Merge pull request #249

Velli20 · web-flow · commit 182aa3bda9c7 · 2026-08-01T15:36:11.000+03:00
pdf-object: remove cross-reference entry wrapper
diff --git a/crates/pdf-document/src/object_loader.rs b/crates/pdf-document/src/object_loader.rs
@@ -19,8 +19,7 @@ use crate::error::PdfReaderError;
 use crate::object_stream::read_object_stream;
 use crate::reader::{EncryptionContext, object_id};
 use pdf_object::{
-    cross_reference_table::{CrossReferenceEntry, CrossReferenceEntryType},
-    error::ObjectError,
+    cross_reference_table::CrossReferenceEntryType, error::ObjectError,
     object_variant::ObjectVariant,
 };
 use pdf_object_collection::object_collection::ObjectCollection;
@@ -45,12 +44,12 @@ struct LoadPlan {
 
 impl LoadPlan {
     /// Partitions live xref entries into normal objects and compressed-stream batches.
-    fn from_entries(entries: &BTreeMap<usize, CrossReferenceEntry>) -> Self {
+    fn from_entries(entries: &BTreeMap<usize, CrossReferenceEntryType>) -> Self {
         let mut normal_objects = entries
             .values()
-            .filter_map(|entry| match entry.entry_type {
-                CrossReferenceEntryType::Normal { byte_offset, .. } if byte_offset != 0 => {
-                    Some(byte_offset)
+            .filter_map(|entry| match entry {
+                CrossReferenceEntryType::Normal { byte_offset, .. } if *byte_offset != 0 => {
+                    Some(*byte_offset)
                 }
                 _ => None,
             })
@@ -67,14 +66,14 @@ impl LoadPlan {
             let CrossReferenceEntryType::Compressed {
                 object_stream_number,
                 index_within_stream,
-            } = entry.entry_type
+            } = entry
             else {
                 continue;
             };
             compressed_streams
-                .entry(object_stream_number)
+                .entry(*object_stream_number)
                 .or_default()
-                .entry(index_within_stream)
+                .entry(*index_within_stream)
                 .or_default()
                 .push(object_number);
         }
@@ -264,7 +263,7 @@ pub(super) struct ObjectLoader<'input, 'loader> {
 impl<'input, 'loader> ObjectLoader<'input, 'loader> {
     /// Creates an object loader for one parsed cross-reference table.
     pub(super) fn new(
-        entries: &BTreeMap<usize, CrossReferenceEntry>,
+        entries: &BTreeMap<usize, CrossReferenceEntryType>,
         parser: &'loader mut PdfParser<'input>,
         encryption: EncryptionContext,
         diagnostics: &'loader mut Vec<PdfReadDiagnostic>,
@@ -476,12 +475,12 @@ mod tests {
     #[test]
     fn load_plan_partitions_entries_and_groups_compressed_indexes() {
         let entries = BTreeMap::from([
-            (0, CrossReferenceEntry::new_free(0, 65_535)),
-            (1, CrossReferenceEntry::new_normal(10, 0)),
-            (2, CrossReferenceEntry::new_compressed(5, 1)),
-            (3, CrossReferenceEntry::new_compressed(5, 1)),
-            (4, CrossReferenceEntry::new_normal(0, 0)),
-            (5, CrossReferenceEntry::new_normal(50, 0)),
+            (0, CrossReferenceEntryType::new_free(0, 65_535)),
+            (1, CrossReferenceEntryType::new_normal(10, 0)),
+            (2, CrossReferenceEntryType::new_compressed(5, 1)),
+            (3, CrossReferenceEntryType::new_compressed(5, 1)),
+            (4, CrossReferenceEntryType::new_normal(0, 0)),
+            (5, CrossReferenceEntryType::new_normal(50, 0)),
         ]);
 
