Wei Dai

mod highlight;

use std::collections::BTreeMap;
use std::fs;
use std::io::Read;
use std::path::{Path, PathBuf};
use std::time::Instant;

use walkdir::WalkDir;

use highlight::highlight_code;

/// Above this threshold the file is read in full and classified
/// (text or binary) based on its complete content.
const FULL_READ_LIMIT: u64 = 4_000_000; // 4 MB
/// For files larger than the threshold above, we only read a preview
/// of this size (so we never load huge files fully into memory).
const PREVIEW_BYTES: usize = 16_384;
/// How many bytes of binary content to show in the hex dump.
const HEXDUMP_BYTES: usize = 4096;

struct FileResult {
    rel_path: PathBuf, // relative path of the original file
    out_rel: PathBuf,  // relative path of the generated .html file (inside out/)
    label: String,     // label shown in the badge (language, or "binary")
}

enum Unreadable {
    IoError(PathBuf),
}

enum Content {
    /// Successfully decoded text. `truncated` indicates whether the file
    /// is larger than what was actually read (very large files: preview only).
    Text { content: String, truncated: bool, total_size: u64 },
    /// Content that isn't valid UTF-8 text: we show a hex dump instead
    /// of trying to decode it.
    Binary { preview: Vec<u8>, total_size: u64, truncated: bool },
}

/// Reads a file safely: for small files, reads everything and classifies
/// text/binary based on the full content; for large files, reads only a
/// prefix, to avoid loading huge files into memory.
fn read_content(path: &Path, total_size: u64) -> std::io::Result<Content> {
    if total_size <= FULL_READ_LIMIT {
        let bytes = fs::read(path)?;
        match String::from_utf8(bytes) {
            Ok(s) => Ok(Content::Text {
                content: s,
                truncated: false,
                total_size,
            }),
            Err(e) => Ok(Content::Binary {
                preview: e.into_bytes(),
                total_size,
                truncated: false,
            }),
        }
    } else {
        let mut f = fs::File::open(path)?;
        let mut buf = vec![0u8; PREVIEW_BYTES];
        let n = f.read(&mut buf)?;
        buf.truncate(n);

        match std::str::from_utf8(&buf) {
            Ok(s) => Ok(Content::Text {
                content: s.to_string(),
                truncated: true,
                total_size,
            }),
            Err(e) => {
                let valid_up_to = e.valid_up_to();
                // If the error is only due to cutting a multi-byte
                // character in half (common when reading an arbitrary
                // prefix of a text file), still treat it as text.
                if valid_up_to > 0 && buf.len() - valid_up_to <= 4 {
                    let s = std::str::from_utf8(&buf[..valid_up_to]).unwrap().to_string();
                    Ok(Content::Text {
                        content: s,
                        truncated: true,
                        total_size,
                    })
                } else {
                    Ok(Content::Binary {
                        preview: buf,
                        total_size,
                        truncated: true,
                    })
                }
            }
        }
    }
}

/// Tree node used to build the index (index.html)
#[derive(Default)]
struct Node {
    dirs: BTreeMap<String, Node>,
    files: Vec<(String, String, PathBuf)>, // (display_name, label, href relative to the index)
}

impl Node {
    fn insert_file(&mut self, components: &[String], display_name: String, label: String, href: PathBuf) {
        if components.is_empty() {
            self.files.push((display_name, label, href));
        } else {
            let head = &components[0];
            let child = self.dirs.entry(head.clone()).or_default();
            child.insert_file(&components[1..], display_name, label, href);
        }
    }

    fn render(&self, out: &mut String) {
        out.push_str("<ul class=\"tree\">\n");
        for (name, node) in &self.dirs {
            out.push_str(&format!(
                "<li class=\"dir\"><span class=\"dir-name\">\u{1F4C1} {}</span>\n",
                escape_html(name)
            ));
            node.render(out);
            out.push_str("</li>\n");
        }
        for (name, label, href) in &self.files {
            out.push_str(&format!(
                "<li class=\"file\"><a href=\"{}\">\u{1F4C4} {}</a> <span class=\"badge\">{}</span></li>\n",
                href.to_string_lossy().replace('\\', "/"),
                escape_html(name),
                escape_html(label)
            ));
        }
        out.push_str("</ul>\n");
    }
}

