實踐 staging 利用16進位排序

PETstaging.py

import re

AJCC7 = {
   'lung': {
        'T': {'x', '0', 'is', '1', '1a', '1b', '2', '2a', '2b', '3', '4'},
        'N': {'x', '0', '1', '2', '3'},
        'M': {'x', '0', '1', '1a', '1b'}
    },
    'colon': {
        'T': {'x', '0', 'is', '1', '2', '3', '4', '4a', '4b'},
        'N': {'x', '0', '1', '1a', '1b', '1c', '2', '2a', '2b'},
        'M': {'x', '0', '1', '1a', '1b'}
    }
}

def stringProcess(string): #去掉空白、()及其內容、去掉 T, N, M, p, y, r...
    if string ==string:
        ans = re.sub("\s", "", string)
        ans = re.sub("\(.*\)|T|N|M|\?|^(p|y|r)*","" , ans) #TNM不管行首了
        ans = re.sub("X", "x", ans) #把大寫 X 轉成小寫 x
    if string != string: #把 NaN 轉成空白
        ans = ""
    if string == "無":
        ans = ""
    return ans


def lowerStage(s): #需往後補0 共3位，輸入已經分割好的點，也作為 definite digitalized
    if s=="is": #in situ 定義為 stage 010
        ans=16
    elif s =="x": #x 定義為 stage 500
        ans=1280
    elif len(s) ==3:
        ans=int(s, 16)
    elif len(s) ==2:
        ans=int(s, 16)*16
    elif len(s) ==1:
        ans=int(s, 16)*256
    return ans

def higherStage(s): #需往後補f 共3位
    if s=="is": #in situ 定義為 stage 010
        ans=16
    elif len(s) ==3:
        ans=int(s, 16)
    elif len(s) ==2:
        ans=int(s, 16)*16 +15
    elif len(s) ==1:
        ans=int(s, 16)*256 + 255
    return ans

def rangeStage(s): #輸入口語回傳16位數範圍
    s = set(s.split("or"))
    ans = set()
    for item in s:
        if "-" in item:
            sp = item.split("-")
            sub = set(range(lowerStage(sp[0]), higherStage(sp[1])+1))
            ans = ans.union(sub)
        else:
            sub = set(range(lowerStage(item), higherStage(item)+1))
            ans = ans.union(sub)
    return ans

def matchingStage(est, ref): #est 是估計的範圍， ref 是病理診斷。輸入口語，回傳是否 match
    if est == "": #未打staging
        ans = "no staging"
    elif est == "x": #無法打 staging
        ans = "x"
    elif ref == "" or ref=="x": #無病理診斷或病理無法判定
        ans = "no patho"
    else:
        est = rangeStage(est)
        ref = rangeStage(ref)
        if ref.issubset(est):
            ans = "match"
        else:
            ans = "mismatch"
    return ans

def revertStage(numb): #輸入10進位數字，回傳 staging 字串
    if numb==16: #in situ 定義為 stage 010
        ans = 'is'
    elif numb==1280: #x 定義為 stage 500
        ans = 'x'
    elif numb==0:
        ans = '0'
    else:
        st = format(numb, "x")
        if len(st)==1: 
            st = '00' +st
        elif len(st)==2:
            st = '0' + st
        ans = re.sub("0+$", "", st)
    return(ans)

def specificStage(stage, cancer, category): #輸入口語化，回傳特定癌症精確描述
    ans = set(map(lambda x:revertStage(x), rangeStage(stage))) & AJCC7[cancer][category]
    return(ans)