clawmates/crates/cm-api/src/pdf_renderer.rs

//! LLM + Chromium PDF renderer worker — Slice 6.
//!
//! Watches `mission_artifacts` for rows with `render_pdf_status =
//! 'pending'`. For each:
//!   1. Read the source MD from `<mission_root>/<path>` on disk
//!   2. Call the configured LLM (default: Gemini 2.5 Flash) with a
//!      "produce styled HTML" prompt anchored to a design-system
//!      example. LLM writes HTML with inline CSS.
//!   3. Print that HTML to PDF via `chromium --headless
//!      --print-to-pdf`
//!   4. Save the PDF alongside the MD, update `rendered_pdf_path` +
//!      status = 'done'
//!
//! Graceful degradation: if `GEMINI_API_KEY` is unset or the
//! chromium binary isn't on PATH, the worker marks the row `failed`
//! with a descriptive error rather than blocking boot. Ops enables
//! rendering by wiring both.
//!
//! The frontend already renders `rendered_pdf_path` as an "Open PDF"
//! button on artifact cards (Slice 2).

use serde_json::json;
use sqlx::PgPool;
use std::path::{Path, PathBuf};
use std::time::Duration;

const POLL_INTERVAL: Duration = Duration::from_secs(30);
const MAX_PARALLEL: usize = 2;
const DEFAULT_MODEL: &str = "gemini-2.5-flash";

/// Where per-mission artifacts land on disk. Overridable so dev vs.
/// prod can move the tree; matches the pattern in
/// `research_container::research_workspace_root`.
fn missions_root() -> PathBuf {
    std::env::var("CLAWMATES_MISSIONS_ROOT")
        .map(PathBuf::from)
        .unwrap_or_else(|_| PathBuf::from("/var/lib/clawmates-missions"))
}

fn chromium_bin() -> String {
    std::env::var("CHROMIUM_BIN").unwrap_or_else(|_| "chromium".to_string())
}

fn renderer_model() -> String {
    std::env::var("CLAWMATES_PDF_RENDERER_MODEL").unwrap_or_else(|_| DEFAULT_MODEL.to_string())
}

/// Spawn the poller. No-op-friendly: if there's nothing pending or
/// no rendering pipeline configured, we still tick + observe.
pub fn spawn(pool: PgPool) {
    tokio::spawn(async move {
        // Small startup delay so migrations + loaders finish first.
        tokio::time::sleep(Duration::from_secs(8)).await;
        let mut ticker = tokio::time::interval(POLL_INTERVAL);
        ticker.tick().await;
        loop {
            ticker.tick().await;
            if let Err(e) = sweep_once(&pool).await {
                eprintln!("pdf_renderer: sweep failed: {e}");
            }
        }
    });
}

async fn sweep_once(pool: &PgPool) -> Result<(), String> {
    let pending = cm_db::repo::missions::next_pdf_pending(pool, MAX_PARALLEL as i64)
        .await
        .map_err(|e| format!("next_pdf_pending: {e}"))?;
    for artifact in pending {
        let pool = pool.clone();
        let id = artifact.id;
        tokio::spawn(async move {
            match render_one(&pool, &artifact).await {
                Ok(pdf_path) => {
                    let _ = cm_db::repo::missions::set_pdf_result(&pool, id, Some(&pdf_path), None)
                        .await;
                    eprintln!("pdf_renderer: rendered {id} → {pdf_path}");
                }
                Err(e) => {
                    let _ = cm_db::repo::missions::set_pdf_result(&pool, id, None, Some(&e)).await;
                    eprintln!("pdf_renderer: {id} failed: {e}");
                }
            }
        });
    }
    Ok(())
}

async fn render_one(
    _pool: &PgPool,
    artifact: &cm_db::repo::missions::MissionArtifact,
) -> Result<String, String> {
    // 1. Locate the source MD on disk.
    let mission_root = missions_root().join(artifact.mission_id.to_string());
    let src_path = mission_root.join(&artifact.path);
    let md = tokio::fs::read_to_string(&src_path)
        .await
        .map_err(|e| format!("read {}: {e}", src_path.display()))?;

    // 2. LLM → styled HTML.
    let html = md_to_html_via_llm(&md, artifact.title.as_deref())
        .await
        .map_err(|e| format!("llm render: {e}"))?;

    // 3. Chromium → PDF.
    let tmp = tempdir_for(artifact.id)?;
    let html_path = tmp.join("in.html");
    let pdf_path = tmp.join("out.pdf");
    tokio::fs::write(&html_path, html)
        .await
        .map_err(|e| format!("write {}: {e}", html_path.display()))?;

    let status = tokio::process::Command::new(chromium_bin())
        .args([
            "--headless=new",
            "--disable-gpu",
            "--no-sandbox",
            "--hide-scrollbars",
            &format!("--print-to-pdf={}", pdf_path.display()),
            "--print-to-pdf-no-header",
            "--virtual-time-budget=10000",
            &format!("file://{}", html_path.display()),
        ])
        .stderr(std::process::Stdio::piped())
        .stdout(std::process::Stdio::piped())
        .status()
        .await
        .map_err(|e| format!("spawn chromium: {e}"))?;
    if !status.success() {
        return Err(format!("chromium exited {status}"));
    }

