Skip to content

A fast, helpful, and open-source document parser

improve file type from bytes detection

Logan Markewich29a35b38be7e
diff --git a/Cargo.lock b/Cargo.lock
index 0a11802..194b2f3 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -264,17 +264,6 @@ dependencies = [
"nom",
]
-[[package]]
-name = "cfb"
-version = "0.7.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d38f2da7a0a2c4ccf0065be06397cc26a81f4e528be095826eee9d4adbb8c60f"
-dependencies = [
- "byteorder",
- "fnv",
- "uuid",
-]
-
[[package]]
name = "cfg-if"
version = "1.0.4"
@@ -598,6 +587,12 @@ dependencies = [
"simd-adler32",
]
+[[package]]
+name = "file-format"
+version = "0.29.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "55d9ccda37e95b4f0978a3074b4a9939979103a7256459cfb449c9c84d1adf23"
+
[[package]]
name = "filetime"
version = "0.2.27"
@@ -1082,15 +1077,6 @@ dependencies = [
"serde_core",
]
-[[package]]
-name = "infer"
-version = "0.19.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a588916bfdfd92e71cacef98a63d9b1f0d74d6599980d11894290e7ddefffcf7"
-dependencies = [
- "cfb",
-]
-
[[package]]
name = "inout"
version = "0.1.4"
@@ -1250,8 +1236,8 @@ version = "2.2.1"
dependencies = [
"blake3",
"clap",
+ "file-format",
"image",
- "infer",
"liteparse-pdfium",
"liteparse-pdfium-sys",
"ordered-float",
@@ -2813,16 +2799,6 @@ version = "0.2.2"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "06abde3611657adf66d383f00b093d7faecc7fa57071cce2578660c9f1010821"
-[[package]]
-name = "uuid"
-version = "1.23.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ddd74a9687298c6858e9b88ec8935ec45d22e8fd5e6394fa1bd4e99a87789c76"
-dependencies = [
- "js-sys",
- "wasm-bindgen",
-]
-
[[package]]
name = "vcpkg"
version = "0.2.15"
diff --git a/crates/liteparse/Cargo.toml b/crates/liteparse/Cargo.toml
index 48477fc..e4b085a 100644
--- a/crates/liteparse/Cargo.toml
+++ b/crates/liteparse/Cargo.toml
@@ -21,8 +21,8 @@ tesseract = ["dep:tesseract-rs"]
[dependencies]
blake3 = "1"
clap = { version = "4.5.55", features = ["derive"] }
+file-format = { version = "0.29.0", features = ["reader"] }
image = { version = "0.25", default-features = false, features = ["png"] }
-infer = "0.19.0"
ordered-float = "5.3.0"
pdfium = { package = "liteparse-pdfium", version = "1.3.0", path = "../pdfium" }
pdfium-sys = { package = "liteparse-pdfium-sys", version = "1.3.0", path = "../pdfium-sys" }
diff --git a/crates/liteparse/src/conversion.rs b/crates/liteparse/src/conversion.rs
index e6e83d5..95d8d2f 100644
--- a/crates/liteparse/src/conversion.rs
+++ b/crates/liteparse/src/conversion.rs
@@ -1,3 +1,4 @@
+use file_format::FileFormat;
use tempfile::TempDir;
use crate::error::LiteParseError;
@@ -553,7 +554,16 @@ pub async fn convert_image_to_pdf(
}
pub fn guess_extension_from_data(data: &[u8]) -> Option<String> {
- infer::get(data).map(|k| k.extension().to_string())
+ // `file-format` inspects ZIP-based containers via their central directory
+ // (requires the `reader` feature), so DOCX/XLSX/PPTX/ODF resolve to their
+ // specific format instead of a generic "zip" regardless of entry ordering
+ // or file size. Unrecognized input falls back to the generic binary format,
