mirror of
https://github.com/dfinke/ImportExcel.git
synced 2025-12-06 00:23:20 +00:00
45ed6a06dc
```powershell
$rows = $h.SelectNodes("//table[$TableIndex]//tr")
```
XPath selector in line 53 uses complex expression that can lead to unexpected result. The problem is that HtmlAgilityPack may have specific issues. In particular, on websites containing multiple tables this selector can find not one table. This is aggravated by the fact that tables can have different structures.
To avoid ambiguity this PR suggests to separate queries. Oneliner simplifies error checking
```powershell
$rows = try {
$h.SelectSingleNode("//table[$TableIndex]").SelectNodes(".//tr")
} catch {}
if (-not $rows) {Write-Warning "Could not find rows for `"//table[$TableIndex]`" in $Url ."}
```
This expression doesn't even need testing, it just works.
86 lines
3.7 KiB
PowerShell
86 lines
3.7 KiB
PowerShell
# https://www.leeholmes.com/blog/2015/01/05/extracting-tables-from-powershells-invoke-webrequest/
|
|
# tweaked from the above code
|
|
function Get-HtmlTable {
|
|
param(
|
|
[Parameter(Mandatory=$true)]
|
|
$Url,
|
|
$TableIndex=0,
|
|
$Header,
|
|
[int]$FirstDataRow=0,
|
|
[Switch]$UseDefaultCredentials
|
|
)
|
|
if ($PSVersionTable.PSVersion.Major -gt 5 -and -not (Get-Command ConvertFrom-Html -ErrorAction SilentlyContinue)) {
|
|
# Invoke-WebRequest on .NET core doesn't have ParsedHtml so we need HtmlAgilityPack or similiar Justin Grote's PowerHTML wraps that nicely
|
|
throw "This version of PowerShell needs the PowerHTML module to process HTML Tables."
|
|
}
|
|
|
|
$r = Invoke-WebRequest $Url -UseDefaultCredentials: $UseDefaultCredentials
|
|
$propertyNames = $Header
|
|
|
|
if ($PSVersionTable.PSVersion.Major -le 5) {
|
|
$table = $r.ParsedHtml.getElementsByTagName("table")[$TableIndex]
|
|
$totalRows=@($table.rows).count
|
|
|
|
for ($idx = $FirstDataRow; $idx -lt $totalRows; $idx++) {
|
|
|
|
$row = $table.rows[$idx]
|
|
$cells = @($row.cells)
|
|
|
|
if(!$propertyNames) {
|
|
if($cells[0].tagName -eq 'th') {
|
|
$propertyNames = @($cells | ForEach-Object {$_.innertext -replace ' ',''})
|
|
} else {
|
|
$propertyNames = @(1..($cells.Count + 2) | Foreach-Object { "P$_" })
|
|
}
|
|
continue
|
|
}
|
|
|
|
$result = [ordered]@{}
|
|
|
|
for($counter = 0; $counter -lt $cells.Count; $counter++) {
|
|
$propertyName = $propertyNames[$counter]
|
|
|
|
if(!$propertyName) { $propertyName= '[missing]'}
|
|
$result.$propertyName= $cells[$counter].InnerText
|
|
}
|
|
|
|
[PSCustomObject]$result
|
|
}
|
|
}
|
|
else {
|
|
$h = ConvertFrom-Html -Content $r.Content
|
|
if ($TableIndex -is [valuetype]) { $TableIndex += 1}
|
|
$rows = try {
|
|
$h.SelectSingleNode("//table[$TableIndex]").SelectNodes(".//tr")
|
|
} catch {}
|
|
if (-not $rows) {Write-Warning "Could not find rows for `"//table[$TableIndex]`" in $Url ."}
|
|
if ( -not $propertyNames) {
|
|
if ( $tableHeaders = $rows[$FirstDataRow].SelectNodes("th")) {
|
|
$propertyNames = $tableHeaders.foreach({[System.Web.HttpUtility]::HtmlDecode( $_.innerText ) -replace '\W+','_' -replace '(\w)_+$','$1' })
|
|
$FirstDataRow += 1
|
|
}
|
|
else {
|
|
$c = 0
|
|
$propertyNames = $rows[$FirstDataRow].SelectNodes("td") | Foreach-Object { "P$c" ; $c ++ }
|
|
}
|
|
}
|
|
Write-Verbose ("Property names: " + ($propertyNames -join ","))
|
|
foreach ($n in $FirstDataRow..($rows.Count-1)) {
|
|
$r = $rows[$n].SelectNodes("td|th")
|
|
if ($r -and $r.innerText -ne "" -and $r.count -gt $rows[$n].SelectNodes("th").count ) {
|
|
$c = 0
|
|
$newObj = [ordered]@{}
|
|
foreach ($p in $propertyNames) {
|
|
$n = $null
|
|
#Join descentandts for cases where the text in the cell is split (e.g with a <BR> ). We also want to remove HTML codes, trim and convert unicode minus sign to "-"
|
|
$cellText = $r[$c].Descendants().where({$_.NodeType -eq "Text"}).foreach({[System.Web.HttpUtility]::HtmlDecode( $_.innerText ).Trim()}) -Join " " -replace "\u2212","-"
|
|
if ([double]::TryParse($cellText, [ref]$n)) {$newObj[$p] = $n }
|
|
else {$newObj[$p] = $cellText }
|
|
$c ++
|
|
}
|
|
[pscustomObject]$newObj
|
|
}
|
|
}
|
|
}
|
|
}
|