def findiobjextract(patt, node, root):
whole, head = "", ""
parent = treeops.getparentphrase(node, root)
current = parent
while current.getnext() is not None:
current = current.getnext()
text = treeops.text(current)
if "advp" in text or "qual" in text or "timex" in text or "pp" in text or "vp" in text or text in "obj" or text in "np" or text in "nps" or text in "ap" or text in "aps":
continue
elif "iobj" in text:
whole, head, excase = treeops.getextractphrase(current, root)
break
else:
break
if not head:
current = parent
while current.getprevious() is not None:
current = current.getprevious()
text = treeops.text(current)
if "advp" in text or "pp" in text or "qual" in text or "timex" in text or "vp" in text or text in "obj" or text in "np" or text in "nps" or text in "ap" or text in "aps":
continue
elif "iobj" in text:
whole, head, excase = treeops.getextractphrase(current, root)
break
else:
break
if not head:
return "", ""
pattcase = helpers.getpatterncase(patt)
if pattcase not in excase and excase not in pattcase:
return "", ""
else:
return whole, head
def findsubjextract(patt, node, root):
whole, head = "", ""
parent = treeops.getparentphrase(node, root)
current = parent
while current.getprevious() is not None:
current = current.getprevious()
text = treeops.text(current)
if "advp" in text or "pp" in text or "vp" in text or "qual" in text or "timex" in text or text in "np" or text in "nps" or text in "ap" or text in "aps":
continue
elif "subj" in text:
whole, head, excase = treeops.getextractphrase(current, root)
break
elif "scp" in text: # =Maðurinn= sem *beit* hestinn
newcurrent = current.getprevious()
ntext = treeops.text(current)
if "advp" in ntext or "pp" in ntext or "qual" in ntext or "timex" in ntext or ntext in "np" or ntext in "nps" or ntext in "ap" or ntext in "aps":
continue
elif "subj" in ntext:
whole, head, excase = treeops.getextractphrase(current, root)
break
else:
break
if not head:
current = parent
while current.getnext() is not None:
current = current.getnext()
text = treeops.text(current)
if "advp" in text or "pp" in text or "qual" in text or "timex" in text or "vp" in text or text in "np" or text in "nps" or text in "ap" or text in "aps":
continue
elif "subj" in text:
whole, head, excase = treeops.getextractphrase(current, root)
break
else:
break
if not head: # Looking for AP-COMP acting as subj
current = parent
while current.getprevious() is not None:
current = current.getprevious()
text = treeops.text(current)
if "advp" in text or "pp" in text or "vp" in text or "qual" in text or "timex" in text or text in "np" or text in "nps":
continue
elif text in "ap-comp" or text in "aps-comp":
whole, head, excase = treeops.getextractphrase(current, root)
break
elif "scp" in text: # =Maðurinn= sem *beit* hestinn
newcurrent = current.getprevious()
ntext = treeops.text(current)
if "advp" in ntext or "pp" in ntext or "qual" in ntext or "timex" in ntext or ntext in "np" or ntext in "nps":
continue
elif ntext in "ap-comp" or ntext in "aps-comp":
whole, head, excase = treeops.getextractphrase(
current, root)
break
else:
break
if head:
print "yay, ap-comp acting as np-subj! {}".format(whole)
if not head:
return "", ""
pattcase = helpers.getpatterncase(patt)
if pattcase not in excase and excase not in pattcase:
return "", ""
else:
return whole, head
def getyestertrigger(node, root):
prep = "í"
trigger, triggertype = "", ""
parent = treeops.getparentphrase(node, root)
current = parent
while current.getprevious() is not None:
current = current.getprevious()
text = treeops.text(current)
if "np-qual" in text or text in "np" or text in "nps" or text in "ap" or text in "aps" or "timex" in text:
continue
elif "np" in text:
triggertype = "noun pp|" + prep
trigger = treeops.gettriggernoun(current)
break
elif "vp" in text:
if "vpp-comp" in text:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(current)
break
elif "vpi" in text or "vps" in text or "vpg" in text:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
break
else: # This verb might be the only one, maybe there's something else to check
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
loopcurrent = current
while loopcurrent.getprevious() is not None:
loopcurrent = loopcurrent.getprevious()
ltext = treeops.text(loopcurrent)
if "advp" in ltext or ltext in "np" or ltext in "nps" or ltext in "ap" or ltext in "aps" or "np-qual" in ltext or "timex" in ltext:
continue
elif "vpp-comp" in ltext:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "vpi" in ltext or "vps" in ltext or "vpg" in ltext:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "pp" in ltext:
continue
else:
break #this verb is the only one we have.
