https://bugs.documentfoundation.org/show_bug.cgi?id=137737
--- Comment #20 from Mike Kaganski <[email protected]> --- My proposal now is: 1. Fix this by excluding special processing of footnote/endnote anchor. This will change the behavior: the footnote anchor will be considered a "character" (non-word, so e.g. regex g\b will find "following" in the first body paragraph). Note: "$ will *not* find the first body paragraph's last " anymore, while "[:control:]$ will (but of course, one would need to use captures or look-ahead assertions to not replace the footnote/endnote character itself, also removing the note itself). 2. In the long run, see how to process the rest of the special characters consistently. Namely, they are RES_TXTATR_FLYCNT (fly anchor), RES_TXTATR_FIELD (field), RES_TXTATR_REFMARK (reference), RES_TXTATR_TOXMARK (index entry), RES_TXTATR_META and RES_TXTATR_METAFIELD (no idea when these are used). The problem here is that each of them will create the same problem; and for those that conceptually can't be considered "control characters", we need to drop them from the string at *all* times, not only when they are at start/end of string, and also when they are outside of the nStart/rEnd boundaries. Please reply what you think specifically about #1 here. Note again, that this would fix it in a very different way compared to what OP wanted. -- You are receiving this mail because: You are the assignee for the bug.
