Merge pull request #1105 from kiwix/string_slugification

Add String Slugification for Generating File Name

include/tools.h

@@ -282,5 +282,14 @@ FeedCategories readCategoriesFromFeed(const std::string& content);
  * @return full language name.
  */
 std::string getLanguageSelfName(const std::string& lang);
+
+/**
+ * Slugifies the filename by converting any characters reserved by the operating
+ * system to '_'. Note filename is only the file name and not a path.
+ *
+ * @param filename Valid UTF-8 encoded file name string. 
+ * @return slugified string.
+ */
+std::string getSlugifiedFileName(const std::string& filename);
 }
 #endif // KIWIX_TOOLS_H

src/tools/stringTools.cpp

@@ -32,6 +32,7 @@
 
 #include <iostream>
 #include <iomanip>
+#include <regex>
 
 /* tell ICU where to find its dat file (tables) */
 void kiwix::loadICUExternalTables()
@@ -439,3 +440,13 @@ template<>
 std::string kiwix::extractFromString(const std::string& str) {
   return str;
 }
+
+std::string kiwix::getSlugifiedFileName(const std::string& filename)
+{
+#ifdef _WIN32
+  const std::regex reservedCharsReg(R"([<>:"/\\|?*])");
+#else
+  const std::regex reservedCharsReg("/");
+#endif
+  return std::regex_replace(filename, reservedCharsReg, "_"); 
+}

test/stringTools.cpp

@@ -19,6 +19,7 @@
 
 #include "gtest/gtest.h"
 #include "../src/tools/stringTools.h"
+#include "../include/tools.h"
 #include <string>
 #include <vector>
 
@@ -170,4 +171,17 @@ TEST(stringTools, stripSuffix)
   EXPECT_EQ(stripSuffix("abc123", "987"), "abc123");
 }
 
+TEST(stringTools, getSlugifiedFileName)
+{
+  EXPECT_EQ(getSlugifiedFileName("abc123.png"), "abc123.png");
+  EXPECT_EQ(getSlugifiedFileName("/"), "_");
+  EXPECT_EQ(getSlugifiedFileName("abc/123.pdf"), "abc_123.pdf");
+  EXPECT_EQ(getSlugifiedFileName("abc//123.yaml"), "abc__123.yaml");
+  EXPECT_EQ(getSlugifiedFileName("//abc//123//"), "__abc__123__");
+#ifdef _WIN32
+  EXPECT_EQ(getSlugifiedFileName(R"(<>:"/\\|?*)"), "__________");
+  EXPECT_EQ(getSlugifiedFileName(R"(<abc>:"/123\\|?*<.txt>)"), "_abc____123______.txt_");
+#endif
+}
+
 };