break
break
elif "pp" in text:
continue
else:
break #nothing found here
if not trigger:
current = parent
while current.getnext() is not None:
current = current.getnext()
text = treeops.text(current)
if "np-qual" in text or "timex" in text or text in "np" or text in "nps" or text in "ap" or text in "aps" or "np-timex" in text:
continue
elif "np" in text:
triggertype = "noun pp|" + prep
trigger = treeops.gettriggernoun(current)
break
elif "vp" in text:
if "vpp-comp" in text:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(current)
break
elif "vpi" in text or "vps" in text or "vpg" in text:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
break
else:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
loopcurrent = current
while loopcurrent.getnext() is not None:
loopcurrent = loopcurrent.getnext()
ltext = treeops.text(loopcurrent)
if "advp" in ltext or ltext in "np" or ltext in "nps" or ltext in "ap" or ltext in "aps" or "np-qual" in ltext or "timex" in ltext:
continue
elif "vpp-comp" in ltext:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "vpi" in ltext or "vps" in ltext or "vpg" in ltext:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "pp" in ltext:
continue
else:
break
break
elif "pp" in text:
continue
else:
break
if not trigger:
return "", ""
return trigger, triggertype
def findppextract(patt, node, root):
whole, head, case = "", "", ""
parts, prep = [], []
parent = treeops.getparentphrase(node, root)
current = parent
while current.getnext() is not None:
current = current.getnext()
text = treeops.text(current)
if treeops.yester(
current
): # Deals with "í gær" being categorized as ADVP, not PP
pp = "í"
whole = "gær"
head = "gær"
case = "nom"
pattcase = helpers.getpatterncase(patt)
if pattcase not in case and case not in pattcase: #Wrong pp phrase
head, case = "", ""
continue
pattprep = helpers.getpatternprep(patt)
if pp not in pattprep or pattprep not in pp:
pp, head, case = "", "", ""
continue
break
elif "advp" in text:
for word in current.findall(".//WORD"):
mytag = treeops.tag(word)
if "a" in mytag[0]:
prep.append(treeops.text(word))
continue
nowcurrent = current
while nowcurrent.getnext() is not None:
nowcurrent = nowcurrent.getnext()
nowtext = treeops.text(nowcurrent)
nowtext = nowtext.replace("[", "")
if nowtext in "np" or nowtext in "nps" or "timex" in nowtext: #Want to treat as a pp phrase
for word in nowcurrent.findall(".//WORD"):
parts.append(treeops.lemma(word))
if not head:
if treeops.isnoun(word):
head = treeops.lemma(word)
case = treeops.headcase(treeops.tag(word))
continue
else:
break
if not parts or not prep: #Shouldn't be treated as pp phrase
head, case, parts, prep = "", "", [], []
continue
pattcase = helpers.getpatterncase(patt)
if pattcase not in case and case not in pattcase: #Wrong pp phrase
head, case, parts, prep = "", "", [], []
continue
pp = " ".join(prep)
pattprep = helpers.getpatternprep(patt)
if pp not in pattprep or pattprep not in pp:
pp, head, case, parts, prep = "", "", "", [], []
continue
whole = " ".join(parts)
break
elif "qual" in text or text in "np" or text in "nps" or text in "ap" or text in "aps" or "timex" in text:
continue
elif "pp" in text:
for word in current.findall(".//WORD"):
mytag = treeops.tag(word)
if "a" in mytag[0]:
prep.append(treeops.text(word))
continue
else:
parts.append(treeops.lemma(word))
if not head:
if treeops.isnoun(word):
head = treeops.lemma(word)
case = treeops.headcase(treeops.tag(word))
if not parts or not prep: #Weird pp phrase, continue search
head, case, parts, prep = "", "", [], []
continue
pattcase = helpers.getpatterncase(patt)
if pattcase not in case and case not in pattcase: #Wrong pp phrase
head, case, parts, prep = "", "", [], []
continue
pp = " ".join(prep)
pattprep = helpers.getpatternprep(patt)
if pp not in pattprep or pattprep not in pp: #wrong preposition phrase
pp, head, case, parts, prep = "", "", "", [], []
continue
if current.getnext(
) is not None and not head: # If I haven't found the head yet
mynext = current.getnext()
mytext = treeops.text(mynext)
if "np" in mytext:
for word in mynext.findall(".//WORD"):
parts.append(treeops.lemma(word))
if treeops.isnoun(word):
head = treeops.lemma(word)
case = treeops.headcase(treeops.tag(word))
pp = " ".join(prep)
