Actual version: v04alpha
Buzzword search with "AND" option within distance. Its like an embedder only with plain txt search!
The maximum text found is never larger than the original text, as overlapping sections are merged!
-> all is in character and percent
-> keep in mind 5000characters ~1200token (aprox one book page)
EXE on huggingface:
https://huggingface.co/kalle07/raw-txt-snippet-creator
- Only windows tested!
- Only txt files, tested with 2MB (one large book) ~10-20sec
- Choose one txt file or a whole folder
- Type a buzzword or more, with AND its connected with in a distance
- All matches found are cut out as a snippet
- All overlaped snippets ar merged
- Two search options "usual wildcard" and "fuzzy-search"
(wildcard search If you have the word “friendship” and search for “friend” it will not be found. You should use “friend*”. "?" is only one character like usual.)
(fuzzy is sometime usefully , but it dont work with any punctuation like ip adresses, but it can handle in some cases * and ?) - All snippets are appended and saved (one for wildcard one for fuzzy - file) in json format with te match and found position
(the position you can see eg: in notepad++) - Output files are always overwritten when you click “Search” again
- Now you can easily copy and paste to your chat