fn main() {
    let args: Vec<String> = std::env::args().collect();
    let input_dir = args.get(1).cloned().unwrap_or_else(|| ".".to_string());
    let output_dir = args.get(2).cloned().unwrap_or_else(|| "out".to_string());
    // Extra directory names to exclude, optional, comma-separated
    // (e.g. "node_modules,.git"). Nothing is excluded by default:
    // the generator includes everything it finds, including
    // __pycache__, compiled files, binaries, etc.
    let extra_excludes: Vec<String> = args
        .get(3)
        .map(|s| s.split(',').map(|p| p.trim().to_string()).filter(|p| !p.is_empty()).collect())
        .unwrap_or_default();

    let input = PathBuf::from(&input_dir);
    let output = PathBuf::from(&output_dir);

    if !input.is_dir() {
        eprintln!("Error: '{}' is not a valid directory.", input.display());
        std::process::exit(1);
    }

    if let Err(e) = fs::create_dir_all(&output) {
        eprintln!("Error creating the output directory: {e}");
        std::process::exit(1);
    }

    // Absolute path of the output directory, so we can exclude it from the
    // scan in case it's nested inside the input directory (e.g. input=".",
    // output="out"). This is the only automatic exclusion and it can't be
    // turned off: it only prevents the program from processing the files
    // it is currently writing.
    let output_abs = fs::canonicalize(&output).unwrap_or_else(|_| output.clone());

    println!("Scanning '{}' -> '{}'", input.display(), output.display());
    if !extra_excludes.is_empty() {
        println!("Directories excluded on request: {}", extra_excludes.join(", "));
    }
    let start = Instant::now();

    let mut results: Vec<FileResult> = Vec::new();
    let mut unreadable: Vec<Unreadable> = Vec::new();
    let mut n_binary = 0usize;
    let mut n_truncated = 0usize;

    for entry in WalkDir::new(&input)
        .into_iter()
        .filter_entry(|e| {
            if e.file_type().is_dir() {
                if let Some(name) = e.file_name().to_str() {
                    if extra_excludes.iter().any(|x| x == name) {
                        return false;
                    }
                }
                if let Ok(abs) = fs::canonicalize(e.path()) {
                    if abs == output_abs {
                        return false;
                    }
                }
            }
            true
        })
    {
        let entry = match entry {
            Ok(e) => e,
            Err(_) => continue,
        };

        if !entry.file_type().is_file() {
            continue;
        }

        let path = entry.path();
        let rel_path = match path.strip_prefix(&input) {
            Ok(p) => p.to_path_buf(),
            Err(_) => path.to_path_buf(),
        };

        let meta = match fs::metadata(path) {
            Ok(m) => m,
            Err(_) => {
                unreadable.push(Unreadable::IoError(rel_path));
                continue;
            }
        };

        let content = match read_content(path, meta.len()) {
            Ok(c) => c,
            Err(_) => {
                unreadable.push(Unreadable::IoError(rel_path));
                continue;
            }
        };

        let syntax_lang = detect_language(path);

        let (body_html, label, note): (String, String, Option<String>) = match &content {
            Content::Text { content, truncated, total_size } => {
                let body = highlight_code(content, syntax_lang);
                let note = if *truncated {
                    n_truncated += 1;
                    Some(format!(
                        "Truncated preview: showing the first {} bytes out of {} total.",
                        content.len(),
                        total_size
                    ))
                } else {
                    None
                };
                (body, syntax_lang.to_string(), note)
            }
            Content::Binary { preview, total_size, truncated } => {
                n_binary += 1;
                let dump = hex_dump(preview, HEXDUMP_BYTES);
                let shown = preview.len().min(HEXDUMP_BYTES);
                let note = Some(format!(
                    "Binary file: {} bytes total. Showing the first {} bytes as hex.{}",
                    total_size,
                    shown,
                    if *truncated { " (only the first bytes of the file were read)" } else { "" }
                ));
                (escape_html(&dump), "binary".to_string(), note)
            }
        };

        let mut out_rel = rel_path.clone();
        let new_name = format!(
            "{}.html",
            out_rel.file_name().unwrap_or_default().to_string_lossy()
        );
        out_rel.set_file_name(new_name);

        let out_path = output.join(&out_rel);
        if let Some(parent) = out_path.parent() {
            let _ = fs::create_dir_all(parent);
        }

        let depth = out_rel.components().count().saturating_sub(1);
        let back_to_index = "../".repeat(depth);

        let html = render_file_html(
            &rel_path.to_string_lossy(),
            &label,
            &body_html,
            note.as_deref(),
            &back_to_index,
        );

        if fs::write(&out_path, html).is_ok() {
            results.push(FileResult {
                rel_path,
                out_rel,
                label,
            });
        }
    }

