如何从Powershell中的一个.csv文件读取多个数据集

Question

I have a temp recorder that (daily) reads multiple sensors and saves the data to a single .csv with a whole bunch of header information before each set of date/time and temperature. the file looks something like this:

"readerinfo","onlylistedonce"
"downloadinfo",YYYY/MM/DD 00:00:00
"timezone",-8
"headerstuff","headersuff"

"sensor1","sensorstuff"
"serial#","0000001"
"about15lines","ofthisstuff"
"header1","header2"
datetime,temp
datetime,temp
datetime,temp

"sensor2","sensorstuff"
"serial#","0000002"
"about15lines","ofthisstuff"
"header1","header2"
datetime,temp
datetime,temp
datetime,temp
"downloadcomplete"

My aim is to pull out the date/time and temp data for each sensor and save it as a new file so that I can run some basic stats(hi/lo/avg temp)on it. (It would be beautiful if I could somehow identify which sensor the data came from based on a serial number listed in the header info, but that's less important than separating out the data into sets) The lengths of the date/time lists change from sensor to sensor based on how long they've been recording and the number of sensors changes daily also. Even if I could just split the sensor data, header info and all, into however many files there are sensors, that would be a good start.

Answer 1

This isn't exactly a CSV file in the traditional sense. I imagine you already know this, given your description of the file contents.

If the lines with datetime,temp truly do not have any double quotes in them, per your example data, then the following script should work. This script is self-containing, since it declares the example data in-line.

IMPORTANT : You will need to modify the line containing the declaration of the $SensorList variable. You will have to populate this variable with the names of the sensors, or you can parameterize the script to accept an array of sensor names.

UPDATE : I changed the script to be parameterized.

Results

The results of the script are as follows:

1. sensor1.csv (with corresponding data)
2. sensor2.csv (with corresponding data)
3. Some green text will be written to the PowerShell host, indicating which sensor is currently detected

Script

The contents of the script should appear as follows. Save the script file to a folder, such as c:\\test\\test.ps1 , and then execute it.

# Declare text as a PowerShell here-string
$Text = @"
"readerinfo","onlylistedonce"
"downloadinfo",YYYY/MM/DD 00:00:00
"timezone",-8
"headerstuff","headersuff"

"sensor1","sensorstuff"
"serial#","0000001"
"about15lines","ofthisstuff"
"header1","header2"
datetime,tempfromsensor1
datetime,tempfromsensor1
datetime,tempfromsensor1

"sensor2","sensorstuff"
"serial#","0000002"
"about15lines","ofthisstuff"
"header1","header2"
datetime,tempfromsensor2
datetime,tempfromsensor2
datetime,tempfromsensor2
"downloadcomplete"
"@.Split("`n");

# Declare the list of sensor names
$SensorList = @('sensor1', 'sensor2');
$CurrentSensor = $null;

# WARNING: Clean up all CSV files in the same directory as the script
Remove-Item -Path $PSScriptRoot\*.csv;

# Iterate over each line in the text file
foreach ($Line in $Text) {
    #region Line matches double quote
    if ($Line -match '"') {
        # Parse the property/value pairs (where double quotes are present)
        if ($Line -match '"(.*?)",("(?<value>.*)"|(?<value>.*))') {
            $Entry = [PSCustomObject]@{
                Property = $matches[1];
                Value = $matches['value'];
            };
            if ($matches[1] -in $SensorList) {
                $CurrentSensor = $matches[1];
                Write-Host -ForegroundColor Green -Object ('Current sensor is: {0}' -f $CurrentSensor);
            }
        }        
    }
    #endregion Line matches double quote
    #region Line does not match double quote
    else {
        # Parse the datetime/temp pairs
        if ($Line -match '(.*?),(.*)') {
            $Entry = [PSCustomObject]@{
                DateTime = $matches[1];
                Temp = $matches[2];
            };
            # Write the sensor's datetime/temp to its file
            Add-Content -Path ('{0}\{1}.csv' -f $PSScriptRoot, $CurrentSensor) -Value $Line;
        }
    }
    #endregion Line does not match double quote
}

Answer 2

Using the data sample you provided, the output of this script would as follows:

C:\\sensoroutput_20140204.csv

sensor1,datetime,temp
sensor1,datetime,temp
sensor1,datetime,temp
sensor2,datetime,temp
sensor2,datetime,temp
sensor2,datetime,temp

I believe this is what you are looking for. The assumption here is the new line characters. The get-content line is reading the data and breaking it into "sets" by using 2 new line characters as the delimiter to split on. I chose to use the environment's (Windows) new line character. Your source file may have different new line characters. You can use Notepad++ to see which characters they are eg \\r\\n, \\n, etc.

$newline = [Environment]::NewLine
$srcfile = "C:\sensordata.log"
$dstpath = 'C:\sensoroutput_{0}.csv' -f (get-date -f 'yyyyMMdd')

# Reads file as a single string with out-string
# then splits with a delimiter of two new line chars
$datasets = get-content $srcfile -delimiter ($newline * 2)

foreach ($ds in $datasets) {
  $lines = ($ds -split $newline)                   # Split dataset into lines
  $setname = $lines[0] -replace '\"(\w+).*', '$1'  # Get the set or sensor name
  $lines | % {
    if ($_ -and $_ -notmatch '"') {                # No empty lines and no lines with quotes
      $data = ($setname, ',', $_ -join '')         # Concats set name, datetime, and temp
      Out-File -filepath $dstpath -inputObject $data -encoding 'ascii' -append
    }
  }
}