高效排列多相似数据

Question

这里显示的数据文件是仪器导出的测量记录。

我上传了here，有兴趣的可以下载。

背景

Sample
RECORD-1
FID1, FID2, front_temperature, laser, laserlow, pressure, mode
-925    284 1452    315 143 16653   He  -28500
-924    281 1462    322 136 16641   He  -28628
-920    281 1455    311 139 16649   He  -28756
-923    279 1454    312 139 16636   He  -28884
......

Sample
RECORD-2
FID1, FID2, front_temperature, laser, laserlow, pressure, mode
-925    284 1452    315 143 16653   He  -28500
......
......  

一般情况下，不同的样品按照检测程序的顺序有多个记录。这些样本的数据记录都是相同的格式。

我的尝试

如果数据文件中只有一个样本（*.txt 格式），我可以将数据文件排列成 pandas。 Dataframe，然后我可以在 Python 中使用更多的分析过程来处理数据。

我的代码在这里显示：

# Whole datafile with several samples record inside
with open("record.txt") as f:
     mylist = f.read().splitlines() 

## The record for each sample length in 803 lines
lines = mylist[0:803]

### The sample_name was extract from the third line
sample_name = lines[2]

### For each sample, the measure record was saved in several aspects, 
### which were regarded as some columns here
columns  = lines[22].split()

### Generate an empty columns for saving data record later.
df  = {columns[0][:-1]:[],columns[1][:-1]:[],columns[2][:-1]:[],columns[3][:-1]:[],columns[4][:-1]:[],
  columns[5][:-1]:[],columns[6][:-1]:[],} #### I only though about this dumb method for now

## Data extracting
### the valid data record of sample 1 was from line 23
for i in range(0, len(lines[23:]),1):
    for j in range(0, len(columns),1):
        df[columns[j][:-1]].append(lines[23+i].split()[j])
pd.DataFrame(df)  

结果如下所示：

我的目标

从上面的代码中，我可以处理一个样本的数据文件。但是当记录文本中表示有几个样本时。我找不到有效处理它的线索。

这是我的目标的插图。生成用于保存所有样本记录的数据框字典。

任何建议将不胜感激！

Answer 1

我认为您正在寻找这样的东西：

import pandas as pd
# Whole datafile with several samples record inside
with open("record.txt",'r') as f:
     mylist = f.read().splitlines() 

dataset = []
while True:

    try:
        ## The record for each sample length in 803 lines
        lines, mylist = mylist[0:803], mylist[803:] #this split your list!!
        ### The sample_name was extract from the third line
        sample_name = lines[2]



        ### For each sample, the measure record was saved in several aspects, 
        ### which were regarded as some columns here
        columns  = lines[22].split()

        ### Generate an empty columns for saving data record later.
        df  = {columns[0][:-1]:[],columns[1][:-1]:[],columns[2][:-1]:[],columns[3][:-1]:[],columns[4][:-1]:[],
               columns[5][:-1]:[],columns[6][:-1]:[],} #### I only though about this dumb method for now

        ## Data extracting
        ### the valid data record of sample 1 was from line 23
        for i in range(0, len(lines[23:]),1):
            for j in range(0, len(columns),1):
                df[columns[j][:-1]].append(lines[23+i].split()[j])

    except IndexError:
        break

    df = pd.DataFrame(df)
    dataset.append(df)

现在dataset[0] 应该包含样本 1 的 df。