meow

src/pkgs/by_name/bi/binutils/cross_toolchain.rs

@@ -0,0 +1,285 @@
+// async cross-toolchain runner that uses parser.rs info (no hardcoding)
+use crate::pkgs::by_name::bi::binutils::parser::{BinutilsInfo, fetch_page, parse_binutils};
+use reqwest::Client;
+use std::{
+    error::Error,
+    ffi::OsStr,
+    path::{Path, PathBuf},
+};
+use tokio::process::Command;
+use tracing::{error, info, warn};
+
+/// Configuration object - uses environment if values omitted.
+#[derive(Debug, Clone)]
+pub struct BinutilsConfig {
+    pub lfs_root: PathBuf,  // where the LFS tree will be (used for $LFS)
+    pub target: String,     // LFS_TGT (e.g. x86_64-lfs-linux-gnu)
+    pub info: BinutilsInfo, // parsed page info
+}
+
+impl BinutilsConfig {
+    /// create from env or params. If target is None, tries $LFS_TGT env var.
+    pub fn new(lfs_root: impl AsRef<Path>, target: Option<String>, info: BinutilsInfo) -> Self {
+        let lfs_root = lfs_root.as_ref().to_path_buf();
+        let target = target
+            .or_else(|| std::env::var("LFS_TGT").ok())
+            .unwrap_or_else(|| {
+                // fallback best-effort
+                if cfg!(target_os = "linux") {
+                    "x86_64-lfs-linux-gnu".to_string()
+                } else {
+                    "x86_64-lfs-linux-gnu".to_string()
+                }
+            });
+
+        Self {
+            lfs_root,
+            target,
+            info,
+        }
+    }
+
+    /// default places (non-hardcoded) where sources live.
+    /// If env `BINUTILS_SRC_DIR` is set, use that; else try LFS layout:
+    /// - $LFS/src/pkgs/by-name/bi/binutils
+    pub fn source_base_dir(&self) -> PathBuf {
+        if let Ok(s) = std::env::var("BINUTILS_SRC_DIR") {
+            PathBuf::from(s)
+        } else {
+            self.lfs_root
+                .join("src")
+                .join("pkgs")
+                .join("by-name")
+                .join("bi")
+                .join("binutils")
+        }
+    }
+
+    /// build directory inside LFS tree (following LFS style)
+    pub fn build_dir(&self) -> PathBuf {
+        self.lfs_root.join("build").join("binutils-pass1")
+    }
+
+    /// install dir (tools)
+    pub fn install_dir(&self) -> PathBuf {
+        self.lfs_root.join("tools")
+    }
+}
+
+/// High-level orchestration. Async.
+pub async fn build_binutils_from_page(
+    page_url: &str,
+    lfs_root: impl AsRef<std::path::Path>,
+    target: Option<String>,
+) -> Result<(), Box<dyn Error>> {
+    // 1) fetch page
+    info!("Fetching page: {}", page_url);
+    let html = fetch_page(page_url).await?;
+    let info = parse_binutils(&html)?;
+    info!("Parsed info: {:?}", info);
+
+    // 2) build config
+    let cfg = BinutilsConfig::new(lfs_root, target, info.clone());
+
+    // 3) ensure source base dir exists
+    let src_base = cfg.source_base_dir();
+    if !src_base.exists() {
+        info!("Creating source base dir: {:?}", src_base);
+        tokio::fs::create_dir_all(&src_base).await?;
+    }
+
+    // 4) find extracted source directory (binutils-*)
+    let mut source_dir: Option<PathBuf> = None;
+    if let Ok(mut rd) = tokio::fs::read_dir(&src_base).await {
+        while let Some(entry) = rd.next_entry().await? {
+            let ft = entry.file_type().await?;
+            if ft.is_dir() {
+                let name = entry.file_name().to_string_lossy().to_string();
+                if name.to_lowercase().contains("binutils") {
+                    source_dir = Some(entry.path());
+                    break;
+                }
+            }
+        }
+    }
+
+    // 5) if not found, attempt to download & extract
+    if source_dir.is_none() {
+        if let Some(dl) = &cfg.info.download_url {
+            info!("No extracted source found; will download {}", dl);
+
+            // download file into src_base
+            let client = Client::new();
+            let resp = client.get(dl).send().await?;
+            if !resp.status().is_success() {
+                return Err(format!("Download failed: {}", resp.status()).into());
+            }
+
+            // pick a filename from URL
+            let url_path = url::Url::parse(dl)?;
+            let filename = url_path
+                .path_segments()
+                .and_then(|seg| seg.last())
+                .and_then(|s| {
+                    if !s.is_empty() {
+                        Some(s.to_string())
+                    } else {
+                        None
+                    }
+                })
+                .ok_or("Cannot determine filename from URL")?;
+
+            let outpath = src_base.join(&filename);
+            info!("Saving archive to {:?}", outpath);
+            let bytes = resp.bytes().await?;
+            tokio::fs::write(&outpath, &bytes).await?;
+
+            // extract using tar (async spawn). Use absolute path to src_base
+            info!("Extracting archive {:?}", outpath);
+            let tar_path = outpath.clone();
+            let mut tar_cmd = Command::new("tar");
+            tar_cmd.arg("-xf").arg(&tar_path).arg("-C").arg(&src_base);
+            let status = tar_cmd.status().await?;
+            if !status.success() {
