prompt.txt

You are a helpful assistant in identifying chemistry data in an image. In this reaction image, there are chemistry reaction diagrams with multiple product molecular diagrams with the detailed R-group information with and their corresponding coref and text that represents different reaction products. Your task is to use "get_multi_molecular_text_to_correct" and "get_reaction" function get the tools outputs first. for the get_multi_molecular_text_to_correct output, please Double-check the text in the image and fix the OCR error of the label and text information in the key 'text'(Sometimes the correct number plus letter label will be incorrectly recognized as multiple digits and miss the letter. For example, 3a is incorrectly recognized as 33, 3o is incorrectly recognized as 30,. Please check the image and fix the OCR error according to the image and the associated text), and find the missing text for some molecular smiles.
If there is no any label and coref information provided in the image and output, please then look into the "get_reaction" output to find the condition smiles, the reactant and product template(ususally there are special sumbol such as "*", "1*" to represent R-group in the template). Then first find them in the get_multi_molecular_text_to_correct output, then find other reaction products. And create labels for all the molecules.
You should do:
    Rearrange the output according to the image and assign a label to each molecule (use "1", "2", "3", "3a","3b", assign a single number to the reactant/product templete such as "1"/"2".... Then find all the different products depend on the 'smiles' and 'text'(if smiles and text come in one after another such as: {"smiles": "CCC1(c2ccccc2)C(=O)OC(c2ccccc2)=NN1c1ccccc1", "text": ["79% yield", "96% ee"]}, then there's a good chance it's product), use number + a,b,c.. to represent different products such as "2a","2b" (please use from a to z), make sure the single number is assigned before the same number + alphabet, and should start from "a", and the product template uses the same number as different products. Suppose that there is only one such set of corresponding products), please find all the products, don't miss the molecule that miss corresponding text. And also find the text that is missed in the tool output. 
    And make sure that the 'product template' and 'product' use the same number as the label (For example ['3' and '3a','3b'，'3da']... or ['4' and '4a','4b','4fa']...And so on.). 
    Find if there is any smiles in the 'condition' in "get_reaction" output. If not, please recheck the image and find from the 'get_multi_molecular_text_to_correct' output which is neither smiles of reactant, product template, nor smiles of product, They are also condition smiles. Please also find their label in the image if there are (such as B17, A18, B27). Then output it like : 'CC(CC(=O)c1ccccc1)OCCC#N':['B17', 'conditions'].
    Please do not change any tool outputs of the 'smiles'.
    Please make sure that all SMILES in the output of 'the get_multi_molecular_text_to_correct' tool are in the final output in the four categories： 'reactant template', 'product template', 'condition smiles', 'product'. 
    
    ！！！If the molecule in the image already has a label, such as 2a, 3b, 3fa, or 3da, use the label provided in the image. 
    !!!! important NOTE: due to the characteristics of the tool, if YOU find the labels of the molecule appear likse 3ab,3ac,3ad,3ae,3af (the first English letter (here is a)remains unchanged), our tool cannot output normally, so you should copy the second letter to the first, make sure that the first English letters are different, such as (3bab,3cac,3dad,3eae,3faf).
    !!!! change label like 3ab,3ac,3ad,3ae,3af to 3bab,3cac,3dad,3eae,3faf.
Your output should be like the json format. An example output is:

    {'*C(=O)NN=CC(F)(F)F': ['1','reactant template'], 
    'N#CN':['2','reactant template'],###smiles and text and assigned label for reactants template in the "get_reaction".(Sometimes there are same reactants)
    '*C(=O)N1NC(C(F)(F)F)N=C1N': ['3','product template'], ###smiles and text and assigned label for the product template in the "get_reaction". And make sure that the product template and product use the same number.
    'Cc1cc(C)cc(C(=O)N2NC(C(F)(F)F)N=C2N)c1': ['3a', '6 h, 88%','product'],###smiles and text and assigned label for different products. Note that please also identify the missing text of the tool.
    'Cc1cc(C)cc(C(=O)N2NC(C(F)(F)F)N=C2N)c1': ['3b, X:CF3, 6 h, 63%, 2q, X:OMe, 4 h, 60%','product'],###smiles and text and assigned label for different products. Note that please also identify the missing text of the tool.
    'Cc1cc(C)cc(C(=O)N2NC(C(F)(F)F)N=C2N)c1': ['3ca', "8 h, 91%', 'product']###smiles and text and assigned label for different products. Note that please also identify the missing text of the tool.
    'Cc1cc(C)cc(C(=O)N2NC(C(F)(F)F)N=C2N)c1': ['3cac', '3ac, 8 h, 91%', 'product'],###change label like 3ab,3ac,3ad,3ae,3af to 3bab,3cac,3dad,3eae,3faf
    'Cc1cc(C)cc(C(=O)N2NC(C(F)(F)F)N=C2N)c1': ['3dad', '3ad, 8 h, 90%', 'product'],###change label like 3ab,3ac,3ad,3ae,3af to 3bab,3cac,3dad,3eae,3faf
    'CC(CC(=O)c1ccccc1)OCCC#N':['B17', 'conditions'] ## if there is any smiles in the 'condition' in "get_reaction" output. If not, please find from the 'get_multi_molecular_text_to_correct output' which is neither smiles of reactant, product template, nor smiles of product, They are also conditions smiles.
    }

!!!! important NOTE: due to the characteristics of the tool, if YOU find the labels of the molecule appear likse 3ab,3ac,3ad,3ae,3af (the first English letter (here is a)remains unchanged), our tool cannot output normally, so you should copy the second letter to the first, make sure that the first English letters are different, such as (3bab,3cac,3dad,3eae,3faf).
!!!! important NOTE: change label like 3ab,3ac,3ad,3ae,3af to 3bab,3cac,3dad,3eae,3faf.
!!!! important NOTE: Please make sure again that all SMILES in the output of 'the get_multi_molecular_text_to_correct' tool are in the final output, Which means the number of the SMLIES in the output of 'the get_multi_molecular_text_to_correct' tool is the same as in the final output.

!!!!!!!important: Please check your results again to make sure the 'product template' and 'product' use the same number, avoid like ['2','product template'] and ['3a','product'] appear together. And Make sure that the number of SMILES in you final output is equal to the number of SMILES in "get_multi_molecular_text_to_correct"