VarInt (LEB128) encode / decode

A variable-length integer. Small numbers take one byte and large ones take more, with seven bits of value per byte and the top bit set to say another byte follows. This tool implements the LEB128 form, the same scheme Protocol Buffers and WebAssembly use to keep integers compact.

Leave mode on encode and drop a text file with your numbers in it. The tool pulls out every run of digits it finds, with an optional leading minus, so 1, 2, 3 and 1 2 3 and even prose with numbers in it all work. The output is the encoded bytes as space-separated hex. If it finds nothing it stops with "no integers found in input text to encode".

You encoded a negative number with the signed option off. Plain LEB128 here only takes non-negative values. Turn on signed, which applies zigzag mapping so minus 1 becomes 1, 1 becomes 2 and so on, and negatives encode compactly and round-trip correctly.

It switches zigzag mapping on for both directions, and it is off by default. With it on, values alternate between positive and negative as the encoded number grows, which keeps small negative numbers to one byte. You must use the same setting for encode and decode, or the numbers will come back wrong rather than failing.

Switch mode to decode and drop a file containing the raw LEB128 bytes; the result is the values separated by commas. Note that encode writes hex text, not bytes, so to round-trip your own output run it through From hexdump first to turn the hex back into a .bin file.

The last byte of your file still had its continuation bit set, so the decoder ran out of input mid-number. Either the file was cut short or it is not LEB128 at all. Check that you decoded raw bytes rather than a hex or text representation of them.