diff -r aec6c1503dbf -r 012d491e219a src/relpipe-in-csv.cpp --- a/src/relpipe-in-csv.cpp Sat Jun 06 01:50:42 2020 +0200 +++ b/src/relpipe-in-csv.cpp Wed Sep 23 11:23:12 2020 +0200 @@ -29,111 +29,12 @@ #include +#include "CSVCommand.h" + using namespace std; using namespace relpipe::cli; using namespace relpipe::writer; - -bool readValue(istream& input, stringstream& currentValue, bool& lastInRecord) { - lastInRecord = false; - char ch; - input.get(ch); - if (ch == '"') { - while (input.get(ch)) { - if (ch == '"') { - input.get(ch); - if (ch == '"') { - currentValue << ch; - } else { - if (ch == '\r') input.get(ch); - if (ch == '\n') lastInRecord = true; - else if (ch != ',') throw RelpipeWriterException(L"Unexpected character (should be „\\n“ or „,“)"); - return true; - } - } else { - currentValue << ch; - } - } - } else if (ch == ',') { - return true; - } else if (ch == '\n') { - lastInRecord = true; - return true; - } else if (ch == '\r') { - input.get(ch); - if (ch == '\n') { - lastInRecord = true; - return true; - } else { - throw RelpipeWriterException(L"Crazy carriage stuck during journey"); - } - } else { - for (currentValue << ch; input.get(ch);) { - switch (ch) { - case ',': return true; - case '\r': break; - case '\n': - lastInRecord = true; - return true; - default: currentValue << ch; - } - } - } - return false; -} - -void processDataStream(ostream &output, istream& input, const vector& args) { - wstring_convert < codecvt_utf8> convertor; // UTF-8 is required for CSV - std::shared_ptr writer(Factory::create(output)); - vector metadata; - bool headerDone = false; - bool lastInRecord = false; - stringstream currentValue; - - - while (readValue(input, currentValue, lastInRecord) && input.good()) { - if (headerDone) { - writer->writeAttribute(convertor.from_bytes(currentValue.str())); - } else { - AttributeMetadata am; - am.attributeName = convertor.from_bytes(currentValue.str()); - am.typeId = TypeId::STRING; - metadata.push_back(am); - if (lastInRecord) { - - /* - * Usage (simple syntax): - * relpipe-in-csv → default relation name, attribute names on the first line, all types are string - * relpipe-in-csv my_relation → custom relation name - * relpipe-in-csv my_relation a b c → custom relation name, custom attribute names (a,b,c), first line contains data - * relpipe-in-csv my_relation a integer b string c boolean → custom relation name, custom attribute names (a,b,c), custom types (integer,string,boolean), first line contains data - */ - - vector firstLine; - if (args.size() == (1 + metadata.size())) { - for (int i = 0; i < metadata.size(); i++) { - firstLine.push_back(metadata[i].attributeName); - metadata[i].attributeName = args[1 + i]; - } - } else if (args.size() == (1 + 2 * metadata.size())) { - for (int i = 0; i < metadata.size(); i++) { - firstLine.push_back(metadata[i].attributeName); - metadata[i].attributeName = args[1 + i * 2]; - metadata[i].typeId = writer->toTypeId(args[1 + i * 2 + 1]); - } - } - - headerDone = true; - writer->startRelation(args.size() > 0 ? args[0] : L"csv", metadata, true); - if (firstLine.size()) { - for (string_t value : firstLine) writer->writeAttribute(value); - } - } - } - - currentValue.str(""); - currentValue.clear(); - } -} +using namespace relpipe::in::csv; int main(int argc, char** argv) { setlocale(LC_ALL, ""); @@ -143,7 +44,8 @@ int resultCode = CLI::EXIT_CODE_UNEXPECTED_ERROR; try { - processDataStream(cout, cin, cli.arguments()); + CSVCommand command; + command.processDataStream(cout, cin, cli.arguments()); resultCode = CLI::EXIT_CODE_SUCCESS; } catch (RelpipeWriterException e) { fwprintf(stderr, L"Caught Writer exception: %ls\n", e.getMessge().c_str());