From cf6a6c8a730afa2724c0c257ccae923f48b0c581 Mon Sep 17 00:00:00 2001
From: Ross Gibb <rosgibb@cisco.com>
Date: Sun, 16 Jul 2017 14:13:14 -0600
Subject: [PATCH] Adding escape-cell function to escape Excel unicode sequences
 if the user desires

---
 src/dk/ative/docjure/spreadsheet.clj       | 27 ++++++++++++++++++++++
 test/dk/ative/docjure/spreadsheet_test.clj | 20 ++++++++++++++--
 2 files changed, 45 insertions(+), 2 deletions(-)

diff --git a/src/dk/ative/docjure/spreadsheet.clj b/src/dk/ative/docjure/spreadsheet.clj
index 6235ebb..cb1e36d 100644
--- a/src/dk/ative/docjure/spreadsheet.clj
+++ b/src/dk/ative/docjure/spreadsheet.clj
@@ -252,6 +252,33 @@
       (if (= (.getCellType cell) Cell/CELL_TYPE_FORMULA) (.setCellType cell Cell/CELL_TYPE_BLANK))
       (.setCellValue cell null))))
 
+(defn escape-cell
+  "When inputting a cell's value in an xlsx workbook, Excel allows 16-bit
+  unicode characters to be input in an escaped form matching this regular
+  expression: _x[0-9A-F]{4}_.
+
+  For example, the character 'A' can be input as '_x0041_', the
+  lowercase greek character pi can be input as '_x03C0_'.
+
+  If your data contains cell values that match this form that you do not
+  want to be interpreted as 16-bit unicode values, the string '_x005F'
+  needs to be prepended to each occurance. For example, to stop
+  '_x03C0_' from being interpreted as the character pi you must input
+  '_x005F_x03C0_'.
+
+  This function will escape all occurances of the Excel 16-bit unicode
+  escape sequence in the specified cell value.
+
+  (escape-cell \"foo _x0041_ bar _x03C0_\")
+  => \"foo _x005F_x0041_ bar _x005F_x03C0_\""
+  [value]
+  (if (string? value)
+    (clojure.string/replace
+     value
+     #"(_x[0-9A-F]{4}_)"
+     "_x005F$1")
+    value))
+
 (defn add-row! [^Sheet sheet values]
   (assert-type sheet Sheet)
   (let [row-num (if (= 0 (.getPhysicalNumberOfRows sheet))
diff --git a/test/dk/ative/docjure/spreadsheet_test.clj b/test/dk/ative/docjure/spreadsheet_test.clj
index bfd5873..bd043e0 100644
--- a/test/dk/ative/docjure/spreadsheet_test.clj
+++ b/test/dk/ative/docjure/spreadsheet_test.clj
@@ -260,8 +260,24 @@
         (is (= 1.0 (.getNumericCellValue a1))))
       (testing "should set double"
         (set-cell! a1 (double 1.2))
-        (is (= 1.2 (.getNumericCellValue a1)))))))
-
+        (is (= 1.2 (.getNumericCellValue a1)))))
+    (testing "set-cell! for unicode"
+      (testing "regular unicode"
+        (set-cell! a1 "foo\u220Fbar")
+        (is (= "foo\u220Fbar"
+               (.getStringCellValue a1))))
+      (testing "unescaped"
+        (set-cell! a1 "foo_x220F_bar")
+        (is (= "foo\u220Fbar"
+               (.getStringCellValue a1))))
+      (testing "escaped"
+        (set-cell! a1 (escape-cell "foo_x220F_bar"))
+        (is (= "foo_x220F_bar"
+               (.getStringCellValue a1))))
+      (testing "multiple escaped"
+        (set-cell! a1 (escape-cell "foo _x0041_ bar _x03C0_"))
+        (is (= "foo _x0041_ bar _x03C0_"
+               (.getStringCellValue a1)))))))
 
 (deftest sheet-seq-test
   (let [sheet-name "Sheet 1"