Quotes support for CSV parser

Signed-off-by: Alexander Borsuk <me@alex.bio>

base/string_utils.cpp

@@ -483,15 +483,66 @@ bool AlmostEqual(std::string const & str1, std::string const & str2, size_t mism
   return false;
 }
 
-void ParseCSVRow(std::string const & s, char const delimiter, std::vector<std::string> & target)
+namespace
+{
+// Trim, unquote the string, and unescape two double quotes.
+std::string & UnescapeCSVColumn(std::string & s)
+{
+  Trim(s);
+
+  if (s.size() < 2)
+    return s;
+
+  if (*s.begin() == '"' && *s.rbegin() == '"')
+    s = s.substr(1, s.size() - 2);
+
+  for (size_t i = 1; i < s.size(); ++i)
+    if (s[i] == '"' && s[i - 1] == '"')
+      s.erase(i, 1);
+
+  return s;
+}
+}  // namespace
+
+void ParseCSVRow(std::string const & row, char const delimiter, std::vector<std::string> & target)
 {
   target.clear();
-  TokenizeIterator<SimpleDelimiter, std::string::const_iterator, true /* KeepEmptyTokens */> it(s.begin(), s.end(), delimiter);
+
-  for (; it; ++it)
+  std::string prevColumns;
+  for (TokenizeIterator<SimpleDelimiter, std::string::const_iterator, true /* KeepEmptyTokens */> it {row.begin(), row.end(), delimiter}; it; ++it)
   {
-    std::string column(*it);
+    std::string_view column = *it;
-    Trim(column);
+    size_t const quotesCount = std::count(column.begin(), column.end(), '"');
-    target.push_back(std::move(column));
+    bool const evenQuotes = quotesCount % 2 == 0;
+    if (prevColumns.empty())
+    {
+      if (evenQuotes)
+      {
+        if (quotesCount == 0)
+          target.emplace_back(column);
+        else
+        {
+          std::string strColumn {column};
+          target.push_back(UnescapeCSVColumn(strColumn));
+        }
+      }
+      else
+      {
+        prevColumns = column;
+        prevColumns.push_back(',');
+      }
+    }
+    else
+    {
+      prevColumns.append(column);
+      if (evenQuotes)
+        prevColumns.push_back(',');
+      else
+      {
+        target.push_back(UnescapeCSVColumn(prevColumns));
+        prevColumns.clear();
+      }
+    }
   }
 
   // Special case: if the string is empty, return an empty array instead of {""}.

coding/coding_tests/csv_reader_test.cpp

@@ -8,6 +8,8 @@
 #include <string>
 #include <vector>
 
+namespace csv_reader_test
+{
 using platform::tests_support::ScopedFile;
 
 using Row = coding::CSVReader::Row;
@@ -179,3 +181,41 @@ UNIT_TEST(CSVReaderIterator)
     TEST_EQUAL(index, answer.size(), ());
   }
 }
+
+UNIT_TEST(CSVReaderEmptyColumns)
+{
+  auto const kContentWithEmptyColumns = ",,2,,4,\n,,,,,";
+  auto const fileName = "test.csv";
+  ScopedFile sf(fileName, kContentWithEmptyColumns);
+  Rows const answer = {{"", "", "2", "", "4", ""}, {"", "", "", "", "", ""}};
+  coding::CSVReader reader(sf.GetFullPath());
+  size_t index = 0;
+  while (auto const optionalRow = reader.ReadRow())
+  {
+    TEST_EQUAL(*optionalRow, answer[index], ());
+    ++index;
+  }
+  TEST_EQUAL(index, answer.size(), ());
+  TEST(!reader.ReadRow(), ());
+  TEST(!reader.ReadRow(), ());
+}
+
+UNIT_TEST(CSVReaderQuotes)
+{
+  auto const kContentWithQuotes = R"(noquotes, "" , "with space","with, comma","""double"" quotes","""double,"", commas", """""",)";
+  auto const fileName = "test.csv";
+  ScopedFile sf(fileName, kContentWithQuotes);
+  Rows const answer = {{"noquotes", "", "with space", "with, comma", "\"double\" quotes", "\"double,\", commas","\"\"", ""}};
+  coding::CSVReader reader(sf.GetFullPath());
+  size_t index = 0;
+  while (auto const optionalRow = reader.ReadRow())
+  {
+    TEST_EQUAL(*optionalRow, answer[index], ());
+    ++index;
+  }
+  TEST_EQUAL(index, answer.size(), ());
+  TEST(!reader.ReadRow(), ());
+  TEST(!reader.ReadRow(), ());
+}
+
+}  // namespace csv_reader_test