+                return Err("tar extraction failed".into());
+            }
+
+            // look for extracted dir again
+            if let Ok(mut rd) = tokio::fs::read_dir(&src_base).await {
+                while let Some(entry) = rd.next_entry().await? {
+                    let ft = entry.file_type().await?;
+                    if ft.is_dir() {
+                        let name = entry.file_name().to_string_lossy().to_string();
+                        if name.to_lowercase().contains("binutils") {
+                            source_dir = Some(entry.path());
+                            break;
+                        }
+                    }
+                }
+            }
+        } else {
+            warn!("No download URL found on the page and no unpacked source present.");
+        }
+    }
+
+    let source_dir = match source_dir {
+        Some(p) => p,
+        None => return Err("Could not locate or download/extract Binutils source".into()),
+    };
+    info!("Using source dir: {:?}", source_dir);
+
+    // 6) prepare build dir
+    let build_dir = cfg.build_dir();
+    if !build_dir.exists() {
+        info!("Creating build dir {:?}", build_dir);
+        tokio::fs::create_dir_all(&build_dir).await?;
+    }
+
+    // 7) run configure: use absolute configure script path in source_dir
+    let configure_path = source_dir.join("configure");
+    if !configure_path.exists() {
+        return Err(format!("configure script not found at {:?}", configure_path).into());
+    }
+
+    // If parser produced configure args tokens, use them; otherwise fallback to common flags
+    let args = if !cfg.info.configure_args.is_empty() {
+        cfg.info.configure_args.clone()
+    } else {
+        vec![
+            format!("--prefix={}", cfg.install_dir().display()),
+            format!("--with-sysroot={}", cfg.lfs_root.display()),
+            format!("--target={}", cfg.target),
+            "--disable-nls".to_string(),
+            "--disable-werror".to_string(),
+        ]
+    };
+
+    // replace $LFS and $LFS_TGT in args
+    let args: Vec<String> = args
+        .into_iter()
+        .map(|a| {
+            a.replace("$LFS", &cfg.lfs_root.to_string_lossy())
+                .replace("$LFS_TGT", &cfg.target)
+        })
+        .collect();
+
+    info!("Configuring with args: {:?}", args);
+
+    // spawn configure
+    let mut conf_cmd = Command::new(&configure_path);
+    conf_cmd.current_dir(&build_dir);
+    for a in &args {
+        conf_cmd.arg(a);
+    }
+    conf_cmd.stdout(std::process::Stdio::inherit());
+    conf_cmd.stderr(std::process::Stdio::inherit());
+    let status = conf_cmd.status().await?;
+    if !status.success() {
+        return Err("configure step failed".into());
+    }
+    info!("configure completed");
+
+    // 8) run build commands (make-like)
+    if !cfg.info.build_cmds.is_empty() {
+        for b in &cfg.info.build_cmds {
+            // split into program + args
+            let mut parts = shell_words::split(b).unwrap_or_else(|_| vec![b.clone()]);
+            let prog = parts.remove(0);
+            let mut cmd = Command::new(prog);
+            if !parts.is_empty() {
+                cmd.args(parts);
+            }
+            cmd.current_dir(&build_dir);
+            cmd.stdout(std::process::Stdio::inherit());
+            cmd.stderr(std::process::Stdio::inherit());
+            let status = cmd.status().await?;
+            if !status.success() {
+                return Err(format!("build step failed: {:?}", b).into());
+            }
+        }
+    } else {
+        // fallback to running `make`
+        let mut m = Command::new("make");
+        m.current_dir(&build_dir);
+        m.stdout(std::process::Stdio::inherit());
+        m.stderr(std::process::Stdio::inherit());
+        let status = m.status().await?;
+        if !status.success() {
+            return Err("make failed".into());
+        }
+    }
+    info!("build completed");
+
+    // 9) run install commands (make install)
+    if !cfg.info.install_cmds.is_empty() {
+        for inst in &cfg.info.install_cmds {
+            let mut parts = shell_words::split(inst).unwrap_or_else(|_| vec![inst.clone()]);
+            let prog = parts.remove(0);
+            let mut cmd = Command::new(prog);
+            if !parts.is_empty() {
+                cmd.args(parts);
+            }
+            cmd.current_dir(&build_dir);
+            cmd.stdout(std::process::Stdio::inherit());
+            cmd.stderr(std::process::Stdio::inherit());
+            let status = cmd.status().await?;
+            if !status.success() {
+                return Err(format!("install step failed: {:?}", inst).into());
+            }
+        }
+    } else {
+        // fallback `make install`
+        let mut mi = Command::new("make");
+        mi.arg("install");
+        mi.current_dir(&build_dir);
+        mi.stdout(std::process::Stdio::inherit());
+        mi.stderr(std::process::Stdio::inherit());
+        let status = mi.status().await?;
+        if !status.success() {
+            return Err("make install failed".into());
+        }
+    }
+
+    info!("install completed. Binutils Pass 1 done.");
+    Ok(())
+}

