/**
* Extracted from "http://scores.nba.com/games/20031029/scoreboard.html" which has a lot of table
* columns with unclosed DIV tags because the closing DIV doesn't have a slash. This caused
* java.lang.StackOverflowError on Windows. Tests the new non-recursive CompositeTagScanner with
* the walk back through the parse stack. See also Bug #750117 StackOverFlow while Node-Iteration
* and others.
*/
public void testInvalidNesting() throws ParserException {
String html =
"<table cellspacing=\"2\" cellpadding=\"0\" border=\"0\" width=\"600\">\n"
+ "<tr>\n"
+ "<td><div class=\"ScoreBoardSec\"> <a target=\"_parent\" class=\"ScoreBoardSec\" href=\"http://www.nba.com/heat/\">Heat</a><div></td>\n"
+ "</tr>\n"
+ "</table>";
createParser(html);
parseAndAssertNodeCount(1);
assertType("table", TableTag.class, node[0]);
TableTag table = (TableTag) node[0];
assertTrue("table should have 3 nodes", 3 == table.getChildCount());
assertType("row", TableRow.class, table.childAt(1));
TableRow row = (TableRow) table.childAt(1);
assertTrue("row should have 3 nodes", 3 == row.getChildCount());
assertType("column", TableColumn.class, row.childAt(1));
TableColumn column = (TableColumn) row.childAt(1);
assertTrue("column should have 1 node", 1 == column.getChildCount());
assertType("element", Div.class, column.childAt(0));
Div div = (Div) column.childAt(0);
assertTrue("div should have 3 nodes", 3 == div.getChildCount());
assertType("link", LinkTag.class, div.childAt(1));
LinkTag link = (LinkTag) div.childAt(1);
assertTrue("link contents", link.getLink().equals("http://www.nba.com/heat/"));
assertType("bogus div", Div.class, div.childAt(2));
assertTrue("bogus div should have no children", 0 == ((Div) div.childAt(2)).getChildCount());
}
private HtmlTable makeMockTable(String tableIdentifier) {
// Create just enough "table" to test if
TableTag tableTag = new TableTag();
TableRow tableRow = new TableRow();
TableColumn tableColumn = new TableColumn();
tableColumn.setChildren(new NodeList(new TextNode(tableIdentifier)));
tableRow.setChildren(new NodeList(tableColumn));
tableTag.setChildren(new NodeList(tableRow));
return new HtmlTable(tableTag);
}
/** Returns the number of columns/cells in the given row, including cell spacing. */
private static int getColumnCount(TableRow row) {
Node[] cells = row.getChildren().extractAllNodesThatMatch(HTML_ROW_FILTER).toNodeArray();
int cols = 0;
for (Node cell : cells) {
Integer colSpan = MathUtils.parseInt(((TagNode) cell).getAttribute("colspan"));
cols += colSpan != null ? colSpan : 1;
}
return cols;
}
/** 从课表处，分课表 */
public List<Courses> parseCourses(String html) {
Parser parser = new Parser();
try {
parser.setInputHTML(html);
parser.setEncoding("utf-8");
} catch (ParserException e) {
e.printStackTrace();
}
NodeFilter filter = new NodeClassFilter(TableTag.class);
NodeList nodeList = null;
try {
nodeList = parser.extractAllNodesThatMatch(filter);
} catch (ParserException e) {
e.printStackTrace();
}
List<Courses> list = new ArrayList<Courses>();
String schoolyear = "";
String semester = "";
for (int i = 0; i < nodeList.size(); i++) {
if (nodeList.elementAt(i) instanceof TableTag) {
TableTag tag = (TableTag) nodeList.elementAt(i);
TableRow[] rows = tag.getRows();
for (int j = 0; j < rows.length; j++) {
TableRow row = (TableRow) rows[j];
TableColumn[] columns = row.getColumns();
Courses courses = null;
boolean isCourse = false;
