> Try these: > > http://www.mail-archive.com/lyx-devel@lists.lyx.org/msg178570.html > http://www.mail-archive.com/lyx-devel@lists.lyx.org/msg178331.html
Going through the discussion there, I have a couple of ideas : If regular expressions have to be enabled, then queries will be regular expressions embedded in latex code/plain text. This makes it necessary to maintain a latex version of the content (not export every time find is invoked). An alternative would be: - We maintain a table that works like a symbol table - it has all the objects that are there in the current document and it's text content. i.e.: object | content -----------|----------- section | Introduction subsection | this is the text in the subsection - The table is updated as and when new content is added to the editor window. Maybe there is some scope for optimization here. - Maybe each entry will also have a mapping to the inset which represents it (I'm not entirely sure how insets work, yet). - Whenever find is invoked, the query is also converted into (a chain of) object and content. - To match, you first filter all objects of the query type (assuming there's only one query term). Next, match the content - here there can be regular expressions in the query. Results are displayed using the mapping between insets and query items. What do you think? On a side note, cant there be a search for text inside the editor window also? This could help a lot in editing long papers - maybe you dont want to put a mathematical symbol in the query after all.
