private FinanceReport ParseReport(string code, string[] lines)
{
FinanceReport report = new FinanceReport();
int lineIndex = 0;
string companyName;
if (!ParseHeader(code, lines, ref lineIndex, out companyName))
{
_errorWriter.WriteLine("Failed to parse report header");
return(null);
}
report.CompanyCode = code;
report.CompanyName = companyName;
while (lineIndex < lines.Length)
{
int currentLineIndex = lineIndex;
FinanceReportTable table = ParseTable(lines, ref lineIndex);
if (table != null && table.Name != "环比分析")
{
report.AddTable(table);
}
else
{
report.Annotations = string.Join(
Environment.NewLine,
Enumerable
.Range(currentLineIndex, lines.Length - currentLineIndex)
.Select(i => lines[i]));
break;
}
}
return(report);
}
private FinanceReportTable ParseTable(string[] lines, ref int lineIndex)
{
const string UnknownTableName = "<unknown table name>";
string tableName;
// parse table name
int startLineIndex = lineIndex;
if (!ParseTableName(lines, ref lineIndex, out tableName))
{
_errorWriter.WriteLine("failed to find table name between line {0}~{1}", startLineIndex, lineIndex);
tableName = UnknownTableName;
}
if (tableName != UnknownTableName)
{
// get normalized table name
tableName = _dataDictionary.GetNormalizedTableName(tableName);
}
// parse table definition, include row definiton and column definition
string rowDefinition;
string[] columnDefinitions;
startLineIndex = lineIndex;
if (!ParseTableDefinition(lines, ref lineIndex, out rowDefinition, out columnDefinitions))
{
_errorWriter.WriteLine("failed to parse table definition from line {0}", startLineIndex);
return(null);
}
// get normalized column definitions
columnDefinitions = columnDefinitions.Select(s => _dataDictionary.GetNormalizedColumnName(tableName, s)).ToArray();
FinanceReportTable table = new FinanceReportTable(tableName, rowDefinition, columnDefinitions);
// find out all possible rows
startLineIndex = lineIndex;
while (lineIndex < lines.Length)
{
string currentLine = lines[lineIndex];
if (currentLine.StartsWith(TableLastRowStartPattern)) // end of table
{
lineIndex++;
break;
}
if (currentLine.StartsWith(EffectiveRowStartPattern) || currentLine.StartsWith(TableSectionSeparatorStartPattern))
{
lineIndex++;
continue;
}
// not a valid line in table
break;
}
int endLineIndex = lineIndex;
if (startLineIndex == endLineIndex)
{
// empty table
return(table);
}
// get all cells
string[][] cells =
Enumerable
.Range(startLineIndex, endLineIndex - startLineIndex)
.Select(i => lines[i])
.SkipWhile(s => !s.StartsWith(EffectiveRowStartPattern))
.Select(s => s.Split(new string[] { RowCellSeparator }, StringSplitOptions.RemoveEmptyEntries).Select(x => x.Trim()).ToArray())
.ToArray();
// clean up cells
cells = CleanUpCells(cells, table.ColumnCount);
FinanceReportColumnDefinition[] tableColumnDefinitions = table.ColumnDefinitions.ToArray();
// create rows and adjust cell values according to unit.
bool isHbfxTable = (table.Name == "环比分析");
// last chance of getting table name according to row names.
if (table.Name == UnknownTableName)
{
var tableNamesList = cells
.Select(rc => isHbfxTable ? table.RowDefinition + rc[0] : rc[0])
.Select(rowName => _dataDictionary.GetPossibleNormalizedTableNameByRowNameAlias(GetCleanedRowName(rowName)))
.ToList();
IEnumerable <string> tableNames = new List <string>();
if (tableNamesList.Count == 1)
{
tableNames = tableNamesList[0];
}
else if (tableNamesList.Count > 1)
{
tableNames = tableNamesList[0];
for (int i = 1; i < tableNamesList.Count; ++i)
{
tableNames = tableNames.Intersect(tableNamesList[i]);
}
}
if (tableNames.Count() == 0)
{
_errorWriter.WriteLine("failed to guess table name from row names, no table contains all row names");
return(null);
}
else if (tableNames.Count() > 1)
{
_errorWriter.WriteLine("failed to guess table name from row names, more than one tables contain all row names");
return(null);
}
// now we can set the table name to the unique possibility
table.ResetTableName(tableNames.First());
_errorWriter.WriteLine("find table name {0} from row names", table.Name);
}
foreach (var rowCells in cells)
{
string rowName = isHbfxTable ? table.RowDefinition + rowCells[0] : rowCells[0];
// get normalized row name
rowName = _dataDictionary.GetNormalizedRowName(table.Name, rowName);
int rowIndex = table.AddRow(rowName);
//if (rowCells[0] == "筹资活动产生的现金流出小计")
//{
// Console.WriteLine("..........");
// Console.ReadKey();
//}
FinanceReportRow row = table[rowIndex];
for (int i = 0; i < row.Length; ++i)
{
row[i].Parse(rowCells[i + 1], tableColumnDefinitions[i].HasUnit ? tableColumnDefinitions[i].Unit : row.Unit);
}
}
return(table);
}
private static void CreateRevenueTableForLast12Months(IEnumerable <FinanceReport> reports)
{
foreach (var report in reports)
{
// assume the tables in report has been expanded and merged.
