Subversion Repositories navi

Compare Revisions

Ignore whitespace Rev 48 → Rev 49

/tsimapiak/parse.py
0,0 → 1,95
#!/usr/bin/python
# -*- coding: utf-8 -*-
 
import re
from tsimapiak.dbconnector import getnavilist
 
wordlist = getnavilist()
 
infixes0 = [ u"awn", u"eyk", u"us", u"äp" ]
infixes1 = [ u"ìyev", u"iyev", u"arm", u"asy", u"ilv", u"ìmv", u"imv" u"ìrm", u"irv", u"ìsy", u"aly", u"ary", u"ìly", u"ìry", u"ìlm", u"alm", u"am", u"ay", u"er", u"ìm", u"iv", u"ìy", u"ol" ]
infixes2 = [ u"äng", u"ats", u"eiy", u"ei", u"uy" ]
 
# Returns array with Word,Infix 0,Infix 1,Infix 2,Case,Gender,Number suffixes,Inclusive,Indefinite,Vocative (suffix),Plural,Adposition,Adject pre,Adject suff,am/ay/tu/vi/yu,adverbial,nominalise,sä,fne,lenited?
def parsefix(original):
realword = u""
infix0 = u""
infix1 = u""
infix2 = u""
infix01 = u""
infix_1 = u""
infix_2 = u""
for eachword in wordlist:
regex = re.sub(u" ",u"[^ ]* [^ ]*",eachword[2])
regex = re.sub(u"^",u"[^ ]*",regex)
regex = re.sub(u"$",u"[^ ]*",regex)
regex = re.sub(u"<0><1>",u"[^ ]*",regex)
regex = re.sub(u"<2>",u"[^ ]*",regex)
if re.match(regex,original):
realword = eachword[2]
break
if realword == u"":
return [u"[" + original + u"]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""]
else:
if re.search(u"<",realword):
beginning = re.sub(u"<0><1>.*",u"",realword)
middle = re.sub(u".*<0><1>(.*)<2>.*",ur"\1",realword)
end = re.sub(u".*<2>",u"",realword)
infix01 = re.sub(u".*?" + re.sub(u"<0><1>",u"([^ ]*)",re.sub(u"<2>",u"[^ ]*",realword)) + u".*?",ur"\1",original)
infix_2 = re.sub(u".*?" + re.sub(u"<2>",u"([^ ]*)",re.sub(u"<0><1>",u"[^ ]*",realword)) + u".*?",ur"\1",original)
for eachinfix in infixes0:
if infix01.startswith(eachinfix):
infix0 = eachinfix
infix_1 = infix01[len(eachinfix):]
break
else:
infix0 = u""
infix_1 = infix01
gotinfix1 = False
for eachinfix in infixes1:
if infix_1.startswith(eachinfix):
infix1 = eachinfix
infix_1 = infix_1[len(eachinfix):]
if infix_1 != u"":
if re.search(u"<0><1><2>",realword):
infix_2 = infix_1
else:
return [u"[" + original + u"]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""]
gotinfix1 = True
break
if gotinfix1 == False:
if re.search(u"<0><1><2>",realword):
if infix_1 == u"":
infix_2 = infix_1
infix1 = u""
elif infix_1 == u"":
infix1 = u""
else:
return [u"[" + original + u"]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""]
gotinfix2 = False
for eachinfix in infixes2:
if infix_2.startswith(eachinfix):
infix2 = infix_2
gotinfix2 = True
break
if gotinfix2 == False:
if infix_2.startswith(end):
suffixes = infix2[len(end) - 1:] + end
elif infix_2 == u"":
infix2 = u""
else:
return [u"[" + original + u"]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""]
# print u"0" + unicode(infix0) + u" 1" + unicode(infix1) + u" 2" + unicode(infix2)
return [realword,infix0,infix1,infix2,u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""]
else:
return [realword,u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""]
 
print parsefix(u"oel")
print parsefix(u"ngati")
print parsefix(u"kameie")
print parsefix(u"kìyevame")
print parsefix(u"English")
print parsefix(u"keykivame")
print parsefix(u"tìsusiti")
print parsefix(u"tayìng nari")
print parsefix(u"tìtusìng")
/tsimapiak/dbconnector.py
12,30 → 12,8
SELECT *, CHAR_LENGTH(navi) AS NL
FROM `metaWords`
ORDER BY NL DESC"""):
if row["partOfSpeech"] in (u"v.", u"vin.", u"vtr."):
current = unicode(row["ipa"])
current = current.replace(ur"ɛ",ur"e").replace(ur".",ur"").replace(ur"ɾ",ur"r") \
.replace(ur"ɪ",ur"ì").replace(ur"ˈ",ur"").replace(ur"'",ur"x") \
.replace(ur"ŋ",ur"ng").replace(ur"j",ur"y").replace(ur"ʔ",ur"'") \
.replace(ur"æ",ur"ä").replace(ur"ˌ",ur"").replace(ur"\t{ts}",ur"ts") \
.replace(ur"ṛ",ur"rr").replace(ur"ḷ",ur"ll").replace(ur"k̚",ur"k ") \
.replace(ur"p̚",ur"p ").replace(ur"t̚",ur"t ").replace(ur"'̚",ur"' ") \
.replace(u"\\",ur"").replace(ur"(",ur"").replace(ur")",ur"") \
.replace(ur"[",ur"").replace(ur"]",ur"").replace(ur" "," ") \
.strip()
current = re.sub(ur" or.*","",current)
current = re.sub(ur"z(.*)engk(.*)e",ur"z\1enk\2e",current)
current = re.sub(ur"t(.*)ì(m|n)\ ",ur"t\1ìng ",current)
current = current.split(ur"$cdot$")
if len(current) == 3:
current = current[0] + u"<0><1>" + current[1] + u"<2>" + current[2]
elif len(current) == 2:
current = current[0] + u"<0><1><2>" + current[1]
else:
current = u"<0><1><2>" + current[0]
else:
current = unicode(row["navi"])
ret.append([row["id"], row["navi"], current, row["partOfSpeech"]])
infix = makeinfix(row)
ret.append([row["id"], row["navi"], infix, row["partOfSpeech"]])
db.close()
return ret
 
