🐛 Fix nested file structure support for ZIP/RAR format

core/Cargo.toml

@@ -44,7 +44,7 @@ uuid = "1.8.0"
 walkdir = "2.4.0"
 webp = "0.2.6"
 xml-rs = "0.8.20" # XML reader/writer
-zip = "0.6.6"
+zip = "2.1.3"
 
 [dev-dependencies]
 temp-env = "0.3.6"

core/src/filesystem/archive.rs

@@ -5,7 +5,7 @@ use std::{
 };
 use tracing::{trace, warn};
 use walkdir::WalkDir;
-use zip::write::FileOptions;
+use zip::{write::FileOptions, CompressionMethod};
 
 /// Creates a new zip file at `destination` from the contents of the folder `unpacked_path`.
 pub(crate) fn zip_dir(
@@ -17,8 +17,8 @@ pub(crate) fn zip_dir(
 
 	let mut zip_writer = zip::ZipWriter::new(zip_file);
 
-	let options = FileOptions::default()
-		.compression_method(zip::CompressionMethod::Stored)
+	let options: FileOptions<'_, ()> = FileOptions::default()
+		.compression_method(CompressionMethod::Stored)
 		.unix_permissions(0o755);
 
 	trace!("Creating zip file at {:?}", destination);
@@ -35,7 +35,6 @@ pub(crate) fn zip_dir(
 		// Some unzip tools unzip files with directory paths correctly, some do not!
 		if path.is_file() {
 			trace!("Adding file to zip file: {:?} as {:?}", path, name);
-			#[allow(deprecated)]
 			zip_writer.start_file_from_path(name, options)?;
 			let mut f = File::open(path)?;
 

core/src/filesystem/common.rs

@@ -71,14 +71,35 @@ pub struct FileParts {
 }
 
 pub trait PathUtils {
+	/// Returns the file name, file stem, and extension of the file.
 	fn file_parts(&self) -> FileParts;
+	/// Returns the result of `infer::get_from_path`.
 	fn infer_kind(&self) -> std::io::Result<Option<infer::Type>>;
+	/// Returns the content type of the file based on the extension.
+	fn naive_content_type(&self) -> ContentType;
+	/// Returns true if the file is hidden (i.e. starts with a dot). Also checks for
+	/// files within a __MACOSX directory.
 	fn is_hidden_file(&self) -> bool;
+	/// Returns true if the file is supported by Stump.
 	fn should_ignore(&self) -> bool;
+	/// Returns true if the file is an image.
 	fn is_supported(&self) -> bool;
+	/// Returns true if the file is an image.
 	fn is_img(&self) -> bool;
+	/// Returns true if the file is a thumbnail image. This calls the `is_img` function
+	/// from the same trait, and then checks if the file name is one of the following:
+	/// - cover
+	/// - thumbnail
+	/// - folder
+	///
+	/// These will *potentially* be reserved filenames in the future... Not sure
+	/// if this functionality will be kept.
 	fn is_thumbnail_img(&self) -> bool;
+	/// Returns true if the directory has any media files in it. This is a shallow
+	/// check, and will not check subdirectories.
 	fn dir_has_media(&self) -> bool;
+	/// Returns true if the directory has any media files in it. This is a deep
+	/// check, and will check *all* subdirectories.
 	fn dir_has_media_deep(&self) -> bool;
 }
 
@@ -120,13 +141,33 @@ impl PathUtils for Path {
 		infer::get_from_path(self)
 	}
 
+	fn naive_content_type(&self) -> ContentType {
+		let extension = self
+			.extension()
+			.and_then(|e| e.to_str())
+			.unwrap_or_default();
+
+		if extension.is_empty() {
+			return ContentType::UNKNOWN;
+		}
+
+		ContentType::from_extension(extension)
+	}
+
 	/// Returns true if the file is hidden (i.e. starts with a dot).
 	fn is_hidden_file(&self) -> bool {
+		// If the file is contained inside of a __MACOSX directory, assume it is hidden.
+		// We don't want to deal with these files.
+		if self.starts_with("__MACOSX") {
+			return true;
+		}
+
 		let FileParts { file_name, .. } = self.file_parts();
 
 		file_name.starts_with('.')
 	}
 
+	// TODO(327): Remove infer usage
 	/// Returns true if the file is a supported media file. This is a strict check when
 	/// infer can determine the file type, and a loose extension-based check when infer cannot.
 	fn is_supported(&self) -> bool {
@@ -148,28 +189,11 @@ impl PathUtils for Path {
 		!self.is_supported()
 	}
 
-	/// Returns true if the file is an image. This is a strict check when infer
-	/// can determine the file type, and a loose extension-based check when infer cannot.
+	/// Returns true if the file is an image. This is a naive check based on the extension.
 	fn is_img(&self) -> bool {
-		if let Ok(Some(file_type)) = infer::get_from_path(self) {
-			return file_type.mime_type().starts_with("image/");
-		}
-
-		let FileParts { extension, .. } = self.file_parts();
-
-		extension.eq_ignore_ascii_case("jpg")
-			|| extension.eq_ignore_ascii_case("png")
-			|| extension.eq_ignore_ascii_case("jpeg")
+		self.naive_content_type().is_image()
 	}
 