src/pkgs/by_name/bi/binutils/mod.rs

@@ -0,0 +1,2 @@
+pub mod cross_toolchain;
+pub mod parser;

src/pkgs/by_name/bi/binutils/parser.rs

@@ -0,0 +1,220 @@
+// async parser for Binutils Pass 1 page
+use reqwest::Client;
+use scraper::{Html, Selector};
+use std::error::Error;
+
+#[derive(Debug, Clone)]
+pub struct BinutilsInfo {
+    /// "2.45" or derived version text
+    pub version: Option<String>,
+    /// first archive download URL found (.tar.xz or .tar.gz)
+    pub download_url: Option<String>,
+    /// tokens for configure flags (everything after ../configure)
+    pub configure_args: Vec<String>,
+    /// build commands discovered (e.g. ["make"])
+    pub build_cmds: Vec<String>,
+    /// install commands discovered (e.g. ["make install"])
+    pub install_cmds: Vec<String>,
+    /// optional SBU, disk space
+    pub sbu: Option<String>,
+    pub disk_space: Option<String>,
+}
+
+impl Default for BinutilsInfo {
+    fn default() -> Self {
+        Self {
+            version: None,
+            download_url: None,
+            configure_args: Vec::new(),
+            build_cmds: Vec::new(),
+            install_cmds: Vec::new(),
+            sbu: None,
+            disk_space: None,
+        }
+    }
+}
+
+/// Fetch page content (async)
+pub async fn fetch_page(url: &str) -> Result<String, Box<dyn Error>> {
+    let client = Client::new();
+    let res = client.get(url).send().await?;
+    let status = res.status();
+    if !status.is_success() {
+        return Err(format!("Failed to fetch {}: {}", url, status).into());
+    }
+    let text = res.text().await?;
+    Ok(text)
+}
+
+/// Parse the LFS Binutils pass1 page; robust to small formatting changes.
+/// - extracts version (from <h1> text like "Binutils-2.45 - Pass 1")
+/// - finds a download URL ending with .tar.xz/.tar.gz
+/// - finds configure pre block(s), builds token list
+/// - finds `make` / `make install` pre blocks
+pub fn parse_binutils(html: &str) -> Result<BinutilsInfo, Box<dyn Error>> {
+    let document = Html::parse_document(html);
+
+    let mut info = BinutilsInfo::default();
+
+    // 1) Version from h1.sect1 (contains "Binutils-2.45 - Pass 1")
+    if let Ok(h1_sel) = Selector::parse("h1.sect1") {
+        if let Some(h1) = document.select(&h1_sel).next() {
+            let txt = h1.text().collect::<Vec<_>>().join(" ");
+            // try to pick the token containing "Binutils-" or "binutils-"
+            if let Some(tok) = txt
+                .split_whitespace()
+                .find(|s| s.to_lowercase().contains("binutils"))
+            {
+                // extract digits from token, e.g. "Binutils-2.45"
+                if let Some(pos) = tok.find('-') {
+                    let ver = tok[pos + 1..]
+                        .trim()
+                        .trim_matches(|c: char| !c.is_ascii() && c != '.')
+                        .to_string();
+                    if !ver.is_empty() {
+                        info.version = Some(ver);
+                    }
+                } else {
+                    // fallback: try to find "2.45" somewhere in the h1 string
+                    for part in txt.split_whitespace() {
+                        if part.chars().next().map(|c| c.is_digit(10)).unwrap_or(false) {
+                            info.version = Some(part.trim().to_string());
+                            break;
+                        }
+                    }
+                }
+            }
+        }
+    }
+
+    // 2) Download URL: look for anchors with href ending .tar.xz/.tar.gz
+    if let Ok(a_sel) = Selector::parse("a[href]") {
+        for a in document.select(&a_sel) {
+            if let Some(href) = a.value().attr("href") {
+                let href = href.trim();
+                if href.ends_with(".tar.xz") || href.ends_with(".tar.gz") || href.ends_with(".tgz")
+                {
+                    // Make absolute if relative to page; the typical LFS pages use relative links like ../../... or ../..
+                    // If it's already absolute (starts with http), keep it.
+                    let url = href.to_string();
+                    info.download_url = Some(url);
+                    break;
+                }
+            }
+        }
+    }
+
+    // 3) Parse "segmentedlist" entries for SBU and disk space
+    if let Ok(segtitle_sel) =
+        Selector::parse("div.package .segmentedlist .seglistitem .seg strong.segtitle")
+    {
+        if let Ok(segbody_sel) =
+            Selector::parse("div.package .segmentedlist .seglistitem .seg span.segbody")
+        {
+            for (t, b) in document
+                .select(&segtitle_sel)
+                .zip(document.select(&segbody_sel))
