Refactor:Mv mod visitor vfs under crates_io

Signed-off-by: Jiahao XU <Jiahao_XU@outlook.com>

src/drivers/crates_io/vfs.rs

@@ -0,0 +1,52 @@
+use std::collections::{hash_map::HashMap, hash_set::HashSet};
+use std::io;
+use std::path::Path;
+
+use cargo_toml::AbstractFilesystem;
+
+use crate::helpers::PathExt;
+
+/// This type stores the filesystem structure for the crate tarball
+/// extracted in memory and can be passed to
+/// `cargo_toml::Manifest::complete_from_abstract_filesystem`.
+#[derive(Debug)]
+pub(super) struct Vfs(HashMap<Box<Path>, HashSet<Box<str>>>);
+
+impl Vfs {
+    pub(super) fn new() -> Self {
+        Self(HashMap::with_capacity(16))
+    }
+
+    /// * `path` - must be canonical, must not be empty.
+    pub(super) fn add_path(&mut self, mut path: &Path) {
+        while let Some(parent) = path.parent() {
+            // Since path has parent, it must have a filename
+            let filename = path.file_name().unwrap();
+
+            // `cargo_toml`'s implementation does the same thing.
+            // https://docs.rs/cargo_toml/0.11.5/src/cargo_toml/afs.rs.html#24
+            let filename = filename.to_string_lossy();
+
+            self.0
+                .entry(parent.into())
+                .or_insert_with(|| HashSet::with_capacity(4))
+                .insert(filename.into());
+
+            path = parent;
+        }
+    }
+}
+
+impl AbstractFilesystem for Vfs {
+    fn file_names_in(&self, rel_path: &str) -> io::Result<HashSet<Box<str>>> {
+        let rel_path = Path::new(rel_path).normalize_path();
+
+        Ok(self.0.get(&*rel_path).map(Clone::clone).unwrap_or_default())
+    }
+}
+
+impl AbstractFilesystem for &Vfs {
+    fn file_names_in(&self, rel_path: &str) -> io::Result<HashSet<Box<str>>> {
+        (*self).file_names_in(rel_path)
+    }
+}

src/drivers/crates_io/visitor.rs

@@ -0,0 +1,81 @@
+use std::io::Read;
+use std::path::{Path, PathBuf};
+
+use cargo_toml::Manifest;
+use log::debug;
+use tar::Entries;
+
+use super::vfs::Vfs;
+use crate::{
+    helpers::{PathExt, TarEntriesVisitor},
+    BinstallError, Meta,
+};
+
+#[derive(Debug)]
+pub(super) struct ManifestVisitor {
+    cargo_toml_content: Vec<u8>,
+    /// manifest_dir_path is treated as the current dir.
+    manifest_dir_path: PathBuf,
+
+    vfs: Vfs,
+}
+
+impl ManifestVisitor {
+    pub(super) fn new(manifest_dir_path: PathBuf) -> Self {
+        Self {
+            // Cargo.toml is quite large usually.
+            cargo_toml_content: Vec::with_capacity(2000),
+            manifest_dir_path,
+            vfs: Vfs::new(),
+        }
+    }
+
+    /// Load binstall metadata using the extracted information stored in memory.
+    pub(super) fn load_manifest(&self) -> Result<Manifest<Meta>, BinstallError> {
+        debug!("Loading manifest directly from extracted file");
+
+        // Load and parse manifest
+        let mut manifest = Manifest::<Meta>::from_slice_with_metadata(&self.cargo_toml_content)?;
+
+        // Checks vfs for binary output names
+        manifest.complete_from_abstract_filesystem(&self.vfs)?;
+
+        // Return metadata
+        Ok(manifest)
+    }
+}
+
+impl TarEntriesVisitor for ManifestVisitor {
+    fn visit<R: Read>(&mut self, entries: Entries<'_, R>) -> Result<(), BinstallError> {
+        for res in entries {
+            let mut entry = res?;
+            let path = entry.path()?;
+            let path = path.normalize_path();
+
+            let path = if let Ok(path) = path.strip_prefix(&self.manifest_dir_path) {
+                path
+            } else {
+                // The path is outside of the curr dir (manifest dir),
+                // ignore it.
+                continue;
+            };
+
+            if path == Path::new("Cargo.toml")
+                || path == Path::new("src/main.rs")
+                || path.starts_with("src/bin")
+            {
+                self.vfs.add_path(path);
+            }
+
+            if path == Path::new("Cargo.toml") {
+                // Since it is possible for the same Cargo.toml to appear
+                // multiple times using `tar --keep-old-files`, here we
+                // clear the buffer first before reading into it.
+                self.cargo_toml_content.clear();
+                entry.read_to_end(&mut self.cargo_toml_content)?;
+            }
+        }
+
+        Ok(())
+    }
+}