Symlink forest in venv

This emulates the behavior where a single site-packages directory
contains all/most PyPi packages. Packages distributed by NVIDIA
currently assume this through the use of rpath as `$ORIGIN/../../`
to reach the `nvidia` package location. Downstream libraries like torch
and jax do not set up the dynamic library search path based on sys.path
either.

This is a casual attempt, but can be refined by others.

TODO:
[] Build binaries for all architectures.

Fixes #274.

py/tools/py/src/pth.rs

@@ -1,5 +1,5 @@
 use std::{
-    fs::{self, File},
+    fs::{self, DirEntry, File},
     io::{BufRead, BufReader, BufWriter, Write},
     path::{Path, PathBuf},
 };
@@ -19,27 +19,11 @@ impl PthFile {
         }
     }
 
-    pub fn copy_to_site_packages(&self, dest: &Path) -> miette::Result<()> {
-        let dest_pth = dest.join(self.src.file_name().expect(".pth must be a file"));
-
-        if let Some(prefix) = self.prefix.as_deref() {
-            self.prefix_and_write_pth(dest_pth, prefix)
-        } else {
-            fs::copy(self.src.as_path(), dest_pth)
-                .map(|_| ())
-                .into_diagnostic()
-                .wrap_err("Unable to copy .pth file to site-packages")
-        }
-    }
-
-    fn prefix_and_write_pth<P>(&self, dest: P, prefix: &str) -> miette::Result<()>
-    where
-        P: AsRef<Path>,
-    {
+    pub fn set_up_site_packages(&self, dest: &Path) -> miette::Result<()> {
         let source_pth = File::open(self.src.as_path())
             .into_diagnostic()
             .wrap_err("Unable to open source .pth file")?;
-        let dest_pth = File::create(dest)
+        let dest_pth = File::create(dest.join(self.src.file_name().expect(".pth must be a file")))
             .into_diagnostic()
             .wrap_err("Unable to create destination .pth file")?;
 
@@ -48,13 +32,69 @@ impl PthFile {
 
         let mut line = String::new();
         while reader.read_line(&mut line).unwrap() > 0 {
-            let entry = Path::new(prefix).join(Path::new(line.trim()));
+            let entry: PathBuf;
+            if self.prefix.is_some() {
+                entry = Path::new(self.prefix.as_deref().unwrap()).join(Path::new(line.trim()));
+            } else {
+                entry = PathBuf::from(line.trim());
+            }
             line.clear();
-            writeln!(writer, "{}", entry.to_string_lossy())
-                .into_diagnostic()
-                .wrap_err("Unable to write new .pth file entry with prefix")?;
+            if entry.file_name().is_some_and(|x| x == "site-packages") {
+                let src_dir = dest.join(entry).canonicalize().unwrap();
+                create_symlinks(&src_dir, &src_dir, &dest)?;
+            } else {
+                writeln!(writer, "{}", entry.to_string_lossy())
+                    .into_diagnostic()
+                    .wrap_err("Unable to write new .pth file entry")?;
+            }
         }
 
         Ok(())
     }
 }
+
+fn create_symlinks(dir: &Path, root_dir: &Path, dst_dir: &Path) -> miette::Result<()> {
+    // Create this directory at the destination.
+    let tgt_dir = dst_dir.join(dir.strip_prefix(root_dir).unwrap());
+    std::fs::create_dir_all(&tgt_dir)
+        .into_diagnostic()
+        .wrap_err(format!(
+            "unable to create parent directory for symlink: {}",
+            tgt_dir.to_string_lossy()
+        ))?;
+
+    // Recurse.
+    let read_dir = fs::read_dir(dir).into_diagnostic().wrap_err(format!(
+        "unable to read directory {}",
+        dir.to_string_lossy()
+    ))?;
+    for entry in read_dir {
+        let entry = entry.into_diagnostic().wrap_err(format!(
+            "unable to read directory entry {}",
+            dir.to_string_lossy()
+        ))?;
+        let path = entry.path();
+        if path.is_dir() {
+            create_symlinks(&path, root_dir, dst_dir)?;
+        } else if dir != root_dir || entry.file_name() != "__init__.py" {
+            create_symlink(&entry, root_dir, dst_dir)?;
+        }
+    }
+    Ok(())
+}
+
+fn create_symlink(e: &DirEntry, root_dir: &Path, dst_dir: &Path) -> miette::Result<()> {
+    let tgt = e.path();
+    let link = dst_dir.join(tgt.strip_prefix(root_dir).unwrap());
+    if link.exists() && link.file_name().is_some_and(|x| x == "__init__.py") && fs::metadata(&tgt).unwrap().len() == 0 {
+        return Ok(());
+    }
+    std::os::unix::fs::symlink(&tgt, &link)
+        .into_diagnostic()
+        .wrap_err(format!(
+            "unable to create symlink: {} -> {}",
+            tgt.to_string_lossy(),
+            link.to_string_lossy()
+        ))?;
+    Ok(())
+}

py/tools/py/src/venv.rs

@@ -51,7 +51,7 @@ pub fn create_venv(
         .into_diagnostic()?;
 
     if let Some(pth) = pth_file {
-        pth.copy_to_site_packages(&venv_location.join(install_paths.platlib()))?
+        pth.set_up_site_packages(&venv_location.join(install_paths.platlib()))?
     }
 
     Ok(())