fix: avoid scanning through all local file headers while opening an a…

…rchive

Fixes #280

src/read.rs

@@ -323,10 +323,7 @@ pub(crate) fn find_content<'a>(
     reader: &'a mut (impl Read + Seek),
 ) -> ZipResult<io::Take<&'a mut dyn Read>> {
     // TODO: use .get_or_try_init() once stabilized to provide a closure returning a Result!
-    let data_start = match data.data_start.get() {
-        Some(data_start) => *data_start,
-        None => find_data_start(data, reader)?,
-    };
+    let data_start = data.data_start(reader)?;
 
     reader.seek(SeekFrom::Start(data_start))?;
     Ok((reader as &mut dyn Read).take(data.compressed_size))
@@ -337,14 +334,14 @@ fn find_content_seek<'a, R: Read + Seek>(
     reader: &'a mut R,
 ) -> ZipResult<SeekableTake<'a, R>> {
     // Parse local header
-    let data_start = find_data_start(data, reader)?;
+    let data_start = data.data_start(reader)?;
     reader.seek(SeekFrom::Start(data_start))?;
 
     // Explicit Ok and ? are needed to convert io::Error to ZipError
     Ok(SeekableTake::new(reader, data.compressed_size)?)
 }
 
-fn find_data_start(
+pub(crate) fn find_data_start(
     data: &ZipFileData,
     reader: &mut (impl Read + Seek + Sized),
 ) -> Result<u64, ZipError> {
@@ -1068,14 +1065,6 @@ pub(crate) fn central_header_to_zip_file<R: Read + Seek>(
         ));
     }
 
-    let data_start = find_data_start(&file, reader)?;
-
-    if data_start > central_directory.directory_start {
-        return Err(InvalidArchive(
-            "File data can't start after the central directory",
-        ));
-    }
-
     reader.seek(SeekFrom::Start(central_header_end))?;
     Ok(file)
 }

src/types.rs

@@ -22,10 +22,12 @@ pub(crate) mod ffi {
 }
 
 use crate::extra_fields::ExtraField;
+use crate::read::find_data_start;
 use crate::result::DateTimeRangeError;
 use crate::spec::is_dir;
 use crate::types::ffi::S_IFDIR;
 use crate::{CompressionMethod, ZIP64_BYTES_THR};
+use std::io::{Read, Seek};
 #[cfg(feature = "time")]
 use time::{error::ComponentRange, Date, Month, OffsetDateTime, PrimitiveDateTime, Time};
 
@@ -488,8 +490,11 @@ pub struct ZipFileData {
 
 impl ZipFileData {
     /// Get the starting offset of the data of the compressed file
-    pub fn data_start(&self) -> u64 {
-        *self.data_start.get().unwrap()
+    pub fn data_start(&self, reader: &mut (impl Read + Seek + Sized)) -> ZipResult<u64> {
+        match self.data_start.get() {
+            Some(data_start) => Ok(*data_start),
+            None => Ok(find_data_start(self, reader)?),
+        }
     }
 
     #[allow(dead_code)]

src/write.rs

@@ -673,7 +673,7 @@ impl<A: Read + Write + Seek> ZipWriter<A> {
         let write_position = self.inner.get_plain().stream_position()?;
         let src_index = self.index_by_name(src_name)?;
         let src_data = &mut self.files[src_index];
-        let src_data_start = src_data.data_start();
+        let src_data_start = src_data.data_start(self.inner.get_plain())?;
         debug_assert!(src_data_start <= write_position);
         let mut compressed_size = src_data.compressed_size;
         if compressed_size > (write_position - src_data_start) {