Add utils::vertical_split()

src/serialization/string_utils.cpp

@@ -26,6 +26,7 @@
 #include "utils/general.hpp"
 #include <cassert>
 #include <array>
+#include <stdexcept>
 
 #include <boost/algorithm/string.hpp>
 
@@ -381,6 +382,35 @@ std::vector<std::string> parenthetical_split(const std::string& val,
 	return res;
 }
 
+std::pair<string_view, string_view> vertical_split(const std::string& val)
+{
+	// Count the number of lines.
+	int num_lines = std::count(val.begin(), val.end(), '\n') + 1;
+
+	if(num_lines < 2) {
+		throw std::logic_error("utils::vertical_split: the string contains only one line");
+	}
+
+	// Split the string at the point where we have encountered
+	// (number of lines / 2 - 1) line separators.
+	int split_point = 0;
+	int num_found_line_separators = 0;
+	for(size_t i = 0; i < val.size(); ++i) {
+		if(val[i] == '\n') {
+			++num_found_line_separators;
+			if(num_found_line_separators >= num_lines / 2 - 1) {
+				split_point = i;
+				break;
+			}
+		}
+	}
+
+	assert(split_point != 0);
+
+	return { string_view(val, 0, split_point),
+		string_view(val, split_point + 1, val.size() - (split_point + 1)) };
+}
+
 // Modify a number by string representing integer difference, or optionally %
 int apply_modifier( const int number, const std::string &amount, const int minimum ) {
 	// wassert( amount.empty() == false );

src/serialization/string_utils.hpp

@@ -23,6 +23,7 @@
 #include <set>
 #include <sstream>
 #include <string>
+#include <utility>
 #include <vector>
 
 class t_string;
@@ -31,6 +32,34 @@ namespace utils {
 
 using string_map = std::map<std::string, t_string>;
 
+// TODO: when https://github.com/aquileia/external has boost::string_view,
+// replace this class with it.
+class string_view
+{
+public:
+	const char* str;
+	const int size;
+
+	string_view(const std::string& str_, int offset, int size_)
+		: str(&str_[offset])
+		, size(size_)
+	{
+	}
+
+	string_view(const std::string& str_)
+		: str(str_.c_str())
+		, size(str_.size())
+	{
+	}
+
+	string_view(const string_view&) = default;
+
+	explicit operator std::string() const
+	{
+		return std::string(str, size);
+	}
+};
+
 bool isnewline(const char c);
 bool portable_isspace(const char c);
 bool notspace(char c);
@@ -145,6 +174,19 @@ std::vector<std::string> square_parenthetical_split(
 	const std::string& right = ")]",
 	const int flags = REMOVE_EMPTY | STRIP_SPACES);
 
+/**
+ * Splits a string into two parts as evenly as possible based on lines.
+ * For example, if the string contains 3288 lines, then both parts will
+ * be 1644 lines long.
+ *
+ * The line separator in between won't be in either of the parts the
+ * function returns.
+ *
+ * Because this function is intended for extremely long strings
+ * (kilobytes long), it returns string_views for performance.
+ */
+std::pair<string_view, string_view> vertical_split(const std::string& val);
+
 /**
  * Generates a new string joining container items in a list.
  *