将多个标头信息字段附加到文件，直到找到下一个标头

Question

我正在尝试使用 bash 解析许多日志文件。日志文件如下所示：

"",1/8/2016
"Timezone",-6

"Serial No.","000001"
"Location:","LS_trap_2c"
"High temperature limit (�C)",-20
"Low temperature limit (�C)",-40
"Date - Time","Temperature (�C)"
"8/11/2015 12:00",28.0
"8/11/2015 14:00",28.5
"8/11/2015 16:00",24.0

"",1/8/2016
"Timezone",-6

"Serial No.","000002"
"Location:","LS_trap_2D"
"High temperature limit (�C)",-20
"Low temperature limit (�C)",-40
"Date - Time","Temperature (�C)"
"8/11/2015 12:00",28.0
"8/11/2015 14:00",28.5

我想将序列号和位置（以后可能还有其他）附加到每一行，直到到达下一个标题，并将其输出到master.csv 文件。该文件最终应如下所示：

"",1/8/2016
"Timezone",-6

"Serial No.","000001"
"Location:","Trap_2c"
"High temperature limit (�C)",-20
"Low temperature limit (�C)",-40
"Date - Time","Temperature (�C)"
LS_trap_2c,000001,"8/11/2015 12:00",28.0
LS_trap_2c,000001,"8/11/2015 14:00",28.5
LS_trap_2c,000001,"8/11/2015 16:00",24.0

"",1/8/2016
"Timezone",-6

"Serial No.","00002"
"Location:","LS_trap_2D"
"High temperature limit (�C)",-20
"Low temperature limit (�C)",-40
"Date - Time","Temperature (�C)"
LS_trap_2D,00002,"8/11/2015 12:00",28.0
LS_trap_2D,00002,"8/11/2015 14:00",28.5

这是一个帮助我使用 bash sed 处理类似文件的问题：

Bash Append header information to each line of a file until next header found

这个oneliner非常适合查找标题，将其存储在holdspace中并将其添加到每行的前面

sed -r '/^"/h;//!{G;s/(.*)\n.*"(.*)"/\2,\1/}' fil.csv  >masfil.csv

这种方法不适用于将多个字符串附加到前面，因为我不确定如何在 sed 中使用多个保留空间。另外，我不确定 sed 是否是最好的方法。我对 sed 不是很熟悉，所以任何指针都将不胜感激。

Answer 1

sed 中只有一个保持空间，但在这种情况下不需要多个：

/^"Serial No."/ {       # If we are on the "Serial No." line...
    N                   # Append next line to pattern space
    h                   # Copy pattern space to hold space
    # Remove everything but location and serial number from pattern space
    s/"[^"]*","([^"]*)"\n"[^"]*","([^"]*)"/\1,\2,/
    x                   # Swap pattern space and hold space
}
/^"[[:digit:]]/ {       # We are on a line where we want to prepend our data
    G                   # Append hold space to pattern space
    s/(.*)\n(.*)/\2\1/  # Move hold space content to front of pattern space
}

如果这是存储在文件sedscr.sed中，可以这样调用

sed -E -f sedscr.sed infile

这确实删除了双引号，如示例输入/输出中所示；它还假设应该在数据前面添加日期的行是带有日期的行，即以双引号和数字开头。

Answer 2

awk 来救援！

假设您的数据是一致的

awk -F, '/"Serial No."/ {sn = $2} 
         /"Location:"/  {loc = $2} 
         /"([0-9]{1,2}\/){2}[0-9]{4} [0-9]{2}:[0-9]{2}"/ 
                        {$0 = loc FS sn FS $0}1' file

在分配 sn 和 loc 时，您也可以使用 gsub(/"/,"",$2) 去掉引号，但由于其余字段都被引用，因此不确定要删除它们。