fix bugs

pdf-rs · Dec 1, 2023 · e54cbe4 · e54cbe4
1 parent 0de4f2d
commit e54cbe4
Show file tree

Hide file tree

Showing 5 changed files with 30 additions and 11 deletions.
diff --git a/pdf/Cargo.toml b/pdf/Cargo.toml
@@ -40,7 +40,7 @@ euclid = { version = "0.22.7", optional = true }
 bitflags = "1.3"
 istring = { version = "0.3.3", features = ["std", "size"] }
 datasize = "0.2.13"
-globalcache = { version = "0.2", features = ["sync"], optional = true }
+globalcache = { version = "0.2.2", features = ["sync"], optional = true }
 indexmap = "2.1.0"
 
 [dev-dependencies]

diff --git a/pdf/src/file.rs b/pdf/src/file.rs
@@ -39,19 +39,24 @@ impl<T> PromisedRef<T> {
 
 pub trait Cache<T: Clone> {
     fn get_or_compute(&self, key: PlainRef, compute: impl FnOnce() -> T) -> T;
+    fn clear(&self);
 }
 pub struct NoCache;
 impl<T: Clone> Cache<T> for NoCache {
     fn get_or_compute(&self, _key: PlainRef, compute: impl FnOnce() -> T) -> T {
         compute()
     }
+    fn clear(&self) {}
 }
 
 #[cfg(feature="cache")]
 impl<T: Clone + ValueSize + Send + 'static> Cache<T> for Arc<SyncCache<PlainRef, T>> {
     fn get_or_compute(&self, key: PlainRef, compute: impl FnOnce() -> T) -> T {
         self.get(key, compute)
     }
+    fn clear(&self) {
+        (**self).clear()
+    }
 }
 
 pub trait Log {
@@ -67,7 +72,7 @@ pub struct Storage<B, OC, SC, L> {
     stream_cache: SC,
 
     // objects that differ from the backend
-    changes:    HashMap<ObjNr, Primitive>,
+    changes:    HashMap<ObjNr, (Primitive, GenNr)>,
 
     refs:       XRefTable,
 
@@ -226,7 +231,7 @@ where
     }
     fn resolve_ref(&self, r: PlainRef, flags: ParseFlags, resolve: &impl Resolve) -> Result<Primitive> {
         match self.changes.get(&r.id) {
-            Some(p) => Ok((*p).clone()),
+            Some((p, _)) => Ok((*p).clone()),
             None => match t!(self.refs.get(r.id)) {
                 XRef::Raw {pos, ..} => {
                     let mut lexer = Lexer::with_offset(t!(self.backend.read(self.start_offset + pos ..)), self.start_offset + pos);
@@ -347,7 +352,7 @@ where
         let id = self.refs.len() as u64;
         self.refs.push(XRef::Promised);
         let primitive = obj.to_primitive(self)?;
-        self.changes.insert(id, primitive);
+        self.changes.insert(id, (primitive, 0));
         let rc = Shared::new(obj);
         let r = PlainRef { id, gen: 0 };
 
@@ -362,7 +367,7 @@ where
             XRef::Invalid => panic!()
         };
         let primitive = obj.to_primitive(self)?;
-        self.changes.insert(old.id, primitive);
+        self.changes.insert(old.id, (primitive, r.gen));
         let rc = Shared::new(obj);
 
         Ok(RcRef::new(r, rc))
@@ -400,14 +405,13 @@ where
 
         let xref_promise = self.promise::<Stream<XRefInfo>>();
 
-
         let mut changes: Vec<_> = self.changes.iter().collect();
         changes.sort_unstable_by_key(|&(id, _)| id);
 
-        for (&id, primitive) in changes.iter() {
+        for &(&id, &(ref primitive, gen)) in changes.iter() {
             let pos = self.backend.len();
-            self.refs.set(id, XRef::Raw { pos: pos as _, gen_nr: 0 });
-            writeln!(self.backend, "{} {} obj", id, 0)?;
+            self.refs.set(id, XRef::Raw { pos: pos as _, gen_nr: gen });
+            writeln!(self.backend, "{} {} obj", id, gen)?;
             primitive.serialize(&mut self.backend)?;
             writeln!(self.backend, "endobj")?;
         }
@@ -431,6 +435,7 @@ where
         write!(self.backend, "\nstartxref\n{}\n%%EOF", xref_pos).unwrap();
 
         // update trailer which may have change now.
+        self.cache.clear();
         *trailer = Trailer::from_dict(trailer_dict, &self.resolver())?;
 
         Ok(&self.backend)

diff --git a/pdf/src/font.rs b/pdf/src/font.rs
@@ -517,7 +517,7 @@ impl ToUnicodeMap {
 pub fn utf16be_to_char(
     data: &[u8],
 ) -> impl Iterator<Item = std::result::Result<char, std::char::DecodeUtf16Error>> + '_ {
-    char::decode_utf16(data.chunks(2).map(|w| u16::from_be_bytes([w[0], w[1]])))
+    char::decode_utf16(data.chunks_exact(2).map(|w| u16::from_be_bytes([w[0], w[1]])))
 }
 /// converts UTF16-BE to a string replacing illegal/unknown characters
 pub fn utf16be_to_string_lossy(data: &[u8]) -> String {

diff --git a/pdf/src/object/types.rs b/pdf/src/object/types.rs
@@ -958,6 +958,9 @@ pub struct FieldDictionary {
 
     #[pdf(key="AA")]
     pub actions: Option<Dictionary>,
+
+    #[pdf(other)]
+    pub other: Dictionary
 }
 
 #[derive(Debug, DataSize)]

diff --git a/pdf_derive/src/lib.rs b/pdf_derive/src/lib.rs
@@ -725,6 +725,17 @@ fn impl_objectwrite_for_struct(ast: &DeriveInput, fields: &Fields) -> SynStream
         None => quote! {}
     };
 
+    let other = parts.iter().filter(|(field, attrs, _)| attrs.other).flat_map(|(field, _, _)| field).next();
+    let init_dict = if let Some(other) = other {
+        quote! {
+            let mut dict = self.#other.clone();
+        }
+    } else {
+        quote! {
+            let mut dict = pdf::primitive::Dictionary::new();
+        }
+    };
+
     quote! {
         impl #impl_generics pdf::object::ObjectWrite for #id #ty_generics #where_clause {
             fn to_primitive(&self, update: &mut impl pdf::object::Updater) -> Result<pdf::primitive::Primitive> {
@@ -733,7 +744,7 @@ fn impl_objectwrite_for_struct(ast: &DeriveInput, fields: &Fields) -> SynStream
         }
         impl #impl_generics pdf::object::ToDict for #id #ty_generics #where_clause {
             fn to_dict(&self, updater: &mut impl pdf::object::Updater) -> Result<pdf::primitive::Dictionary> {
-                let mut dict = pdf::primitive::Dictionary::new();
+                #init_dict
                 #pdf_type
                 #( #checks_code )*
                 #(#fields_ser)*