Cleanup and document rust code (#44)

* Cleanup and document rust code

* Python compile

* bump bytes

* bump pyo3-bytes

Author: kylebarron

Cargo.toml

@@ -1,7 +1,12 @@
 [package]
 name = "async-tiff"
-version = "0.1.0"
+version = "0.1.0-beta.1"
 edition = "2021"
+authors = ["Kyle Barron <kyle@developmentseed.org>"]
+license = "MIT OR Apache-2.0"
+repository = "https://github.com/developmentseed/async-tiff"
+description = "Low-level asynchronous TIFF reader."
+readme = "README.md"
 
 [dependencies]
 byteorder = "1"
@@ -11,6 +16,7 @@ futures = "0.3.31"
 jpeg = { package = "jpeg-decoder", version = "0.3.0", default-features = false }
 num_enum = "0.7.3"
 # Match the version used by pyo3-object-store
+# We'll upgrade to object_store 0.12 ASAP when it comes out
 object_store = { git = "https://github.com/apache/arrow-rs", rev = "7a15e4b47ca97df2edef689c9f2ebd2f3888b79e" }
 thiserror = "1"
 tokio = { version = "1.43.0", optional = true }

python/Cargo.toml

@@ -1,5 +1,5 @@
 [package]
-name = "async-tiff"
+name = "py-async-tiff"
 version = "0.1.0-beta.1"
 authors = ["Kyle Barron <kyle@developmentseed.org>"]
 edition = "2021"
@@ -18,12 +18,13 @@ crate-type = ["cdylib"]
 
 [dependencies]
 async-tiff = { path = "../" }
-bytes = "1.8"
+bytes = "1.10.1"
+# We'll upgrade to object_store 0.12 ASAP
 # Match the version used by pyo3-object-store
 object_store = { git = "https://github.com/apache/arrow-rs", rev = "7a15e4b47ca97df2edef689c9f2ebd2f3888b79e" }
 pyo3 = { version = "0.23.0", features = ["macros"] }
 pyo3-async-runtimes = "0.23"
-pyo3-bytes = "0.1.2"
+pyo3-bytes = "0.1.3"
 pyo3-object_store = { git = "https://github.com/developmentseed/obstore", rev = "28ba07a621c1c104f084fb47ae7f8d08b1eae3ea" }
 rayon = "1.10.0"
 tokio-rayon = "2.1.0"

python/src/decoder.rs

@@ -2,7 +2,7 @@ use std::collections::HashMap;
 use std::sync::Arc;
 
 use async_tiff::decoder::{Decoder, DecoderRegistry};
-use async_tiff::error::AiocogeoError;
+use async_tiff::error::{AsyncTiffError, AsyncTiffResult};
 use async_tiff::tiff::tags::PhotometricInterpretation;
 use bytes::Bytes;
 use pyo3::exceptions::PyTypeError;
@@ -74,12 +74,12 @@ impl<'py> FromPyObject<'py> for PyDecoder {
 impl Decoder for PyDecoder {
     fn decode_tile(
         &self,
-        buffer: bytes::Bytes,
+        buffer: Bytes,
         _photometric_interpretation: PhotometricInterpretation,
         _jpeg_tables: Option<&[u8]>,
-    ) -> async_tiff::error::Result<bytes::Bytes> {
+    ) -> AsyncTiffResult<Bytes> {
         let decoded_buffer = Python::with_gil(|py| self.call(py, buffer))
-            .map_err(|err| AiocogeoError::General(err.to_string()))?;
+            .map_err(|err| AsyncTiffError::General(err.to_string()))?;
         Ok(decoded_buffer.into_inner())
     }
 }

python/src/tiff.rs

@@ -1,4 +1,5 @@
-use async_tiff::{AsyncFileReader, COGReader, ObjectReader, PrefetchReader};
+use async_tiff::reader::{AsyncFileReader, ObjectReader, PrefetchReader};
+use async_tiff::TIFF;
 use pyo3::prelude::*;
 use pyo3::types::PyType;
 use pyo3_async_runtimes::tokio::future_into_py;
@@ -7,7 +8,7 @@ use pyo3_object_store::PyObjectStore;
 use crate::PyImageFileDirectory;
 
 #[pyclass(name = "TIFF", frozen)]
-pub(crate) struct PyTIFF(COGReader);
+pub(crate) struct PyTIFF(TIFF);
 
