From 308b85852e7c99dfe61fe2495dec6f29386c46aa Mon Sep 17 00:00:00 2001
From: Elliot Crosby-McCullough <elliot@dexory.com>
Date: Tue, 21 Apr 2026 09:55:38 +0100
Subject: [PATCH] Add a patch to handle badly formed XLS files

---
 lib/spreadsheet/excel/reader.rb           |   9 +++++++++
 lib/spreadsheet/excel/worksheet.rb        |   2 +-
 test/data/test_row_record_empty_range.xls | Bin 0 -> 5632 bytes
 test/integration.rb                       |  14 ++++++++++++++
 4 files changed, 24 insertions(+), 1 deletion(-)
 create mode 100644 test/data/test_row_record_empty_range.xls

diff --git a/lib/spreadsheet/excel/reader.rb b/lib/spreadsheet/excel/reader.rb
index ac034f1..54a6972 100755
--- a/lib/spreadsheet/excel/reader.rb
+++ b/lib/spreadsheet/excel/reader.rb
@@ -1309,6 +1309,15 @@ def set_row_address worksheet, work, pos, len
           xf = (flags & 0x0fff0000) >> 16
           attrs.store :default_format, @workbook.format(xf)
         end
+        # When a ROW record claims no cells (`first_used` == `first_unused`) but
+        # `set_missing_row_address` already recorded a valid offset from cell
+        # records found earlier in the stream, preserve the original offset.
+        # Some XLS writers emit ROW records after cell data with zeroed column
+        # ranges; without this fix, `read_row` would seek to the wrong position.
+        if first_used == first_unused && (existing = worksheet.row_addresses[index])
+          attrs[:offset] = existing[:offset]
+          attrs[:row_block] = existing[:row_block]
+        end
         # TODO: Row spacing
         worksheet.set_row_address index, attrs
       end
diff --git a/lib/spreadsheet/excel/worksheet.rb b/lib/spreadsheet/excel/worksheet.rb
index 0797edc..5854c9d 100644
--- a/lib/spreadsheet/excel/worksheet.rb
+++ b/lib/spreadsheet/excel/worksheet.rb
@@ -12,7 +12,7 @@ class Worksheet < Spreadsheet::Worksheet
       include Spreadsheet::Excel::Offset
 
       offset :dimensions
-      attr_reader :offset, :ole, :links, :guts, :notes
+      attr_reader :offset, :ole, :links, :guts, :notes, :row_addresses
       def initialize opts = {}
         @row_addresses = nil
         super
diff --git a/test/data/test_row_record_empty_range.xls b/test/data/test_row_record_empty_range.xls
new file mode 100644
index 0000000000000000000000000000000000000000..c5b2b998cb19ba2eb9d2c0306088eceac72e2d9f
GIT binary patch
literal 5632
zcmeHLO-NKx6#m}m_*4FksgQvTLRMxtGyX4h)GSwtKMNy*8l6`S+Kfyu!WL<n%OGl1
zv=Q92vD)-QEo#>yFlrG9f!4BBP|!k~&UfC7sgO(rQt{5rJNNwD``vrbz4yF3^LpO1
zIP;)<2^**l8Tg*aMb;X-A#!d?FM{|J396|wM`STM^bbWKH%AsS%Z;b`ud=z8T<;~g
za6j{faNvVH$1#SUSk&}xA$95)LI%gsD%`*wu@qpZj;dJsu<{+s?`h3v%G=d)O?fMd
zv1lB_e6XjwS+F2TM_rZbD-dqrk+xY-iyx(UDRmMP*SXon>qbXB5*qzG9Tk~*n1!rc
zw;46#q0vQ@$O6vdYr+eB;RHy}y^eir<6yuGd}E&^JO0_SE*tGPU75fn5+`+oPs{e6
z5Cae4z!}rD{L5H?98Z13(GxmrN^vk2Hn#&^=DFc`WXy`hqLP&R%~NLFj1IFxSI9CY
z7x$eHMXiW6F5+M$eA=|U9c~d1_<a5{V89AR!=ZTCd%^Zu<71|`LwfTE>gpRBo0?mE
zwzLiK`D^7H@C6!vVs$350>8rQ6s(r!rpAW)x`5wjOM>NZsP#1}mME~(Ptm&clds1z
zPk0!GDtq*B79j`AcR#;5GSnM9p_WOpka$*iGnPo;GFwRF0y^X+Mq*N{))xDfUtt~W
zQn_4ti5zrEHC4T(N#XF6*14+QD(R<A>1}Y&8hNkC)1UetS4@8l?h|>gd+B1y+`Hfm
zD&<8PmOl8ILXA;l+`dJ65=@nbBqk=e_sZ?Gwl<-yfbju5%5BXMjdQg%muN(=wG7cX
zTU*N%jq$KGw`j6ls<FLCBSIg?=V1m(`iRMx+CF0Pg>P%b<d`nG*p8St1dKk8!UNfp
zNsX9!s~R!$S2bdCZR~c$EEJ8HF2$^rNwm3&@sOZsCreU}i)VqPw+>nM!eoqWtHYfL
zI1z9n;6%WQfD-{H0!{?dhybsHylL_3$O|K{iM%fJO343pR^(yn(Ek~Me#GQs!HYwP
zN*TwvtbF=Z2Dg!l4EItxl={DrbqA&QI7Tp};}}V&`MNstBWksOj$@s3GbQy~VzWBw
T`F^|qwewG}i%o0&QU5;xv7!DI

literal 0
HcmV?d00001

diff --git a/test/integration.rb b/test/integration.rb
index a82d49c..a9ecc08 100644
--- a/test/integration.rb
+++ b/test/integration.rb
@@ -1217,6 +1217,20 @@ def test_missing_row_op
       assert_not_nil sheet[2, 1]
     end
 
+    def test_row_record_with_empty_cell_range
+      # Some XLS writers emit ROW records with `first_used` == `first_unused` (claiming
+      # no cells) even though cell records (e.g. `LABELSST`) exist for that row
+      # earlier in the stream. Previously this caused `read_row` to seek to the wrong
+      # offset, returning an empty row despite valid cell data being present.
+      path = File.join @data, "test_row_record_empty_range.xls"
+      book = Spreadsheet.open path
+      sheet = book.worksheet 0
+      row0 = sheet.row(0).to_a.compact
+      assert_operator row0.length, :>, 0, "Row 0 should not be empty"
+      assert_equal "Name", row0[0]
+      assert_equal ["Name", "Code", "Description", "Reference", "Date", "Quantity"], row0
+    end
+
     def test_changes
       path = File.join @data, "test_changes.xls"
       book = Spreadsheet.open path