From d568d52e1710f7c5a5359cc395dc15f1d3762503 Mon Sep 17 00:00:00 2001
From: Paul Harrington <phrrngtn@panix.com>
Date: Sat, 18 Jul 2026 15:36:17 -0400
Subject: [PATCH] Allow newlines and tabs as whitespace (#8)

---
 include/xlfparser.h | 17 +++++++++++++----
 tests/tests.cpp     | 43 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+), 4 deletions(-)

diff --git a/include/xlfparser.h b/include/xlfparser.h
index ef841fe..216e895 100644
--- a/include/xlfparser.h
+++ b/include/xlfparser.h
@@ -81,6 +81,16 @@ namespace xlfparser {
         return std::strlen(str);
     }
 
+    /* True for any character Excel treats as insignificant whitespace between tokens. */
+    template <typename char_type>
+    inline bool _is_whitespace(char_type c)
+    {
+        return c == XLFP_CHAR(' ') ||
+               c == XLFP_CHAR('\t') ||
+               c == XLFP_CHAR('\n') ||
+               c == XLFP_CHAR('\r');
+    }
+
     /* thrown by tokenize for any invalid formula */
     class invalid_formula: public std::runtime_error
     {
@@ -404,7 +414,6 @@ namespace xlfparser {
         const char_type QUOTE_SINGLE  = XLFP_CHAR('\'');
         const char_type PAREN_OPEN    = XLFP_CHAR('(');
         const char_type PAREN_CLOSE   = XLFP_CHAR(')');
-        const char_type WHITESPACE    = XLFP_CHAR(' ');
         const char_type ERROR_START   = XLFP_CHAR('#');
 
         // Some chars can be changed in the options
@@ -652,8 +661,8 @@ namespace xlfparser {
                 continue;
             }
 
-            // trim white-space
-            if (formula[index] == WHITESPACE)
+            // trim white-space (spaces, tabs and newlines)
+            if (_is_whitespace(formula[index]))
             {
                 if (index > start)
                 {
@@ -661,7 +670,7 @@ namespace xlfparser {
                     start = index;
                 }
 
-                while ((formula[index] == WHITESPACE) && (index < size))
+                while ((index < size) && _is_whitespace(formula[index]))
                     index++;
 
                 tokens.push_back(Token(start, index-1, Token::Type::Whitespace, Token::Subtype::None));
diff --git a/tests/tests.cpp b/tests/tests.cpp
index 9d37b94..db3a76f 100644
--- a/tests/tests.cpp
+++ b/tests/tests.cpp
@@ -236,6 +236,49 @@ TEST_CASE("Implicit intersection parsed correctly", "[xlfparser]")
 }
 
 
+TEST_CASE("Newlines and tabs are treated as whitespace", "[xlfparser]")
+{
+    // Multi-line formulas (e.g. entered with Alt+Enter) should tokenize the
+    // same as their single-line equivalents. See issue #8.
+    std::string formula("=SUM(\n\t1,\n\t2\n)");
+    auto result = tokenize(formula);
+
+    REQUIRE(result.size() == 5);
+
+    CHECK_THAT(result[0].value(formula), Equals("SUM"));
+    CHECK(result[0].type() == Token::Type::Function);
+    CHECK(result[0].subtype() == Token::Subtype::Start);
+
+    CHECK_THAT(result[1].value(formula), Equals("1"));
+    CHECK(result[1].type() == Token::Type::Operand);
+
+    CHECK_THAT(result[2].value(formula), Equals(","));
+    CHECK(result[2].type() == Token::Type::Argument);
+
+    CHECK_THAT(result[3].value(formula), Equals("2"));
+    CHECK(result[3].type() == Token::Type::Operand);
+
+    CHECK(result[4].type() == Token::Type::Function);
+    CHECK(result[4].subtype() == Token::Subtype::Stop);
+}
+
+
+TEST_CASE("Newline between operands is an intersection operator", "[xlfparser]")
+{
+    // As with a space, a newline separating two ranges is an implicit
+    // intersection operator.
+    std::string formula("=A1:A10\nB1:B10");
+    auto result = tokenize(formula);
+
+    REQUIRE(result.size() == 3);
+
+    CHECK(result[0].type() == Token::Type::Operand);
+    CHECK(result[1].type() == Token::Type::OperatorInfix);
+    CHECK(result[1].subtype() == Token::Subtype::Intersection);
+    CHECK(result[2].type() == Token::Type::Operand);
+}
+
+
 TEST_CASE("Arrays are parsed correctly", "[xlfparser]")
 {
     std::string formula("={1;2}");