在每个时间戳拆分评论

Question

嘿，我在一个单元格中有一个带有各种时间戳的评论，如下所示：-

2019-07-26 20:36:19-（工作笔记）通知呼叫者交易已从 Concur 中删除。将 INC 解决为没有待处理的操作。向来电者发送解决方案电子邮件，复制粘贴响应和 simplify/summarize 从 Eng 团队收到的信息是更新工作说明是将状态更新为等待用户是

2019-07-26 10:32:05 - oneflow（工作笔记）[code] 嗨 Team.We 已经删除了那些 gits。

我想要的是将这个单元格拆分成行，以便每个时间戳都与其各自的文本拆分。

请帮忙。 R 或 Python 中的任何代码都会有所帮助。

Answer 1

Python 选项使用 regex:

import re

s = """2019-07-26 20:36:19 - (Work notes) Informed the caller that the [...]
line without timestamp!
2019-07-26 10:32:05 - oneflow (Work notes)[code] Hi Team.We have removed those gits."""

# search for the timestamps
timestamps = re.findall(r'\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}', s)

# if timestamps were found, obtain their indices in the string:
if timestamps:
    idx = [s.index(t) for t in timestamps] + [None] # add None to get the last part...

    # split the string and put the results in tuples:
    text_tuples = []
    l = len(timestamps[0]) # how many characters to expect for the timestamp
    for i, j in zip(idx[:-1], idx[1:]): # use zip to iterate over two sequences at once
        text_tuples.append((s[i:i+l], # timestamp
                            s[i+l:j].strip(' - '))) # part before next timestamp

# text_tuples
# [('2019-07-26 20:36:19',
#   '(Work notes) Informed the caller that the [...]\nline without timestamp!\n'),
#  ('2019-07-26 10:32:05',
#   'oneflow (Work notes)[code] Hi Team.We have removed those gits.')]

在此示例中，您将获得包含时间戳和该行相应其余部分的元组列表。如果一行没有时间戳，则不会进入输出。

编辑：pandas DataFrame 在 OP 评论后的扩展：

import re
import pandas as pd

# create a custom function to split the comments:
def split_comment(s):
    # search for the timestamps
    timestamps = re.findall(r'\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}', s)

    # if timestamps were found, obtain their indices in the string:
    if timestamps:
        idx = [s.index(t) for t in timestamps] + [None] # add None to get the last part...
        # split the string and put the results in tuples:
        splitted = []
        l = len(timestamps[0]) # how many characters to expect for the timestamp
        for i, j in zip(idx[:-1], idx[1:]): # use zip to iterate over two sequences at once
            splitted.append([s[i:i+l], # timestamp
                             s[i+l:j].strip(' - ')]) # part before next timestamp
        return splitted
    return ['NaT', s] # no timestamp found, return s

s0 = """2019-07-26 20:36:19 - (Work notes) Informed the caller that the [...]
line without timestamp!
2019-07-26 10:32:05 - oneflow (Work notes)[code] Hi Team.We have removed those gits."""
s1 = "2019-07-26 20:36:23  another comment"

# create example df
df = pd.DataFrame({'s': [s0, s1], 'id': [0, 1]})

# create a dummy column that holds the resulting series we get if we apply the function:
df['tmp'] = df['s'].apply(split_comment)

# explode the df so we have one row for each timestamp / comment pair:
df = df.explode('tmp').reset_index(drop=True)

# create two columns from the dummy column, 'timestamp' and 'comment':
df[['timestamp', 'comment']] = pd.DataFrame(df['tmp'].to_list(), index=df.index)

# drop stuff we dont need anymore:
df = df.drop(['s', 'tmp'], axis=1)

# so now we have:
# df
#    id            timestamp                                            comment
# 0   0  2019-07-26 20:36:19  (Work notes) Informed the caller that the [......
# 1   0  2019-07-26 10:32:05  oneflow (Work notes)[code] Hi Team.We have rem...
# 2   1  2019-07-26 20:36:23                                    another comment

在每个时间戳拆分评论

Split Comment at Each TimeStamp

r

datetime

text

split

comments