From 6ed6757702eea4f775ad06e293550b99a499a415 Mon Sep 17 00:00:00 2001
From: bnewbold <bnewbold@robocracy.org>
Date: Fri, 26 May 2017 15:10:52 -0700
Subject: WIP

---
 src/bin/bhash.rs | 95 +++++++++++++++++++++++++++++++++++++++++++-------------
 src/lib.rs       |  3 ++
 2 files changed, 77 insertions(+), 21 deletions(-)

diff --git a/src/bin/bhash.rs b/src/bin/bhash.rs
index b5d2937..819d9d0 100644
--- a/src/bin/bhash.rs
+++ b/src/bin/bhash.rs
@@ -12,18 +12,68 @@ extern crate tree_magic;
 extern crate flate2;
 extern crate tar;
 extern crate crypto;
+extern crate zip;
 
 use clap::App;
 use bad_hashish::Result;
 use std::path::Path;
-use std::io::BufReader;
+use std::io::{Read, BufReader};
 use std::io::prelude::*;
 use std::fs::File;
 use flate2::read::GzDecoder;
 use tar::Archive;
 use crypto::digest::Digest;
 use crypto::sha1::Sha1;
+use crypto::sha2::Sha256;
+use crypto::blake2b::Blake2b;
+use zip::read::ZipArchive;
 
+fn hash_stream<T: Read>(mut stream: T, name: &str, total_size: u64) -> Result<()> {
+
+    let mut hash_sha1 = Sha1::new();
+    let mut hash_sha256 = Sha256::new();
+    let mut hash_blake2b = Blake2b::new(32);
+    let mut buf: [u8; 1*1024*1024] = [0; 1*1024*1024];
+    loop {
+        let got = stream.read(&mut buf[..])?;
+        if got <= 0 { break };
+        hash_sha1.input(&buf[0..got]);
+        //hash_sha256.input(&buf[0..got]);
+        //hash_blake2b.input(&buf[0..got]);
+    }
+    /*
+    println!("{} {} {} {} {}",
+        hash_sha1.result_str(),
+        hash_sha256.result_str(),
+        hash_blake2b.result_str(),
+        total_size,
+        name);
+    */
+    println!("{}  {}",
+        hash_sha1.result_str(),
+        name);
+    Ok(())
+}
+
+fn do_tar<T: Read>(stream: T) -> Result<()> {
+    //println!("It's a tar inside");
+    let mut a = Archive::new(stream);
+    for inner in a.entries().unwrap() {
+        let mut inner = inner.unwrap();
+        let (inner_name, inner_size) = {
+            let header = inner.header();
+            // Only do actual files ("regular", not directories, fifo, etc)
+            if header.entry_type() != tar::EntryType::Regular {
+                continue;
+            }
+            let inner_name = header.path()?.to_string_lossy().to_string();
+            let inner_size = header.entry_size()?;
+            (inner_name, inner_size)
+        };
+        hash_stream(&mut inner, &inner_name, inner_size)?;
+    }
+    Ok(())
+}
 
 fn run() -> Result<()> {
 
@@ -38,7 +88,22 @@ fn run() -> Result<()> {
         println!("{} ({})", f, tree_magic::from_filepath(path));
 
         if tree_magic::match_filepath("application/zip", path) {
-            //println!("It's a zip.");
+            let f = File::open(path)?;
+            let mut zip = ZipArchive::new(f)?;
+
+            for i in 0..zip.len() {
+                let mut inner = zip.by_index(i).unwrap();
+                if (0o40000u32 & inner.unix_mode().unwrap()) != 0 {
+                    // This is a directory
+                    continue
+                }
+                let (name, size) = (inner.name().to_string(), inner.size());
+                hash_stream(&mut inner, &name, size)?;
+            }
+
+        } else if tree_magic::match_filepath("application/x-tar", path) {
+            let f = File::open(path)?;
+            do_tar(f)?;
         } else if tree_magic::match_filepath("application/gzip", path) {
             //println!("It's gzip.");
             let f = File::open(path)?;
@@ -52,26 +117,14 @@ fn run() -> Result<()> {
             };
 
             if is_tar {
-                //println!("It's a tar inside");
-                let mut a = Archive::new(reader);
-                for inner in a.entries().unwrap() {
-                    let mut inner = inner.unwrap();
-                    // Only do actual files ("regular", not directories, fifo, etc)
-                    if inner.header().entry_type() != tar::EntryType::Regular {
-                        continue;
-                    }
-                    let mut hasher = Sha1::new();
-                    let mut buf: [u8; 1*1024*1024] = [0; 1*1024*1024];
-                    loop {
-                        let got = inner.read(&mut buf[..])?;
-                        if got <= 0 { break };
-                        hasher.input(&buf[0..got]);
-                    }
-                    println!("{}  {}",
-                             hasher.result_str(),
-                             inner.header().path()?.to_str().unwrap());
-                }
+                do_tar(reader)?;
             }
+        } else if path.is_dir() {
+            continue
+        } else {
+            let mut f = File::open(path)?;
+            let (name, size) = (&path.to_string_lossy().to_string(), f.metadata()?.len());
+            hash_stream(&mut f, name, size)?;
         }
     }
 
diff --git a/src/lib.rs b/src/lib.rs
index 3897634..6f6b241 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -8,12 +8,15 @@ extern crate log;
 #[macro_use]
 extern crate error_chain;
 
+extern crate zip;
+
 
 mod errors {
     // Create the Error, ErrorKind, ResultExt, and Result types
     error_chain! {
         foreign_links {
             Io(::std::io::Error);
+            Zip(::zip::result::ZipError);
         }
     }
 }
-- 
cgit v1.2.3