import _ from 'lodash';
import flatten from '../../../core/utils/flatten';
import TimeSeries from '../../../core/time_series2';
import TableModel from '../../../core/table_model';

var transformers = {};

transformers['timeseries_to_rows'] = {
  description: 'Time series to rows',
  getColumns: function() {
    return [];
  },
  transform: function(data, panel, model) {
    model.columns = [{ text: 'Time', type: 'date' }, { text: 'Metric' }, { text: 'Value' }];

    for (var i = 0; i < data.length; i++) {
      var series = data[i];
      for (var y = 0; y < series.datapoints.length; y++) {
        var dp = series.datapoints[y];
        model.rows.push([dp[1], series.target, dp[0]]);
      }
    }
  },
};

transformers['timeseries_to_columns'] = {
  description: 'Time series to columns',
  getColumns: function() {
    return [];
  },
  transform: function(data, panel, model) {
    model.columns.push({ text: 'Time', type: 'date' });

    // group by time
    var points = {};

    for (let i = 0; i < data.length; i++) {
      var series = data[i];
      model.columns.push({ text: series.target });

      for (var y = 0; y < series.datapoints.length; y++) {
        var dp = series.datapoints[y];
        var timeKey = dp[1].toString();

        if (!points[timeKey]) {
          points[timeKey] = { time: dp[1] };
          points[timeKey][i] = dp[0];
        } else {
          points[timeKey][i] = dp[0];
        }
      }
    }

    for (var time in points) {
      var point = points[time];
      var values = [point.time];

      for (let i = 0; i < data.length; i++) {
        var value = point[i];
        values.push(value);
      }

      model.rows.push(values);
    }
  },
};

transformers['timeseries_aggregations'] = {
  description: 'Time series aggregations',
  getColumns: function() {
    return [
      { text: 'Avg', value: 'avg' },
      { text: 'Min', value: 'min' },
      { text: 'Max', value: 'max' },
      { text: 'Total', value: 'total' },
      { text: 'Current', value: 'current' },
      { text: 'Count', value: 'count' },
    ];
  },
  transform: function(data, panel, model) {
    var i, y;
    model.columns.push({ text: 'Metric' });

    for (i = 0; i < panel.columns.length; i++) {
      model.columns.push({ text: panel.columns[i].text });
    }

    for (i = 0; i < data.length; i++) {
      var series = new TimeSeries({
        datapoints: data[i].datapoints,
        alias: data[i].target,
      });

      series.getFlotPairs('connected');
      var cells = [series.alias];

      for (y = 0; y < panel.columns.length; y++) {
        cells.push(series.stats[panel.columns[y].value]);
      }

      model.rows.push(cells);
    }
  },
};

transformers['annotations'] = {
  description: 'Annotations',
  getColumns: function() {
    return [];
  },
  transform: function(data, panel, model) {
    model.columns.push({ text: 'Time', type: 'date' });
    model.columns.push({ text: 'Title' });
    model.columns.push({ text: 'Text' });
    model.columns.push({ text: 'Tags' });

    if (!data || !data.annotations || data.annotations.length === 0) {
      return;
    }

    for (var i = 0; i < data.annotations.length; i++) {
      var evt = data.annotations[i];
      model.rows.push([evt.time, evt.title, evt.text, evt.tags]);
    }
  },
};

transformers['table'] = {
  description: 'Table',
  getColumns: function(data) {
    if (!data || data.length === 0) {
      return [];
    }

    // Single query returns data columns as is
    if (data.length === 1) {
      return [...data[0].columns];
    }

    // Track column indexes: name -> index
    const columnNames = {};

    // Union of all columns
    const columns = data.reduce((acc, series) => {
      series.columns.forEach(col => {
        const { text } = col;
        if (columnNames[text] === undefined) {
          columnNames[text] = acc.length;
          acc.push(col);
        }
      });
      return acc;
    }, []);

    return columns;
  },
  transform: function(data, panel, model) {
    if (!data || data.length === 0) {
      return;
    }

    const noTableIndex = _.findIndex(data, d => d.type !== 'table');
    if (noTableIndex > -1) {
      throw {
        message: `Result of query #${String.fromCharCode(
          65 + noTableIndex
        )} is not in table format, try using another transform.`,
      };
    }

    // Single query returns data columns and rows as is
    if (data.length === 1) {
      model.columns = [...data[0].columns];
      model.rows = [...data[0].rows];
      return;
    }

    // Track column indexes of union: name -> index
    const columnNames = {};

    // Union of all non-value columns
    const columnsUnion = data.reduce((acc, series) => {
      series.columns.forEach(col => {
        const { text } = col;
        if (columnNames[text] === undefined) {
          columnNames[text] = acc.length;
          acc.push(col);
        }
      });
      return acc;
    }, []);

