Finding Old File on PC

Needs a classifier that is better than a "this EXE is a PE32 and that EXE is PE32+".

There are utilities that can determine "percentage Delphi, percentage Watcom compiler" and so on. Such a utility stands a better chance of identifying the item.

You can see a reference to such things, on virustotal.com entries.

I can start with a Xananews installer, and get the Virustotal page for it.

Name: XanaNews.exe Size: 4073984 bytes (3978 KiB) SHA256: BC4E157D7B4FDFD90DDFEFFA4AB417BB57E0566706BA404C1C9A0CCD726E2A25

DetectItEasy PE32 Linker: Turbo Linker (2.25*,Delphi) [GUI32] Compiler: Embarcadero Delphi (XE) [Enterprise] <=== the installer is not all Delphi, but it still detected the "theme"

So then, "TrID" and "DetectItEasy" are examples of tools that could classify all the executables in the collection.

I don't know the details for running a scan on a lot of files, and you may want a way of limiting the number of items they scan. If it takes 11 seconds each and 200,000 files, it could take a while.

And the reason you're doing this, is to look for things that contain compiler references you've used in the past for making home tools.

You could also look for telltale file extensions of the source files. For example, Agent Ransack accepts REGEX during file searches (mythicsoftware.com , their free one).

\.c$ # all the C language source cpp$ # The C++ source jar$ # Java jar files

And so on. While Windows federated search has some extension search capability, it may not be enough.

Paul

Simple, non-techie me, uses the free "Agent Ransack" to search for files on all drives. It uses various text searches and also Regex. Excuse me if I'm not understanding the question.

No, that's spot on :-)

Now if I could only remember what I might have called the file 18 years ago!

FINDSTR is in most Windows installs.

Run it from a command prompt. FINDSTR /? shows the parameters. You can use it to find likely words/expressions in your program. You can do recursive searches down the directories. I don't how well it works and how fast it is as I use grep on Unix type OSes now. It saves having to install anything.

If you're fairly sure that you haven't amended the file since, could you not limit the search to all files created between say 2008 and 2010? Once listed, copy them to a temporary folder and search through the contents in turn for "Barca" and/or "Access". Even if they were *.exe files, you /might/ have left a comment which included those words.

It searches content for key words and terms too.

It can.

The search box has

Filename: txt$ # Regex for "files ending in .txt"

Content: turkey dinner # Text files are easy to search for text, binaries not so much

TreeToSearch: C:\ # Can load a dozen drive letters here, can search the entire machine

and it would report

Found one file, "christmas.txt"

The other product, File Locator Pro by Mythicsoft, would presumably have more features than this, but it is a paid product. Agent Ransack is the free one.

You can run the utility "strings" on binary files, looking for text in them, but that does not always work as well as you would expect. You can change the length of string the thing extracts, which can cut down the noise. If you ask for "all strings four text characters or longer", that generates a huge amount of noise. As you make the string size larger, the noise drops off. But you can also easily miss a file by making the string capture rely on a too long string.

It's also possible to search a drive at the physical level, with HxD hex editor.

Paul

Another free program for searching for key words within a file is Everything

menu/search/advanced search

File names containing section leave as default (blank)

A word or phrase in a file = your keyword

Located in = c:/ (example)

Note: It will indicate "querying" in the bottom LH corner of the main window and take rather a long time if checking 277k files. It may not find a key word in a binary file.

Thanks, will check it out :-)

Knowing what I'm looking for would be a good start...

From the FAQ

Yes, "Everything" can search file content with the content: search function.File content is not indexed, searching content is slow.

The free programmers sledgehammer approach I've taken is to install Apache Solr and get it to build an index.

Then querying that, which is lightning fast - using very similar deep searching terms that are used with Google, for instance. It knows about common document file formats and is able to index than search inside them too.

I have it (currently broken.... needs fixing) on a local webserver, it kind of looks like Google - but is not :).

One day, I'll feed it to a local LLM, and turn my brain to cabbage mode....

Solr looks interesting/useful, thanks for the heads up

J^n

Solr uses an inverted index.

The search may be blazing fast, but building the database that enabled the blazing fast result takes time.

Somewhere in the documentation, it should tell you the maximum number of files it supports. That will be a recommendation based on how much it slows down if you go past that point (a performance recommendation, not an absolute limit). For example, one million files would be my guess at a limit, but check the documentation and see what it says for a limit.

The standard training set for an AI, is 420TB (forty two, ten terabyte disk drives). That's entirely another level of file examination. And the AI is "forgetful" and can actually not deliver all the things you asked for. The inverted index is deterministic, by comparison.

Paul