public void testTagsComplexMethods() {
try {
NodeFilter filterLink = new NodeClassFilter(LinkTag.class);
NodeFilter filterDiv = new NodeClassFilter(Div.class);
OrFilter filterLinkDiv = new OrFilter(filterLink, filterDiv);
NodeFilter filterTable = new NodeClassFilter(TableColumn.class);
OrFilter filter = new OrFilter(filterLinkDiv, filterTable);
String[] tmpSplitTags =
ParserUtils.splitTags(
"OutsideLeft<A>AInside</A><DIV><DIV>DivInside</DIV></DIV><TD>TableColoumnInside</TD>OutsideRight",
filter);
assertStringEquals(
"modified text",
"OutsideLeft*OutsideRight",
new String(tmpSplitTags[0] + '*' + tmpSplitTags[1]));
tmpSplitTags =
ParserUtils.splitTags(
"OutsideLeft<A>AInside</A><DIV><DIV>DivInside</DIV></DIV><TD>TableColoumnInside</TD>OutsideRight",
filter,
false,
false);
assertStringEquals(
"modified text",
"OutsideLeft*AInside*<DIV>DivInside</DIV>*TableColoumnInside*OutsideRight",
new String(
tmpSplitTags[0]
+ '*'
+ tmpSplitTags[1]
+ '*'
+ tmpSplitTags[2]
+ '*'
+ tmpSplitTags[3]
+ '*'
+ tmpSplitTags[4]));
tmpSplitTags =
ParserUtils.splitTags(
"OutsideLeft<A>AInside</A><DIV><DIV>DivInside</DIV></DIV><TD>TableColoumnInside</TD>OutsideRight",
filter,
true,
false);
assertStringEquals(
"modified text",
"OutsideLeft*AInside*DivInside*TableColoumnInside*OutsideRight",
new String(
tmpSplitTags[0]
+ '*'
+ tmpSplitTags[1]
+ '*'
+ tmpSplitTags[2]
+ '*'
+ tmpSplitTags[3]
+ '*'
+ tmpSplitTags[4]));
tmpSplitTags =
ParserUtils.splitTags(
"OutsideLeft<A>AInside</A><DIV><DIV>DivInside</DIV></DIV><TD>TableColoumnInside</TD>OutsideRight",
filter,
false,
true);
assertStringEquals(
"modified text",
"OutsideLeft*OutsideRight",
new String(tmpSplitTags[0] + '*' + tmpSplitTags[1]));
tmpSplitTags =
ParserUtils.splitTags(
"OutsideLeft<A>AInside<DIV><DIV>DivInside</DIV></DIV></A><TD>TableColoumnInside</TD>OutsideRight",
new String[] {"DIV", "TD", "A"});
assertStringEquals(
"modified text",
"OutsideLeft*OutsideRight",
new String(tmpSplitTags[0] + '*' + tmpSplitTags[1]));
assertStringEquals(
"modified text",
"OutsideLeftOutsideRight",
ParserUtils.trimTags(
"OutsideLeft<A>AInside<DIV><DIV>DivInside</DIV></DIV></A><TD>TableColoumnInside</TD>OutsideRight",
new String[] {"DIV", "TD", "A"}));
} catch (Exception e) {
String msg = e.getMessage();
if (null == msg) msg = e.getClass().getName();
fail(msg);
}
}
public void testTagsMethods() {
try {
String[] tmpSplitTags =
ParserUtils.splitTags(
"Begin <DIV><DIV> +12.5 </DIV></DIV> ALL OK", new String[] {"DIV"});
assertStringEquals(
"modified text", "Begin * ALL OK", new String(tmpSplitTags[0] + '*' + tmpSplitTags[1]));
tmpSplitTags =
ParserUtils.splitTags(
"Begin <DIV><DIV> +12.5 </DIV></DIV> ALL OK", new String[] {"DIV"}, false, false);
assertStringEquals(
"modified text",
"Begin *<DIV> +12.5 </DIV>* ALL OK",
new String(tmpSplitTags[0] + '*' + tmpSplitTags[1] + '*' + tmpSplitTags[2]));
tmpSplitTags =
ParserUtils.splitTags(
"Begin <DIV><DIV> +12.5 </DIV></DIV> ALL OK", new String[] {"DIV"}, true, false);
assertStringEquals(
"modified text",
"Begin * +12.5 * ALL OK",
new String(tmpSplitTags[0] + '*' + tmpSplitTags[1] + '*' + tmpSplitTags[2]));
