From 1dedb46d5eda0b2433ef58ac5d6759c76c5bdb28 Mon Sep 17 00:00:00 2001
From: Nathan Zimmerberg <39104088+nhz2@users.noreply.github.com>
Date: Tue, 8 Oct 2024 19:00:22 -0400
Subject: [PATCH] Port new go test and change EOCD search to match yauzl (#83)

* add new go test file

* Use yauzl eocd search method
---
 src/reader.jl                                 |  36 +++++++-----------
 .../testdata/comment-truncated.zip            | Bin 0 -> 216 bytes
 test/test_ported-go-tests.jl                  |   1 +
 test/test_reader.jl                           |   2 +-
 4 files changed, 16 insertions(+), 23 deletions(-)
 create mode 100644 test/examples from go/testdata/comment-truncated.zip

diff --git a/src/reader.jl b/src/reader.jl
index eefc537..7ac82c2 100644
--- a/src/reader.jl
+++ b/src/reader.jl
@@ -345,40 +345,32 @@ end
 
 # If this fails, io isn't a zip file, io isn't seekable, 
 # or the end of the zip file was corrupted
+# Using yauzl method https://github.com/thejoshwolfe/yauzl/blob/51010ce4e8c7e6345efe195e1b4150518f37b393/index.js#L111-L113
 function find_end_of_central_directory_record(io::IO)::Int64
     seekend(io)
     fsize = position(io)
-    # First assume comment is length zero
     fsize ≥ 22 || throw(ArgumentError("io isn't a zip file. Too small"))
-    seek(io, fsize-22)
-    b = read!(io, zeros(UInt8, 22))
-    check_comment_len_valid(b, comment_len) = (
-        EOCDSig == @view(b[end-21-comment_len:end-18-comment_len]) &&
-        comment_len%UInt8 == b[end-1-comment_len] &&
-        UInt8(comment_len>>8) == b[end-comment_len]
-    )
-    if check_comment_len_valid(b, 0)
-        # No Zip comment fast path
-        fsize-22
-    else
-        # There maybe is a Zip comment slow path
-        fsize > 22 || throw(ArgumentError("io isn't a zip file."))
-        max_comment_len::Int = min(0xFFFF, fsize-22)
-        seek(io, fsize - (max_comment_len+22))
-        b = read!(io, zeros(UInt8, (max_comment_len+22)))
-        comment_len = 1
-        while comment_len < max_comment_len && !check_comment_len_valid(b, comment_len)
-            comment_len += 1
+    for comment_len in 0:Int(min(0xFFFF, fsize-22))
+        seek(io, fsize-22-comment_len)
+        if readle(io, UInt32) != 0x06054b50
+            continue
         end
-        if !check_comment_len_valid(b, comment_len)
+        skip(io, 16)
+        if readle(io, UInt16) == comment_len
+            return fsize-22-comment_len
+        else
             throw(ArgumentError("""
                 io isn't a zip file. 
                 It may be a zip file with a corrupted ending.
                 """
             ))
         end
-        fsize-22-comment_len
     end
+    throw(ArgumentError("""
+        io isn't a zip file. 
+        It may be a zip file with a corrupted ending.
+        """
+    ))
 end
 
 function check_EOCD64_used(io::IO, eocd_offset)::Bool
diff --git a/test/examples from go/testdata/comment-truncated.zip b/test/examples from go/testdata/comment-truncated.zip
new file mode 100644
index 0000000000000000000000000000000000000000..1bc19a85575964f378a8a30f198ed6ba5360aa7d
GIT binary patch
literal 216
zcmWIWW@cf4gUWrGI~jpI5C#dmdHT2ppr}ax{CO=%28Pozb5c_hOA-UT8JU2>aDc83
pE&*nMbOm^`vVk~^KxhP{)xa|7=AgR>tO!m(+=pt;1fVP<0{|c)7RUeq

literal 0
HcmV?d00001

diff --git a/test/test_ported-go-tests.jl b/test/test_ported-go-tests.jl
index e41db7c..edd9507 100644
--- a/test/test_ported-go-tests.jl
+++ b/test/test_ported-go-tests.jl
@@ -30,6 +30,7 @@ end
         "readme.notzip",
         "test-baddirsz.zip",
         "test-badbase.zip",
+        "comment-truncated.zip",
     ]
 
     for filename in same_content_files
diff --git a/test/test_reader.jl b/test/test_reader.jl
index 2e4d678..9e40321 100644
--- a/test/test_reader.jl
+++ b/test/test_reader.jl
@@ -66,7 +66,7 @@ using SHA: sha256
     @test_throws ArgumentError find_eocd(io)
 
     io = IOBuffer("PK\x05\x06"^30000*"ab")
-    @test find_eocd(io) == 100700
+    @test_throws ArgumentError find_eocd(io)
 
     io = IOBuffer("PK\x05\x06"*"\0"^16*"\xff\xff"*"a"^(2^16-1))
     @test find_eocd(io) == 0