GenericQtClientCore/formats/csvparser.cpp

#include "csvparser.h"

#include <QHash>
#include <QLocale>
#include <QVariant>

#include "../../3rdParty/rapidcsv/src/rapidcsv.h"
#include "../model/metadata.h"

using namespace rapidcsv;

QList<ModelItemValues> CsvParser::getItemsFromCSVFile(const QString& fileName) {
  Document doc(fileName.toStdString());

  const bool isCompatible = isCsvCompatible(doc);
  if (isCompatible) {
    const QList<ModelItemValues> result = createListItemsFromCsvEntries(doc);
    return result;
  } else {
    return QList<ModelItemValues>();
  }
}

bool CsvParser::exportToCSVFile(const QList<QStringList>& rows, const QString& filePath) {
  Document doc(std::string(), LabelParams(0, -1));
  const QList<QString> headerNames = GET_HEADER_NAMES();
  for (int column = 0; column < headerNames.size(); ++column) {
    doc.SetColumnName(column, headerNames.at(column).toStdString());
  }
  for (int row = 0; row < rows.size(); ++row) {
    QStringList rowValues = rows.at(row);
    std::vector<std::string> rowValueStrings;
    for (int column = 0; column < rowValues.size(); ++column) {
      rowValueStrings.push_back(rowValues.at(column).toStdString());
    }
    doc.InsertRow(row, rowValueStrings);
  }
  doc.Save(filePath.toStdString());
  return true;
}

CsvParser::CsvParser() {}

/** A CSV file is compatible if the following is true:
 *  - there is a CSV column for every column in the table model
 *  (except there are optional columns defined in the model)
 * @brief CsvParser::isCsvCompatible
 * @param doc
 * @return
 */
bool CsvParser::isCsvCompatible(const rapidcsv::Document& doc) {
  qInfo() << "Checking CSV document for compatiblity...";
  const std::vector<std::string> columnNames = doc.GetColumnNames();
  for (const QString& headerName : GET_HEADER_NAMES()) {
    bool isHeaderNameFound = false;
    if (std::find(columnNames.begin(), columnNames.end(), headerName) != columnNames.end()) {
      qDebug() << QString("Header found in column names: %1").arg(headerName);
    } else {
      const QString errorString =
          QString("Couldn't find header name '%1' in CSV file. Aborting...").arg(headerName);
      qWarning() << errorString;
      return false;
    }
  }
  return true;
}

QList<ModelItemValues> CsvParser::createListItemsFromCsvEntries(const rapidcsv::Document& doc) {
  QList<ModelItemValues> result;
  const int rowCount               = doc.GetRowCount();
  const QList<QString> headerNames = GET_HEADER_NAMES();

  /// get the values for all columns
  QHash<QString, std::vector<std::string>> columnValueMap = extractColumnValues(headerNames, doc);

  /// get item values for each row
  for (int row = 0; row < rowCount; ++row) {
    const ModelItemValues itemValues = getItemValuesForRow(headerNames, columnValueMap, row);
    result.append(itemValues);
  }
  return result;
}

QHash<QString, std::vector<std::string>> CsvParser::extractColumnValues(
    const QList<QString> headerNames,
    const rapidcsv::Document& doc) {
  QHash<QString, std::vector<std::string>> columnValueMap;
  for (const QString& columnName : headerNames) {
    // NEXT add support for optional columns
    // if (optionalCsvHeaderNames.contains(columnName)) {
    //   const std::vector<std::string> columnNames = doc.GetColumnNames();
    //   int columnIdx                              = doc.GetColumnIdx(columnName.toStdString());
    //   if (columnIdx == -1) {
    //     continue;
    //   }
    // }
    const std::vector<std::string> columnValues =
        doc.GetColumn<std::string>(columnName.toStdString());
    columnValueMap.insert(columnName, columnValues);
  }

  return columnValueMap;
}

ModelItemValues CsvParser::getItemValuesForRow(
    const QList<QString>& headerNames,
    const QHash<QString, std::vector<std::string>>& columnValueMap,
    const int row) {
  ModelItemValues result;
  for (const QString& columnName : headerNames) {
    if (!columnValueMap.contains(columnName)) {
      continue;
    }
    int role                = ROLE_NAMES.key(columnName.toLatin1());
    std::string valueString = columnValueMap.value(columnName).at(row);

    QVariant value = parseItemValue(role, valueString);
    if (value.isValid()) {
      result[role] = value;
    }
  }
  return result;
}

QVariant CsvParser::parseItemValue(const int role, const std::string& valueString) {
  QVariant result;
  if (STRING_ROLES.contains(role)) {
    /// string values
    result = QString::fromStdString(valueString);
  } else if (INT_ROLES.contains(role)) {
    /// int values

    /// GetColumn<int> crashed (probably because of the empty values)
    /// so the strings will be processed later
    const QString intAsString = QString::fromStdString(valueString);
    result                    = intAsString.toInt();
  } else if (DOUBLE_ROLES.contains(role)) {
    /// double values
    const QString doubleAsString = QString::fromStdString(valueString);
    double doubleValue;
    if (doubleAsString.contains(',')) {
      QLocale german(QLocale::German);
      doubleValue = german.toDouble(doubleAsString);
    } else {
      doubleValue = doubleAsString.toDouble();
    }
    result = doubleValue;

    // } else if (typeColumns.contains(columnName)) {
    //   // NEXT validate string is allowed
    //   values[role] = QString::fromStdString(columnValueMap.value(columnName).at(row));
  } else {
    /// no type recognized for column
    QString errorString = QString("Couldn't find value type for role '%1'").arg(role);
    qCritical() << errorString;
  }
  return result;
}