Is there any library that can convert a circuit drawn as text to a qiskit QuantumCircuit?

Question

Does anyone know of a library to convert a parametrized qiskit circuit drawn as 'text' back to a QuantumCircuit or similar?

I have see [this] 1 question, but unfortunately I cannot determine in which format the circuit has been saved. I need to start of a file in which the drawn circuit has been saved as text like this:

              ┌────┐    ┌────────┐ ┌──────────┐┌───┐ ░ ┌─┐
   q_0 -> 1 ──┤ √X ├────┤ Rz(3π) ├─┤ Rz(p[2]) ├┤ X ├─░─┤M├───────── 
            ┌─┴────┴─┐ ┌┴────────┴┐└──────────┘└─┬─┘ ░ └╥┘   ┌─┐    
... 

 meas: 4/════════════════════════════════════════════╩══╩══╩══╩═ 

In more detail the saved file in my case is a text file looking like this: It is a dictionary converted to a string using json. The values of the dictionary are lists of the ascii-art strings of the circuits generated by QuantumCircuit.draw():

{'ansatz 1':"['           ┌───┐     ┌──────────┐ ┌──────────┐┌──────────┐┌───┐      ░ ┌─┐   »\n   q_0: ───┤ X ├─────┤ Rz(p[0]) ├─┤ Ry(p[1]) ├┤ Rz(p[2]) ├┤ X ├──────░─┤M├───»\n           ├───┤     ├──────────┤ ├──────────┤├──────────┤└─┬─┘┌───┐ ░ └╥┘┌─┐»\n   q_1: ───┤ X ├─────┤ Rz(p[3]) ├─┤ Ry(p[4]) ├┤ Rz(p[5]) ├──┼──┤ X ├─░──╫─┤M├»\n        ┌──┴───┴───┐ ├──────────┤ ├──────────┤└──────────┘  │  └─┬─┘ ░  ║ └╥┘»\n   q_2: ┤ Rz(p[6]) ├─┤ Ry(p[7]) ├─┤ Rz(p[8]) ├─────■────────■────┼───░──╫──╫─»\n        ├──────────┤┌┴──────────┤┌┴──────────┤   ┌─┴─┐           │   ░  ║  ║ »\n   q_3: ┤ Rz(p[9]) ├┤ Ry(p[10]) ├┤ Rz(p[11]) ├───┤ X ├───────────■───░──╫──╫─»\n        └──────────┘└───────────┘└───────────┘   └───┘               ░  ║  ║ »\nmeas: 4/════════════════════════════════════════════════════════════════╩══╩═»\n                                                                        0  1 »\n«              \n«   q_0: ──────\n«              \n«   q_1: ──────\n«        ┌─┐   \n«   q_2: ┤M├───\n«        └╥┘┌─┐\n«   q_3: ─╫─┤M├\n«         ║ └╥┘\n«meas: 4/═╩══╩═\n«         2  3 ', '...}

Answer 1

You can try ChatGpt, this is what I got according to your text:
Note that, it doesn't work accurately every time, you can try to specify more in detail, or point out the error.
I said:

              ┌────┐    ┌────────┐ ┌──────────┐┌───┐ ░ ┌─┐
   q_0 -> 1 ──┤ √X ├────┤ Rz(3π) ├─┤ Rz(p[2]) ├┤ X ├─░─┤M├───────── 
            ┌─┴────┴─┐ ┌┴────────┴┐└──────────┘└─┬─┘ ░ └╥┘   ┌─┐    
... 

 meas: 4/════════════════════════════════════════════╩══╩══╩══╩═
can you convert it into qiskit quantum circuit code

ChatGPT reply:

from qiskit import QuantumCircuit, QuantumRegister, execute, Aer

# Create a quantum circuit with a single qubit and a single classical bit
qreg = QuantumRegister(1)
creg = ClassicalRegister(1)
qc = QuantumCircuit(qreg, creg)

# Apply the gates in the circuit
qc.u3(3*pi, 0, 0, qreg[0])  # √X gate
qc.rz(3*pi, qreg[0])  # Rz(3π) gate
qc.rz(p[2], qreg[0])  # Rz(p[2]) gate
qc.x(qreg[0])  # X gate

# Measure the qubit
qc.measure(qreg, creg)

# Execute the circuit using a simulator
backend = Aer.get_backend('qasm_simulator')
job = execute(qc, backend, shots=1)

# Print the measurement result
print(job.result().get_counts())

```

Answer 2

As explained in the comments, you can export QuantumCircuit object as qasm string or file and reload it later into QuantumCircuit again using the QuantumCircuit.from_qasm_string and Quantumcircuit.from_qasm_file methods.

from qiskit import QuantumCircuit
qc = QuantumCircuit(1)
qc.x(0)

# export qasm string to a variable
qasm_string = qc.qasm()
qc_from_string = QuantumCircuit.from_qasm_str(qasm_string)
print(qc == qc_from_string)
# Output: True

# store qasm string into a file
with open('qasm_file', 'w') as output_file:
    output_file.write(qasm_string)

qc_from_file = QuantumCircuit.from_qasm_file('qasm_file')
print(qc == qc_from_file)
# Output: True

You can find the jupyter notebook with full code and output here: https://gist.github.com/HuangJunye/02bdcd927de4aab0cfc214428fa1fdac

Do note that there are limitations of converting QuantumCircuit as qasm as OpenQASM2 does not support all features such as unbound parameters. Exporting to Qobj or qpy as explained in this answer is better because the conversion is lossless.