Zip: explain the meaning of some numeric constants throught the code

crystal-lang · Jan 19, 2017 · 6952e76 · 6952e76
1 parent 4e7c5c7
commit 6952e76
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 5 deletions.
diff --git a/src/zip/file.cr b/src/zip/file.cr
@@ -81,6 +81,8 @@ class Zip::File
     end
   end
 
+  # Try to find the directory end offset (by searching its signature)
+  # in the last 64, 1024 and 65K bytes (in that order)
   private def find_directory_end_offset
     find_directory_end_offset(64) ||
       find_directory_end_offset(1024) ||
@@ -100,6 +102,8 @@ class Zip::File
 
     i = buf_size - 1 - 4
     while i >= 0
+      # These are the bytes the make up the directory end signature,
+      # according to the spec
       break if buf[i] == 0x50 && buf[i + 1] == 0x4b && buf[i + 2] == 0x05 && buf[i + 3] == 0x06
       i -= 1
     end
@@ -172,17 +176,30 @@ class Zip::File
       # Apparently a zip entry might have different extra bytes
       # in the local file header and central directory header,
       # so to know the data offset we must read them again.
+      #
+      # The bytes inside a local file header, from the signature
+      # and up to the extra length field, sum up 30 bytes.
+      #
+      # This 30 and 22 constants are burned inside the zip spec and
+      # will never change.
       @io.read_at(offset.to_i32, 30) do |io|
-        # at least check that the signature is OK
+        # at least check that the signature is OK (these are 4 bytes)
         signature = read(io, UInt32)
         if signature != FileInfo::SIGNATURE
           raise Zip::Error.new("wrong local file header signature (expected 0x#{FileInfo::SIGNATURE.to_s(16)}, got 0x#{signature.to_s(16)})")
         end
 
         # Skip most of the headers except filename length and extra length
+        # (skip 22, so we already read 26 bytes)
         io.skip(22)
+
+        # With these two we read 4 bytes more, so we are at 30 bytes
         filename_length = read(io, UInt16)
         extra_length = read(io, UInt16)
+
+        # The data of this entry comes at the local file header offset
+        # plus 30 bytes (the ones we just skipped) plus skipping the
+        # filename's bytes plus skipping the extra bytes.
         @offset + 30 + filename_length + extra_length
       end
     end

diff --git a/src/zip/file_info.cr b/src/zip/file_info.cr
@@ -95,15 +95,15 @@ module Zip::FileInfo
     write io, @uncompressed_size        # 4
     write io, @filename.bytesize.to_u16 # filename length (2)
     write io, extra.size.to_u16         # extra field length (2)
-    24
+    24                                  # the 24 bytes we just wrote
   end
 
   protected def write_data_descriptor(io : IO)
     io.write FileInfo::DEFLATE_END_SIGNATURE # 4
     write io, @crc32                         # 4
     write io, @compressed_size               # 4
     write io, @uncompressed_size             # 4
-    16
+    16                                       # the 16 bytes we just wrote
   end
 
   protected def decompressor_for(io, is_sized = false)

diff --git a/src/zip/writer.cr b/src/zip/writer.cr
@@ -34,9 +34,12 @@ class Zip::Writer
   # Creates a new writer to the given *io*.
   def initialize(@io : IO, @sync_close = false)
     @entries = [] of Entry
-    @written = 0_u32
     @compressed_size_counter = ChecksumWriter.new
     @uncompressed_size_counter = ChecksumWriter.new(compute_crc32: true)
+
+    # Keep track of how many bytes we write, because we need
+    # that to write the offset of each local file header and some other info
+    @written = 0_u32
   end
 
   # Creates a new writer to the given *filename*.
@@ -188,16 +191,25 @@ class Zip::Writer
       write Zip::CENTRAL_DIRECTORY_HEADER_SIGNATURE # 4
       write Zip::VERSION                            # version made by (2)
       write Zip::VERSION                            # version needed to extract (2)
+      @written += 8                                 # the 8 bytes we just wrote
+
       @written += entry.meta_to_io(@io)
+
       write entry.comment.bytesize.to_u16 # file comment length (2)
       write 0_u16                         # disk number start (2)
       write 0_u16                         # internal file attribute (2)
       write 0_u32                         # external file attribute (4)
       write entry.offset                  # relative offset of local header (4)
+      @written += 14                      # the 14 bytes we just wrote
+
       @io << entry.filename
+      @written += entry.filename.bytesize
+
       @io.write(entry.extra)
+      @written += entry.extra.size
+
       @io << entry.comment
-      @written += 22 + entry.filename.bytesize + entry.extra.size + entry.comment.bytesize
+      @written += entry.comment.bytesize
     end
   end