Added splitting at multiple whitespace characters

src/parser/SparqlParser.cpp

@@ -63,7 +63,7 @@ void SparqlParser::parsePrologue(string str, ParsedQuery& query) {
 
 // _____________________________________________________________________________
 void SparqlParser::addPrefix(const string& str, ParsedQuery& query) {
-  auto parts = ad_utility::split(ad_utility::strip(str, ' '), ' ');
+  auto parts = ad_utility::splitWs(ad_utility::strip(str, ' '));
   if (parts.size() != 3) {
     throw ParseException(string("Invalid PREFIX statement: ") + str);
   }
@@ -73,12 +73,12 @@ void SparqlParser::addPrefix(const string& str, ParsedQuery& query) {
   }
   SparqlPrefix p{ad_utility::strip(parts[1], " :\t\n"), uri};
   query._prefixes.emplace_back(p);
-};
+}
 
 // _____________________________________________________________________________
 void SparqlParser::parseSelect(const string& str, ParsedQuery& query) {
   assert(ad_utility::startsWith(str, "SELECT"));
-  auto vars = ad_utility::split(str, ' ');
+  auto vars = ad_utility::splitWs(str);
   size_t i = 1;
   if (vars.size() > i && vars[i] == "DISTINCT") {
     ++i;

src/util/StringUtils.h

@@ -8,6 +8,7 @@
 #include <stdlib.h>
 #include <assert.h>
 #include <iostream>
+#include <cctype>
 #include <clocale>
 #include <cstring>
 #include <cwchar>
@@ -108,6 +109,9 @@ inline string rstrip(const string& text, const char *s);
 //! Splits a string at the separator, kinda like python.
 inline vector<string> split(const string& orig, const char sep);
 
+//! Splits a string at any maximum length sequence of whitespace
+inline vector<string> splitWs(const string &orig);
+
 //! Splits a string a any character inside the seps string.
 inline vector<string> splitAny(const string& orig, const char *seps);
 
@@ -381,6 +385,33 @@ vector<string> split(const string& orig, const char sep) {
   return result;
 }
 
+// _____________________________________________________________________________
+vector<string> splitWs(const string &orig) {
+  vector<string> result;
+  if (orig.size() > 0) {
+    size_t start = 0;
+    size_t pos = 0;
+    while (pos < orig.size()) {
+      if (isspace(orig[pos])) {
+        if (start != pos) {
+          result.emplace_back(orig.substr(start, pos - start));
+        }
+        // skip any whitespace
+        while (pos < orig.size() && isspace(orig[pos])) {
+          pos++;
+        }
+        start = pos;
+      }
+      pos++;
+    }
+    // avoid adding whitespace at the back of the string
+    if (!isspace(orig[orig.size() - 1])) {
+      result.emplace_back(orig.substr(start, pos - start));
+    }
+  }
+  return result;
+}
+
 // _____________________________________________________________________________
 vector<string> splitAny(const string& orig, const char *seps) {
   return splitAny(orig, string(seps));

test/StringUtilsTest.cpp

@@ -211,6 +211,45 @@ TEST(StringUtilsTest, strip) {
   ASSERT_EQ(u8"äö", strip(u8"xxaxaxaxaäöaaaxxx", "xa"));
   ASSERT_EQ("xxaxaxaxa♥", rstrip("xxaxaxaxa♥aaaxxx", "xa"));
 }
+
+
+TEST(StringUtilsTest, splitWs) {
+  string s1 = "  this\nis\t  \nit  ";
+  string s2 = "\n   \t  \n \t";
+  string s3 = "thisisit";
+  string s4 = "this is\nit";
+  string s5 = "a";
+  auto v1 = splitWs(s1);
+  ASSERT_EQ(size_t(3), v1.size());
+  ASSERT_EQ("this", v1[0]);
+  ASSERT_EQ("is", v1[1]);
+  ASSERT_EQ("it", v1[2]);
+
+  auto v2 = splitWs(s2);
+  ASSERT_EQ(size_t(0), v2.size());
+
+
+  auto v3 = splitWs(s3);
+  ASSERT_EQ(size_t(1), v3.size());
+  ASSERT_EQ("thisisit", v3[0]);
+
+  auto v4 = splitWs(s4);
+  ASSERT_EQ(size_t(3), v4.size());
+  ASSERT_EQ("this", v4[0]);
+  ASSERT_EQ("is", v4[1]);
+  ASSERT_EQ("it", v4[2]);
+
+  auto v5 = splitWs(s5);
+  ASSERT_EQ(size_t(1), v5.size());
+  ASSERT_EQ("a", v5[0]);
+
+  // and with unicode
+  string s6 = u8"Spaß \t ❤ \n漢字  ";
+  auto v6 = splitWs(s6);
+  ASSERT_EQ(u8"Spaß", v6[0]);
+  ASSERT_EQ(u8"❤", v6[1]);
+  ASSERT_EQ(u8"漢字", v6[2]);
+}
 }  // namespace