Assuming I have the following multiindex DF import pandas as pd
import numpy as np
import pandas as pd
input_id = np.array(['12345'])
docType = np.array(['pre','pub','app','dw'])
docId = np.array(['34455667'])
sec_type = np.array(['bib','abs','cl','de'])
sec_ids = np.array(['x-y','z-k'])
index = pd.MultiIndex.from_product([input_id,docType,docId,sec_type,sec_ids])
content= [str(randint(1,10))+ '##' + str(randint(1,10)) for i in range(len(index))]
df = pd.DataFrame(content, index=index, columns=['content'])
df.rename_axis(index=['input_id','docType','docId','secType','sec_ids'], inplace=True)
df
I know that I can query a multiindex DF as follows:
# querying a multiindex DF
idx = pd.IndexSlice
df.loc[idx[:,['pub','pre'],:,'de',:]]
basically with the help of pd.IndexSlice I can pass the values I want for every of the indexes. In the above case I want the resulting DF where the second index is 'pub' OR 'pre' and the 4th one is 'de'.
I am looking for the way to pass a range of values to the query. something like multiindex 3 beeing between 34567 and 45657. Assume those are integers.
pseudocode: df.loc[idx[:,['pub','pre'],XXXXX,'de',:]]
XXXX = ?
EDIT 1: docId column index is of text type, probably its necessary to change it first to int
Turns out query is very powerful:
df.query('docType in ["pub","pre"] and ("34455667" <= docId <= "3445568") and (secType=="de")')
Output:
content
input_id docType docId secType sec_ids
12345 pre 34455667 de x-y 2##9
z-k 6##1
pub 34455667 de x-y 6##5
z-k 9##8