         let plan = LoadPlan::from_entries(&entries);
diff --git a/crates/pdf-document/src/reader.rs b/crates/pdf-document/src/reader.rs
@@ -14,7 +14,7 @@ use pdf_object::object_id::PdfObjectId;
 use pdf_object::object_lookup::ObjectLookupExt;
 use pdf_object::object_resolver::{ObjectResolver, PassthroughResolver};
 use pdf_object::{
-    cross_reference_table::{CrossReferenceEntry, CrossReferenceTable},
+    cross_reference_table::{CrossReferenceEntryType, CrossReferenceTable},
     error::ObjectError,
     object_variant::ObjectVariant,
     trailer::Trailer,
@@ -94,7 +94,7 @@ impl EncryptionContext {
     /// Creates decryption state from the trailer's optional encryption entry.
     fn from_trailer(
         trailer: &mut Trailer,
-        entries: &BTreeMap<usize, CrossReferenceEntry>,
+        entries: &BTreeMap<usize, CrossReferenceEntryType>,
         parser: &mut PdfParser,
         password: &[u8],
         diagnostics: &mut Vec<PdfReadDiagnostic>,
@@ -171,7 +171,7 @@ fn extract_page_tree(
 /// Resolves and parses the trailer's encryption dictionary without decrypting it.
 fn load_encrypt_dictionary(
     encrypt_reference: ObjectVariant,
-    entries: &BTreeMap<usize, CrossReferenceEntry>,
+    entries: &BTreeMap<usize, CrossReferenceEntryType>,
     parser: &mut PdfParser,
 ) -> Result<EncryptDictionary, PdfReaderError> {
     let object = match encrypt_reference {
diff --git a/crates/pdf-object-collection/src/object_collection.rs b/crates/pdf-object-collection/src/object_collection.rs
@@ -216,7 +216,7 @@ impl ObjectCollection {
                         (
                             k.to_string(),
                             json!({
-                                "entry_type": format!("{:?}", v.entry_type)
+                                "entry_type": format!("{v:?}")
                             }),
                         )
                     })
diff --git a/crates/pdf-object/src/cross_reference_table.rs b/crates/pdf-object/src/cross_reference_table.rs
@@ -11,18 +11,18 @@ use crate::trailer::Trailer;
 #[derive(Debug, PartialEq, Clone)]
 pub struct CrossReferenceTable {
     /// The map of object numbers to cross-reference entries.
-    pub entries: BTreeMap<usize, CrossReferenceEntry>,
+    pub entries: BTreeMap<usize, CrossReferenceEntryType>,
     /// The trailer associated with this cross-reference table.
     pub trailer: Trailer,
 }
 
 impl CrossReferenceTable {
-    pub fn new(entries: BTreeMap<usize, CrossReferenceEntry>, trailer: Trailer) -> Self {
+    pub fn new(entries: BTreeMap<usize, CrossReferenceEntryType>, trailer: Trailer) -> Self {
         CrossReferenceTable { entries, trailer }
     }
 }
 
-/// Distinguishes the type of a cross-reference entry.
+/// Represents a cross-reference entry.
 #[derive(Debug, PartialEq, Clone)]
 pub enum CrossReferenceEntryType {
     /// Type 1: object at a byte offset in the file (traditional).
@@ -42,60 +42,52 @@ pub enum CrossReferenceEntryType {
     },
 }
 
-#[derive(Debug, PartialEq, Clone)]
-pub struct CrossReferenceEntry {
-    pub entry_type: CrossReferenceEntryType,
-}
-
-impl CrossReferenceEntry {
+impl CrossReferenceEntryType {
+    /// Creates a normal entry for an uncompressed object at a byte offset.
     pub fn new_normal(byte_offset: usize, generation_number: usize) -> Self {
-        CrossReferenceEntry {
-            entry_type: CrossReferenceEntryType::Normal {
-                byte_offset,
-                generation_number,
-            },
+        CrossReferenceEntryType::Normal {
+            byte_offset,
+            generation_number,
         }
     }
 
+    /// Creates an entry for an object stored within a compressed object stream.
     pub fn new_compressed(object_stream_number: usize, index_within_stream: usize) -> Self {
-        CrossReferenceEntry {
-            entry_type: CrossReferenceEntryType::Compressed {
-                object_stream_number,
-                index_within_stream,
-            },
+        CrossReferenceEntryType::Compressed {
+            object_stream_number,
+            index_within_stream,
         }
     }
 
+    /// Creates an entry for a free object.
     pub fn new_free(next_free_object: usize, generation_number: usize) -> Self {
-        CrossReferenceEntry {
-            entry_type: CrossReferenceEntryType::Free {
-                next_free_object,
-                generation_number,
-            },
+        CrossReferenceEntryType::Free {
+            next_free_object,
+            generation_number,
         }
     }
 
     /// Returns the byte offset if this is a Normal entry.
     pub fn byte_offset(&self) -> Option<usize> {
-        match &self.entry_type {
+        match self {
             CrossReferenceEntryType::Normal { byte_offset, .. } => Some(*byte_offset),
             _ => None,
         }
     }
 
     /// Returns true if this is a Normal (in-use, uncompressed) entry.
     pub fn is_normal(&self) -> bool {
-        matches!(self.entry_type, CrossReferenceEntryType::Normal { .. })
+        matches!(self, CrossReferenceEntryType::Normal { .. })
     }
 
     /// Returns true if this is a Free entry.
     pub fn is_free(&self) -> bool {
-        matches!(self.entry_type, CrossReferenceEntryType::Free { .. })
+        matches!(self, CrossReferenceEntryType::Free { .. })
     }
 
