mdparser.stub.php

<?php

/** @generate-class-entries */

namespace MdParser;

final class Exception extends \RuntimeException
{
}

/**
 * IMPORTANT: default values below MUST agree with the `default_value`
 * column in `mdparser_options_fields[]` inside `mdparser_options.c`.
 * ZPP does not auto-apply arginfo defaults to internal methods, so
 * the C constructor seeds values[] from the C table and this stub is
 * only used for reflection / IDE signatures. If you change a default,
 * change it in both places.
 */
final readonly class Options
{
    public bool $sourcepos;
    public bool $hardbreaks;
    public bool $nobreaks;
    public bool $smart;
    public bool $unsafe;
    public bool $validateUtf8;
    public bool $githubPreLang;
    public bool $liberalHtmlTag;
    public bool $footnotes;
    public bool $strikethroughDoubleTilde;
    public bool $tablePreferStyleAttributes;
    public bool $fullInfoString;
    public bool $tables;
    public bool $strikethrough;
    public bool $tasklist;
    public bool $autolink;
    public bool $tagfilter;
    public bool $headingAnchors;
    public bool $nofollowLinks;

    public function __construct(
        bool $sourcepos = false,
        bool $hardbreaks = false,
        bool $nobreaks = false,
        bool $smart = false,
        bool $unsafe = false,
        bool $validateUtf8 = true,
        bool $githubPreLang = true,
        bool $liberalHtmlTag = false,
        bool $footnotes = false,
        bool $strikethroughDoubleTilde = false,
        bool $tablePreferStyleAttributes = false,
        bool $fullInfoString = false,
        bool $tables = true,
        bool $strikethrough = true,
        bool $tasklist = true,
        bool $autolink = true,
        bool $tagfilter = true,
        bool $headingAnchors = false,
        bool $nofollowLinks = false,
    ) {}

    /**
     * Maximum-safety preset: the standard defaults plus autolink off,
     * so bare URLs in untrusted input do not get wrapped in live <a>
     * tags. Use for forum comments, email rendering, or any rendering
     * path where the source is untrusted and link creation should be
     * explicit.
     */
    public static function strict(): Options {}

    /**
     * GitHub-flavored preset: standard defaults plus footnotes on, to
     * match the feature set github.com renders for README files and
     * issue comments. Everything else (tables, strikethrough,
     * tasklist, autolink, tagfilter) is already on in the default
     * constructor.
     */
    public static function github(): Options {}

    /**
     * Trusted-input preset: raw HTML passthrough (unsafe: true),
     * tagfilter disabled, and liberal HTML tag parsing. Use only when
     * the markdown source is authored by you or comes from a trusted
     * pipeline; this preset explicitly disables the XSS safety net.
     */
    public static function permissive(): Options {}
}

final class Parser
{
    public readonly Options $options;

    public function __construct(?Options $options = null) {}

    public function toHtml(string $source): string {}

    public function toXml(string $source): string {}

    /**
     * Returns a structural representation of the markdown source as
     * a nested array. Link URLs and raw HTML literals are preserved
     * verbatim -- the `unsafe`, `tagfilter`, and URL-scheme defenses
     * apply to the rendering paths (`toHtml` / `toXml` / `toInlineHtml`),
     * NOT to `toAst`. Consumers that emit HTML from the AST must
     * apply their own URL scheme allowlist and HTML sanitization.
     */
    public function toAst(string $source): array {}

    /**
     * Render `$source` as inline-only HTML: no `<p>` wrapper and no
     * block-level constructs. Block markers like `#`, `-`, `>`, `1.`
     * are emitted as literal text instead of being parsed as
     * headings / lists / blockquotes. Matches the semantics of
     * Parsedown::line() and cebe/markdown::parseParagraph() so users
     * migrating from those libraries have a drop-in path for
     * rendering short strings (chat messages, table cell contents,
     * user display names) without the surrounding paragraph tags.
     *
     * `headingAnchors` is silently a no-op for this method (no headings
     * are ever emitted in inline mode); `nofollowLinks` still applies.
     * On empty or whitespace-only input the return value is the empty
     * string. Literal U+200B (zero-width space) bytes in the source
     * are stripped as collateral of the per-line sentinel mechanism.
     */
    public function toInlineHtml(string $source): string {}

    /**
     * Static shortcut: parse `$source` with the default Options and
     * return HTML. Equivalent to `(new Parser)->toHtml($source)` but
     * without the object boilerplate for one-off conversions. Mirrors
     * `Markdown::defaultTransform()` from michelf/php-markdown.
     */
    public static function html(string $source): string {}

    /**
     * Static shortcut: parse `$source` with the default Options and
     * return CommonMark XML.
     */
    public static function xml(string $source): string {}

    /**
     * Static shortcut: parse `$source` with the default Options and
     * return the nested-array AST.
     */
    public static function ast(string $source): array {}
}