pyexcel · PLPeeters · Jun 26, 2025 · Jun 26, 2025 · Jun 26, 2025 · Jun 26, 2025
diff --git a/CONTRIBUTORS.rst b/CONTRIBUTORS.rst
@@ -6,3 +6,4 @@
 In alphabetical order:
 
 * `Mark Skelton <https://github.com/mtskelton>`_
+* `Pierre-Louis Peeters <https://github.com/PLPeeters>`_
diff --git a/changelog.yml b/changelog.yml
@@ -2,7 +2,12 @@ name: pyexcel-xlsxr
 organisation: pyexcel
 releases:
 - changes:
-  - action: Updated
+- action: Updated
+    details:
+    - 'Fix freeze when parsing certain corrupt XLSX files'
+  date: 26.06.2025
+  version: 0.6.2
+- action: Updated
     details:
     - '#9: Potential fix for incorrect reading of data with empty cells when used with pyexcel '
   date: 11.11.2024

diff --git a/pyexcel_xlsxr/__init__.py b/pyexcel_xlsxr/__init__.py
@@ -1,10 +1,11 @@
 """
-    pyexcel_xlsxr
-    ~~~~~~~~~~~~~~~~~~~
-    The lower level xlsx file format handler using lxml
-    :copyright: (c) 2015-2020 by Onni Software Ltd & its contributors
-    :license: New BSD License
+pyexcel_xlsxr
+~~~~~~~~~~~~~~~~~~~
+The lower level xlsx file format handler using lxml
+:copyright: (c) 2015-2020 by Onni Software Ltd & its contributors
+:license: New BSD License
 """
+
 from pyexcel_io.io import get_data as read_data
 from pyexcel_io.io import isstream
 from pyexcel_io.plugins import IOPluginInfoChainV2

diff --git a/pyexcel_xlsxr/_version.py b/pyexcel_xlsxr/_version.py
@@ -1,2 +1,2 @@
-__version__ = '0.6.1'
-__author__ = 'C.W.'
+__version__ = "0.6.1"
+__author__ = "C.W."
diff --git a/pyexcel_xlsxr/messy_xlsx.py b/pyexcel_xlsxr/messy_xlsx.py
@@ -9,17 +9,12 @@
 STYLE_FILENAME = "xl/styles.xml"
 SHARED_STRING = "xl/sharedStrings.xml"
 WORK_BOOK = "xl/workbook.xml"
-SHEET_MATCHER = "xl/worksheets/(work)?sheet([0-9]+)?.xml"
-SHEET_INDEX_MATCHER = "xl/worksheets/(work)?sheet(([0-9]+)?).xml"
-XLSX_ROW_MATCH = re.compile(rb".*?(<row.*?<\/.*?row>).*?", re.MULTILINE)
-NUMBER_FMT_MATCHER = re.compile(
-    rb".*?(<numFmts.*?<\/.*?numFmts>).*?", re.MULTILINE
-)
-XFS_FMT_MATCHER = re.compile(
-    rb".*?(<cellXfs.*?<\/.*?cellXfs>).*?", re.MULTILINE
-)
-SHEET_FMT_MATCHER = re.compile(rb".*?(<sheet .*?\/>).*?", re.MULTILINE)
-DATE_1904_MATCHER = re.compile(rb".*?(<workbookPr.*?\/>).*?", re.MULTILINE)
+SHEET_MATCHER = re.compile(r"xl/worksheets/(?:work)?sheet([0-9]+)?.xml")
+XLSX_ROW_MATCH = re.compile(rb"<row\b[^>]*>.*?</row>", re.DOTALL)
+NUMBER_FMT_MATCHER = re.compile(rb"<numFmts\b[^>]*>.*?</numFmts>", re.DOTALL)
+XFS_FMT_MATCHER = re.compile(rb"<cellXfs\b[^>]*>.*?</cellXfs>", re.DOTALL)
+SHEET_FMT_MATCHER = re.compile(rb"<sheet\b.*?/>", re.DOTALL)
+DATE_1904_MATCHER = re.compile(rb"<workbookPr\b.*?/>", re.DOTALL)
 # "xmlns:x14ac="http://schemas.microsoft.com/office/spreadsheetml/2009/9/ac"
 # But it not used for now
 X14AC_NAMESPACE = b'xmlns:x14ac="http://not.used.com/"'
@@ -158,14 +153,15 @@ def find_sheets(file_list):
     return [
         sheet_file
         for sheet_file in file_list
-        if re.match(SHEET_MATCHER, sheet_file)
+        if SHEET_MATCHER.match(sheet_file)
     ]
 
 
 def get_sheet_index(file_name):
-    if re.match(SHEET_MATCHER, file_name):
-        result = re.search(SHEET_INDEX_MATCHER, file_name)
-        index = int(result.group(3)) if result.group(3) else 1
+    sheet_match = SHEET_MATCHER.match(file_name)