    // 4. Move next to the source MD so the artifact tree stays self-
    // contained. Filename derived from the MD path (foo.md → foo.pdf).
    let out_rel = pdf_sibling(&artifact.path);
    let out_abs = mission_root.join(&out_rel);
    if let Some(parent) = out_abs.parent() {
        tokio::fs::create_dir_all(parent)
            .await
            .map_err(|e| format!("mkdir {}: {e}", parent.display()))?;
    }
    tokio::fs::copy(&pdf_path, &out_abs)
        .await
        .map_err(|e| format!("copy pdf: {e}"))?;
    // Best-effort tmp cleanup — the temp dir lives under /tmp so the
    // OS will reap it anyway.
    let _ = tokio::fs::remove_dir_all(&tmp).await;
    Ok(out_rel)
}

/// Ask the configured LLM to turn `md` into a fully self-contained
/// styled HTML doc. Uses whichever provider `CLAWMATES_PDF_RENDERER_MODEL`
/// resolves to. Defaults to Gemini 2.5 Flash + GEMINI_API_KEY.
async fn md_to_html_via_llm(md: &str, title: Option<&str>) -> Result<String, String> {
    let model = renderer_model();
    // For now we hardcode the Gemini path — anthropic + openai
    // variants land when the design-system template stabilizes.
    if !model.starts_with("gemini") {
        return Err(format!(
            "renderer model {model} not yet wired (only gemini-* supported in Slice 6)"
        ));
    }
    let api_key =
        std::env::var("GEMINI_API_KEY").map_err(|_| "GEMINI_API_KEY unset".to_string())?;

    let system = r#"You are a document typesetter. Given a Markdown source,
produce ONE self-contained HTML document that:
- Has ALL styles inline in a single <style> block in <head>. No external
  fonts, no external CSS. System font stack only.
- Uses a clean, modern, readable serif for body copy (Georgia / "Iowan Old
  Style" / "Charter" / serif) and a sans for headings.
- Uses ONLY these accent colors: #ff8a7a (heading), #5ec8d8 (link),
  #101014 (body text), #f7f7f8 (page bg).
- Renders code blocks with a monospace stack and a subtle background.
- Uses page-break-inside: avoid on headings and images.
- Puts a document title in an <h1> at the top if provided.
- Includes NOTHING outside the HTML — no ```html fence, no commentary."#;

    let prompt = match title {
        Some(t) => format!("Document title: {t}\n\nMarkdown:\n\n{md}"),
        None => md.to_string(),
    };

    let url = format!(
        "https://generativelanguage.googleapis.com/v1beta/models/{}:generateContent?key={}",
        model, api_key
    );
    let body = json!({
        "system_instruction": { "parts": [{ "text": system }] },
        "contents": [{ "role": "user", "parts": [{ "text": prompt }] }],
        "generationConfig": {
            "temperature": 0.2,
            "maxOutputTokens": 32000,
        }
    });

    let client = reqwest::Client::builder()
        .timeout(Duration::from_secs(120))
        .build()
        .map_err(|e| format!("http client: {e}"))?;
    let resp = client
        .post(&url)
        .json(&body)
        .send()
        .await
        .map_err(|e| format!("gemini call: {e}"))?;
    if !resp.status().is_success() {
        let code = resp.status();
        let body = resp.text().await.unwrap_or_default();
        return Err(format!("gemini {code}: {}", &body[..body.len().min(500)]));
    }
    let json: serde_json::Value = resp.json().await.map_err(|e| format!("gemini json: {e}"))?;
    let text = json
        .pointer("/candidates/0/content/parts/0/text")
        .and_then(|v| v.as_str())
        .ok_or_else(|| "gemini response missing text".to_string())?;
    // Strip a stray ```html fence if the model added one despite the
    // system prompt — cheap belt to the suspenders.
    let cleaned = text
        .trim()
        .strip_prefix("```html")
        .and_then(|s| s.strip_suffix("```"))
        .map(|s| s.trim())
        .unwrap_or(text.trim())
        .to_string();
    Ok(cleaned)
}

fn tempdir_for(id: uuid::Uuid) -> Result<PathBuf, String> {
    let dir = std::env::temp_dir().join(format!("clawmates-pdf-{id}"));
    std::fs::create_dir_all(&dir).map_err(|e| format!("mkdir tmp: {e}"))?;
    Ok(dir)
}

/// `research/v3/spec.md` → `research/v3/spec.pdf`.
/// `foo/bar/without_ext` → `foo/bar/without_ext.pdf` (rare — parser
/// never emits an extension-less MD, but we're defensive).
fn pdf_sibling(md_path: &str) -> String {
    let p = Path::new(md_path);
    let stem = p.file_stem().and_then(|s| s.to_str()).unwrap_or("output");
    let parent = p.parent().map(|x| x.to_string_lossy().to_string());
    let base = format!("{stem}.pdf");
    match parent {
        Some(pp) if !pp.is_empty() => format!("{pp}/{base}"),
        _ => base,
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn pdf_sibling_paths() {
        assert_eq!(pdf_sibling("research/v3/spec.md"), "research/v3/spec.pdf");
        assert_eq!(pdf_sibling("spec.md"), "spec.pdf");
        assert_eq!(pdf_sibling("no_ext"), "no_ext.pdf");
    }
}