    // Map old column index to union index per series, e.g.,
    // given columnNames {A: 0, B: 1} and
    // data [{columns: [{ text: 'A' }]}, {columns: [{ text: 'B' }]}] => [[0], [1]]
    const columnIndexMapper = data.map(series => series.columns.map(col => columnNames[col.text]));

    // Flatten rows of all series and adjust new column indexes
    const flattenedRows = data.reduce((acc, series, seriesIndex) => {
      const mapper = columnIndexMapper[seriesIndex];
      series.rows.forEach(row => {
        const alteredRow = [];
        // Shifting entries according to index mapper
        mapper.forEach((to, from) => {
          alteredRow[to] = row[from];
        });
        acc.push(alteredRow);
      });
      return acc;
    }, []);

    // Returns true if both rows have matching non-empty fields as well as matching
    // indexes where one field is empty and the other is not
    function areRowsMatching(columns, row, otherRow) {
      let foundFieldToMatch = false;
      for (let columnIndex = 0; columnIndex < columns.length; columnIndex++) {
        if (row[columnIndex] !== undefined && otherRow[columnIndex] !== undefined) {
          if (row[columnIndex] !== otherRow[columnIndex]) {
            return false;
          }
        } else if (row[columnIndex] === undefined || otherRow[columnIndex] === undefined) {
          foundFieldToMatch = true;
        }
      }
      return foundFieldToMatch;
    }

    // Merge rows that have same values for columns
    const mergedRows = {};
    const compactedRows = flattenedRows.reduce((acc, row, rowIndex) => {
      if (!mergedRows[rowIndex]) {
        // Look from current row onwards
        let offset = rowIndex + 1;
        // More than one row can be merged into current row
        while (offset < flattenedRows.length) {
          // Find next row that could be merged
          const match = _.findIndex(flattenedRows, otherRow => areRowsMatching(columnsUnion, row, otherRow), offset);
          if (match > -1) {
            const matchedRow = flattenedRows[match];
            // Merge values from match into current row if there is a gap in the current row
            for (let columnIndex = 0; columnIndex < columnsUnion.length; columnIndex++) {
              if (row[columnIndex] === undefined && matchedRow[columnIndex] !== undefined) {
                row[columnIndex] = matchedRow[columnIndex];
              }
            }
            // Dont visit this row again
            mergedRows[match] = matchedRow;
            // Keep looking for more rows to merge
            offset = match + 1;
          } else {
            // No match found, stop looking
            break;
          }
        }
        acc.push(row);
      }
      return acc;
    }, []);

    model.columns = columnsUnion;
    model.rows = compactedRows;
  },
};

transformers['json'] = {
  description: 'JSON Data',
  getColumns: function(data) {
    if (!data || data.length === 0) {
      return [];
    }

    var names: any = {};
    for (var i = 0; i < data.length; i++) {
      var series = data[i];
      if (series.type !== 'docs') {
        continue;
      }

      // only look at 100 docs
      var maxDocs = Math.min(series.datapoints.length, 100);
      for (var y = 0; y < maxDocs; y++) {
        var doc = series.datapoints[y];
        var flattened = flatten(doc, null);
        for (var propName in flattened) {
          names[propName] = true;
        }
      }
    }

    return _.map(names, function(value, key) {
      return { text: key, value: key };
    });
  },
  transform: function(data, panel, model) {
    var i, y, z;

    for (let column of panel.columns) {
      var tableCol: any = { text: column.text };

      // if filterable data then set columns to filterable
      if (data.length > 0 && data[0].filterable) {
        tableCol.filterable = true;
      }

      model.columns.push(tableCol);
    }

    if (model.columns.length === 0) {
      model.columns.push({ text: 'JSON' });
    }

    for (i = 0; i < data.length; i++) {
      var series = data[i];

      for (y = 0; y < series.datapoints.length; y++) {
        var dp = series.datapoints[y];
        var values = [];

        if (_.isObject(dp) && panel.columns.length > 0) {
          var flattened = flatten(dp, null);
          for (z = 0; z < panel.columns.length; z++) {
            values.push(flattened[panel.columns[z].value]);
          }
        } else {
          values.push(JSON.stringify(dp));
        }

        model.rows.push(values);
      }
    }
  },
};

function transformDataToTable(data, panel) {
  var model = new TableModel();

  if (!data || data.length === 0) {
    return model;
  }

  var transformer = transformers[panel.transform];
  if (!transformer) {
    throw { message: 'Transformer ' + panel.transform + ' not found' };
  }

  transformer.transform(data, panel, model);
  return model;
}

export { transformers, transformDataToTable };