var tables = report.Tables.Where(t => t.Name == "利润表");
var financeReportTables = tables as FinanceReportTable[] ?? tables.ToArray();
if (!financeReportTables.Any())
{
continue;
}
if (financeReportTables.Count() > 1)
{
throw new InvalidOperationException(
string.Format("there are more than one revenue table in the report for company {0}", report.CompanyCode));
}
var table = financeReportTables.First();
var columns = table.ColumnDefinitions.ToArray();
for (var i = 0; i < columns.Length; ++i)
{
columns[i].Tag = i;
}
var dateColumns = columns
.Where(c => c.Type == FinanceReportColumnDefinition.ColumnType.Date)
.OrderByDescending(c => c.Date)
.ToArray();
if (dateColumns.Length == 0)
{
continue;
}
// find first non-empty column (it is important to avoid outliers)
FinanceReportColumnDefinition firstActiveColumn = null;
for (var i = 0; i < dateColumns.Length; ++i)
{
int i1 = i;
if (table.Rows.Any(r => r[dateColumns[i1].Tag].Type == FinanceReportCell.CellType.Decimal) &&
dateColumns[i].Date < DateTime.Now)
{
firstActiveColumn = dateColumns[i];
break;
}
}
if (firstActiveColumn == null)
{
return;
}
var newColumns = new[] { firstActiveColumn };
var newRevenueTable
= new FinanceReportTable(
"跨年度利润表",
table.RowDefinition,
table.Unit,
newColumns);
if (firstActiveColumn.Date.Month == 12) // 年报
{
// just copy rows
foreach (var row in table.Rows)
{
var rowIndex = newRevenueTable.AddRow(row.Name);
newRevenueTable[rowIndex][0].Copy(row[firstActiveColumn.Tag]);
}
report.AddTable(newRevenueTable);
}
else
{
// not yearly report, it could be seasonal report or half year report
// so we need to get last 12 month revenue data by 3 data:
// Last 12 month data = latest data + lastest annual report data - last year corresponding month data.
// for example:
// data(2012/9~2013/9) = data(2013/9)+data(2012/12)-data(2012/9)
var firstColumnIndex = firstActiveColumn.Tag;
var secondColumnIndex = -1;
foreach (FinanceReportColumnDefinition t in dateColumns)
{
if (t.Date.Year == firstActiveColumn.Date.Year - 1 &&
t.Date.Month == 12)
{
secondColumnIndex = t.Tag;
break;
}
}
if (secondColumnIndex < 0)
{
// skip new table, just return;
return;
}
var thirdColumnIndex = -1;
foreach (FinanceReportColumnDefinition t in dateColumns)
{
if (t.Date.Year == firstActiveColumn.Date.Year - 1 &&
t.Date.Month == firstActiveColumn.Date.Month)
{
thirdColumnIndex = t.Tag;
break;
}
}
if (thirdColumnIndex < 0)
{
// skip new table, just return;
return;
}
foreach (var row in table.Rows)
{
var v1 = GetCellDecimalValue(row[firstColumnIndex]);
var v2 = GetCellDecimalValue(row[secondColumnIndex]);
var v3 = GetCellDecimalValue(row[thirdColumnIndex]);
var rowIndex = newRevenueTable.AddRow(row.Name);
newRevenueTable[rowIndex][0].DecimalValue = v1 + v2 - v3;
}
report.AddTable(newRevenueTable);
}
}
}