+
+    if sheet_match:
+        index = int(sheet_match.group(1)) if sheet_match.group(1) else 1
         return index - 1
     else:
         raise Exception("Invalid sheet file name")
@@ -338,10 +334,10 @@ def parse_book_properties(book_content):
     )
     namespaces = {"r": ns}
 
-    xlsx_header = u"<wrapper {0}>".format(
+    xlsx_header = "<wrapper {0}>".format(
         " ".join('xmlns:{0}="{1}"'.format(k, v) for k, v in namespaces.items())
     ).encode("utf-8")
-    xlsx_footer = u"</wrapper>".encode("utf-8")
+    xlsx_footer = "</wrapper>".encode("utf-8")
     sheets = SHEET_FMT_MATCHER.findall(book_content)
     for sheet in sheets:
         block = xlsx_header + sheet + xlsx_footer

diff --git a/requirements-dev.txt b/requirements-dev.txt
@@ -0,0 +1,4 @@
+black~=25.1
+coverage~=7.9
+pytest~=8.4
+pyexcel~=0.7.3
diff --git a/test.bat b/test.bat
@@ -1,2 +1,2 @@
 pip freeze
-nosetests --with-coverage --cover-package pyexcel_xlsxr --cover-package tests tests --with-doctest --doctest-extension=.rst README.rst  pyexcel_xlsxr
+coverage run -m --source=pyexcel_xlsxr pytest && coverage report --show-missing
diff --git a/test.sh b/test.sh
@@ -1,3 +1,3 @@
 #/bin/bash
 pip freeze
-nosetests --with-coverage --cover-package pyexcel_xlsxr --cover-package tests tests --with-doctest --doctest-extension=.rst README.rst  pyexcel_xlsxr
+coverage run -m --source=pyexcel_xlsxr pytest && coverage report --show-missing
diff --git a/tests/test_bug_fixes.py b/tests/test_bug_fixes.py
@@ -4,123 +4,118 @@
 from pyexcel_xlsxr import get_data
 from pyexcel_io.reader import EncapsulatedSheetReader
 
-from nose.tools import eq_
-
 
 def test_issue_1():
     test_file = get_fixture("issue_1.xlsx")
     data = get_data(test_file)
     data_array = [list(map(str, row)) for row in data["dataSheet1"]]
-    eq_(
-        data_array,
-        [
-            ["", "D0"],
-            ["Pads", "PADA"],
-            ["Timestamp", "13:26:26.375087"],
-            ["I", "V"],
-            ["0.0", "0.7830809999999999"],
-            ["1.0", "1.11145"],
-            ["2.0", "1.176147"],
-            ["3.0", "1.222229"],
-            ["4.0", "1.25946"],
-            ["5.0", "1.293334"],
-            ["6.0", "1.323852"],
-            ["7.0", "1.351623"],
-            ["8.0", "1.3778679999999999"],
-            ["9.0", "1.402893"],
-            ["10.0", "1.427001"],
-            ["11.0", "1.449279"],
-            ["12.0", "1.471252"],
-            ["13.0", "1.4923089999999999"],
-            ["14.0", "1.512451"],
-            ["15.0", "1.531982"],
-            ["16.0", "1.551513"],
-            ["17.0", "1.5701289999999999"],
-            ["18.0", "1.588134"],
-            ["19.0", "1.606445"],
-            ["20.0", "1.623535"],
-            ["21.0", "1.64093"],
-            ["22.0", "1.657714"],
-            ["23.0", "1.674804"],
-            ["24.0", "1.6906729999999999"],
-            ["25.0", "1.707153"],
-            ["26.0", "1.7233269999999998"],
-            ["27.0", "1.738586"],
-            ["28.0", "1.7544549999999999"],
-            ["29.0", "1.769104"],
-            ["30.0", "1.784667"],
-            ["31.0", "1.799316"],
-            ["32.0", "1.8148799999999998"],
-            ["33.0", "1.8286129999999998"],
-            ["34.0", "1.8432609999999998"],
