text cleaner

it is mojibake: text saved as UTF-8 and then read as an older western encoding, so é turns into é and an apostrophe into ’. the fix reverses that exact mistake, and only where the result is valid text, so real accented words are left alone.

text copied out of a PDF breaks at the end of every printed line. it joins those lines back into paragraphs, keeps the blank lines between paragraphs, keeps lists on their own lines, and rejoins words that were hyphenated across a line break.

it removes the symbols that chat assistants and notes apps use for formatting: # headings, ** bold **, `code` and [link](url) syntax, so the text pastes cleanly into places that do not render markdown. bullets stay as plain dashes.