 #[pymethods]
 impl PyTIFF {
@@ -32,7 +33,7 @@ impl PyTIFF {
             } else {
                 Box::new(reader)
             };
-            Ok(PyTIFF(COGReader::try_open(reader).await.unwrap()))
+            Ok(PyTIFF(TIFF::try_open(reader).await.unwrap()))
         })?;
         Ok(cog_reader)
     }

src/cog.rs

@@ -1,18 +1,19 @@
-use crate::async_reader::AsyncCursor;
-use crate::error::Result;
+use crate::error::AsyncTiffResult;
 use crate::ifd::ImageFileDirectories;
+use crate::reader::{AsyncCursor, AsyncFileReader};
 use crate::tiff::{TiffError, TiffFormatError};
-use crate::AsyncFileReader;
 
-#[derive(Debug)]
-pub struct COGReader {
-    #[allow(dead_code)]
-    cursor: AsyncCursor,
+/// A TIFF file.
+#[derive(Debug, Clone)]
+pub struct TIFF {
     ifds: ImageFileDirectories,
 }
 
-impl COGReader {
-    pub async fn try_open(reader: Box<dyn AsyncFileReader>) -> Result<Self> {
+impl TIFF {
+    /// Open a new TIFF file.
+    ///
+    /// This will read all the Image File Directories (IFDs) in the file.
+    pub async fn try_open(reader: Box<dyn AsyncFileReader>) -> AsyncTiffResult<Self> {
         let mut cursor = AsyncCursor::try_open_tiff(reader).await?;
         let version = cursor.read_u16().await?;
 
@@ -44,9 +45,10 @@ impl COGReader {
 
         let ifds = ImageFileDirectories::open(&mut cursor, first_ifd_location, bigtiff).await?;
 
-        Ok(Self { cursor, ifds })
+        Ok(Self { ifds })
     }
 
+    /// Access the underlying Image File Directories.
     pub fn ifds(&self) -> &ImageFileDirectories {
         &self.ifds
     }
@@ -58,7 +60,7 @@ mod test {
     use std::sync::Arc;
 
     use crate::decoder::DecoderRegistry;
-    use crate::ObjectReader;
+    use crate::reader::ObjectReader;
 
     use super::*;
     use object_store::local::LocalFileSystem;
@@ -72,7 +74,7 @@ mod test {
         let store = Arc::new(LocalFileSystem::new_with_prefix(folder).unwrap());
         let reader = ObjectReader::new(store, path);
 
-        let cog_reader = COGReader::try_open(Box::new(reader.clone())).await.unwrap();
+        let cog_reader = TIFF::try_open(Box::new(reader.clone())).await.unwrap();
 
         let ifd = &cog_reader.ifds.as_ref()[1];
         let decoder_registry = DecoderRegistry::default();

src/decoder.rs

@@ -1,11 +1,13 @@
+//! Decoders for different TIFF compression methods.
+
 use std::collections::HashMap;
 use std::fmt::Debug;
 use std::io::{Cursor, Read};
 
 use bytes::Bytes;
 use flate2::bufread::ZlibDecoder;
 
-use crate::error::Result;
+use crate::error::AsyncTiffResult;
 use crate::tiff::tags::{CompressionMethod, PhotometricInterpretation};
 use crate::tiff::{TiffError, TiffUnsupportedError};
 
@@ -46,14 +48,16 @@ impl Default for DecoderRegistry {
 
 /// A trait to decode a TIFF tile.
 pub trait Decoder: Debug + Send + Sync {
+    /// Decode a TIFF tile.
     fn decode_tile(
         &self,
         buffer: Bytes,
         photometric_interpretation: PhotometricInterpretation,
         jpeg_tables: Option<&[u8]>,
-    ) -> Result<Bytes>;
+    ) -> AsyncTiffResult<Bytes>;
 }
 
+/// A decoder for the Deflate compression method.
 #[derive(Debug, Clone)]
 pub struct DeflateDecoder;
 
@@ -63,14 +67,15 @@ impl Decoder for DeflateDecoder {
         buffer: Bytes,
         _photometric_interpretation: PhotometricInterpretation,
         _jpeg_tables: Option<&[u8]>,
-    ) -> Result<Bytes> {
+    ) -> AsyncTiffResult<Bytes> {
         let mut decoder = ZlibDecoder::new(Cursor::new(buffer));
         let mut buf = Vec::new();
         decoder.read_to_end(&mut buf)?;
         Ok(buf.into())
     }
 }
 
+/// A decoder for the JPEG compression method.
 #[derive(Debug, Clone)]
 pub struct JPEGDecoder;
 
