diff options
| author | ivanmorozov <ivanmorozov@yandex-team.com> | 2022-12-23 17:22:10 +0300 |
|---|---|---|
| committer | ivanmorozov <ivanmorozov@yandex-team.com> | 2022-12-23 17:22:10 +0300 |
| commit | f6fd83a9bc9c2d6bb9502d3dd77adb4f5c31ceae (patch) | |
| tree | cca463108bcf236b61344602b47f474ce42b1ff4 /library/cpp/string_utils/csv/csv.cpp | |
| parent | af6d4963aeec3543ef895865a0a24fa2e039efa3 (diff) | |
| download | ydb-f6fd83a9bc9c2d6bb9502d3dd77adb4f5c31ceae.tar.gz | |
parse line on deploy by option
Diffstat (limited to 'library/cpp/string_utils/csv/csv.cpp')
| -rw-r--r-- | library/cpp/string_utils/csv/csv.cpp | 82 |
1 files changed, 82 insertions, 0 deletions
diff --git a/library/cpp/string_utils/csv/csv.cpp b/library/cpp/string_utils/csv/csv.cpp new file mode 100644 index 0000000000..218473c62c --- /dev/null +++ b/library/cpp/string_utils/csv/csv.cpp @@ -0,0 +1,82 @@ +#include "csv.h" + +TStringBuf NCsvFormat::CsvSplitter::Consume() { + if (Begin == End) { + return nullptr; + } + TString::iterator TokenStart = Begin; + TString::iterator TokenEnd = Begin; + if (Quote == '\0') { + while (1) { + if (TokenEnd == End || *TokenEnd == Delimeter) { + Begin = TokenEnd; + return TStringBuf(TokenStart, TokenEnd); + } + ++TokenEnd; + } + } else { + bool Escape = false; + if (*Begin == Quote) { + Escape = true; + ++TokenStart; + ++TokenEnd; + Y_ENSURE(TokenStart != End, TStringBuf("RFC4180 violation: quotation mark must be followed by something")); + } + while (1) { + if (TokenEnd == End || (!Escape && *TokenEnd == Delimeter)) { + Begin = TokenEnd; + return TStringBuf(TokenStart, TokenEnd); + } else if (*TokenEnd == Quote) { + Y_ENSURE(Escape, TStringBuf("RFC4180 violation: quotation mark must be in the escaped string only")); + if (TokenEnd + 1 == End) { + Begin = TokenEnd + 1; + } else if (*(TokenEnd + 1) == Delimeter) { + Begin = TokenEnd + 1; + } else if (*(TokenEnd + 1) == Quote) { + CustomStringBufs.push_back(TStringBuf(TokenStart, (TokenEnd + 1))); + TokenEnd += 2; + TokenStart = TokenEnd; + continue; + } else { + Y_ENSURE(false, TStringBuf("RFC4180 violation: in escaped string quotation mark must be followed by a delimiter, EOL or another quotation mark")); + } + if (CustomStringBufs.size()) { + CustomString.clear(); + for (auto CustomStringBuf : CustomStringBufs) { + CustomString += TString{ CustomStringBuf }; + } + CustomString += TString{ TStringBuf(TokenStart, TokenEnd) }; + CustomStringBufs.clear(); + return TStringBuf(CustomString); + } else { + return TStringBuf(TokenStart, TokenEnd); + } + } + ++TokenEnd; + } + } +}; + +TString NCsvFormat::TLinesSplitter::ConsumeLine() { + bool Escape = false; + TString result; + TString line; + while (Input.ReadLine(line)) { + for (auto it = line.begin(); it != line.end(); ++it) { + if (*it == Quote) { + Escape = !Escape; + } + } + if (!result) { + result = line; + } else { + result += line; + } + if (!Escape) { + break; + } else { + result += "\n"; + } + } + return result; +}; |