gardenserver/src/main.rs

use actix_files::NamedFile;
use actix_web::error::ErrorInternalServerError;
use actix_web::{error, get, http, middleware, web, App, Error, HttpResponse, HttpServer};
use anyhow::anyhow;
use chrono::{DateTime, Local};
use clap::{App as ClapApp, Arg};
use linkify::LinkFinder;
use log::{info, trace};
use percent_encoding::utf8_percent_encode;
use pulldown_cmark::{html, Options, Parser};
use regex::{Captures, Regex};
use std::collections::HashMap;
use std::fs::File;
use std::io::Read;
use std::net::SocketAddr;
use std::path::{Path, PathBuf};
use std::sync::Mutex;
use std::time::SystemTime;
use std::{env, fs};
use tera::{Context, Tera};

#[derive(Clone)]
struct State {
    garden_dir: PathBuf,
    index_file: Option<String>,
    title: Option<String>,
    tera: Tera,
}

struct MutableState {
    garden_cache: Mutex<GardenCache>,
}

#[derive(Clone, Debug)]
struct GardenCache {
    pages: HashMap<PathBuf, ParsedPage>,
    files: Vec<PathBuf>,
}

impl Default for GardenCache {
    fn default() -> Self {
        GardenCache {
            pages: HashMap::new(),
            files: vec![],
        }
    }
}

#[derive(Clone, Debug)]
struct ParsedPage {
    timestamp: Option<SystemTime>,
    title: String,
    html: String,
    links: Vec<String>,
}

const VERSION: &str = env!("CARGO_PKG_VERSION");

fn main() -> anyhow::Result<()> {
    let env = env_logger::Env::default().filter_or(env_logger::DEFAULT_FILTER_ENV, "info");
    env_logger::init_from_env(env);

    let app = ClapApp::new("gardenserver")
        .version(VERSION)
        .author("Tomáš Mládek <t@mldk.cz>")
        .arg(Arg::with_name("DIRECTORY").required(true).index(1))
        .arg(
            Arg::with_name("BIND")
                .long("bind")
                .default_value("127.0.0.1:8642")
                .help("address and port to bind the Web interface on")
                .required(true),
        )
        .arg(
            Arg::with_name("INDEX_FILE")
                .takes_value(true)
                .short("i")
                .long("index")
                .help("File to be served at the root."),
        )
        .arg(
            Arg::with_name("TITLE")
                .takes_value(true)
                .short("t")
                .long("title")
                .help("Title of this digital garden."),
        );

    let matches = app.get_matches();

    let directory = Path::new(matches.value_of("DIRECTORY").unwrap());

    info!(
        "Starting GardenServer {} of {}...",
        VERSION,
        directory.display()
    );

    let tera = Tera::new("templates/**/*.html")?;

    let sys = actix::System::new("gardenserver");

    let bind: SocketAddr = matches
        .value_of("BIND")
        .unwrap()
        .parse()
        .expect("Incorrect bind format.");
    info!("Starting server at: {}", &bind);

    let mutable_state = web::Data::new(MutableState {
        garden_cache: Mutex::new(update_garden(directory, GardenCache::default())?),
    });

    let state = State {
        garden_dir: directory.to_path_buf(),
        index_file: matches.value_of("INDEX_FILE").map(|s| s.to_string()),
        title: matches.value_of("TITLE").map(|s| s.to_string()),
        tera,
    };

    // Start HTTP server
    HttpServer::new(move || {
        App::new()
            .wrap(middleware::Logger::default())
            .data(state.clone())
            .app_data(mutable_state.clone())
            .service(actix_files::Files::new("/static", "templates"))
            .service(render)
    })
    .bind(&bind)?
    .run();

    Ok(sys.run()?)
}

#[get("{path:.*}")]
async fn render(
    request: web::HttpRequest,
    data: web::Data<State>,
    state: web::Data<MutableState>,
    path: web::Path<String>,
) -> Result<HttpResponse, Error> {
    let mut cache = state.garden_cache.lock().unwrap();
    *cache = update_garden(&data.garden_dir, (*cache).clone())
        .map_err(error::ErrorInternalServerError)?;

    // Redirect to index if path is empty.
    if path.is_empty() {
        let location = match data.index_file.as_ref() {
            Some(index_file) => index_file.clone(),
            None => cache
                .files
                .iter()
                .filter(|f| f.to_str().unwrap().ends_with(".md"))
                .collect::<Vec<&PathBuf>>()
                .first()
                .unwrap_or(&cache.files.first().unwrap())
                .display()
                .to_string(),
        };

        return Ok(HttpResponse::Found()
            .header(http::header::LOCATION, location.as_str())
            .finish());
    }

    let full_path = data.garden_dir.join(path.as_str());