for (int k = 0; k < columns.length; k++) {
Node columnNode = columns[k];
String info = columnNode.toPlainTextString().trim();
String temp = "学年学期：";
int start = info.indexOf(temp);
int len = "2012-2013".length();
if (start != -1) {
start = start + temp.length();
schoolyear = info.substring(start, start + len);
// semester = info.substring(start+len+2);
// 网络正常时候测试学期改为数字
semester = info.substring(start + len + 3, start + len + 4);
if ("一".equals(semester)) {
semester = "1";
} else if ("二".equals(semester)) {
semester = "2";
}
}
if (k == 1 && info.indexOf("[") != -1) {
courses = new Courses();
String courseCode = info.substring(1, 9);
String coursesname = info.substring(10);
courses.setCourseCode(courseCode);
courses.setCoursesname(coursesname);
isCourse = true;
}
if (k == 2 && isCourse) {
double credit = Double.parseDouble(info);
courses.setCredit(credit);
}
if (k == 3 && isCourse) {
courses.setType(info);
}
if (k == 4 && isCourse) {
courses.setLeanType(info);
}
if (k == 5 && isCourse) {
courses.setCheckType(info);
}
if (k == 6 && isCourse) {
courses.setGetType(info);
}
if (k == 7 && isCourse) {
// double score=Double.parseDouble(info);
courses.setScore(info);
}
if (k == 8 && isCourse) {
courses.setRemark(info);
}
} // end for k
if (courses != null) {
courses.setSchoolYear(schoolyear);
courses.setSemester(semester);
list.add(courses);
}
} // end for j
}
}
return list;
}
public List<TimeTable> parseTimeTables(String html) {
Parser parser = new Parser();
try {
parser.setInputHTML(html);
parser.setEncoding("utf-8");
} catch (ParserException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
List<TimeTable> list = new ArrayList<TimeTable>();
NodeFilter filter = new NodeClassFilter(TableTag.class);
NodeList nodeList = null;
try {
nodeList = parser.extractAllNodesThatMatch(filter);
} catch (ParserException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
for (int i = 0; i < nodeList.size(); i++) {
if (nodeList.elementAt(i) instanceof TableTag) {
TableTag tag = (TableTag) nodeList.elementAt(i);
if (tag.getText().indexOf("[课程号]") == -1) {
continue;
}
TableRow[] rows = tag.getRows();
for (int j = 1; j < rows.length; j++) {
TableRow row = (TableRow) rows[j];
TableColumn[] columns = row.getColumns();
boolean isCourse = false;
TimeTable timeTable = null;
for (int k = 0; k < columns.length; k++) {
Node columnNode = columns[k];
String info = columnNode.toPlainTextString().trim();
System.out.println(info + "===" + k);
switch (k) {
case 1:
int start = info.indexOf("[");
int end = info.indexOf("]");
timeTable = new TimeTable();
timeTable.setCourseCode(info.substring(start + 1, end));
timeTable.setCourseName(info.substring(end + 1));
break;
case 3:
timeTable.setCredit(Double.parseDouble(info));
break;
case 4:
timeTable.setType(info);
break;
case 5:
int t_start = info.indexOf("]");
timeTable.setTeacher(info.substring(t_start + 1));
break;
case 8:
List<TimeAndAdress> ta_list = praseStr(info);
for (TimeAndAdress ta : ta_list) {
timeTable.setAddress(ta.getAddress());
timeTable.setTime(ta.getTime());
timeTable.setCycle(ta.getCycle());
timeTable.setSingleDouble(ta.getSingleDouble());
timeTable.setWeek(ta.getWeek());
list.add(timeTable.clone());
}
break;
default:
break;
}
}
} // end for j
}
}
return list;
}
/**
* Creates a list of Grids based on the given HTML string. This works only for table-based HTML
* documents.
*
* @param html the HTML string.
* @return a list of Grids.
