implement better hashing

- call `file::Checked::hash` in `AppState::from_args`

src/appstate.rs

@@ -32,6 +32,20 @@ fn new_http(timeout: Option<Duration>) -> ureq::Agent {
         .into()
 }
 
+fn new_progressbar() -> ProgressBar {
+    ProgressBar::hidden().with_style(
+        ProgressStyle::with_template(&format!(
+            concat!(
+                "{{bar:50.cyan/blue}} {{msg:.magenta}}: ",
+                "{{binary_bytes:.yellow}}{}{{binary_total_bytes:.yellow}} ",
+                "({{eta}})",
+            ),
+            style("/").magenta(),
+        ))
+        .expect("invalid style template"),
+    )
+}
+
 impl AppState {
     fn new(http: ureq::Agent, inner: CacheFile) -> Self {
         Self {
@@ -50,27 +64,38 @@ impl AppState {
     }
 
     pub fn from_args(args: &Cli) -> sharry::Result<Self> {
+        let mut files = args.files.clone();
+
+        // TODO CLI switch begin
+
+        let bar = new_progressbar();
+        bar.set_draw_target(ProgressDrawTarget::stderr());
+        // BOOKMARK assumption: total file size < 2 EiB
+        bar.set_length(files.iter().map(|f| f.get_size()).sum());
+        bar.enable_steady_tick(Duration::from_millis(50));
+
+        for chk in &mut files {
+            bar.set_message(format!("hashing {:?}", chk.get_name()));
+            chk.hash(|bytes| bar.inc(bytes))?;
+            debug!("{chk:?}");
+        }
+
+        bar.finish();
+
+        // TODO CLI switch end
+
         let http = new_http(args.get_timeout());
 
         let share_id = http.share_create(&args.get_uri(), &args.alias, args.get_share_request())?;
 
-        Ok(Self::new(http, CacheFile::from_args(args, share_id)))
+        Ok(Self::new(
+            http,
+            CacheFile::from_args(args, share_id).replace_files(files),
+        ))
     }
 
     fn with_progressbar(&mut self, f: impl FnOnce(&ProgressBar), drop_bar: bool) {
-        let bar = &*self.progress.get_or_insert_with(|| {
+        let bar = &*self.progress.get_or_insert_with(new_progressbar);
-            ProgressBar::hidden().with_style(
-                ProgressStyle::with_template(&format!(
-                    concat!(
-                        "{{bar:50.cyan/blue}} {{msg:.magenta}}: ",
-                        "{{binary_bytes:.yellow}}{}{{binary_total_bytes:.yellow}} ",
-                        "({{eta}})",
-                    ),
-                    style("/").magenta(),
-                ))
-                .expect("style template is not valid"),
-            )
-        });
 
         if let Some(upl) = self.inner.peek_uploading() {
             if bar.length().is_none() {

src/cachefile.rs

@@ -67,6 +67,13 @@ impl CacheFile {
         }
     }
 
+    pub fn replace_files(self, files: Vec<file::Checked>) -> Self {
+        Self {
+            files: files.into(),
+            ..self
+        }
+    }
+
     pub fn queue_empty(&self) -> bool {
         self.files.is_empty()
     }

src/file/checked.rs

@@ -20,6 +20,8 @@ pub struct Checked {
     pub(super) path: PathBuf,
     /// size of that file
     pub(super) size: u64,
+    /// hash of that file
+    pub(super) hash: Option<String>,
 }
 
 impl AsRef<[u8]> for Checked {
@@ -41,6 +43,7 @@ impl Checked {
             Ok(Self {
                 path: fs::canonicalize(&value)?,
                 size: meta.len(),
+                hash: None,
             })
         } else {
             Err(io::Error::new(
@@ -50,6 +53,19 @@ impl Checked {
         }
     }
 
+    pub fn hash(&mut self, f: impl Fn(u64)) -> io::Result<()> {
+        if self.hash.is_some() {
+            return Err(io::Error::other(format!(
+                "file {:?} is already hashed!",
+                self.path.display()
+            )));
+        }
+
+        self.hash = Some(super::compute_file_hash(&self.path, self.size, f)?);
+
+        Ok(())
+    }
+
     /// start uploading this file
     ///
     /// - tries to create a new entry in a share
@@ -68,7 +84,7 @@ impl Checked {
     ) -> sharry::Result<Uploading> {
         let file_id = client.file_create(uri, alias_id, share_id, &self)?;
 
-        Ok(Uploading::new(self.path, self.size, file_id))
+        Ok(Uploading::new(self.path, self.size, self.hash, file_id))
     }
 }
 
@@ -84,4 +100,8 @@ impl<'t> FileTrait<'t> for Checked {
     fn get_size(&self) -> u64 {
         self.size
     }
+
+    fn check_hash(&self, on_progress: impl Fn(u64)) -> io::Result<bool> {
+        super::check_file_hash(&self.path, self.size, &self.hash, on_progress)
+    }
 }

src/file/mod.rs

@@ -2,12 +2,62 @@ mod checked;
 mod chunk;
 mod uploading;
 
-use std::{ffi::OsStr, path::Path};
+use std::{
+    ffi::OsStr,
+    fs,
+    io::{self, Read},
+    path::Path,
+};
+
+use base64ct::{Base64, Encoding};
+use blake2b_simd::Params as Blake2b;
 
 pub use checked::Checked;
 pub use chunk::Chunk;
 pub use uploading::Uploading;
 