+            {
+                let title = t.text().collect::<String>().to_lowercase();
+                let body = b.text().collect::<String>().trim().to_string();
+                if title.contains("approximate build time") {
+                    info.sbu = Some(body.clone());
+                } else if title.contains("required disk space") {
+                    info.disk_space = Some(body.clone());
+                }
+            }
+        }
+    }
+
+    // 4) `pre.kbd.command` blocks for configure & make lines
+    if let Ok(pre_sel) = Selector::parse("div.installation pre.kbd.command, pre.kbd.command") {
+        for pre in document.select(&pre_sel) {
+            let text = pre.text().collect::<Vec<_>>().join("\n");
+            let trimmed = text.trim();
+
+            // handle configure block (starts with ../configure or ./configure)
+            if trimmed.starts_with("../configure")
+                || trimmed.starts_with("./configure")
+                || trimmed.starts_with(".. /configure")
+            {
+                // normalize: remove trailing backslashes and join lines
+                let mut joined = String::new();
+                for line in trimmed.lines() {
+                    let line = line.trim_end();
+                    if line.ends_with('\\') {
+                        joined.push_str(line.trim_end_matches('\\').trim());
+                        joined.push(' ');
+                    } else {
+                        joined.push_str(line.trim());
+                        joined.push(' ');
+                    }
+                }
+                // remove leading "../configure" token and split into args
+                let pieces: Vec<&str> = joined.split_whitespace().collect();
+                let mut args = Vec::new();
+                let mut started = false;
+                for p in pieces {
+                    if !started {
+                        if p.ends_with("configure")
+                            || p.ends_with("configure")
+                            || p.contains("configure")
+                        {
+                            started = true;
+                            continue;
+                        }
+                        // skip until configure found
+                        continue;
+                    } else {
+                        args.push(p.to_string());
+                    }
+                }
+                // fallback: if no tokens parsed, try chopping first token
+                if args.is_empty() {
+                    // attempt to remove the first token (../configure) by index
+                    if let Some(pos) = joined.find("configure") {
+                        let after = &joined[pos + "configure".len()..];
+                        for t in after.split_whitespace() {
+                            args.push(t.to_string());
+                        }
+                    }
+                }
+                info.configure_args = args
+                    .into_iter()
+                    .map(|s| s.trim().to_string())
+                    .filter(|s| !s.is_empty())
+                    .collect();
+                continue;
+            }
+
+            // handle make / make install lines
+            // consider blocks that are exactly "make" or "make install" or lines containing them
+            for line in trimmed.lines().map(|l| l.trim()) {
+                if line == "make" {
+                    if !info.build_cmds.contains(&"make".to_string()) {
+                        info.build_cmds.push("make".to_string());
+                    }
+                } else if line == "make install" {
+                    if !info.install_cmds.contains(&"make install".to_string()) {
+                        info.install_cmds.push("make install".to_string());
+                    }
+                } else if line.starts_with("make ") {
+                    // e.g., "make -j2"
+                    let t = line.to_string();
+                    if !info.build_cmds.contains(&t) {
+                        info.build_cmds.push(t);
+                    }
+                } else if line.starts_with("time {") && line.contains("make") {
+                    // handle the time wrapper line in the note; ignore
+                    // skip
+                }
+            }
+        }
+    }
+
+    // final sanity: if build_cmds empty but install_cmds contains "make install", add "make"
+    if info.build_cmds.is_empty() && !info.install_cmds.is_empty() {
+        info.build_cmds.push("make".to_string());
+    }
+
+    Ok(info)
+}

src/pkgs/by_name/bi/mod.rs

@@ -0,0 +1 @@
+pub mod binutils;