*/
public static List<Grid> fromHtml(String html) throws Exception {
if (html == null || html.trim().isEmpty()) {
return null;
}
List<Grid> grids = new ArrayList<>();
Parser parser = Parser.createParser(html, "UTF-8");
Node[] tables = parser.extractAllNodesThatMatch(new TagNameFilter("table")).toNodeArray();
for (Node t : tables) {
Grid grid = new ListGrid();
TableTag table = (TableTag) t;
TableRow[] rows = table.getRows();
Integer firstColumnCount = null;
for (TableRow row : rows) {
if (getColumnCount(row) == 0) // Ignore if no cells
{
log.warn("Ignoring row with no columns");
continue;
}
Node[] cells = row.getChildren().extractAllNodesThatMatch(HTML_ROW_FILTER).toNodeArray();
if (firstColumnCount == null) // First row becomes header
{
firstColumnCount = getColumnCount(row);
for (Node c : cells) {
TagNode cell = (TagNode) c;
grid.addHeader(new GridHeader(getValue(cell), false, false));
Integer colSpan = MathUtils.parseInt(cell.getAttribute("colspan"));
if (colSpan != null && colSpan > 1) {
grid.addEmptyHeaders((colSpan - 1));
}
}
} else // Rest becomes rows
{
if (firstColumnCount != getColumnCount(row)) // Ignore
{
log.warn(
"Ignoring row which has "
+ row.getColumnCount()
+ " columns since table has "
+ firstColumnCount
+ " columns");
continue;
}
grid.addRow();
for (Node c : cells) {
// TODO row span
TagNode cell = (TagNode) c;
grid.addValue(getValue(cell));
Integer colSpan = MathUtils.parseInt(cell.getAttribute("colspan"));
if (colSpan != null && colSpan > 1) {
grid.addEmptyValues((colSpan - 1));
}
}
}
}
grids.add(grid);
}
return grids;
}
public List<TimeTable> parseTimeTables(String html) {
Parser parser = new Parser();
try {
parser.setInputHTML(html);
parser.setEncoding("utf-8");
} catch (ParserException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
List<TimeTable> list = new ArrayList<TimeTable>();
NodeFilter tagfilter = new NodeClassFilter(TableTag.class);
NodeFilter idFilter = new HasAttributeFilter("id", "reportArea");
NodeFilter filter = new AndFilter(tagfilter, idFilter);
NodeList nodeList = null;
try {
nodeList = parser.extractAllNodesThatMatch(filter);
} catch (ParserException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
for (int i = 0; i < nodeList.size(); i++) {
if (nodeList.elementAt(i) instanceof TableTag) {
TableTag tag = (TableTag) nodeList.elementAt(i);
TableRow[] rows = tag.getRows();
for (int j = 0; j < rows.length; j++) {
TableRow row = (TableRow) rows[j];
TableColumn[] columns = row.getColumns();
boolean isCourse = false;
TimeTable timeTable = null;
for (int k = 0; k < columns.length; k++) {
Node columnNode = columns[k];
String info = columnNode.toPlainTextString().trim();
// System.out.println(info+"=="+k);
if (k == 1 && info.indexOf("[") != -1) {
timeTable = new TimeTable();
String courseCode = info.substring(1, 9);
String coursesname = info.substring(10);
timeTable.setCourseName(coursesname);
timeTable.setCourseCode(courseCode);
isCourse = true;
}
if (k == 2 && isCourse) {
double credit = Double.parseDouble(info);
timeTable.setCredit(credit);
}
if (k == 3 && isCourse) {
timeTable.setType(info);
}
if (k == 4 && isCourse) {
timeTable.setTeacher(info);
}
if (k == 5 && isCourse) {
timeTable.setClassId(info);
}
if (k == 6 && isCourse) {
timeTable.setClassNum(info);
}
if (k == 11 && isCourse) {
List<TimeAndAdress> ta_list = praseStr(info);
for (TimeAndAdress ta : ta_list) {
timeTable.setAddress(ta.getAddress());
timeTable.setTime(ta.getTime());
timeTable.setCycle(ta.getCycle());
timeTable.setSingleDouble(ta.getSingleDouble());
timeTable.setWeek(ta.getWeek());
list.add(timeTable.clone());
}
}
} // end for k
} // end for j
}
}
return list;
}