whole = " ".join(parts)
break
else:
break
if not head:
return "", ""
pp = " ".join(prep)
whole = " ".join(parts)
break
else:
current = parent
if current.getprevious() is not None:
current = current.getprevious()
text = treeops.text(current)
if "scp" in text:
if current.getprevious() is not None:
current = current.getprevious()
text = treeops.text(current)
if "pp" in text:
for word in current.findall(".//WORD"):
mytag = treeops.tag(word)
if "a" in mytag[0]:
prep.append(treeops.text(word))
continue
else:
parts.append(treeops.lemma(word))
if not head:
if treeops.isnoun(word):
head = treeops.lemma(word)
case = treeops.headcase(
treeops.tag(word))
if not parts or not prep:
return "", ""
pattcase = helpers.getpatterncase(patt)
if pattcase not in case and case not in pattcase: #Wrong pp phrase
return "", ""
pp = " ".join(prep)
pattprep = helpers.getpatternprep(patt)
if pp not in pattprep or pattprep not in pp:
return "", ""
if current.getnext(
) is not None and not head: # If I haven't found the head yet
mynext = current.getnext()
mytext = treeops.text(mynext)
if "np" in text:
for word in current.findall(".//WORD"):
parts.append(treeops.lemma(word))
if treeops.isnoun(word):
head = treeops.lemma(word)
case = treeops.headcase(
treeops.tag(word))
break
else:
break
whole = " ".join(parts)
break
else:
break
else:
break
else:
break
else:
break
if not head:
current = parent
while current.getprevious() is not None:
current = current.getprevious()
text = treeops.text(current)
text = text.replace("[", "")
if treeops.yester(current):
pp = "í"
whole = "gær"
head = "gær"
case = "nom"
pattcase = helpers.getpatterncase(patt)
if pattcase not in case and case not in pattcase: #Wrong pp phrase
head, case = "", ""
continue
pattprep = helpers.getpatternprep(patt)
if pp not in pattprep or pattprep not in pp:
pp, head, case = "", "", ""
continue
break
elif "advp" in text:
for word in current.findall(".//WORD"):
mytag = treeops.tag(word)
if "a" in mytag[0]:
prep.append(treeops.text(word))
continue
nowcurrent = current
while nowcurrent.getnext() is not None:
nowcurrent = nowcurrent.getnext()
nowtext = treeops.text(nowcurrent)
nowtext = nowtext.replace("[", "")
if nowtext in "np" or nowtext in "nps" or "timex" in nowtext: #Want to treat as a pp phrase
for word in nowcurrent.findall(".//WORD"):
parts.append(treeops.lemma(word))
if not head:
if treeops.isnoun(word):
head = treeops.lemma(word)
case = treeops.headcase(treeops.tag(word))
continue
else:
break
if not parts or not prep: #Shouldn't be treated as pp phrase
head, case, parts, prep = "", "", [], []
continue
pattcase = helpers.getpatterncase(patt)
if pattcase not in case and case not in pattcase: #Wrong pp phrase
head, case, parts, prep = "", "", [], []
continue
pp = " ".join(prep)
pattprep = helpers.getpatternprep(patt)
if pp not in pattprep or pattprep not in pp:
pp, head, case, parts, prep = "", "", "", [], []
continue
whole = " ".join(parts)
break
elif "qual" in text or text in "np" or text in "nps" or text in "ap" or text in "aps" or "timex" in text:
continue
elif "pp" in text:
for word in current.findall(".//WORD"):
mytag = treeops.tag(word)
if "a" in mytag[0]:
prep.append(treeops.text(word))
continue
else:
parts.append(treeops.lemma(word))
if not head:
if treeops.isnoun(word):
head = treeops.lemma(word)
case = treeops.headcase(treeops.tag(word))
if not parts or not prep: #Weird pp phrase, continue search
head, case, prep = "", "", []
continue
pattcase = helpers.getpatterncase(patt)
if pattcase not in case and case not in pattcase: #Wrong pp phrase
head, case, parts, prep = "", "", [], []
continue
pp = " ".join(prep)
pattprep = helpers.getpatternprep(patt)
if pp not in pattprep or pattprep not in pp: #wrong preposition phrase
pp, head, case, parts, prep = "", "", "", [], []
continue
whole = " ".join(parts)
break
if not head:
return "", ""
return whole, head
def getpreptrigger(node, root):
parts = [] #To account for multi word prepositions
trigger, triggertype = "", ""
parent = treeops.getparentphrase(node, root)
for word in parent.findall(".//WORD"):
newword = treeops.text(word)
mytag = treeops.tag(word)
if "a" in mytag[0]: #Found preposition
parts.append(newword)
if not parts: #No preposition found, have unmarked np to find correct trigger for.