+ // which we surface as `None` to match the prior contract.
+ let fmt = FileFormat::from_bytes(data);
+ if fmt == FileFormat::ArbitraryBinaryData {
+ return None;
+ }
+ Some(fmt.extension().to_string())
}
pub async fn convert_data_to_pdf(
@@ -602,6 +612,74 @@ mod tests {
assert!(!is_supported_extension("noext"));
}
+ /// Build a minimal but structurally valid ZIP (stored, empty entries) whose
+ /// central directory lists `names`. This exercises the same code path
+ /// `file-format` uses to disambiguate ZIP-based office formats.
+ fn build_zip(names: &[&str]) -> Vec<u8> {
+ let mut out = Vec::new();
+ let mut offsets = Vec::new();
+ for name in names {
+ offsets.push(out.len() as u32);
+ let nb = name.as_bytes();
+ out.extend_from_slice(&0x0403_4b50u32.to_le_bytes()); // local header sig
+ out.extend_from_slice(&20u16.to_le_bytes()); // version needed
+ out.extend_from_slice(&[0u8; 20]); // flags..uncompressed size (all zero)
+ out.extend_from_slice(&(nb.len() as u16).to_le_bytes());
+ out.extend_from_slice(&0u16.to_le_bytes()); // extra len
+ out.extend_from_slice(nb);
+ }
+ let cd_offset = out.len() as u32;
+ let mut central = Vec::new();
+ for (i, name) in names.iter().enumerate() {
+ let nb = name.as_bytes();
+ central.extend_from_slice(&0x0201_4b50u32.to_le_bytes()); // central header sig
+ central.extend_from_slice(&20u16.to_le_bytes()); // version made by
+ central.extend_from_slice(&20u16.to_le_bytes()); // version needed
+ central.extend_from_slice(&[0u8; 20]); // flags..uncompressed size
+ central.extend_from_slice(&(nb.len() as u16).to_le_bytes());
+ central.extend_from_slice(&[0u8; 8]); // extra/comment/disk/internal attrs
+ central.extend_from_slice(&0u32.to_le_bytes()); // external attrs
+ central.extend_from_slice(&offsets[i].to_le_bytes()); // local header offset
+ central.extend_from_slice(nb);
+ }
+ let cd_size = central.len() as u32;
+ out.extend_from_slice(&central);
+ out.extend_from_slice(&0x0605_4b50u32.to_le_bytes()); // EOCD sig
+ out.extend_from_slice(&[0u8; 4]); // disk numbers
+ out.extend_from_slice(&(names.len() as u16).to_le_bytes());
+ out.extend_from_slice(&(names.len() as u16).to_le_bytes());
+ out.extend_from_slice(&cd_size.to_le_bytes());
+ out.extend_from_slice(&cd_offset.to_le_bytes());
+ out.extend_from_slice(&0u16.to_le_bytes()); // comment len
+ out
+ }
+
+ #[test]
+ fn test_guess_extension_disambiguates_zip_office_formats() {
+ // ZIP-based office formats must resolve to their specific type, not a
+ // generic "zip", even though they share the PK signature.
+ assert_eq!(
+ guess_extension_from_data(&build_zip(&["[Content_Types].xml", "word/document.xml"]))
+ .as_deref(),
+ Some("docx")
+ );
+ assert_eq!(
+ guess_extension_from_data(&build_zip(&["[Content_Types].xml", "xl/workbook.xml"]))
+ .as_deref(),
+ Some("xlsx")
+ );
+ assert_eq!(
+ guess_extension_from_data(&build_zip(&["[Content_Types].xml", "ppt/presentation.xml"]))
+ .as_deref(),
+ Some("pptx")
+ );
+ // A plain ZIP with no office markers stays "zip".
+ assert_eq!(
+ guess_extension_from_data(&build_zip(&["random/file.txt"])).as_deref(),
+ Some("zip")
+ );
+ }
+
#[test]
fn test_conversion_tool_display() {
assert_eq!(ConversionTool::ImageMagick.to_string(), "ImageMagick");