有没有办法用多条行写评论pandas SQL查询?

时间:2016-02-01 19:32:51

标签: python sql pandas

编写正则表达式时,可以跨多行编写表达式并包含注释,然后在传递编译版本之前使用re.VERBOSE选项编译表达式。我想用pandas.read_sql_query做类似的事情。

例如,而不是:

result = pd.read_sql_query('select a.gvkey, a.tic, a.datadate as fyearend, year(a.datadate) as year, month(a.datadate) as fyrc, b.datadate, month(b.datadate) as month, b.trt1m from COMPM.FUNDA a join COMPM.SECM b on a.gvkey = b.gvkey and year(a.datadate) = year(b.datadate) where a.TIC = "IBM" and a.datafmt = "STD" and a.consol="C" and a.indfmt = "INDL" and year(a.datadate)>1980', engine)

我想写一些像:

q = """select a.gvkey, 
    a.tic,                      #COMMENTS
    a.datadate as fyearend,     #COMMENTS
    year(a.datadate) as year,   #COMMENTS
    month(a.datadate) as fyrc, b.datadate, 
    month(b.datadate) as month, 
    b.trt1m 
    from COMPM.FUNDA a join COMPM.SECM b on a.gvkey = b.gvkey and year(a.datadate) = year(b.datadate) 
    where a.TIC = "IBM" 
        and a.datafmt = "STD" 
        and a.consol="C" 
        and a.indfmt = "INDL" 
        and year(a.datadate)>1980
"""

result = p.read_sql_query(q ,engine)

我的问题与this有关跨多行拆分python命令的问题有关,但我想在查询中添加注释。

正如我所提到的,我在pandas / SQL案例中想要做的事情类似于使用re.VERBOSE的正则表达式案例中可以做的事情。以下是regex的示例:

pattern = r'''\s(shares?| #COMMENTS
            warrants?|       #COMMENTS
            stock|           #AND SO ON...
            (non)?vest(ed)?
            )\b             
            '''
crit = re.compile(pattern_nopt, re.VERBOSE)
match=re.search(crit, string)

这将使查询更具可读性,我发现在与共同作者共享代码时,详尽地对查询进行注释很重要。

1 个答案:

答案 0 :(得分:5)

是的它会起作用,但你必须使用正确的comment delimiter for SQLite
  --代表内联评论   多行注释的/* foo.. */(如C)

所以它看起来像:

q = """select a.gvkey, 
    a.tic,                      -- COMMENTS
    a.datadate as fyearend,     -- COMMENTS
    year(a.datadate) as year,   /* Another very long
    and multi-lines comment... */
    month(a.datadate) as fyrc, b.datadate, 
    month(b.datadate) as month, 
    b.trt1m from COMPM.FUNDA a join COMPM.SECM b on a.gvkey = b.gvkey and year(a.datadate) = year(b.datadate) 
    where a.TIC = "IBM" 
        and a.datafmt = "STD" 
        and a.consol="C" 
        and a.indfmt = "INDL" 
        and year(a.datadate)>1980
"""

result = p.read_sql_query(q, conn)