if parent.getprevious() is not None:
previousnode = parent.getprevious()
ptext = treeops.text(previousnode)
if "pp" in ptext or "advp" in ptext:
found = False #Checking if I find a noun in the phrase
for aword in previousnode.findall(".//WORD"):
newword = treeops.text(aword)
mytag = treeops.tag(aword)
if "a" in mytag[0]: #Found preposition
parts.append(newword)
elif "n" in mytag[0]: #Only applies to PP phrases
found = True
if found:
return "", ""
elif "iobj" in ptext:
trigger, triggertype = getiobjtrigger(previousnode, root)
return trigger, triggertype
elif "obj" in ptext:
trigger, triggertype = getobjtrigger(previousnode, root)
return trigger, triggertype
elif "subj" in ptext:
trigger, triggertype = getsubjtrigger(previousnode, root)
return trigger, triggertype
elif "ap" in ptext:
if previousnode.getprevious() is not None:
moreprevious = previousnode.getprevious()
mtext = treeops.text(moreprevious)
found = False
if not "np" in mtext:
return "", ""
else:
for every in moreprevious.findall(".//WORD"):
etag = treeops.tag(every)
if "a" in etag[0]:
getpreptrigger(moreprevious, root)
elif "n" in etag[0]:
found = True
else:
continue
if not found: #No noun in NP, can add ap and unmarked np to it
if "subj" in mtext:
trigger, triggertype = getsubjtrigger(
moreprevious, root)
return trigger, triggertype
elif "iobj" in mtext:
trigger, triggertype = getiobjtrigger(
moreprevious, root)
return trigger, triggertype
elif "obj" in mtext:
trigger, triggertype = getobjtrigger(
moreprevious, root)
return trigger, triggertype
else:
return "", ""
else:
return "", ""
else:
return "", ""
else:
return "", ""
if len(parts) == 1:
prep = parts[0]
else:
prep = " ".join(parts)
# Getting the trigger
current = parent
while current.getprevious() is not None:
current = current.getprevious()
text = treeops.text(current)
if "np-qual" in text or "timex" in text or "advp" in text or text in "np" or text in "nps" or text in "ap" or text in "aps":
continue
elif "np" in text:
triggertype = "noun pp|" + prep
trigger = treeops.gettriggernoun(current)
break
elif "vp" in text:
if "vpp-comp" in text:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(current)
break
elif "vpi" in text or "vps" in text or "vpg" in text:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
break
else: # This verb might be the only one, maybe there's something else to check
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
loopcurrent = current
while loopcurrent.getprevious() is not None:
loopcurrent = loopcurrent.getprevious()
ltext = treeops.text(loopcurrent)
if "advp" in ltext or "np-qual" in ltext or "timex" in ltext or ltext in "np" or ltext in "nps" or ltext in "ap" or ltext in "aps":
continue
elif "vpp-comp" in ltext:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "vpi" in ltext or "vps" in ltext or "vpg" in ltext:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "pp" in ltext:
continue
else:
break #this verb is the only one we have.
break
break
elif "pp" in text:
continue
else:
break #nothing found here
if not trigger: # Looking after the PP
current = parent
while current.getnext() is not None:
current = current.getnext()
text = treeops.text(current)
if "np-qual" in text or "nps-qual" in text or "timex" in text or "advp" in text or text in "np" or text in "nps" or text in "ap" or text in "aps": #Change: advp and unmarked NPs added
continue
elif "np" in text:
triggertype = "noun pp|" + prep
trigger = treeops.gettriggernoun(current)
break
elif "vp" in text:
if "vpp-comp" in text:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(current)
break
elif "vpi" in text or "vps" in text or "vpg" in text:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
break
else:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
loopcurrent = current
while loopcurrent.getnext() is not None:
loopcurrent = loopcurrent.getnext()
ltext = treeops.text(loopcurrent)
if "advp" in ltext or ltext in "np" or ltext in "nps" or ltext in "ap" or ltext in "aps" or "np-qual" in ltext or "timex" in ltext:
continue
elif "vpp-comp" in ltext:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "vpi" in ltext or "vps" in ltext or "vpg" in ltext:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "pp" in ltext:
continue
else:
break
break
elif "scp" in text:
newnext = current.getnext()
newtext = treeops.text(newnext)
if "subj" in newtext:
newnext = newnext.getnext()
newtext = treeops.text(newnext)
if "vp" in newtext:
if "vpp-comp" in newtext:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(current)
break
elif "vpi" in newtext or "vps" in newtext or "vpg" in newtext:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
break
else:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(current)
loopcurrent = current
while loopcurrent.getnext() is not None:
loopcurrent = loopcurrent.getnext()
ltext = treeops.text(loopcurrent)
if "advp" in ltext or ltext in "np" or ltext in "nps" or ltext in "ap" or ltext in "aps" or "np-qual" in ltext or "timex" in ltext:
continue
elif "vpp-comp" in ltext:
triggertype = "passive pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "vpi" in ltext or "vps" in ltext or "vpg" in ltext:
triggertype = "active pp|" + prep
trigger = treeops.gettriggerverb(loopcurrent)
break
elif "pp" in ltext:
continue
else:
break
elif "pp" in text:
continue
else:
break
if not trigger:
return "", ""
return trigger, triggertype