due to the "high cost" (at least from my point of view) of a thermal printer I was wondering whether it was possible/worth the effort to buy a compared cheaper device with it integrated already (a calculator for example). I was thinking of buying a "P1-DTSC, Canon" calculator for my "misuse". So would it be easier to replace/reprogram the microcontroller or just take the needed printer and control it with an arduino bord.
It would be essentially impossible for you to reprogram the microcontroller in a commercial product like a Canon calculator. First of all, the chip is probably propriety, and not something you could find documentation on. Even if you could, it is unlikely it is able to be reprogrammed within the device -- the chips are likely preprogrammed at the factory before being soldered in place. You would need to buy the necessary hardware to reprogram the chip -- that alone would cost more than your printer.
Then you have the issue of writing new firmware. Without source code, you would have to write the entire firmware from scratch. We're not talking a few days, but several weeks, at the least, of effort. Anyway, without a schematic, and a knowledge of all of the other chips on the board, you would have no idea where to start.
Is it possible? Very unlikely. Is it worth the effort? Absolutely not.
Possible sure. Worth the effort: doubtful. You're probably better off figuring out what kind of thermal printer is in the calculator and then sourcing that thermal printer as that would likely be cheaper.
I doubt you'd be able to reprogram the microcontroller. Replacing it with uC is the same thing as controlling it with an arduino. Both are as much work as buying just the thermal printer and making your own controller.
Someone with sufficient skill and time could probably manage to hack into a typical printing calculator and use it as a printer, but might have to destroy a few calculators in the process before ending up with something that works. It is also possible that some printing calculators might be constructed in such fashion as to make such hacking fairly straightforward.
Probably the easiest approach to hacking would be to tap into the keyboard matrix with some logic that can fake keystrokes, though unless you tap into the display as well your speed will probably be very limited (if you tap into the display, you can type each key as soon as the previous one has registered; otherwise you would have to delay for however long it might take each key to register). Tapping into the printer mechanics would give you more versatility, but would likely require some experimentation to get useful results.
An alternative might be to see if you can find a print mechanism from a surplus house. If you get really lucky, you might get one with a proper data sheet. Otherwise, you would have to experiment. I once bought for US$5 a mechanism that was designed to produce text on 4" wide paper. It had a small motor which when run continuously would cycle the print head back and forth; each cycle would advance the paper by one line, and a contact closure would indicate when the printhead reached or left the home position. Additionally, the motor had another couple wires which seem to have been connected to a commutator-less winding, since they output a somewhat noisy AC waveform when the motor was running, but were unconnected to the motor's power leads. I'm not sure that was intended for use in a speed-control feedback loop, or whether it was intended to be a dot-output frequency reference. I never actually got so far as printing anything.
If you can find something surplus, it will probably be cheaper and easier to work with than a calculator with its guts ripped out. Note also that a printhead is apt to have two power limits you should be mindful of: there's a lower limit of power which you could feed the printhead all day without anything bad happening, but at that level of power printing would be very slow. When the printhead is moving, it's possible to supply it more power, but supplying that much power when the printhead isn't moving could cause it to overhead and be destroyed (when the printhead is moving, it loses a lot of heat to the paper). Depending upon your requirements, the lower power limit might suffice; if it doesn't, you should expect to destroy a few print heads before your project is done.