clicking the file icon now either triggers a download or copies the result's path to the clipboard. Which of these two actions happens can be set in the Admin Area.
  • In the result table, the context menu now has "Copy Selection" and "Copy All" entries like DocFetcher Pro. These entries copy the paths of the selected results and of all results to the clipboard, respectively.
• Minor improvements:
  • When the "message delivery time" attribute of an Outlook email is not available, fall back to indexing other attributes such as "client submit time".
  • There's now a user-specifiec setting to enable and disable using the browser's local storage to store the Search Scope checkbox and expansion states. This is helpful in cases where the folder hierarchy in the Search Scope is so large that the checkbox and expansion states cannot be stored in the browser's local storage. If using the latter is disabled, the checkbox and expansion states are reset after the web client is reopened.
• Major bugfixes:
  • After an index was updated or rebuilt, checkbox states in the Search Scope pane were reset.
  • There were some PDF annotations that were indexed, but not shown in the preview pane.
  • PDF annotations on empty pages were not indexed.
  • RTF bodies of Outlook emails are now indexed. Previously, they were ignored.
• Minor bugfixes:
  • Fixed a "VCRUNTIME140.dll was not found" error that occurred with some of the executables the application ships with.
  • While indexing certain very large files without file extension, the program would try to determine the file's text encoding, but run out of memory and crash.
  • Fixed an indexing crash that occurred when trying to index files whose paths were longer than 32 KB. This could happen when the files resided in infinitely deep folder structures.

DocFetcher Server 1.2 -- 2022-12-21

• Fixed a crash related to the indexing of bzip2 archives.
• At the end of indexing, the server crashed if it took longer than 30s to write the index files to disk. This could happen if the index files were big and had to be written over a slow network connection.
• If a subfolder within an indexed folder was renamed, moved or removed, the files in that subfolder were not properly updated in the index, causing obsolete files to show up in the search results along the current ones. The obsolete files will disappear once you upgrade to the new release and update all your indexes.
• Added a setting HtmlEncodingOverride in the server-conf.txt file to allow forcing the application to use a particular encoding for indexing HTML files.

DocFetcher Server 1.1 -- 2022-08-30

• The application could not be run on older Linux distributions due to a glibc compatibility issue. It now runs on Linux distributions with glibc 2.17 or newer.
• Among the indexing settings, there's now a new setting for skipping content indexing for all files that are bigger than a certain maximum file size. The filenames of these files can still be indexed. With this new setting, it's now possible to skip large files that may cause the application to run out of memory.
• In the Admin Area in the indexes table, there's a new column "Visible". By ticking and unticking the checkboxes in that column, you can control which indexes are transmitted to the clients, and which indexes are kept only on the server side. This is useful if you have some very large and rarely used indexes; keeping them on the server side when they're not needed will make the web interface load faster.
• In the result table, you can now press the arrow-up and arrow-down keys to navigate to the previous or next result, with the contents of the preview pane updated accordingly. However, for this to work properly, you need to turn off "Automatically scroll to first match in preview pane" in the user preferences.
• For tablet users, there's now a button above the result table for downloading the first selected result. Before, downloading results was only possible via the result table's context menu, which is difficult to open on tablets.
• For technical reasons, the users that are counted towards the application's client limit are identified by browser session rather than by IP address. Among other things, this means accessing the web interface from multiple browsers on the same computer counts as multiple clients rather than as a single client. For some users, this can be quite inconvenient. In addition, it was also inconvenient for a single user to switch between different computers. For these use cases, there's now a workaround called "session stealing", which means that when the client limit is reached, new users may take over existing sessions, thus kicking their previous owners out of the web interface. The intended use is for users to kick "themselves" out in order to more easily switch between browsers and/or computers. For instances of DocFetcher Server with a client limit greater than 1, session stealing is off by default and must be enabled in the Admin Area on the Access tab.
• It's now possible to inject custom CSS and JavaScript into the web interface, on the server side, via the files misc/custom.css and misc/custom.js. After changing these files, the server must be restarted. Note that no support or HTML stability guarantees for such customizations is provided, as explained in the comments in the custom.css and custom.js files.

DocFetcher Server 1.0 -- 2022-07-30

First release.