     /// Returns true if this is a Compressed entry.
     pub fn is_compressed(&self) -> bool {
-        matches!(self.entry_type, CrossReferenceEntryType::Compressed { .. })
+        matches!(self, CrossReferenceEntryType::Compressed { .. })
     }
 }
 
diff --git a/crates/pdf-parser/src/cross_reference_stream.rs b/crates/pdf-parser/src/cross_reference_stream.rs
@@ -1,7 +1,7 @@
 use std::collections::BTreeMap;
 
 use pdf_object::{
-    cross_reference_table::{CrossReferenceEntry, CrossReferenceTable},
+    cross_reference_table::{CrossReferenceEntryType, CrossReferenceTable},
     object_lookup::ObjectLookupExt,
     object_resolver::ObjectResolver,
     stream::StreamObject,
@@ -162,7 +162,7 @@ impl<'data> XrefStreamEntryDecoder<'data> {
     }
 
     /// Returns the next complete entry, or `None` when the data is truncated.
-    fn next_entry(&mut self) -> Option<CrossReferenceEntry> {
+    fn next_entry(&mut self) -> Option<CrossReferenceEntryType> {
         let end = self.position.saturating_add(self.layout.entry_width);
         let bytes = self.data.get(self.position..end)?;
         let entry = self.decode_entry(bytes)?;
@@ -171,7 +171,7 @@ impl<'data> XrefStreamEntryDecoder<'data> {
     }
 
     /// Decodes one entry according to the `/W` field widths.
-    fn decode_entry(&self, bytes: &[u8]) -> Option<CrossReferenceEntry> {
+    fn decode_entry(&self, bytes: &[u8]) -> Option<CrossReferenceEntryType> {
         let type_end = self.layout.type_width;
         let second_end = type_end.checked_add(self.layout.second_field_width)?;
         let third_end = second_end.checked_add(self.layout.third_field_width)?;
@@ -189,10 +189,10 @@ impl<'data> XrefStreamEntryDecoder<'data> {
         let third_value = read_field(third_field);
 
         Some(match entry_type {
-            0 => CrossReferenceEntry::new_free(second_value, third_value),
-            1 => CrossReferenceEntry::new_normal(second_value, third_value),
-            2 => CrossReferenceEntry::new_compressed(second_value, third_value),
-            _ => CrossReferenceEntry::new_free(0, 0),
+            0 => CrossReferenceEntryType::new_free(second_value, third_value),
+            1 => CrossReferenceEntryType::new_normal(second_value, third_value),
+            2 => CrossReferenceEntryType::new_compressed(second_value, third_value),
+            _ => CrossReferenceEntryType::new_free(0, 0),
         })
     }
 }
@@ -284,7 +284,7 @@ mod tests {
 
         let e2 = &table.entries[&2];
         assert!(e2.is_compressed());
-        match &e2.entry_type {
+        match e2 {
             CrossReferenceEntryType::Compressed {
                 object_stream_number,
                 index_within_stream,
@@ -349,7 +349,7 @@ mod tests {
 
         let table = parse_xref_stream(&stream, &PassthroughResolver).unwrap();
 
-        assert_eq!(table.entries[&0], CrossReferenceEntry::new_free(0, 0));
+        assert_eq!(table.entries[&0], CrossReferenceEntryType::new_free(0, 0));
     }
 
     /// Builds a compressed xref stream with FlateDecode + PNG Up predictor,
@@ -469,7 +469,7 @@ mod tests {
 
         let e3 = &table.entries[&3];
         assert!(e3.is_compressed());
-        match &e3.entry_type {
+        match e3 {
             CrossReferenceEntryType::Compressed {
                 object_stream_number,
                 index_within_stream,
diff --git a/crates/pdf-parser/src/cross_reference_table.rs b/crates/pdf-parser/src/cross_reference_table.rs
@@ -1,9 +1,7 @@
 use std::collections::BTreeMap;
 
 use pdf_object::{
-    cross_reference_table::{
-        CrossReferenceEntry, CrossReferenceEntryType, CrossReferenceStatus, CrossReferenceTable,
-    },
+    cross_reference_table::{CrossReferenceEntryType, CrossReferenceStatus, CrossReferenceTable},
     object_resolver::ObjectResolver,
 };
 use pdf_tokenizer::PdfToken;
@@ -37,7 +35,7 @@ impl PdfParser<'_> {
     /// Parses the contiguous subsections that follow an `xref` keyword.
     pub(crate) fn parse_cross_reference_subsections(
         &mut self,
-    ) -> Result<BTreeMap<usize, CrossReferenceEntry>, ParserError> {
+    ) -> Result<BTreeMap<usize, CrossReferenceEntryType>, ParserError> {
         let mut entries = BTreeMap::new();
 