46,7 → 24,47
SELECT *
FROM `metaWords`
WHERE navi = ?""",word):
ret.append([row["id"],row["navi"], row["infix"], row["partOfSpeech"]])
infix = makeinfix(row)
ret.append([row["id"],row["navi"], infix, row["partOfSpeech"]])
db.close()
return ret
 
def makeinfix(row):
if row["partOfSpeech"] in (u"v.", u"vin.", u"vtr."):
current = unicode(row["ipa"])
current = current.replace(ur"ɛ",ur"e").replace(ur".",ur"").replace(ur"ɾ",ur"r") \
.replace(ur"ɪ",ur"ì").replace(ur"ˈ",ur"").replace(ur"'",ur"x") \
.replace(ur"ŋ",ur"ng").replace(ur"j",ur"y").replace(ur"ʔ",ur"'") \
.replace(ur"æ",ur"ä").replace(ur"ˌ",ur"").replace(ur"\t{ts}",ur"ts") \
.replace(ur"ṛ",ur"rr").replace(ur"ḷ",ur"ll").replace(ur"k̚",ur"k ") \
.replace(ur"p̚",ur"p ").replace(ur"t̚",ur"t ").replace(ur"'̚",ur"' ") \
.replace(u"\\",ur"").replace(ur"(",ur"").replace(ur")",ur"") \
.replace(ur"[",ur"").replace(ur"]",ur"").replace(ur" "," ") \
.strip()
current = re.sub(ur" or.*","",current)
current = re.sub(ur"z(.*)engk(.*)e",ur"z\1enk\2e",current)
current = re.sub(ur"t(.*)ì(m|n)\ ",ur"t\1ìng ",current)
current = current.split(ur"$cdot$")
if len(current) == 3:
current = current[0] + u"<0><1>" + current[1] + u"<2>" + current[2]
elif len(current) == 2:
current = current[0] + u"<0><1><2>" + current[1]
else:
current = u"<0><1><2>" + current[0]
else:
current = unicode(row["navi"])
return current
 
#def gettrans(id, cod):
#ret = []
#if cod not in (u"est",u"ptbr",u"de",u"eng",u"all"):
#return ret
#db = tornado.database.Connection("127.0.0.1", "navi", user="navi", password="navi")
#if cod == "all":
#for row in db.query("""
#SELECT *
#FROM `metaWords`
#WHERE id = ?""",idd):
#infix = makeinfix(row)
#ret.append([row["id"],row["navi"], infix, row["partOfSpeech"]])
#db.close()
/dev/naviparse.py
File deleted
/webapp/main.py
9,8 → 9,9
import os
import re
 
from tsimapiak.parsenum import parse
from tsimapiak.dbconnector import getnavilist
from tsimapiak import parsenum
from tsimapiak import dbconnector
from tsimapiak import parse
 
class Index(tornado.web.RequestHandler):
def get(self):
25,7 → 26,7
num = self.get_argument("num").strip()
except:
self.redirect("/number")
numout = parse(num.replace(" ",""))
numout = parsenum.parse(num.replace(" ",""))
if numout == None:
numout = -1
self.render("templates/number.html", last=num, numout=numout)
37,16 → 38,29
 
class TestDB(tornado.web.RequestHandler):
def get(self):
lis = getnavilist()
lis = dbconnector.getnavilist()
text = u"id | navi | infix | partofspeech<br />"
text += u"<br />".join(u" | ".join(unicode(y) for y in x) for x in lis)
self.write(text)
 
class ParseTest(tornado.web.RequestHandler):
def get(self):
self.render("templates/number.html", last="", out=None)
def post(self):
try:
word = self.get_argument("word").strip()
except:
self.redirect("/parse")
out = parse.parsefix(word)
self.render("templates/number.html", last=word, out=out)
 
application = tornado.web.Application([
("/", Index),
("/number", Number),
("/restart", Restart),
("/testdb", TestDB)
("/testdb", TestDB),
("/parse", ParseTest)
])
 
if __name__ == "__main__":
/webapp/templates/parse.html
0,0 → 1,20
{% extends "base.html" %}
 
{% block title %}Number parser{% end %}
 
{% block body %}
<b>Na'vi word:</b><br />
<form action="/parse" method="post">
<input id="word" name="word" type="text" value="{{last}}" style="width: 100%;" />
<input name="btn" type="submit" value="Parse!" />
</form>
{% if type(out) == list %}
out[0]<br />
out[1]<br />
out[2]<br />
out[3]
{% end %}
<script type="text/javascript">
document.getElementById("word").focus();
</script>
{% end %}