def cleanMarkupSingle(raw, cleanerMethod): tree = getTreeFromMarkup(raw) buildAdvancedTree(tree) tc = TreeCleaner(tree, save_reports=True) tc.clean([cleanerMethod]) reports = tc.getReports() return (tree, reports)
def preprocess(root): #advtree.buildAdvancedTree(root) #xmltreecleaner.removeChildlessNodes(root) #xmltreecleaner.fixLists(root) #xmltreecleaner.fixParagraphs(root) #xmltreecleaner.fixBlockElements(root) #print"*** parser raw "*5 #parser.show(sys.stdout, root) #print"*** new TreeCleaner "*5 advtree.buildAdvancedTree(root) tc = TreeCleaner(root) tc.cleanAll()
def renderMW(txt, filesuffix=None): parseTree = uparser.parseString(title='Test', raw=txt) advtree.buildAdvancedTree(parseTree) tc = TreeCleaner(parseTree) tc.cleanAll() tmpdir = tempfile.mkdtemp() rw = RlWriter(test_mode=True) rw.wikiTitle = 'testwiki' rw.tmpdir = tmpdir rw.imgDB = dummyImageDB(basedir=tmpdir) elements = rw.write(parseTree) renderElements(elements, filesuffix, tmpdir)
def cleanMarkup(raw): print "Parsing %r" % (raw, ) tree = getTreeFromMarkup(raw) print "before treecleaner: >>>" showTree(tree) print "<<<" print '=' * 20 buildAdvancedTree(tree) tc = TreeCleaner(tree, save_reports=True) tc.cleanAll(skipMethods=[]) reports = tc.getReports() print "after treecleaner: >>>" showTree(tree) print "<<<" return (tree, reports)
def cleanMarkup(raw): print "Parsing %r" % (raw,) tree = getTreeFromMarkup(raw) print "before treecleaner: >>>" showTree(tree) print "<<<" print '=' * 20 buildAdvancedTree(tree) tc = TreeCleaner(tree, save_reports=True) tc.cleanAll(skipMethods=[]) reports = tc.getReports() print "after treecleaner: >>>" showTree(tree) print "<<<" return (tree, reports)
def buildAdvTree(raw): tree = getTreeFromMarkup(raw) advtree.buildAdvancedTree(tree) tc = TreeCleaner(tree, save_reports=True) tc.cleanAll(skipMethods=[]) return tree
def preprocess(root): advtree.buildAdvancedTree(root) tc = TreeCleaner(root) tc.cleanAll()