    // Build the tree for the index
    let mut root = Node::default();
    for r in &results {
        let components: Vec<String> = r
            .rel_path
            .parent()
            .map(|p| {
                p.components()
                    .map(|c| c.as_os_str().to_string_lossy().to_string())
                    .filter(|s| !s.is_empty())
                    .collect()
            })
            .unwrap_or_default();
        let display_name = r
            .rel_path
            .file_name()
            .unwrap_or_default()
            .to_string_lossy()
            .to_string();
        root.insert_file(
            &components,
            display_name,
            r.label.clone(),
            r.out_rel.clone(),
        );
    }

    let mut tree_html = String::new();
    root.render(&mut tree_html);

    let index_html = render_index_html(&tree_html, results.len(), n_binary, n_truncated);
    let _ = fs::write(output.join("index.html"), index_html);

    let elapsed = start.elapsed();
    println!(
        "Done in {:.2?}: {} files included ({} binary, {} with truncated preview), {} unreadable.",
        elapsed,
        results.len(),
        n_binary,
        n_truncated,
        unreadable.len()
    );
    for u in &unreadable {
        match u {
            Unreadable::IoError(p) => println!("  unreadable (I/O error): {}", p.display()),
        }
    }
    println!("Open {}/index.html in your browser.", output.display());
}

/// Produces a hex dump (in the style of `xxd -C`) of the first `limit` bytes.
fn hex_dump(bytes: &[u8], limit: usize) -> String {
    let show = &bytes[..bytes.len().min(limit)];
    let mut out = String::with_capacity(show.len() * 4);
    for (i, chunk) in show.chunks(16).enumerate() {
        let offset = i * 16;
        out.push_str(&format!("{:08x}  ", offset));
        for (j, b) in chunk.iter().enumerate() {
            out.push_str(&format!("{:02x} ", b));
            if j == 7 {
                out.push(' ');
            }
        }
        let missing = 16usize.saturating_sub(chunk.len());
        for _ in 0..missing {
            out.push_str("   ");
        }
        out.push_str(" |");
        for b in chunk {
            let c = *b;
            if (0x20..0x7f).contains(&c) {
                out.push(c as char);
            } else {
                out.push('.');
            }
        }
        out.push_str("|\n");
    }
    out
}

/// Determines the language (used to pick syntax-highlighting rules)
/// from the file's name/extension. This is only used when the content
/// turns out to actually be text; binary files are classified as such
/// regardless by read_content.
fn detect_language(path: &Path) -> &'static str {
    let file_name = path
        .file_name()
        .map(|s| s.to_string_lossy().to_lowercase())
        .unwrap_or_default();

    match file_name.as_str() {
        "dockerfile" => return "dockerfile",
        "makefile" | "gnumakefile" => return "makefile",
        "cmakelists.txt" => return "cmake",
        ".gitignore" | ".dockerignore" | ".npmignore" => return "plaintext",
        "cargo.toml" | "cargo.lock" => return "ini",
        _ => {}
    }

    let ext = path
        .extension()
        .map(|s| s.to_string_lossy().to_lowercase())
        .unwrap_or_default();

    match ext.as_str() {
        "py" | "pyw" | "pyi" => "python",
        "rs" => "rust",
        "js" | "mjs" | "cjs" => "javascript",
        "jsx" => "javascript",
        "ts" => "typescript",
        "tsx" => "typescript",
        "java" => "java",
        "c" => "c",
        "h" => "c",
        "hpp" | "hh" | "hxx" => "cpp",
        "cpp" | "cc" | "cxx" => "cpp",
        "cs" => "csharp",
        "go" => "go",
        "rb" => "ruby",
        "php" => "php",
        "html" | "htm" => "xml",
        "css" => "css",
        "scss" | "sass" => "scss",
        "less" => "less",
        "json" => "json",
        "xml" => "xml",
        "yaml" | "yml" => "yaml",
        "toml" | "ini" | "cfg" | "conf" => "ini",
        "sh" | "bash" | "zsh" => "bash",
        "sql" => "sql",
        "md" | "markdown" => "markdown",
        "kt" | "kts" => "kotlin",
        "swift" => "swift",
        "lua" => "lua",
        "r" => "r",
        "pl" | "pm" => "perl",
        "groovy" | "gradle" => "groovy",
        "vb" => "vbnet",
        "graphql" | "gql" => "graphql",
        "diff" | "patch" => "diff",
        "bat" | "cmd" => "dos",
        "ps1" => "powershell",
        "vue" => "xml",
        "txt" => "plaintext",
        _ => "plaintext",
    }
}

