BigTIFF support (#30)

* Add bigtiff support

* add open helper

* Remove underlying tiff dependency

Author: kylebarron

Cargo.toml

@@ -13,9 +13,9 @@ num_enum = "0.7.3"
 # Match the version used by pyo3-object-store
 object_store = { git = "https://github.com/apache/arrow-rs", rev = "7a15e4b47ca97df2edef689c9f2ebd2f3888b79e" }
 thiserror = "1"
-tiff = "0.9"
 tokio = { version = "1.43.0", optional = true }
 weezl = "0.1.0"
 
 [dev-dependencies]
+tiff = "0.9.1"
 tokio = { version = "1.9", features = ["macros", "fs", "rt-multi-thread"] }

src/async_reader.rs

@@ -169,7 +169,7 @@ pub enum Endianness {
 #[derive(Debug)]
 pub(crate) struct AsyncCursor {
     reader: Box<dyn AsyncFileReader>,
-    offset: usize,
+    offset: u64,
     endianness: Endianness,
 }
 
@@ -212,7 +212,7 @@ impl AsyncCursor {
     }
 
     /// Read the given number of bytes, advancing the internal cursor state by the same amount.
-    pub(crate) async fn read(&mut self, length: usize) -> Result<EndianAwareReader> {
+    pub(crate) async fn read(&mut self, length: u64) -> Result<EndianAwareReader> {
         let range = self.offset as _..(self.offset + length) as _;
         self.offset += length;
         let bytes = self.reader.get_bytes(range).await?;
@@ -281,15 +281,15 @@ impl AsyncCursor {
     }
 
     /// Advance cursor position by a set amount
-    pub(crate) fn advance(&mut self, amount: usize) {
+    pub(crate) fn advance(&mut self, amount: u64) {
         self.offset += amount;
     }
 
-    pub(crate) fn seek(&mut self, offset: usize) {
+    pub(crate) fn seek(&mut self, offset: u64) {
         self.offset = offset;
     }
 
-    pub(crate) fn position(&self) -> usize {
+    pub(crate) fn position(&self) -> u64 {
         self.offset
     }
 }

src/cog.rs

@@ -1,28 +1,56 @@
 use crate::async_reader::AsyncCursor;
 use crate::error::Result;
 use crate::ifd::ImageFileDirectories;
+use crate::tiff::{TiffError, TiffFormatError};
 use crate::AsyncFileReader;
 
 #[derive(Debug)]
 pub struct COGReader {
     #[allow(dead_code)]
     cursor: AsyncCursor,
     ifds: ImageFileDirectories,
+    #[allow(dead_code)]
+    bigtiff: bool,
 }
 
 impl COGReader {
     pub async fn try_open(reader: Box<dyn AsyncFileReader>) -> Result<Self> {
         let mut cursor = AsyncCursor::try_open_tiff(reader).await?;
         let version = cursor.read_u16().await?;
 
-        // Assert it's a standard non-big tiff
-        assert_eq!(version, 42);
-
-        let first_ifd_location = cursor.read_u32().await?;
-
-        let ifds = ImageFileDirectories::open(&mut cursor, first_ifd_location as usize).await?;
-
-        Ok(Self { cursor, ifds })
+        let bigtiff = match version {
+            42 => false,
+            43 => {
+                // Read bytesize of offsets (in bigtiff it's alway 8 but provide a way to move to 16 some day)
+                if cursor.read_u16().await? != 8 {
+                    return Err(
+                        TiffError::FormatError(TiffFormatError::TiffSignatureNotFound).into(),
+                    );
+                }
+                // This constant should always be 0
+                if cursor.read_u16().await? != 0 {
+                    return Err(
+                        TiffError::FormatError(TiffFormatError::TiffSignatureNotFound).into(),
+                    );
+                }
+                true
+            }
+            _ => return Err(TiffError::FormatError(TiffFormatError::TiffSignatureInvalid).into()),
+        };
+
+        let first_ifd_location = if bigtiff {
+            cursor.read_u64().await?
+        } else {
+            cursor.read_u32().await?.into()
+        };
+
+        let ifds = ImageFileDirectories::open(&mut cursor, first_ifd_location, bigtiff).await?;
+
+        Ok(Self {
+            cursor,
+            ifds,
+            bigtiff,
+        })
     }
 
     pub fn ifds(&self) -> &ImageFileDirectories {

src/decoder.rs

@@ -4,10 +4,10 @@ use std::io::{Cursor, Read};
 
 use bytes::Bytes;
 use flate2::bufread::ZlibDecoder;
-use tiff::tags::{CompressionMethod, PhotometricInterpretation};
-use tiff::{TiffError, TiffUnsupportedError};
 
 use crate::error::Result;
+use crate::tiff::tags::{CompressionMethod, PhotometricInterpretation};
+use crate::tiff::{TiffError, TiffUnsupportedError};
 
 /// A registry of decoders.
 #[derive(Debug)]

src/error.rs

@@ -22,7 +22,7 @@ pub enum AiocogeoError {
     ObjectStore(#[from] object_store::Error),
 
     #[error(transparent)]
-    TIFFError(#[from] tiff::TiffError),
+    InternalTIFFError(#[from] crate::tiff::TiffError),
 }
 
 /// Crate-specific result type.

src/geo/geo_key_directory.rs

@@ -3,8 +3,9 @@
 use std::collections::HashMap;
 
 use num_enum::{IntoPrimitive, TryFromPrimitive};
-use tiff::decoder::ifd::Value;
-use tiff::{TiffError, TiffResult};
+
+use crate::tiff::Value;
+use crate::tiff::{TiffError, TiffResult};
 
 /// Geospatial TIFF tag variants
 #[derive(Clone, Copy, Debug, PartialEq, TryFromPrimitive, IntoPrimitive, Eq, Hash)]