I'm confused by the phrase “one way hash value” and so is probably whoever specified this requirement. In cryptography, a one-way hash function has a well-defined meaning.
Let \$\Sigma\$ be an alphabet, for example, \$\Sigma=\{\mathtt{0}, \mathtt{1}\}\$. A family of functions \$(f_k)_{k\in\mathbb{N}}\$ with
$$f_k: \Sigma^k \times \Sigma^* \to \Sigma^k$$
is a family of one-way hash functions if and only if for each \$k\in\mathbb{N}\$, the \$f_k\$ is computable by a polynomial-time algorithm and for every probabilistic polynomial-time algorithm \$\mathcal{A}\$, given random \$K\in\Sigma^k\$ and \$h\in\Sigma^k\$, the output \$m^*=\mathcal{A}(K,h)\$ will satisfy \$f_k(K,m^*)=h\$ only with probability negligible in \$k\$.
\$K\$ is called key or salt. It should be chosen randomly but need not be kept secret.
It is unknown whether such functions exist at all but there are some functions for which it hasn't been proved yet that they are not one-way functions.
If your function should compute such a cryptographic hash, I recommend that you look into one of the existing cryptographic hash function even though I doubt that you're supposed to implement one of these. (Although returning the result as a string would be meaningful in this context because for \$k>64\$, you usually cannot return the hash as an integer.) MD5 and SHA1 are two classical examples of cryptographic hash functions that are reasonably straight-forward to implement. The collision resistance (which is a stronger requirement than the one-way property) of both of them is known to be broken in theory. (Both are also only specified for fixed \$k\$ which means that the above definition doesn't really apply.)
Since I see the word “hash table” in your code, I suspect that instead of a cryptographic hash function, they wanted to see a “simple” hash function that is fast to compute and has low probability of collisions but is not collision-resistant against malicious attackers. Again, have a look at algorithms other people already invented.
- The Fowler-Noll-Vo hash function gives good hashes and is easy to implement. It can also be parametrized on the length of the result. I would recommend the FNV-1a variant as your go-to hash function unless you have reasons to chose differently.
- Jenkins' one-at-a-time hash function is equally easy to implement and will probably be faster because it only uses additions and bit-wise operations as opposed to multiplications.
- Weinberger's PJW hash function is known for its use in the UNIX ELF format and for being mentioned in the Red Dragon Book.
- The lose-lose hash function is the one you've actually implemented. It's name should give you a hint. To your rescue, it was once used as an example by Kernighan and Ritchie.
There are numerous other hash functions and the literature / internet is full of comparisons between them. As you can imagine, while there are clearly better and less good functions available, there is also a lot of personal preference involved in deciding which function to use.
Even if you're planning to use the hash as index for a hash table, don't take the hash modulo the table size in the hash function itself. This will needlessly limit its usefulness. Whoever uses the function should be responsible for wrapping the computed hash into the table size.