tmpSplitTags =
ParserUtils.splitTags(
"Begin <DIV><DIV> +12.5 </DIV></DIV> ALL OK", new String[] {"DIV"}, false, true);
assertStringEquals(
"modified text", "Begin * ALL OK", new String(tmpSplitTags[0] + '*' + tmpSplitTags[1]));
assertStringEquals(
"modified text",
" ALL OK",
ParserUtils.trimTags("<DIV><DIV> +12.5 </DIV></DIV> ALL OK", new String[] {"DIV"}));
assertStringEquals(
"modified text",
"<DIV> +12.5 </DIV> ALL OK",
ParserUtils.trimTags(
"<DIV><DIV> +12.5 </DIV></DIV> ALL OK", new String[] {"DIV"}, false, false));
assertStringEquals(
"modified text",
" +12.5 ALL OK",
ParserUtils.trimTags(
"<DIV><DIV> +12.5 </DIV></DIV> ALL OK", new String[] {"DIV"}, true, false));
assertStringEquals(
"modified text",
" ALL OK",
ParserUtils.trimTags(
"<DIV><DIV> +12.5 </DIV></DIV> ALL OK", new String[] {"DIV"}, false, true));
// Test trimAllTags method
assertStringEquals(
"modified text",
" +12.5 ALL OK",
ParserUtils.trimAllTags("<DIV><DIV> +12.5 </DIV></DIV> ALL OK", false));
assertStringEquals(
"modified text",
" ALL OK",
ParserUtils.trimAllTags("<DIV><DIV> +12.5 </DIV></DIV> ALL OK", true));
assertStringEquals(
"modified text",
" +12.5 ",
ParserUtils.trimAllTags("<DIV><DIV> +12.5 </DIV></DIV>", false));
assertStringEquals(
"modified text", "", ParserUtils.trimAllTags("<DIV><DIV> +12.5 </DIV></DIV>", true));
assertStringEquals(
"modified text", " YYY ", ParserUtils.trimAllTags("<XXX> YYY <ZZZ>", false));
assertStringEquals("modified text", "YYY", ParserUtils.trimAllTags("YYY", false));
assertStringEquals("modified text", "> OK <", ParserUtils.trimAllTags("> OK <", true));
} catch (Exception e) {
String msg = e.getMessage();
if (null == msg) msg = e.getClass().getName();
fail(msg);
}
}
public void testTagsClassMethods() {
try {
NodeFilter filter = new NodeClassFilter(Div.class);
String[] tmpSplitTags =
ParserUtils.splitTags("Begin <DIV><DIV> +12.5 </DIV></DIV> ALL OK", filter);
assertStringEquals(
"modified text", "Begin * ALL OK", new String(tmpSplitTags[0] + '*' + tmpSplitTags[1]));
tmpSplitTags =
ParserUtils.splitTags(
"Begin <DIV><DIV> +12.5 </DIV></DIV> ALL OK", filter, false, false);
assertStringEquals(
"modified text",
"Begin *<DIV> +12.5 </DIV>* ALL OK",
new String(tmpSplitTags[0] + '*' + tmpSplitTags[1] + '*' + tmpSplitTags[2]));
tmpSplitTags =
ParserUtils.splitTags("Begin <DIV><DIV> +12.5 </DIV></DIV> ALL OK", filter, true, false);
assertStringEquals(
"modified text",
"Begin * +12.5 * ALL OK",
new String(tmpSplitTags[0] + '*' + tmpSplitTags[1] + '*' + tmpSplitTags[2]));
tmpSplitTags =
ParserUtils.splitTags("Begin <DIV><DIV> +12.5 </DIV></DIV> ALL OK", filter, false, true);
assertStringEquals(
"modified text", "Begin * ALL OK", new String(tmpSplitTags[0] + '*' + tmpSplitTags[1]));
assertStringEquals(
"modified text",
" ALL OK",
ParserUtils.trimTags("<DIV><DIV> +12.5 </DIV></DIV> ALL OK", filter));
assertStringEquals(
"modified text",
"<DIV> +12.5 </DIV> ALL OK",
ParserUtils.trimTags("<DIV><DIV> +12.5 </DIV></DIV> ALL OK", filter, false, false));
assertStringEquals(
"modified text",
" +12.5 ALL OK",
