Support detecting doc/xls/ppt (#38)

* Check docProps in msooxml matcher

* Allow clippy::upper-case-acronyms lint

* Fix typo: OOXLM -> OOXML

* Support detecting doc/xls/ppt

Author: messense

Cargo.toml

@@ -14,10 +14,13 @@ exclude = ["testdata/*", "tests/*"]
 
 [features]
 default = ["std"]
-std = ["alloc"]
+std = ["alloc", "cfb"]
 alloc = []
 
 [[example]]
 name = "file"
 path = "examples/file.rs"
 required-features = ["std"]
+
+[dependencies]
+cfb = { version = "0.4.0", optional = true }

README.md

@@ -209,7 +209,6 @@ assert_eq!(kind.extension(), "foo");
 
 ## Known Issues
 
-- `doc`, `ppt`, `xls`, `msi` all have the same magic number so it's not possible to tell which one just based on the binary data. `doc` is returned for all.
 - `exe` and `dll` have the same magic number so it's not possible to tell which one just based on the binary data. `exe` is returned for all.
 
 ## License

src/matchers/doc.rs

@@ -3,26 +3,18 @@ use core::convert::TryInto;
 #[allow(clippy::upper_case_acronyms)]
 #[derive(Debug, Eq, PartialEq)]
 enum DocType {
-    // DOC,
+    DOC,
     DOCX,
-    // XLS,
+    XLS,
     XLSX,
-    // PPT,
+    PPT,
     PPTX,
-    OOXLM,
+    OOXML,
 }
 
 /// Returns whether a buffer is Microsoft Word Document (DOC) data.
 pub fn is_doc(buf: &[u8]) -> bool {
-    buf.len() > 7
-        && buf[0] == 0xD0
-        && buf[1] == 0xCF
-        && buf[2] == 0x11
-        && buf[3] == 0xE0
-        && buf[4] == 0xA1
-        && buf[5] == 0xB1
-        && buf[6] == 0x1A
-        && buf[7] == 0xE1
+    ole2(buf) == Some(DocType::DOC)
 }
 
 /// Returns whether a buffer is Microsoft Word Open XML Format Document (DOCX) data.
@@ -32,15 +24,7 @@ pub fn is_docx(buf: &[u8]) -> bool {
 
 /// Returns whether a buffer is Microsoft Excel 97-2003 Worksheet (XLS) data.
 pub fn is_xls(buf: &[u8]) -> bool {
-    buf.len() > 7
-        && buf[0] == 0xD0
-        && buf[1] == 0xCF
-        && buf[2] == 0x11
-        && buf[3] == 0xE0
-        && buf[4] == 0xA1
-        && buf[5] == 0xB1
-        && buf[6] == 0x1A
-        && buf[7] == 0xE1
+    ole2(buf) == Some(DocType::XLS)
 }
 
 /// Returns whether a buffer is Microsoft Excel Open XML Format Spreadsheet (XLSX) data.
@@ -50,15 +34,7 @@ pub fn is_xlsx(buf: &[u8]) -> bool {
 
 /// Returns whether a buffer is Microsoft PowerPoint 97-2003 Presentation (PPT) data.
 pub fn is_ppt(buf: &[u8]) -> bool {
-    buf.len() > 7
-        && buf[0] == 0xD0
-        && buf[1] == 0xCF
-        && buf[2] == 0x11
-        && buf[3] == 0xE0
-        && buf[4] == 0xA1
-        && buf[5] == 0xB1
-        && buf[6] == 0x1A
-        && buf[7] == 0xE1
+    ole2(buf) == Some(DocType::PPT)
 }
 
 /// Returns whether a buffer is Microsoft PowerPoint Open XML Presentation (PPTX) data.
@@ -108,15 +84,43 @@ fn msooxml(buf: &[u8]) -> Option<DocType> {
     let idx = search(buf, start_offset, 6000);
     match idx {
         Some(idx) => start_offset += idx + 4 + 26,
-        None => return Some(DocType::OOXLM),
+        None => return Some(DocType::OOXML),
     };
 
     let typo = check_msooml(buf, start_offset);
     if typo.is_some() {
         return typo;
     }
 
-    Some(DocType::OOXLM)
+    Some(DocType::OOXML)
+}
+
+#[cfg(feature = "std")]
+fn ole2(buf: &[u8]) -> Option<DocType> {
+    use std::io::Cursor;
+
+    if !compare_bytes(buf, &[0xD0, 0xCF, 0x11, 0xE0, 0xA1, 0xB1, 0x1A, 0xE1], 0) {
+        return None;
+    }
+    if let Ok(file) = cfb::CompoundFile::open(Cursor::new(buf)) {
+        return match file.root_entry().clsid().to_string().as_str() {
+            "00020810-0000-0000-c000-000000000046" | "00020820-0000-0000-c000-000000000046" => {
+                Some(DocType::XLS)
+            }
+            "00020906-0000-0000-c000-000000000046" => Some(DocType::DOC),
+            "64818d10-4f9b-11cf-86ea-00aa00b929e8" => Some(DocType::PPT),
+            _ => None,
+        };
+    }
+    None
+}
+
+#[cfg(not(feature = "std"))]
+fn ole2(buf: &[u8]) -> Option<DocType> {
+    if !compare_bytes(buf, &[0xD0, 0xCF, 0x11, 0xE0, 0xA1, 0xB1, 0x1A, 0xE1], 0) {
+        return None;
+    }
+    Some(DocType::DOC)
 }
 
 fn compare_bytes(slice: &[u8], sub_slice: &[u8], start_offset: usize) -> bool {

tests/doc.rs

@@ -1,6 +1,30 @@
 mod common;
 
-test_format!(DOC, "application/msword", "doc", doc, "sample.doc");
+#[cfg(feature = "std")]
+macro_rules! test_format_get_only {
+    ($exp_matchert:ident, $exp_mimet:expr, $exp_ext:expr, $format:ident, $file:expr) => {
+        mod $format {
+            use infer::{MatcherType, Type};
+
+            fn matcher(_buf: &[u8]) -> bool {
+                false
+            }
+
+            #[test]
+            fn get() {
+                let expected_kind =
+                    Type::new(MatcherType::$exp_matchert, $exp_mimet, $exp_ext, matcher);
+                let buf = include_bytes!(concat!("../testdata/", $file));
+                let kind = infer::get(buf).expect("test file matches");
+
+                assert_eq!(expected_kind, kind);
+            }
+        }
+    };
+}
+
+#[cfg(feature = "std")]
+test_format_get_only!(DOC, "application/msword", "doc", doc, "sample.doc");
 
 test_format!(
     DOC,
@@ -10,6 +34,9 @@ test_format!(
     "sample.docx"
 );
 
+#[cfg(feature = "std")]
+test_format_get_only!(DOC, "application/vnd.ms-excel", "xls", xls, "sample.xls");
+
 test_format!(
     DOC,
     "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
@@ -18,6 +45,15 @@ test_format!(
     "sample.xlsx"
 );
 
+#[cfg(feature = "std")]
+test_format_get_only!(
+    DOC,
+    "application/vnd.ms-powerpoint",
+    "ppt",
+    ppt,
+    "sample.ppt"
+);
+
 test_format!(
     DOC,
     "application/vnd.openxmlformats-officedocument.presentationml.presentation",