-	/// Returns true if the file is a thumbnail image. This calls the `is_img` function
-	/// from the same trait, and then checks if the file name is one of the following:
-	/// - cover
-	/// - thumbnail
-	/// - folder
-	///
-	/// These will *potentially* be reserved filenames in the future... Not sure
-	/// if this functionality will be kept.
 	fn is_thumbnail_img(&self) -> bool {
 		if !self.is_img() {
 			return false;
@@ -180,8 +204,6 @@ impl PathUtils for Path {
 		is_accepted_cover_name(&file_stem)
 	}
 
-	/// Returns true if the directory has any media files in it. This is a shallow
-	/// check, and will not check subdirectories.
 	fn dir_has_media(&self) -> bool {
 		if !self.is_dir() {
 			return false;
@@ -205,8 +227,6 @@ impl PathUtils for Path {
 		}
 	}
 
-	/// Returns true if the directory has any media files in it. This is a deep
-	/// check, and will check *all* subdirectories.
 	fn dir_has_media_deep(&self) -> bool {
 		if !self.is_dir() {
 			return false;

core/src/filesystem/media/epub.rs

@@ -117,8 +117,8 @@ impl FileProcessor for EpubProcessor {
 		for chapter in pages {
 			if chapter == 1 {
 				// Assume this is the cover page
-				// FIXME: This is wrong. I just don't want to deal with it right now...
-				content_types.insert(chapter, ContentType::JPEG);
+				let (content_type, _) = Self::get_cover_internal(&mut epub_file)?;
+				content_types.insert(chapter, content_type);
 				continue;
 			}
 
@@ -153,20 +153,9 @@ impl EpubProcessor {
 		EpubDoc::new(path).map_err(|e| FileError::EpubOpenError(e.to_string()))
 	}
 
-	/// Returns the cover image for the epub file. If a cover image cannot be extracted via the
-	/// metadata, it will go through two rounds of fallback methods:
-	///
-	/// 1. Attempt to find a resource with the default ID of "cover"
-	/// 2. Attempt to find a resource with a mime type of "image/jpeg" or "image/png", and weight the
-	/// results based on how likely they are to be the cover. For example, if the cover is named
-	/// "cover.jpg", it's probably the cover. The entry with the heighest weight, if any, will be
-	/// returned.
-	pub fn get_cover(path: &str) -> Result<(ContentType, Vec<u8>), FileError> {
-		let mut epub_file = EpubDoc::new(path).map_err(|e| {
-			tracing::error!("Failed to open epub file: {}", e);
-			FileError::EpubOpenError(e.to_string())
-		})?;
-
+	fn get_cover_internal(
+		epub_file: &mut EpubDoc<BufReader<File>>,
+	) -> Result<(ContentType, Vec<u8>), FileError> {
 		let cover_id = epub_file.get_cover_id().unwrap_or_else(|| {
 			tracing::debug!("Epub file does not contain cover metadata");
 			DEFAULT_EPUB_COVER_ID.to_string()
@@ -214,6 +203,23 @@ impl EpubProcessor {
 		))
 	}
 
+	/// Returns the cover image for the epub file. If a cover image cannot be extracted via the
+	/// metadata, it will go through two rounds of fallback methods:
+	///
+	/// 1. Attempt to find a resource with the default ID of "cover"
+	/// 2. Attempt to find a resource with a mime type of "image/jpeg" or "image/png", and weight the
+	/// results based on how likely they are to be the cover. For example, if the cover is named
+	/// "cover.jpg", it's probably the cover. The entry with the heighest weight, if any, will be
+	/// returned.
+	pub fn get_cover(path: &str) -> Result<(ContentType, Vec<u8>), FileError> {
+		let mut epub_file = EpubDoc::new(path).map_err(|e| {
+			tracing::error!("Failed to open epub file: {}", e);
+			FileError::EpubOpenError(e.to_string())
+		})?;
+
+		EpubProcessor::get_cover_internal(&mut epub_file)
+	}
+
 	pub fn get_chapter(
 		path: &str,
 		chapter: usize,

core/src/filesystem/media/mod.rs

@@ -59,4 +59,13 @@ mod tests {
 			.to_string_lossy()
 			.to_string()
 	}
+
+	// Note: each page should be 96623 bytes. The macOS metadata files should be 220 bytes, but
+	// ignored by the processor. Commenting the sizes for posterity.
+	pub fn get_nested_macos_compressed_cbz_path() -> String {
+		PathBuf::from(env!("CARGO_MANIFEST_DIR"))
+			.join("integration-tests/data/nested-macos-compressed.cbz")
+			.to_string_lossy()
+			.to_string()
+	}
 }