优化两个简单的嵌套循环

Question

我一直在试图优化以下两个嵌套的循环：

def startbars(query_name, commodity_name): 

    global h_list 
    nc, s, h_list = [], {}, {} 
    query = """ SELECT wbcode, Year, """+query_name+""" 
       FROM innovotable WHERE commodity='"""+commodity_name+"""' and 

       """+query_name+""" != 'NULL' """ 
    rows = cursor.execute(query) 
    for row in rows: 
     n = float(row[2]) 
     s[str(row[0])+str(row[1])] = n 
     nc.append(n) 
    for iso in result: 
     try: 
      for an_year in xrange(1961, 2031, 1): 
       skey = iso+str(an_year) 
       h_list[skey] = 8.0/max(nc) * s[skey] 
     except: 
      pass 

任何想法？谢谢。

Answer 1

您的代码是不完整的，这使得它很难给出很好的建议，但：

1. 内环不依赖于外循环，这样拉出来的外循环。
2. max（nc）是第一个循环后的常量，因此将其从循环中拉出。

您还需要知道当前代码的速度有多慢，以及您需要多快，否则您的优化可能会错位。

你的数据结构都搞砸了。也许一些名单，这将是更快：

def startbars(query_name, commodity_name): 

    assert query_name in INNOVOTABLE_FIELD_NAMES 

    ## TODO: Replace with proper SQL query 
    query = """ SELECT wbcode, Year, """+query_name+""" 
      FROM innovotable WHERE commodity='"""+commodity_name+"""' and 

      """+query_name+""" != 'NULL' """ 
    rows = cursor.execute(query) 

    mapYearToWbcodeToField = {} 
    nc = [] 
    global h_list 
    h_list = {} 

    for row in rows: 
     n = float(row[2]) 
     wbCodeToField = mapYearToWbcodeToField.setdefault(int(row[1]),{}) 
     wbCodeToField[str(row[0])] = n 
     nc.append(n) 

    constant = 8.0/max(nc) 


    for (an_year,wbCodeToField) in mapYearToWbcodeToField.iteritems(): 
     if an_year < 1961 or an_year > 2031: 
      continue 

     for (wbCode,value) in wbCodeToField.iteritems(): 
      if wbCode not in result: 
       continue 

      skey = wbCode+str(an_year) 
      h_list[skey] = constant * value 

或移动所有检查进入第一循环：

def startbars(query_name, commodity_name): 

    assert query_name in INNOVOTABLE_FIELD_NAMES 

    ## TODO: Replace with proper SQL query 
    query = """ SELECT wbcode, Year, """+query_name+""" 
      FROM innovotable WHERE commodity='"""+commodity_name+"""' and 

      """+query_name+""" != 'NULL' """ 
    rows = cursor.execute(query) 

    data = [] 
    maxField = None 

    for row in rows: 
     an_year = int(row[1]) 
     if an_year < 1961 or an_year > 2031: 
      continue 

     wbCode = str(row[0]) 
     if wbCode not in result: 
      continue 

     n = float(row[2]) 

     data.append((wbCode+str(an_year),n)) 
     if maxField is None or n > maxField: 
      maxField = n 

    constant = 8.0/maxField 

    global h_list 
    h_list = {} 

    for (skey,n) in data: 
     h_list[skey] = constant * n