    // Redirect to ".md" version if requested path matches a .md file without the extension
    if !full_path.exists() && full_path.extension().is_none() {
        let md_path = format!("{}.md", path.to_string());
        if Path::new(&md_path).exists() {
            return Ok(HttpResponse::Found()
                .header(http::header::LOCATION, md_path)
                .finish());
        }
    }

    if full_path.exists() && !path.ends_with(".md") {
        return Ok(NamedFile::open(full_path)?.into_response(&request)?);
    }

    let page = cache.pages.get(&full_path);

    let mut context = Context::new();
    context.insert("version", VERSION);
    context.insert(
        "garden_title",
        data.title.as_ref().unwrap_or(&"Digital Garden".to_string()),
    );
    context.insert("files", &cache.files);
    context.insert(
        "page_title",
        &match page {
            Some(page) => page.title.clone(),
            None => full_path
                .components()
                .last()
                .unwrap()
                .as_os_str()
                .to_str()
                .unwrap()
                .to_string(),
        },
    );
    context.insert(
        "content",
        &match page {
            Some(page) => page.html.clone(),
            None => data
                .tera
                .render("_not_found.html", &Context::new())
                .map_err(ErrorInternalServerError)?,
        },
    );
    context.insert(
        "mtime",
        &match page {
            Some(page) => {
                if let Some(timestamp) = page.timestamp {
                    let mtime: DateTime<Local> = timestamp.into();
                    Some(mtime.format("%c").to_string())
                } else {
                    None
                }
            }
            None => None,
        },
    );

    Ok(HttpResponse::Ok().body(
        data.tera
            .render("main.html", &context)
            .map_err(ErrorInternalServerError)?,
    ))
}

fn update_garden<P: AsRef<Path>>(
    garden_path: P,
    current: GardenCache,
) -> anyhow::Result<GardenCache> {
    let garden_path = garden_path.as_ref().clone();

    let mut files: Vec<PathBuf> = fs::read_dir(&garden_path)?
        .filter_map(|entry| {
            if let Ok(entry) = entry {
                let path = entry.path();
                if path.is_file() {
                    let stripped_path = path.strip_prefix(&garden_path).unwrap().to_path_buf();
                    if !stripped_path.to_str().unwrap().starts_with(".") {
                        return Some(stripped_path);
                    }
                }
            }
            None
        })
        .collect();
    files.sort();

    if files.is_empty() {
        return Err(anyhow!("Garden is empty."));
    }

    let mut pages = current.pages.clone();

    let markdown_paths = files
        .iter()
        .filter(|p| p.to_str().unwrap_or("").ends_with(".md"))
        .map(|p| garden_path.join(p));
    for path in markdown_paths {
        trace!("Loading {} into cache...", path.display());
        let mtime = path.metadata().unwrap().modified().ok();
        if let Some(page) = pages.get(&path) {
            match (mtime, page.timestamp) {
                (Some(fs_time), Some(last_time)) => {
                    if fs_time == last_time {
                        continue;
                    }
                }
                _ => {}
            }
        }

        let mut file = File::open(&path)?;
        let mut file_string = String::new();
        file.read_to_string(&mut file_string)?;
        let markdown_source = preprocess_markdown(file_string);
        let parser = Parser::new_ext(markdown_source.as_str(), Options::all());
        let mut html_output = String::new();
        html::push_html(&mut html_output, parser);

        // TODO!
        let h1_regex = Regex::new(r"<h1>([^>]+)</h1>").unwrap();
        let title = match h1_regex.captures(&html_output) {
            Some(h1_match) => h1_match.get(1).unwrap().as_str(),
            _ => &path
                .components()
                .last()
                .unwrap()
                .as_os_str()
                .to_str()
                .unwrap_or("???"),
        };

        pages.insert(
            path.clone(),
            ParsedPage {
                timestamp: mtime,
                html: html_output.clone(),
                title: String::from(title),
                links: vec![], // todo!,
            },
        );
    }

    let result = GardenCache { files, pages };
    trace!("{:#?}", result);
    Ok(result)
}

fn preprocess_markdown(string: String) -> String {
    let double_brackets = Regex::new(r"\[\[(?P<inner>[\w .]+)\]\]").unwrap();
    let finder = LinkFinder::new();

    let result = double_brackets
        .replace_all(string.as_str(), |caps: &Captures| {
            format!(
                "[{}]({})",
                &caps[1],
                utf8_percent_encode(&caps[1], percent_encoding::NON_ALPHANUMERIC)
            )
        })
        .to_string();

    let result_vec = Vec::from(result.as_str());
    let start_delims = vec![b'(', b'<'];
    let end_delims = vec![b')', b'>'];
    // link.end() is the first char AFTER the link!
    let links = finder.links(result.as_str()).filter(|link| {
        link.start() == 0
            || link.end() == result.len()
            || !start_delims.contains(&result_vec[link.start() - 1])
            || !end_delims.contains(&result_vec[link.end()])
    });

    let mut offset = 0;
    let mut result_string = result.to_string();
    for link in links {
        let orig = link.as_str();
        let new = format!("<{}>", orig);
        result_string.replace_range((link.start() + offset)..(link.end() + offset), new.as_str());
        offset += new.len() - orig.len();
    }

    result_string
}