-            ["35.0", "1.85791"],
-            ["36.0", "1.871948"],
-            ["37.0", "1.885986"],
-            ["38.0", "1.900329"],
-            ["39.0", "1.913452"],
-            ["40.0", "1.92749"],
-            ["41.0", "1.941223"],
-            ["42.0", "1.954345"],
-            ["43.0", "1.967773"],
-            ["44.0", "1.9808949999999999"],
-            ["45.0", "1.9940179999999998"],
-            ["46.0", "2.007446"],
-            ["47.0", "2.019958"],
-            ["48.0", "2.03247"],
-            ["49.0", "2.0455929999999998"],
-            ["50.0", "2.05841"],
-            ["51.0", "2.071228"],
-            ["52.0", "2.083129"],
-            ["53.0", "2.095336"],
-            ["54.0", "2.1072379999999997"],
-            ["55.0", "2.120056"],
-            ["56.0", "2.131652"],
-            ["57.0", "2.143859"],
-            ["58.0", "2.156066"],
-            ["59.0", "2.167663"],
-            ["60.0", "2.1795649999999998"],
-            ["61.0", "2.191162"],
-            ["62.0", "2.2021479999999998"],
-            ["63.0", "2.214355"],
-            ["64.0", "2.225646"],
-            ["65.0", "2.236633"],
-            ["66.0", "2.247009"],
-            ["67.0", "2.258911"],
-            ["68.0", "2.269897"],
-            ["69.0", "2.2808829999999998"],
-            ["70.0", "2.2915639999999997"],
-            ["71.0", "2.302246"],
-            ["72.0", "2.3138419999999997"],
-            ["73.0", "2.3245229999999997"],
-            ["74.0", "2.334899"],
-            ["75.0", "2.3455809999999997"],
-            ["76.0", "2.356262"],
-            ["77.0", "2.366333"],
-            ["78.0", "2.376708"],
-            ["79.0", "2.3864739999999998"],
-            ["80.0", "2.3971549999999997"],
-            ["81.0", "2.407531"],
-            ["82.0", "2.417602"],
-            ["83.0", "2.427673"],
-            ["84.0", "2.438354"],
-            ["85.0", "2.4472039999999997"],
-            ["86.0", "2.457885"],
-            ["87.0", "2.467956"],
-            ["88.0", "2.477722"],
-            ["89.0", "2.487487"],
-            ["90.0", "2.4978629999999997"],
-            ["91.0", "2.506408"],
-            ["92.0", "2.515869"],
-            ["93.0", "2.5256339999999997"],
-            ["94.0", "2.535095"],
-            ["95.0", "2.54425"],
-            ["96.0", "2.5537099999999997"],
-            ["97.0", "2.562866"],
-            ["98.0", "2.572021"],
-            ["99.0", "2.5805659999999997"],
-            ["100.0", "2.589721"],
-        ],
-    )
+    assert data_array == [
+        ["", "D0"],
+        ["Pads", "PADA"],
+        ["Timestamp", "13:26:26.375087"],
+        ["I", "V"],
+        ["0.0", "0.7830809999999999"],
+        ["1.0", "1.11145"],
+        ["2.0", "1.176147"],
+        ["3.0", "1.222229"],
+        ["4.0", "1.25946"],
+        ["5.0", "1.293334"],
+        ["6.0", "1.323852"],
+        ["7.0", "1.351623"],
+        ["8.0", "1.3778679999999999"],
+        ["9.0", "1.402893"],
+        ["10.0", "1.427001"],
+        ["11.0", "1.449279"],
+        ["12.0", "1.471252"],
+        ["13.0", "1.4923089999999999"],
+        ["14.0", "1.512451"],
+        ["15.0", "1.531982"],
+        ["16.0", "1.551513"],
+        ["17.0", "1.5701289999999999"],
+        ["18.0", "1.588134"],
