Ajout de Orkis proc et utilisation pour tagger

Killian
1 parent 8b1fb953fc
Showing 3 changed files with 43 additions and 14 deletions Side-by-side Diff
processor/Orkis.py
static/js/application.js
webtagger.py
+from BaseProcessor import baseProcessor
+import nltk
+from LiaTools import *
+class Orkis(baseProcessor):
+    """ Processor for Orkis """
+    def __init__(self,dirtyString):
+        self.tagger=Tagger()
+        self.phoner=Phoner()
+        self.dico ={}
+        self.string=dirtyString
+    def isReady(self):
+        self.phoner.isReady()
+        self.tagger.isReady()
+    def clean(self):
+        stopword=StopWord()
+        self.string=stopword.RemoveStopList(self.string)
+    def insertLem(self):
+        self.cleanString=self.tagger.clean(self.string)
+        taggedString=self.tagger.tagg(self.cleanString)
+        self.tableLem = taggedString.rstrip().split("\n")
+        for line in taggedString.rstrip().split("\n"):
+            table = line.rstrip().split(" ")
+            if not table[0] in self.dico :
+                self.dico[table[0]]=[set(),set()]
+            self.dico[table[0]][0].add(table[2])
+    def insertPhon(self):
+        phonedString=self.phoner.phon(self.cleanString)
+        self.tablephon= phonedString.rstrip().split("\n")
+        for line in phonedString.rstrip().split("\n"):
+            table = line.rstrip().split(" ")
+            if table[0] in self.dico:
+                self.dico[table[0]][1].add(table[1])
+    def getDico(self):
+        self.clean()
+        self.insertLem()
+        self.insertPhon()
+        return self.dico
@@ -15,7 +15,7 @@
 		data: data,
 		success: function(data){
 			result=$('#result');
-			result.append(_.escape(data));
+			result.html(_.escape(data));
 			console.log(_.escape(data));
 			console.log("resultat");
 		},
@@ -4,6 +4,7 @@
 import json
 from flask import Flask, request, render_template
 from processor.LiaTools import *
+from processor.Orkis import Orkis
 app = Flask(__name__)
  
 @app.route("/")
  
  
  
  
@@ -12,24 +13,15 @@
  
 @app.route("/tagger",methods=['POST'])
 def cleaner():
-    # Charging Processor et check if they are okay ( aim is to dynamic charge later ) 
-    tagger = Tagger()
-    tagger.isReady()
-    phoner = Phoner()
-    phoner.isReady()
-    stoplist = StopWord()
-    stoplist.isReady()
     # Receive String from post parametre Raw text
     dirtyString= request.values[u'string']
+    # Charging Processor et check if they are okay ( aim is to dynamic charge later ) 
+    orkisProc = Orkis(dirtyString)
     # Processing 
-    dirtyString = stoplist.RemoveStopList(dirtyString)
-    lemm = tagger.lemm(tagger.clean(dirtyString))
     # Adding lemm of each words cause we went ther phonem too
-    dirtyString = dirtyString+" "+ lemm
-    cleanString= phoner.clean(dirtyString)
-    taggedString= phoner.phon(cleanString)
+    taggedTable= orkisProc.getDico()
     # Returning a row text to be parse client side
-    return taggedString
+    return unicode(taggedTable)
 if __name__ == '__main__':
     app.debug = True
     app.run(host='0.0.0.0')
	1	+from BaseProcessor import baseProcessor
	2	+import nltk
	3	+from LiaTools import *
	4	+class Orkis(baseProcessor):
	5	+ """ Processor for Orkis """
	6	+ def __init__(self,dirtyString):
	7	+ self.tagger=Tagger()
	8	+ self.phoner=Phoner()
	9	+ self.dico ={}
	10	+ self.string=dirtyString
	11	+ def isReady(self):
	12	+ self.phoner.isReady()
	13	+ self.tagger.isReady()
	14	+ def clean(self):
	15	+ stopword=StopWord()
	16	+ self.string=stopword.RemoveStopList(self.string)
	17	+ def insertLem(self):
	18	+ self.cleanString=self.tagger.clean(self.string)
	19	+ taggedString=self.tagger.tagg(self.cleanString)
	20	+ self.tableLem = taggedString.rstrip().split("\n")
	21	+ for line in taggedString.rstrip().split("\n"):
	22	+ table = line.rstrip().split(" ")
	23	+ if not table[0] in self.dico :
	24	+ self.dico[table[0]]=[set(),set()]
	25	+ self.dico[table[0]][0].add(table[2])
	26	+ def insertPhon(self):
	27	+ phonedString=self.phoner.phon(self.cleanString)
	28	+ self.tablephon= phonedString.rstrip().split("\n")
	29	+ for line in phonedString.rstrip().split("\n"):
	30	+ table = line.rstrip().split(" ")
	31	+ if table[0] in self.dico:
	32	+ self.dico[table[0]][1].add(table[1])
	33	+ def getDico(self):
	34	+ self.clean()
	35	+ self.insertLem()
	36	+ self.insertPhon()
	37	+ return self.dico
...	...	@@ -15,7 +15,7 @@
15	15	data: data,
16	16	success: function(data){
17	17	result=$('#result');
18		- result.append(_.escape(data));
	18	+ result.html(_.escape(data));
19	19	console.log(_.escape(data));
20	20	console.log("resultat");
21	21	},
...	...	@@ -4,6 +4,7 @@
4	4	import json
5	5	from flask import Flask, request, render_template
6	6	from processor.LiaTools import *
	7	+from processor.Orkis import Orkis
7	8	app = Flask(__name__)
8	9
9	10	@app.route("/")
10	11
11	12
12	13
13	14
...	...	@@ -12,24 +13,15 @@
12	13
13	14	@app.route("/tagger",methods=['POST'])
14	15	def cleaner():
15		- # Charging Processor et check if they are okay ( aim is to dynamic charge later )
16		- tagger = Tagger()
17		- tagger.isReady()
18		- phoner = Phoner()
19		- phoner.isReady()
20		- stoplist = StopWord()
21		- stoplist.isReady()
22	16	# Receive String from post parametre Raw text
23	17	dirtyString= request.values[u'string']
	18	+ # Charging Processor et check if they are okay ( aim is to dynamic charge later )
	19	+ orkisProc = Orkis(dirtyString)
24	20	# Processing
25		- dirtyString = stoplist.RemoveStopList(dirtyString)
26		- lemm = tagger.lemm(tagger.clean(dirtyString))
27	21	# Adding lemm of each words cause we went ther phonem too
28		- dirtyString = dirtyString+" "+ lemm
29		- cleanString= phoner.clean(dirtyString)
30		- taggedString= phoner.phon(cleanString)
	22	+ taggedTable= orkisProc.getDico()
31	23	# Returning a row text to be parse client side
32		- return taggedString
	24	+ return unicode(taggedTable)
33	25	if __name__ == '__main__':
34	26	app.debug = True
35	27	app.run(host='0.0.0.0')