def refineML_TopLevelMatch(ml):
if PRINT_DEBUG:
for m in ml:
print "Phrase match\t%s\t-> %s\t\t(%s)\t-> (%s)" % (
m[0].tag(), m[1].tag(), nodeText(m[0]), nodeText(
m[1])), topLevelPhrasesMatch(m)
return [m for m in ml if topLevelPhrasesMatch(m)]
def linkIdenticalWords(n1, n2, matchList):
"""
Matches nodes in the two trees where the words match, even if the node labels do not.
Using wordsMatch which includes stemming.
@return Pair of matched nodes
"""
assert not n1.isLeaf(), "Original sentence node is a single leaf"
assert not n2.isLeaf(), "Highlight sentence node is a single leaf"
ch1List = [(ch, nodeText(ch)) for ch in n1]
ch2List = [(ch, nodeText(ch)) for ch in n2]
if False and PRINT_DEBUG:
print "ch1"
for ch in ch1List:
print ch[0].tag(), ch[0], ch[1]
print "ch2"
for ch in ch2List:
print ch[0], ch[1]
# do loop explicitly to catch recursion errors and carry on
matching = []
for (ch1, txt1) in ch1List:
for (ch2, txt2) in ch2List:
try:
if (ch1, ch2) not in matchList and wordsMatch(
txt1, txt2, ch1.tag(), ch2.tag()):
matching.append((ch1, ch2))
except RuntimeError:
# recursion problem in matching trees
pass
# if len(matching)>0:
# print "\n\n\nMatching: ", matching
# raise SystemExit, "Need to add POS tags"
return matching
def linkParentNodes(matchList, matchListNoStopList, maxDepth=1):
"""
Considers pairs of matched nodes in the match list.
If two nodes have the same parent, then this parent
is matched to the ancestor of the corresponding
highlight nodes.
A maxDepth=1 allows siblings to be matched
"""
matchingParentNodes = []
if PRINT_DEBUG:
print "\nInvestigating node distances"
assert len(matchList) > 0, "No matches to work with"
srcroot = matchList[0][0].root()
hiroot = matchList[0][1].root()
print "Roots: ", id(srcroot), "---", id(hiroot)
for ms, mh in matchList:
print id(ms.root()), "---", id(mh.root())
if not id(ms.root()) == id(srcroot): print "src!!!"
if not id(mh.root()) == id(hiroot):
print "hi!!!"
print "Original tree:"
print nodeText(hiroot)
print hiroot
print "New tree involving:", mh
print nodeText(mh.root())
print mh.root()
for (s1, h1) in matchList: # was matchListNoStopList
# for (s2, h2) in matchListNoStopList[i+1:]:
for (s2, h2) in matchListNoStopList:
d = nodeDistance(s1, s2)
if PRINT_DEBUG:
print "nodeDistance: ", d, "\t\t", s1.treeposition(
), "---", s2.treeposition(
), "\t; highlights\t", h1.treeposition(
), "---", h2.treeposition(),
if d <= maxDepth:
ca1 = s1.root()[commonAncestor(s1, s2)]
ca2 = h1.root()[commonAncestor(h1, h2)]
if (ca1, ca2) in matchList: continue
if (ca1, ca2) in matchingParentNodes: continue
if PRINT_DEBUG:
print
print "Original: ", nodeText(s1), " --- ", nodeText(
s2), "\tdistance ", d
print "Highlight: ", nodeText(h1), " --- ", nodeText(
h2), nodeDistance(h1, h2)
print "Ancestors in match list? ", (ca1, ca2) in matchList
print "Original parent phrase: ", nodeText(ca1)
print "Highlight parent phrase:", nodeText(ca2)
matchingParentNodes.append((ca1, ca2))
return matchingParentNodes
def _refineML_RemoveMissedProperNounsTest(m):
"""
Return True if the match conveys any NP present in the source tree
"""
if m[0].isLeaf() or m[1].isLeaf(): return True
# looking for S where NP isn't included
# if m[0].tag()=="S" and m[1].tag():
srcTags = [ch.tag() for ch in m[0]]
tgtTags = [ch.tag() for ch in m[1]]
if "NP" not in srcTags and "NP" in tgtTags:
# TODO: check that NP actually contains proper nouns as children
if PRINT_DEBUG:
# print "Missing NP"
print nodeText(m[0])
print nodeText(m[1])
#raise SystemExit
return False
return True
def printAllMatchListInfo(matchList):
print "\nMatch list:"
for (ph1, ph2) in matchList:
print ph1.treeposition(), "---", ph2.treeposition(), "\t\t", ph1.tag(
), nodeText(ph1), " --- ", ph2.tag(), nodeText(ph2)
print