ParserUtils.trimTags("<DIV><DIV> +12.5 </DIV></DIV> ALL OK", filter, true, false));
assertStringEquals(
"modified text",
" ALL OK",
ParserUtils.trimTags("<DIV><DIV> +12.5 </DIV></DIV> ALL OK", filter, false, true));
NodeFilter filterTableRow = new NodeClassFilter(TableRow.class);
NodeFilter filterTableColumn = new NodeClassFilter(TableColumn.class);
OrFilter filterOr = new OrFilter(filterTableRow, filterTableColumn);
assertStringEquals(
"modified text",
" ALL OK",
ParserUtils.trimTags("<TR><TD> +12.5 </TD></TR> ALL OK", filterOr));
assertStringEquals(
"modified text",
"<TD> +12.5 </TD> ALL OK",
ParserUtils.trimTags("<TR><TD> +12.5 </TD></TR> ALL OK", filterOr, false, false));
assertStringEquals(
"modified text",
" +12.5 ALL OK",
ParserUtils.trimTags("<TR><TD> +12.5 </TD></TR> ALL OK", filterOr, true, false));
assertStringEquals(
"modified text",
" ALL OK",
ParserUtils.trimTags("<TR><TD> +12.5 </TD></TR> ALL OK", filterOr, false, true));
} catch (Exception e) {
String msg = e.getMessage();
if (null == msg) msg = e.getClass().getName();
fail(msg);
}
}
private static boolean handleURL(String address) {
Main.status(String.format("Processing page \"%s\".", address));
try {
NodeList posts = getPosts(address);
if (posts.toNodeArray().length == 0) {
return false;
}
for (Node post_node : posts.toNodeArray()) {
if (post_node instanceof TagNode) {
TagNode post = (TagNode) post_node;
Post new_post = new Post(Long.parseLong(post.getAttribute("id").substring(5)));
if (!Main.post_post_hash.containsKey(new_post)) {
NodeList photo_posts = getPhotoPosts(post.getChildren());
NodeList remarks = getRemarks(photo_posts);
for (Node node : remarks.toNodeArray()) {
Matcher matcher = lores.matcher(node.getText());
String media_url = "";
if (matcher.find()) {
media_url = matcher.group();
media_url = media_url.substring(17, media_url.length() - 1);
}
String thumb =
media_url.replace(
media_url.substring(media_url.lastIndexOf("_"), media_url.lastIndexOf(".")),
"_75sq");
URL thumb_url = new URL(thumb);
new_post.pictures.add(new Picture(new URL(media_url), thumb_url));
}
NodeList photoset_posts = getPhotosetPosts(post.getChildren());
NodeList iframes = getIFrames(photoset_posts);
for (Node node : iframes.toNodeArray()) {
if (node instanceof TagNode) {
String iframe_url = ((TagNode) node).getAttribute("src");
Parser parser2 = new Parser(iframe_url);
NodeList a_list = parser2.extractAllNodesThatMatch(new TagNameFilter("a"));
Node[] a_array = a_list.toNodeArray();
Node[] img_array =
a_list.extractAllNodesThatMatch(new TagNameFilter("img"), true).toNodeArray();
String media_url;
for (int i = 0; i < a_array.length; i++) {
media_url = ((TagNode) img_array[i]).getAttribute("src");
String thumb =
media_url.replace(
media_url.substring(
media_url.lastIndexOf("_"), media_url.lastIndexOf(".")),
"_75sq");
URL thumb_url = new URL(thumb);
new_post.pictures.add(new Picture(new URL(media_url), thumb_url));
}
}
}
Main.handlePost(new_post);
} else {
new_post = post_post_hash.get(new_post);
handleNonDownloadPost(new_post);
}
}
}
} catch (Exception ex) {
ex.printStackTrace();
Main.status("Error handling post.");
}
return true;
}