+        ["19.0", "1.606445"],
+        ["20.0", "1.623535"],
+        ["21.0", "1.64093"],
+        ["22.0", "1.657714"],
+        ["23.0", "1.674804"],
+        ["24.0", "1.6906729999999999"],
+        ["25.0", "1.707153"],
+        ["26.0", "1.7233269999999998"],
+        ["27.0", "1.738586"],
+        ["28.0", "1.7544549999999999"],
+        ["29.0", "1.769104"],
+        ["30.0", "1.784667"],
+        ["31.0", "1.799316"],
+        ["32.0", "1.8148799999999998"],
+        ["33.0", "1.8286129999999998"],
+        ["34.0", "1.8432609999999998"],
+        ["35.0", "1.85791"],
+        ["36.0", "1.871948"],
+        ["37.0", "1.885986"],
+        ["38.0", "1.900329"],
+        ["39.0", "1.913452"],
+        ["40.0", "1.92749"],
+        ["41.0", "1.941223"],
+        ["42.0", "1.954345"],
+        ["43.0", "1.967773"],
+        ["44.0", "1.9808949999999999"],
+        ["45.0", "1.9940179999999998"],
+        ["46.0", "2.007446"],
+        ["47.0", "2.019958"],
+        ["48.0", "2.03247"],
+        ["49.0", "2.0455929999999998"],
+        ["50.0", "2.05841"],
+        ["51.0", "2.071228"],
+        ["52.0", "2.083129"],
+        ["53.0", "2.095336"],
+        ["54.0", "2.1072379999999997"],
+        ["55.0", "2.120056"],
+        ["56.0", "2.131652"],
+        ["57.0", "2.143859"],
+        ["58.0", "2.156066"],
+        ["59.0", "2.167663"],
+        ["60.0", "2.1795649999999998"],
+        ["61.0", "2.191162"],
+        ["62.0", "2.2021479999999998"],
+        ["63.0", "2.214355"],
+        ["64.0", "2.225646"],
+        ["65.0", "2.236633"],
+        ["66.0", "2.247009"],
+        ["67.0", "2.258911"],
+        ["68.0", "2.269897"],
+        ["69.0", "2.2808829999999998"],
+        ["70.0", "2.2915639999999997"],
+        ["71.0", "2.302246"],
+        ["72.0", "2.3138419999999997"],
+        ["73.0", "2.3245229999999997"],
+        ["74.0", "2.334899"],
+        ["75.0", "2.3455809999999997"],
+        ["76.0", "2.356262"],
+        ["77.0", "2.366333"],
+        ["78.0", "2.376708"],
+        ["79.0", "2.3864739999999998"],
+        ["80.0", "2.3971549999999997"],
+        ["81.0", "2.407531"],
+        ["82.0", "2.417602"],
+        ["83.0", "2.427673"],
+        ["84.0", "2.438354"],
+        ["85.0", "2.4472039999999997"],
+        ["86.0", "2.457885"],
+        ["87.0", "2.467956"],
+        ["88.0", "2.477722"],
+        ["89.0", "2.487487"],
+        ["90.0", "2.4978629999999997"],
+        ["91.0", "2.506408"],
+        ["92.0", "2.515869"],
+        ["93.0", "2.5256339999999997"],
+        ["94.0", "2.535095"],
+        ["95.0", "2.54425"],
+        ["96.0", "2.5537099999999997"],
+        ["97.0", "2.562866"],
+        ["98.0", "2.572021"],
+        ["99.0", "2.5805659999999997"],
+        ["100.0", "2.589721"],
+    ]
 
 
 def test_issue_5():
@@ -131,7 +126,7 @@ def test_issue_5():
 
     sheet = EncapsulatedSheetReader(XLSXSheet(native_sheet))
     data = sheet.to_array()
-    eq_(list(data), [[None, 11, 11]])
+    assert list(data) == [[None, 11, 11]]
 
 
 def get_fixture(file_name):
Original file line number	Diff line number	Diff line change
Expand Up		@@ -6,3 +6,4 @@
		In alphabetical order:

		* `Mark Skelton <https://github.com/mtskelton>`_
		* `Pierre-Louis Peeters <https://github.com/PLPeeters>`_