         loop {
@@ -60,7 +58,7 @@ impl PdfParser<'_> {
     /// Parses one xref subsection header and its declared number of entries.
     fn parse_cross_reference_subsection(
         &mut self,
-    ) -> Result<(usize, Vec<CrossReferenceEntry>), ParserError> {
+    ) -> Result<(usize, Vec<CrossReferenceEntryType>), ParserError> {
         let start_object_number = self.read_number::<usize>(true)?;
         self.skip_whitespace_and_comments();
         let entry_count = self.read_number::<usize>(true)?;
@@ -76,7 +74,7 @@ impl PdfParser<'_> {
     /// Parses a traditional xref row into its corresponding entry.
     pub(crate) fn parse_cross_reference_entry(
         &mut self,
-    ) -> Result<CrossReferenceEntry, ParserError> {
+    ) -> Result<CrossReferenceEntryType, ParserError> {
         self.skip_whitespace_and_comments();
         let field1 = self.read_number::<usize>(true)?;
         self.skip_whitespace_and_comments();
@@ -93,9 +91,9 @@ impl PdfParser<'_> {
         })?;
 
         Ok(match status {
-            CrossReferenceStatus::Normal => CrossReferenceEntry::new_normal(field1, field2),
+            CrossReferenceStatus::Normal => CrossReferenceEntryType::new_normal(field1, field2),
             CrossReferenceStatus::Free | CrossReferenceStatus::Old => {
-                CrossReferenceEntry::new_free(field1, field2)
+                CrossReferenceEntryType::new_free(field1, field2)
             }
         })
     }
@@ -105,8 +103,8 @@ impl PdfParser<'_> {
 /// leading free-object-zero pattern encountered in some PDFs.
 fn normalize_xref_subsection_entries(
     start_object_number: usize,
-    entries: Vec<CrossReferenceEntry>,
-) -> Vec<(usize, CrossReferenceEntry)> {
+    entries: Vec<CrossReferenceEntryType>,
+) -> Vec<(usize, CrossReferenceEntryType)> {
     let has_leading_free_object_zero = start_object_number > 0
         && entries
             .first()
@@ -132,14 +130,12 @@ fn normalize_xref_subsection_entries(
 
 /// Returns whether an entry is the object-zero free entry incorrectly included
 /// at the start of a non-zero xref subsection.
-fn is_malformed_leading_free_object_zero(entry: &CrossReferenceEntry) -> bool {
+fn is_malformed_leading_free_object_zero(entry: &CrossReferenceEntryType) -> bool {
     matches!(
         entry,
-        CrossReferenceEntry {
-            entry_type: CrossReferenceEntryType::Free {
-                next_free_object: 0,
-                generation_number: 65_535,
-            }
+        CrossReferenceEntryType::Free {
+            next_free_object: 0,
+            generation_number: 65_535,
         }
     )
 }
@@ -271,8 +267,8 @@ mod tests {
     #[test]
     fn test_normalize_xref_subsection_entries_uses_declared_range() {
         let entries = vec![
-            CrossReferenceEntry::new_normal(17, 0),
-            CrossReferenceEntry::new_normal(81, 0),
+            CrossReferenceEntryType::new_normal(17, 0),
+            CrossReferenceEntryType::new_normal(81, 0),
         ];
 
         let normalized = normalize_xref_subsection_entries(4, entries);
@@ -284,9 +280,9 @@ mod tests {
     #[test]
     fn test_normalize_xref_subsection_entries_handles_leading_free_object_zero() {
         let entries = vec![
-            CrossReferenceEntry::new_free(0, 65_535),
-            CrossReferenceEntry::new_normal(17, 0),
-            CrossReferenceEntry::new_normal(81, 0),
+            CrossReferenceEntryType::new_free(0, 65_535),
+            CrossReferenceEntryType::new_normal(17, 0),
+            CrossReferenceEntryType::new_normal(81, 0),
         ];
 
         let normalized = normalize_xref_subsection_entries(4, entries);
diff --git a/crates/pdf-parser/src/xref_builder.rs b/crates/pdf-parser/src/xref_builder.rs
diff --git a/crates/pdf-parser/tests/xref_builder.rs b/crates/pdf-parser/tests/xref_builder.rs

Original file line number	Diff line number	Diff line change
`@@ -216,7 +216,7 @@ impl ObjectCollection {`
`216`	`216`	`(`
`217`	`217`	`k.to_string(),`
`218`	`218`	`json!({`
`219`		`- "entry_type": format!("{:?}", v.entry_type)`
	`219`	`+ "entry_type": format!("{v:?}")`
`220`	`220`	`}),`
`221`	`221`	`)`
`222`	`222`	`})`