[Pranav-error]

Fixes #7982.

G_get_datumparams_from_projinfo() wrote the datum, datumparams, nadgrids, and towgs84/dx/dy/dz values straight from a PROJ_INFO-format Key_Value set into caller-supplied buffers with plain sprintf, and the function signature had no size parameter at all — so it had no way to bound the write regardless of how big the caller's buffer actually was.

general/g.setproj/main.c calls it with two 100-byte stack buffers. Any datum value over 99 characters overflows the first; any towgs84/nadgrids value, or the three dx/dy/dz values combined (joined as "towgs84=%s,%s,%s"), over roughly 90 characters overflows the second. These values come from a location's PROJ_INFO file, which is plain text and not length-validated anywhere — hand-edited, imported from an external CRS/WKT definition via g.proj, or shipped in a downloaded sample or shared location.

lib/gis/proj3.c's own caller uses 256-byte buffers — so the function already has two different "safe" sizes depending on which caller you look at, which is the real problem: there is no size the function itself can assume.

Fixed by adding explicit size_t parameters for each buffer and switching both sprintf calls to snprintf, updating both call sites (g.setproj and proj3.c) to pass sizeof() their own buffers.

Verified with a small program linked against a local build of libgrass_gis: a 499-character towgs84 value, which previously overflowed the 100-byte g.setproj buffer via plain sprintf, now truncates safely to 99 characters plus the terminator instead. Built the gis library (grass_gis target) clean with the change — general/g.setproj has no CMake target in this tree to build standalone, so I also compiled and linked a small driver program against the installed header and built library using the exact call shape g.setproj uses, to confirm the signature change is fully consistent end to end.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[github-actions]

[pre-commit] _{reported by reviewdog 🐶}

Suggested change

	if (exist && (G_get_datumparams_from_projinfo(
	old_proj_keys, lbuf, sizeof(lbuf), lbufa,
	sizeof(lbufa)) == 2)) {
	if (exist &&
	(G_get_datumparams_from_projinfo(old_proj_keys, lbuf, sizeof(lbuf),
	lbufa, sizeof(lbufa)) == 2)) {

[github-actions]

[pre-commit] _{reported by reviewdog 🐶}

Suggested change

	G_find_key_value("dy", projinfo),
	G_find_key_value("dz", projinfo));
	G_find_key_value("dy", projinfo),
	G_find_key_value("dz", projinfo));

[Valyrian-Code]

Not a maintainer, just had a look. Both callers are updated and the sizeof arguments are real arrays, not pointers. Nothing in grass-addons calls this one either.

One thing: snprintf truncates quietly and the return is still 1 or 2, so a caller cannot tell a full value from a cut one. Hard to hit at 256 bytes though.

[Pranav-error]

Thanks for checking. Right, a cut value isn't reported to the caller. I kept the return values as they are, since reporting truncation would change what the function returns; I can add it if the maintainers want that.