fn escape_html(s: &str) -> String {
    let mut out = String::with_capacity(s.len());
    for c in s.chars() {
        match c {
            '&' => out.push_str("&amp;"),
            '<' => out.push_str("&lt;"),
            '>' => out.push_str("&gt;"),
            '"' => out.push_str("&quot;"),
            '\'' => out.push_str("&#39;"),
            _ => out.push(c),
        }
    }
    out
}

const STYLE: &str = r#"
:root {
  --bg: #0d1117;
  --panel: #161b22;
  --text: #c9d1d9;
  --muted: #8b949e;
  --accent: #58a6ff;
  --border: #30363d;
}
* { box-sizing: border-box; }
body {
  margin: 0;
  background: var(--bg);
  color: var(--text);
  font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Helvetica, Arial, sans-serif;
}
header {
  padding: 14px 20px;
  background: var(--panel);
  border-bottom: 1px solid var(--border);
  display: flex;
  align-items: center;
  gap: 12px;
  flex-wrap: wrap;
}
header a { color: var(--accent); text-decoration: none; font-weight: 600; }
header a:hover { text-decoration: underline; }
.path { color: var(--muted); font-size: 0.9em; word-break: break-all; }
.note {
  padding: 10px 20px;
  background: #21262d;
  border-bottom: 1px solid var(--border);
  color: var(--muted);
  font-size: 0.85em;
}
main { padding: 0; }
pre {
  margin: 0;
  padding: 16px 20px 40px 20px;
  overflow-x: auto;
  font-size: 13px;
  line-height: 1.5;
}
code { font-family: "SFMono-Regular", Consolas, "Liberation Mono", Menlo, monospace; }
.container { max-width: 1000px; margin: 30px auto; padding: 0 20px 60px; }
h1 { font-size: 1.6em; }
.meta { color: var(--muted); font-size: 0.9em; margin-bottom: 20px; }
ul.tree { list-style: none; padding-left: 18px; margin: 6px 0; }
ul.tree > li { margin: 4px 0; }
.dir-name { font-weight: 600; color: var(--text); }
.file a { color: var(--accent); text-decoration: none; }
.file a:hover { text-decoration: underline; }
.badge {
  display: inline-block;
  font-size: 0.72em;
  color: var(--muted);
  border: 1px solid var(--border);
  border-radius: 10px;
  padding: 1px 8px;
  margin-left: 6px;
}
.skipped { color: var(--muted); font-size: 0.85em; margin-top: 30px; }

/* Syntax highlighting: generated in Rust, no JavaScript involved. */
.c { color: #8b949e; font-style: italic; }
.s { color: #a5d6ff; }
.n { color: #79c0ff; }
.k { color: #ff7b72; font-weight: 600; }
"#;

fn render_file_html(rel_path: &str, label: &str, body_html: &str, note: Option<&str>, back_to_index: &str) -> String {
    let title = escape_html(rel_path);
    let note_html = match note {
        Some(n) => format!("<div class=\"note\">{}</div>\n", escape_html(n)),
        None => String::new(),
    };
    format!(
        r#"<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>{title}</title>
<style>{STYLE}</style>
</head>
<body>
<header>
  <a href="{back_to_index}index.html">&larr; Index</a>
  <span class="path">{title}</span>
  <span class="badge">{label}</span>
</header>
{note_html}<main>
<pre><code>{body_html}</code></pre>
</main>
</body>
</html>
"#
    )
}

fn render_index_html(tree_html: &str, n_ok: usize, n_binary: usize, n_truncated: usize) -> String {
    let mut extra = String::new();
    if n_binary > 0 {
        extra.push_str(&format!(
            "<p class=\"skipped\">{} binary files included with a hex-dump preview.</p>",
            n_binary
        ));
    }
    if n_truncated > 0 {
        extra.push_str(&format!(
            "<p class=\"skipped\">{} very large files shown as a truncated preview only.</p>",
            n_truncated
        ));
    }

    let folder_emoji = '\u{1F4C2}';
    format!(
        r#"<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>Source index</title>
<style>{STYLE}</style>
</head>
<body>
<div class="container">
<h1>{folder_emoji} Source index</h1>
<p class="meta">{n_ok} files included in total.</p>
{tree_html}
{extra}
</div>
</body>
</html>
"#
    )
}