「検索する」から「AIに聞く」時代へ——
ChatGPT・Perplexity・Google AIに信頼情報源として引用されるWebサイト構造化（GEO/AEO）実践ガイド

#![forbid(unsafe_code)]

use std::fmt::Write;

/// Represents a canonical Q&A pair shared identically between visible HTML and JSON-LD FAQPage.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct CanonicalFaqItem<'a> {
    pub question: &'a str,
    pub answer: &'a str,
}

/// Represents a verified technical article node for Schema.org `@graph` serialization.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct CanonicalArticleNode<'a> {
    pub canonical_url: &'a str,
    pub headline: &'a str,
    pub description: &'a str,
    pub author_name: &'a str,
    pub author_credential: &'a str,
    pub date_published: &'a str,
    pub date_modified: &'a str,
    pub faqs: &'a [CanonicalFaqItem<'a>],
}

#[derive(Debug, Clone, PartialEq, Eq)]
pub enum GeoValidationError {
    InvalidCanonicalScheme,
    EmptyHeadlineOrDescription,
    InsufficientFactDensity { found_numeric_tokens: usize, minimum_required: usize },
    FaqAnswerTooShort { index: usize },
}

/// Escapes a UTF-8 string for safe inclusion inside a JSON string literal,
/// additionally escaping `<`, `>`, and `&` to prevent `</script>` breakout in HTML.
pub fn write_html_safe_json_escaped(out: &mut String, input: &str) {
    for ch in input.chars() {
        match ch {
            '"' => out.push_str("\\\""),
            '\\' => out.push_str("\\\\"),
            '\n' => out.push_str("\\n"),
            '\r' => out.push_str("\\r"),
            '\t' => out.push_str("\\t"),
            '<' => out.push_str("\\u003c"),
            '>' => out.push_str("\\u003e"),
            '&' => out.push_str("\\u0026"),
            c if c.is_control() => {
                let _ = write!(out, "\u{:04x}", c as u32);
            }
            c => out.push(c),
        }
    }
}

/// Counts numeric/specification tokens in text to ensure RAG citation readiness.
pub fn count_verifiable_fact_tokens(text: &str) -> usize {
    text.split(|c: char| c.is_whitespace() || c == '、' || c == '。' || c == '（' || c == '）')
        .filter(|tok| !tok.is_empty() && tok.chars().any(|c| c.is_ascii_digit()))
        .count()
}

/// Validates fact density and serializes a deterministic Schema.org `@graph` JSON-LD payload.
pub fn serialize_verified_jsonld_graph(
    node: &CanonicalArticleNode<'_>,
    min_fact_tokens: usize,
) -> Result<String, GeoValidationError> {
    if !node.canonical_url.starts_with("https://ok.st/") {
        return Err(GeoValidationError::InvalidCanonicalScheme);
    }
    if node.headline.trim().is_empty() || node.description.trim().is_empty() {
        return Err(GeoValidationError::EmptyHeadlineOrDescription);
    }

    let mut total_facts = count_verifiable_fact_tokens(node.description);
    for (idx, faq) in node.faqs.iter().enumerate() {
        if faq.answer.chars().count() < 20 {
            return Err(GeoValidationError::FaqAnswerTooShort { index: idx });
        }
        total_facts += count_verifiable_fact_tokens(faq.answer);
    }

    if total_facts < min_fact_tokens {
        return Err(GeoValidationError::InsufficientFactDensity {
            found_numeric_tokens: total_facts,
            minimum_required: min_fact_tokens,
        });
    }

    let mut buf = String::with_capacity(2048);
    buf.push_str("{\"@context\":\"https://schema.org\",\"@graph\":[{\"@type\":[\"TechArticle\",\"LearningResource\"],\"@id\":\"");
    write_html_safe_json_escaped(&mut buf, node.canonical_url);
    buf.push_str("#article\",\"headline\":\"");
    write_html_safe_json_escaped(&mut buf, node.headline);
    buf.push_str("\",\"description\":\"");
    write_html_safe_json_escaped(&mut buf, node.description);
    buf.push_str("\",\"author\":{\"@type\":\"Person\",\"name\":\"");
    write_html_safe_json_escaped(&mut buf, node.author_name);
    buf.push_str("\",\"jobTitle\":\"");
    write_html_safe_json_escaped(&mut buf, node.author_credential);
    buf.push_str("\"},\"datePublished\":\"");
    write_html_safe_json_escaped(&mut buf, node.date_published);
    buf.push_str("\",\"dateModified\":\"");
    write_html_safe_json_escaped(&mut buf, node.date_modified);
    buf.push_str("\"},{\"@type\":\"FAQPage\",\"@id\":\"");
    write_html_safe_json_escaped(&mut buf, node.canonical_url);
    buf.push_str("#faq\",\"mainEntity\":[");

    for (i, faq) in node.faqs.iter().enumerate() {
        if i > 0 {
            buf.push(',');
        }
        buf.push_str("{\"@type\":\"Question\",\"name\":\"");
        write_html_safe_json_escaped(&mut buf, faq.question);
        buf.push_str("\",\"acceptedAnswer\":{\"@type\":\"Answer\",\"text\":\"");
        write_html_safe_json_escaped(&mut buf, faq.answer);
        buf.push_str("\"}}");
    }
    buf.push_str("]}]}");
    Ok(buf)
}