Which parts of function are executed externally and which are executed "on chain"?

Question

I am still a solidity noob, so I apologize if the question is too basic.

I am trying to write a contract which saves the hash of a (potentially) large bunch of data on chain.

For this reason, I would like to write a function which performs the hash externally (i.e. on the caller machine, something like a view function), and then it saves back the result on the chain. Is it possible? A code example here:

contract Test {
  bytes32[] hashes;

  function extCall(bytes _data) external view {
    bytes32 hash = keccak256(_data);
    _intCall(hash);
  }

  function _intCall(bytes32 _hash) private {
    hashes.push(_hash)
  }
}

The intention here was to exclude the possibility to save the hash on the chain without having the original data. For this reason, the only way to call the function _intCall would be by means of calling it through extCall. But, it seems that by doing this trick, also extCall is called on chain and consumes gas. In particular, i get this warning while testing on truffle:

Warning: Function declared as view, but this expression (potentially) modifies the state and thus requires non-payable (the default) or payable.

P.s. there is a (BIG) chance that I totally misunderstood the topic, so every help is really accepted!

Answer 1

Anything you do as part of a transaction happens on chain. Anything that modifies a contract's state needs to be done as part of a transaction.

It's fine to call a view function as a helper; that code will only run on a single node (the one you're talking to) and can return a value. So this would not incur a cost for all the data you're hashing:

contract Test {
    bytes32[] hashes;

    function hash(bytes data) public pure {
        return keccak256(data);
    }

    function store(bytes32 hash) public {
        hashes.push(hash);
    }
}

This would be used something like this (web3.py-ish pseudocode):

hash = contract.call().hash("my document here")  # not a transaction
contract.transact().store(hash)  # a transaction

But then you may as well just do the hashing outside of the contract altogether (e.g. hash = keccak256("my document here")).

There's no way to prove to anyone that you have a document matching the hash unless you send the entire document as part of a transaction.

Answer 2

pure -> To operate with data (literal) examples:

function helloo() public/external pure {

a) return 4 + 5;

b) uint number = 6;
   return number;

c) string memory hello = "hi"; // 'memory' -> refers to the temporary memory of the virtual machine of ethereum.
   return hello;

}

view -> to read contract data

examples:

string hello = "hi"; -> GLOBAL IN THE CONTRACT (BLOCKCHAIN) if it has been declared outside the function.

function helloo() public/external view {

a) return hello; 

b) string storage str = hello; -> Refers to the state variable 
(pointer).   
return str;

}

*None of these modifiers consumes gas because these function not write or save data in the blockchain.