extra_total_from_invoice.php

<?php

function extractTotalAPagar(string $text): ?array
{
    $norm = mb_strtolower(removeAccents($text));

    // 1) Same-line: "Total a pagar: RD$ 1,234.56" or "TOTAL A PAGAR 1.234,56"
    $patternSameLine = '/total\s*a\s*pagar[^\S\r\n]*[:\-]?[^\S\r\n]*([a-z]{0,3}\$)?[^\S\r\n]*([\d\.\,\s]+)\b/u';
    if (preg_match($patternSameLine, $norm, $m, PREG_OFFSET_CAPTURE)) {
        $raw   = trim(matchRawSlice($text, $m[1][1], $m[2][1] + strlen($m[2][0]) - $m[1][1]));
        $value = parseMoneyAmount($raw);
        if ($value !== null) {
            return ['label' => 'Total a pagar', 'raw' => $raw, 'value' => $value];
        }
    }

    // 2) Next-line: 
    // "Total a pagar:"
    // "RD$ 1.234,56"
    $patternNextLine = '/total\s*a\s*pagar[^\S\r\n]*[:\-]?[^\S\r\n]*\R[^\S\r\n]*([a-z]{0,3}\$)?[^\S\r\n]*([\d\.\,\s]+)\b/u';
    if (preg_match($patternNextLine, $norm, $m, PREG_OFFSET_CAPTURE)) {
        $raw   = trim(matchRawSlice($text, $m[1][1], $m[2][1] + strlen($m[2][0]) - $m[1][1]));
        $value = parseMoneyAmount($raw);
        if ($value !== null) {
            return ['label' => 'Total a pagar', 'raw' => $raw, 'value' => $value];
        }
    }

    // 3) Line-by-line fallback: find a line with the label, grab the nearest number
    $lines = preg_split("/\R/u", $text);
    foreach ($lines as $i => $line) {
        if (preg_match('/total\s*a\s*pagar/i', removeAccents($line))) {
            // Same line
            if (preg_match('/([A-Z]{0,3}\$)?\s*([\d\.\,]+)\b/u', $line, $mm)) {
                $raw   = trim($mm[0]);
                $value = parseMoneyAmount($raw);
                if ($value !== null) return ['label'=>'Total a pagar','raw'=>$raw,'value'=>$value];
            }
            // Next 3 lines
            for ($j = $i + 1; $j <= $i + 3 && $j < count($lines); $j++) {
                if (preg_match('/([A-Z]{0,3}\$)?\s*([\d\.\,]+)\b/u', $lines[$j], $mm)) {
                    $raw   = trim($mm[0]);
                    $value = parseMoneyAmount($raw);
                    if ($value !== null) return ['label'=>'Total a pagar','raw'=>$raw,'value'=>$value];
                }
            }
        }
    }

    return null;
}

/** Remove accents for matching (á -> a, é -> e, etc.) */
function removeAccents(string $s): string
{
    $t = iconv('UTF-8', 'ASCII//TRANSLIT//IGNORE', $s);
    return $t !== false ? $t : $s;
}

/** Slice the original text using byte offsets from a regex on a normalized string */
function matchRawSlice(string $text, int $start, int $length): string
{
    if ($start < 0 || $length <= 0) return '';
    return mb_substr($text, mb_strlen(mb_strcut($text, 0, $start, 'UTF-8'), 'UTF-8'), 
                     mb_strlen(mb_strcut($text, $start, $length, 'UTF-8'), 'UTF-8'), 'UTF-8');
}

/** Parse "RD$ 1,234.56" or "$ 1.234,56" to float(1234.56) safely */
function parseMoneyAmount(string $raw): ?float
{
    $s = trim($raw);
    // Remove currency words/symbols/spaces
    $s = preg_replace('/(?i)(rd\$|usd|\$|dop|ars|mxn|cop|pen|clp|\s)+/u', '', $s);
    $s = trim($s);

    // Decide decimal separator
    $hasDot   = strpos($s, '.') !== false;
    $hasComma = strpos($s, ',') !== false;

    if ($hasDot && $hasComma) {
        // Assume thousands = dot, decimal = comma (1.234,56)
        $s = str_replace('.', '', $s);
        $s = str_replace(',', '.', $s);
    } elseif ($hasComma && !$hasDot) {
        // Assume comma is decimal (1234,56)
        $s = str_replace('.', '', $s); // just in case
        $s = str_replace(',', '.', $s);
    } else {
        // Only dot or plain digits: dot is decimal
        $s = str_replace(',', '', $s);
    }

    // Remove any remaining spaces
    $s = preg_replace('/\s+/', '', $s);

    return is_numeric($s) ? (float)$s : null;
}