+fn compute_file_hash<P>(path: P, size: u64, on_progress: impl Fn(u64)) -> io::Result<String>
+where
+    P: AsRef<Path>,
+{
+    let mut file = fs::File::open(path)?;
+    let mut hasher = Blake2b::new().hash_length(64).to_state();
+
+    let mut buf = vec![0u8; 4 * 1024 * 1024];
+    let mut bytes_read = 0;
+
+    loop {
+        let n = file.read(&mut buf)?;
+        if n == 0 {
+            break;
+        }
+        hasher.update(&buf[..n]);
+
+        bytes_read += n as u64;
+        on_progress(n as u64);
+    }
+
+    if bytes_read != size {
+        return Err(io::Error::other(format!(
+            "Hashed {bytes_read:?} bytes, known file size {:?}!",
+            size
+        )));
+    }
+
+    Ok(Base64::encode_string(hasher.finalize().as_bytes()))
+}
+
+fn check_file_hash(
+    path: impl AsRef<Path>,
+    size: u64,
+    hash: &Option<String>,
+    on_progress: impl Fn(u64),
+) -> io::Result<bool> {
+    let Some(hash) = hash else { return Ok(false) };
+
+    Ok(*hash == compute_file_hash(path, size, on_progress)?)
+}
+
 pub trait FileTrait<'t> {
     /// extract the filename part of a `Path` reference
     ///
@@ -25,4 +75,6 @@ pub trait FileTrait<'t> {
 
     /// get the file's size
     fn get_size(&self) -> u64;
+
+    fn check_hash(&self, on_progress: impl Fn(u64)) -> io::Result<bool>;
 }

src/file/uploading.rs

@@ -11,8 +11,12 @@ use super::{Checked, Chunk, FileTrait};
 
 #[derive(Serialize, Deserialize, Debug)]
 pub struct Uploading {
+    /// canonical path to a regular file
     path: PathBuf,
+    /// size of that file
     size: u64,
+    /// hash of that file
+    hash: Option<String>,
     file_id: String,
     #[serde(skip)]
     last_offset: Option<u64>,
@@ -20,10 +24,11 @@ pub struct Uploading {
 }
 
 impl Uploading {
-    pub(super) fn new(path: PathBuf, size: u64, file_id: String) -> Self {
+    pub(super) fn new(path: PathBuf, size: u64, hash: Option<String>, file_id: String) -> Self {
         Self {
             path,
             size,
+            hash,
             file_id,
             last_offset: None,
             offset: 0,
@@ -79,6 +84,7 @@ impl Uploading {
         Checked {
             path: self.path,
             size: self.size,
+            hash: self.hash,
         }
     }
 }
@@ -94,4 +100,8 @@ impl<'t> FileTrait<'t> for Uploading {
     fn get_size(&self) -> u64 {
         self.size
     }
+
+    fn check_hash(&self, on_progress: impl Fn(u64)) -> io::Result<bool> {
+        super::check_file_hash(&self.path, self.size, &self.hash, on_progress)
+    }
 }

src/main.rs

@@ -24,6 +24,32 @@ use output::{Log, SHRUPL};
 use sharry::{ClientError, Parameter};
 
 fn main() {
+    let args = Cli::parse();
+
+    env_logger::Builder::new()
+        .filter_module("shrupl", args.get_level_filter())
+        .parse_default_env()
+        .init();
+
+    info!("args: {args:#?}");
+
+    println!("{} to {}!", style("Welcome").magenta().bold(), *SHRUPL);
+
+    let mut state = AppState::try_resume(&args)
+        .and_then(|state| output::prompt_continue().then_some(state))
+        .unwrap_or_else(|| match AppState::from_args(&args) {
+            Ok(state) => {
+                state.save().unwrap_or_else(|e| {
+                    Log::warning(format_args!("Failed to save state: {e}"));
+                });
+                state
+            }
+            Err(e) => {
+                Log::handle(&e);
+                Log::error(format_args!("Failed to create state: {e}"));
+            }
+        });
+
     let check_ctrlc = {
         let stop = Arc::new(AtomicBool::new(false));
         let stop_ctrlc = stop.clone();
@@ -41,36 +67,6 @@ fn main() {
         }
     };
 
-    let args = Cli::parse();
-
-    env_logger::Builder::new()
-        .filter_module("shrupl", args.get_level_filter())
-        .parse_default_env()
-        .init();
-
-    info!("args: {args:#?}");
-
-    println!("{} to {}!", style("Welcome").magenta().bold(), *SHRUPL);
-
-    let mut state = AppState::try_resume(&args)
-        .and_then(|state| output::prompt_continue().then_some(state))
-        .unwrap_or_else(|| {
-            check_ctrlc();
-
-            match AppState::from_args(&args) {
-                Ok(state) => {
-                    state.save().unwrap_or_else(|e| {
-                        Log::warning(format_args!("Failed to save state: {e}"));
-                    });
-                    state
-                }
-                Err(e) => {
-                    Log::handle(&e);
-                    Log::error(format_args!("Failed to create state: {e}"));
-                }
-            }
-        });
-
     info!("continuing with state: {state:#?}");
 
     let fns_magenta = output::style_all(&args.file_names(), StyledObject::magenta).join(", ");