在熊猫中合并NaN行而不会丢失单元格值

Question

我有与此相似的数据框

status  time  text
  1     12:25 some text
  NaN   NaN   status 1 txt
  NaN   NaN   s1
  2     15:23 some text
  NaN   NaN   status 2 txt
  NaN   NaN   s2

我想按状态合并行，但是我不想丢失文本单元格，像这样。

status  time  text
1       12:25 some text status 1 txt s1
2       15:23 some text status 2 txt s2

我已经尝试过按这种状态进行分组，但是我丢失了文本单元格。

df = df.groupby("status")[["time", "text"]].first().reset_index()

Answer 1

尝试：

df["grp"]=(~df.status.isna()|~df.time.isna()).cumsum()
df=df.groupby("grp").agg({"status": "first", "time": "first", "text": " ".join})
#optionally:
#df=df.groupby("grp").agg({"status": "first", "time": "first", "text": " ".join}).reset_index(drop=True)

输出：

     status   time                       text
grp
1       1.0  12:25  some text status 1 txt s1
2       2.0  15:23  some text status 2 txt s2