| 3 | #include "csv.h" |
| 4 | |
| 5 | TStringBuf NCsvFormat::CsvSplitter::Consume() { |
| 6 | if (Begin == End) { |
| 7 | return nullptr; |
| 8 | } |
| 9 | TString::const_iterator TokenStart = Begin; |
| 10 | TString::const_iterator TokenEnd = Begin; |
| 11 | if (Quote == '\0') { |
| 12 | while (1) { |
| 13 | if (TokenEnd == End || *TokenEnd == Delimeter) { |
| 14 | Begin = TokenEnd; |
| 15 | return TStringBuf(TokenStart, TokenEnd); |
| 16 | } |
| 17 | ++TokenEnd; |
| 18 | } |
| 19 | } else { |
| 20 | bool Escape = false; |
| 21 | if (*Begin == Quote) { |
| 22 | Escape = true; |
| 23 | ++TokenStart; |
| 24 | ++TokenEnd; |
| 25 | Y_ENSURE(TokenStart != End, TStringBuf("RFC4180 violation: quotation mark must be followed by something")); |
| 26 | } |
| 27 | while (1) { |
| 28 | if (TokenEnd == End || (!Escape && *TokenEnd == Delimeter)) { |
| 29 | Begin = TokenEnd; |
| 30 | return TStringBuf(TokenStart, TokenEnd); |
| 31 | } else if (*TokenEnd == Quote) { |
| 32 | Y_ENSURE(Escape, TStringBuf("RFC4180 violation: quotation mark must be in the escaped string only")); |
| 33 | if (TokenEnd + 1 == End) { |
| 34 | Begin = TokenEnd + 1; |
| 35 | } else if (*(TokenEnd + 1) == Delimeter) { |
| 36 | Begin = TokenEnd + 1; |
| 37 | } else if (*(TokenEnd + 1) == Quote) { |
| 38 | TempResultParts.push_back(TStringBuf(TokenStart, (TokenEnd + 1))); |
| 39 | TokenEnd += 2; |
| 40 | TokenStart = TokenEnd; |
| 41 | continue; |
| 42 | } else { |
| 43 | Y_ENSURE(false, TStringBuf("RFC4180 violation: in escaped string quotation mark must be followed by a delimiter, EOL or another quotation mark")); |
| 44 | } |
| 45 | if (TempResultParts.size()) { |
| 46 | auto newEscapedStringPtr = std::make_unique<TString>(); |
| 47 | size_t newStringSize = 0; |
| 48 | for (auto tempResultPart : TempResultParts) { |
| 49 | newStringSize += tempResultPart.size(); |
| 50 | } |
| 51 | newStringSize += TokenEnd - TokenStart; |
| 52 | newEscapedStringPtr->reserve(newStringSize); |
| 53 | for (auto tempResultPart : TempResultParts) { |
| 54 | *newEscapedStringPtr += TString{ tempResultPart }; |
| 55 | } |
| 56 | *newEscapedStringPtr += TString{ TStringBuf(TokenStart, TokenEnd) }; |
| 57 | TempResultParts.clear(); |
| 58 | // Storing built string so that returned TStringBuf won't change until this splitter is destroyed |
| 59 | TempResults.push_back(std::move(newEscapedStringPtr)); |
| 60 | return TStringBuf(*TempResults.back()); |
| 61 | } else { |
| 62 | return TStringBuf(TokenStart, TokenEnd); |