Merge pull request #67 from jmcnamara/merged_cells

Read xlsx merged cell elements.

Author: cjw296

tests/merged_cells.xlsx

@@ -39,3 +39,26 @@ def test_merged_cells(self):
         row_lo, row_hi, col_lo, col_hi = sheet3.merged_cells[0]
         self.assertEqual(sheet3.cell(row_lo, col_lo).value, 'MERGED')
         self.assertEqual((row_lo, row_hi, col_lo, col_hi), (3, 7, 2, 5))
+
+    def test_merged_cells_xlsx(self):
+        book = xlrd.open_workbook(from_this_dir('merged_cells.xlsx'))
+
+        sheet1 = book.sheet_by_name('Sheet1')
+        expected = []
+        got = sheet1.merged_cells
+        self.assertEqual(expected, got)
+
+        sheet2 = book.sheet_by_name('Sheet2')
+        expected = [(0, 1, 0, 2)]
+        got = sheet2.merged_cells
+        self.assertEqual(expected, got)
+
+        sheet3 = book.sheet_by_name('Sheet3')
+        expected = [(0, 1, 0, 2), (0, 1, 2, 4), (1, 4, 0, 2), (1, 9, 2, 4)]
+        got = sheet3.merged_cells
+        self.assertEqual(expected, got)
+
+        sheet4 = book.sheet_by_name('Sheet4')
+        expected = [(0, 1, 0, 2), (2, 20, 0, 1), (1, 6, 2, 5)]
+        got = sheet4.merged_cells
+        self.assertEqual(expected, got)

tests/test_cell.py

@@ -511,6 +511,7 @@ def __init__(self, sheet, logfile=DLF, verbosity=0):
         self.rowx = -1 # We may need to count them.
         self.bk = sheet.book
         self.sst = self.bk._sharedstrings
+        self.merged_cells = sheet.merged_cells
         self.warned_no_cell_name = 0
         self.warned_no_row_num = 0
         if ET_has_iterparse:
@@ -528,6 +529,8 @@ def own_process_stream(self, stream, heading=None):
                 elem.clear() # destroy all child elements (cells)
             elif elem.tag == U_SSML12 + "dimension":
                 self.do_dimension(elem)
+            elif elem.tag == U_SSML12 + "mergeCell":
+                self.do_merge_cell(elem)
         self.finish_off()
 
     def do_dimension(self, elem):
@@ -539,6 +542,16 @@ def do_dimension(self, elem):
             self.sheet._dimnrows = rowx + 1
             self.sheet._dimncols = colx + 1
 
+    def do_merge_cell(self, elem):
+        # The ref attribute should be a cell range like "B1:D5".
+        ref = elem.get('ref')
+        if ref:
+            first_cell_ref, last_cell_ref = ref.split(':')
+            first_rowx, first_colx = cell_name_to_rowx_colx(first_cell_ref)
+            last_rowx, last_colx = cell_name_to_rowx_colx(last_cell_ref)
+            self.merged_cells.append((first_rowx, last_rowx + 1,
+                                      first_colx, last_colx + 1))
+
     def do_row(self, row_elem):
 
         def bad_child_tag(child_tag):