120 lines
4.2 KiB
PowerShell
120 lines
4.2 KiB
PowerShell
# Test parsing on local 315cuisine HTML
|
|
|
|
$html = [System.IO.File]::ReadAllText('C:\Users\DEREKC~1.CHN\AppData\Local\Temp\test-restaurant.html', [System.Text.Encoding]::UTF8)
|
|
|
|
$areaMap = [ordered]@{
|
|
"AIRWAY HEIGHTS" = "Airway Heights"
|
|
"ATHOL" = "Athol"
|
|
"COEUR D'ALENE" = "Coeur d'Alene"
|
|
"POST FALLS" = "Post Falls"
|
|
"HAYDEN" = "Hayden"
|
|
"LIBERTY LAKE" = "Liberty Lake"
|
|
"NORTH SPOKANE" = "North Spokane"
|
|
"SOUTH SPOKANE" = "South Spokane"
|
|
"SPOKANE VALLEY" = "Spokane Valley"
|
|
"WEST SPOKANE" = "West Spokane"
|
|
"WORLEY" = "Worley"
|
|
"DOWNTOWN" = "Downtown"
|
|
}
|
|
|
|
function Get-CleanText($rawHtml) {
|
|
$t = $rawHtml -replace '<[^>]+>', ' '
|
|
$t = $t -replace '&', '&'
|
|
$t = $t -replace '<', '<'
|
|
$t = $t -replace '>', '>'
|
|
$t = $t -replace '"', '"'
|
|
$t = $t -replace ''', "'"
|
|
$t = $t -replace ' ', ' '
|
|
$t = $t -replace '\s+', ' '
|
|
$t.Trim()
|
|
}
|
|
|
|
function Extract-Dishes($courseHtml) {
|
|
$dishes = [System.Collections.ArrayList]@()
|
|
$opts = [System.Text.RegularExpressions.RegexOptions]::Singleline
|
|
|
|
$pMatches = [regex]::Matches($courseHtml, '<p[^>]*>(.*?)</p>', $opts)
|
|
|
|
foreach ($pm in $pMatches) {
|
|
$pContent = $pm.Groups[1].Value
|
|
if ($pContent -notmatch '<strong>') { continue }
|
|
|
|
$nameM = [regex]::Match($pContent, '<strong>(.*?)</strong>', $opts)
|
|
if (-not $nameM.Success) { continue }
|
|
$name = Get-CleanText $nameM.Groups[1].Value
|
|
|
|
if ($name -match '^(GF|GFA|V\+?|DF|V:)$') { continue }
|
|
if ($name.Length -lt 3) { continue }
|
|
if ($name -match '^[A-Z]{1,3}:') { continue }
|
|
if ($name.Length -gt 80) { continue }
|
|
|
|
$afterBr = ''
|
|
if ($pContent -match '<br\s*/?>(.*?)$') {
|
|
$afterBr = $matches[1]
|
|
} else {
|
|
$afterStrong = [regex]::Match($pContent, '</strong>(.*?)$', $opts)
|
|
if ($afterStrong.Success) { $afterBr = $afterStrong.Groups[1].Value }
|
|
}
|
|
|
|
$desc = Get-CleanText $afterBr
|
|
$null = $dishes.Add([PSCustomObject]@{ name = $name; desc = $desc })
|
|
}
|
|
|
|
return ,$dishes
|
|
}
|
|
|
|
function Extract-CourseBlock($html, $courseLabel) {
|
|
$opts = [System.Text.RegularExpressions.RegexOptions]::Singleline
|
|
$pattern = [regex]::Escape($courseLabel) + '.{0,300}?et_pb_text_inner">(.+?)(?=<h[123]|et_pb_text_inner"><h|V:\s*<|Vegetarian item|et_pb_row_inner_[23])'
|
|
$m = [regex]::Match($html, $pattern, $opts)
|
|
if ($m.Success) { return $m.Groups[1].Value }
|
|
|
|
$idx = $html.IndexOf($courseLabel)
|
|
if ($idx -ge 0) {
|
|
$sub = $html.Substring($idx, [Math]::Min(4000, $html.Length - $idx))
|
|
$innerM = [regex]::Match($sub, 'et_pb_text_inner">(.*?)(?=et_pb_text_inner|</div></div></div>)', $opts)
|
|
if ($innerM.Success) { return $innerM.Groups[1].Value }
|
|
}
|
|
return ''
|
|
}
|
|
|
|
$nameM = [regex]::Match($html, '<title>(.+?) \| Inlander')
|
|
Write-Host "Name: $($nameM.Groups[1].Value.Trim())"
|
|
|
|
$priceM = [regex]::Match($html, '<strong>\$(\d+)</strong>')
|
|
Write-Host "Price: $($priceM.Groups[1].Value)"
|
|
|
|
$cuisineM = [regex]::Match($html, 'CUISINE:\s*([A-Z][A-Za-z /]+?)(?:\s*</|\s*<)')
|
|
$cuisine = (Get-Culture).TextInfo.ToTitleCase($cuisineM.Groups[1].Value.Trim().ToLower())
|
|
Write-Host "Cuisine: $cuisine"
|
|
|
|
$phoneM = [regex]::Match($html, '\((?:208|509)\) \d{3}-\d{4}')
|
|
Write-Host "Phone: $($phoneM.Value)"
|
|
|
|
$hoursM = [regex]::Match($html, 'Menu served [^<]+')
|
|
Write-Host "Hours: $($hoursM.Value.Trim())"
|
|
|
|
$areas = @()
|
|
foreach ($aKey in $areaMap.Keys) {
|
|
if ($html.ToUpper().Contains($aKey)) { $areas += $areaMap[$aKey] }
|
|
}
|
|
Write-Host "Areas: $($areas -join ', ')"
|
|
|
|
Write-Host ""
|
|
Write-Host "--- FIRST COURSE ---"
|
|
$fc = Extract-CourseBlock $html 'First Course'
|
|
$firstCourse = Extract-Dishes $fc
|
|
foreach ($d in $firstCourse) { Write-Host " [$($d.name)] | $($d.desc)" }
|
|
|
|
Write-Host ""
|
|
Write-Host "--- SECOND COURSE ---"
|
|
$sc = Extract-CourseBlock $html 'Second Course'
|
|
$secondCourse = Extract-Dishes $sc
|
|
foreach ($d in $secondCourse) { Write-Host " [$($d.name)] | $($d.desc)" }
|
|
|
|
Write-Host ""
|
|
Write-Host "--- THIRD COURSE ---"
|
|
$tc = Extract-CourseBlock $html 'Third Course'
|
|
$thirdCourse = Extract-Dishes $tc
|
|
foreach ($d in $thirdCourse) { Write-Host " [$($d.name)] | $($d.desc)" }
|