@@ -80,11 +85,12 @@ impl Decoder for JPEGDecoder {
         buffer: Bytes,
         photometric_interpretation: PhotometricInterpretation,
         jpeg_tables: Option<&[u8]>,
-    ) -> Result<Bytes> {
+    ) -> AsyncTiffResult<Bytes> {
         decode_modern_jpeg(buffer, photometric_interpretation, jpeg_tables)
     }
 }
 
+/// A decoder for the LZW compression method.
 #[derive(Debug, Clone)]
 pub struct LZWDecoder;
 
@@ -94,14 +100,15 @@ impl Decoder for LZWDecoder {
         buffer: Bytes,
         _photometric_interpretation: PhotometricInterpretation,
         _jpeg_tables: Option<&[u8]>,
-    ) -> Result<Bytes> {
+    ) -> AsyncTiffResult<Bytes> {
         // https://github.com/image-rs/image-tiff/blob/90ae5b8e54356a35e266fb24e969aafbcb26e990/src/decoder/stream.rs#L147
         let mut decoder = weezl::decode::Decoder::with_tiff_size_switch(weezl::BitOrder::Msb, 8);
         let decoded = decoder.decode(&buffer).expect("failed to decode LZW data");
         Ok(decoded.into())
     }
 }
 
+/// A decoder for uncompressed data.
 #[derive(Debug, Clone)]
 pub struct UncompressedDecoder;
 
@@ -111,7 +118,7 @@ impl Decoder for UncompressedDecoder {
         buffer: Bytes,
         _photometric_interpretation: PhotometricInterpretation,
         _jpeg_tables: Option<&[u8]>,
-    ) -> Result<Bytes> {
+    ) -> AsyncTiffResult<Bytes> {
         Ok(buffer)
     }
 }
@@ -121,7 +128,7 @@ fn decode_modern_jpeg(
     buf: Bytes,
     photometric_interpretation: PhotometricInterpretation,
     jpeg_tables: Option<&[u8]>,
-) -> Result<Bytes> {
+) -> AsyncTiffResult<Bytes> {
     // Construct new jpeg_reader wrapping a SmartReader.
     //
     // JPEG compression in TIFF allows saving quantization and/or huffman tables in one central

src/error.rs

@@ -1,29 +1,36 @@
+//! Error handling.
+
 use std::fmt::Debug;
 use thiserror::Error;
 
 /// Enum with all errors in this crate.
 #[derive(Error, Debug)]
 #[non_exhaustive]
-pub enum AiocogeoError {
+pub enum AsyncTiffError {
+    /// End of file error.
     #[error("End of File: expected to read {0} bytes, got {1}")]
     EndOfFile(usize, usize),
 
     /// General error.
     #[error("General error: {0}")]
     General(String),
 
+    /// IO Error.
     #[error(transparent)]
     IOError(#[from] std::io::Error),
 
+    /// Error while decoding JPEG data.
     #[error(transparent)]
     JPEGDecodingError(#[from] jpeg::Error),
 
+    /// Error while fetching data using object store.
     #[error(transparent)]
     ObjectStore(#[from] object_store::Error),
 
+    /// An error during TIFF tag parsing.
     #[error(transparent)]
     InternalTIFFError(#[from] crate::tiff::TiffError),
 }
 
 /// Crate-specific result type.
-pub type Result<T> = std::result::Result<T, AiocogeoError>;
+pub type AsyncTiffResult<T> = std::result::Result<T, AsyncTiffError>;

src/geo/affine.rs

@@ -5,30 +5,37 @@ use crate::ImageFileDirectory;
 pub struct AffineTransform(f64, f64, f64, f64, f64, f64);
 
 impl AffineTransform {
+    /// Construct a new Affine Transform
     pub fn new(a: f64, b: f64, xoff: f64, d: f64, e: f64, yoff: f64) -> Self {
         Self(a, b, xoff, d, e, yoff)
     }
 
+    /// a
     pub fn a(&self) -> f64 {
         self.0
     }
 
+    /// b
     pub fn b(&self) -> f64 {
         self.1
     }
 
+    /// c
     pub fn c(&self) -> f64 {
         self.2
     }
 
+    /// d
     pub fn d(&self) -> f64 {
         self.3
     }
 
+    /// e
     pub fn e(&self) -> f64 {
         self.4
     }
 
+    /// f
     pub fn f(&self) -> f64 {
         self.5
     }

src/geo/geo_key_directory.rs

@@ -1,4 +1,5 @@
 #![allow(dead_code)]
+#![allow(missing_docs)]
 
 use std::collections::HashMap;