xml2json.php

<?php
/*
 *  xml2json.php: Convert publist XML database to JSON format
 *  Usage: php xml2json.php <input.xml> [output.json]
 *  If output.json is omitted, writes to stdout.
 *
 *  JSON format differences from XML:
 *  - author and editor are arrays of name strings instead of " and "-delimited strings
 *  - [[...]] HTML escaping is expanded to actual HTML characters
 *  - All other fields are plain strings (month and year as integers)
 *
 *  Copyright 2026--2026 by Eitan Frachtenberg (publist@frachtenberg.org)
 *  This program is distributed under the terms of the GNU General Public License
  */

if ($argc < 2) {
    fwrite(STDERR, "Usage: php xml2json.php <input.xml> [output.json]\n");
    exit(1);
}

$fn = $argv[1];
$data = file_get_contents($fn);
if ($data === false) {
    fwrite(STDERR, "Error: cannot read file '$fn'\n");
    exit(1);
}

$parser = xml_parser_create();
xml_parser_set_option($parser, XML_OPTION_CASE_FOLDING, 0);
xml_parser_set_option($parser, XML_OPTION_SKIP_WHITE, 1);
$ok = xml_parse_into_struct($parser, $data, $values, $tags);
xml_parser_free($parser);

if (!$ok) {
    fwrite(STDERR, "Error: failed to parse XML in '$fn'\n");
    exit(1);
}

$pubs = [];

if (isset($tags['publication'])) {
    $ranges = $tags['publication'];
    for ($i = 0; $i < count($ranges); $i += 2) {
        $offset = $ranges[$i] + 1;
        $len    = $ranges[$i + 1] - $offset;
        $slice  = array_slice($values, $offset, $len);

        $pub = [];
        foreach ($slice as $item) {
            $value = isset($item['value']) ? trim($item['value']) : '';
            if ($value === '') continue;
            // Expand [[...]] HTML escaping to real HTML
            $value = str_replace('[[', '<', $value);
            $value = str_replace(']]', '>', $value);
            $pub[$item['tag']] = $value;
        }

        // Normalize author(s)/editor(s) to arrays; keep canonical field names
        foreach (['author' => ['author', 'authors'],
                  'editor' => ['editor', 'editors']] as $dst => $srcs) {
            foreach ($srcs as $src) {
                if (!isset($pub[$src])) continue;
                $names = array_values(array_filter(array_map('trim', explode(' and ', $pub[$src]))));
                $pub[$dst] = $names;
                if ($src !== $dst) unset($pub[$src]);
                break;
            }
        }

        // Store month and year as integers for natural JSON
        foreach (['month', 'year'] as $numfield) {
            if (isset($pub[$numfield]) && ctype_digit($pub[$numfield])) {
                $pub[$numfield] = (int) $pub[$numfield];
            }
        }

        $pubs[] = $pub;
    }
}

$json = json_encode($pubs, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES);

if ($argc >= 3) {
    if (file_put_contents($argv[2], $json . "\n") === false) {
        fwrite(STDERR, "Error: cannot write to '$argv[2]'\n");
        exit(1);
    }
} else {
    echo $json . "\n";
}