Extract the first table from an HTML document into a CSV using WordPress’ HTML API

table-to-csv.php

<?php

// Ensure you load the HTML API
//  - require_once __DIR__ . '/wp-load.php';
//  - Manual requires
//  - PHP extension
//  - any other approach

// Extracts the first table in the HTML, if one is found.
// Usage: cat water-report.html | php table-to-csv.php


$html = file_get_contents( 'php://stdin' );

$p = WP_HTML_Processor::create_fragment( $html );

// Find the table
$p->next_tag( 'TABLE' );

while ( $p->next_tag( 'TR' ) ) {
	$col = 0;

	while ( $p->next_token() ) {
		$token_name = $p->get_token_name();

		if ( 'TABLE' === $token_name ) {
			exit(0);
		}

		if ( 'TR' === $token_name ) {
			break;
		}

		if ( 'TD' !== $token_name && 'TH' !== $token_name ) {
			continue;
		}

		$cell = '';
		if ( $col++ > 0 ) {
			echo ",";
		}

		while ( $p->next_token() && $token_name !== $p->get_token_name() ) {
			switch ( $p->get_token_name() ) {
				case '#text':
					$cell .= $p->get_modifiable_text();
					break;
			}
		}

		$cell = trim( str_replace( "\n", '', $cell ) );
		if ( str_contains( $cell, ',' ) ) {
			echo '"' . str_replace( '"', '""', $cell ) . '"';
		} else {
			echo $cell;
		}
	}

	echo "\n";
}