experiments-in-summarization/preprocess.py at master · mtthss/experiments-in-summarization · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
import os
import re
import nltk
import random
import cPickle as pk
import xml.etree.ElementTree as ET

from collections import defaultdict


__author__ = 'matteo'


ref = False
rivals = False
LM = True
learnDic = False
random_UKN = True

sent_detector = nltk.data.load('tokenizers/punkt/english.pickle')

def clean(txt, stop=False, stem=False):
    txt = re.sub('"|\'|-|\||\n|<|>|\*|\\\\+', ' ', txt)
    txt = txt.lower()
    txt = ''.join(i for i in txt if not i.isdigit())
    txt = re.sub('\s+', ' ', txt)
    return txt.strip()

test_refs = ["d632i","d426a","D0628A","d324e","D0644H","D0601A","D0630C","D0647B","D0606F",
             "D0637A","D0607G","d407b","d346h","d393f","d350a","d400b","d332h","D0602B","D0623E"]


# references
if ref:

    year = 2005
    ref_in_path = "./data/references/"+str(year)
    ref_out_path = "./data2/references/"+str(year)+"/"
    sent_detector = nltk.data.load('tokenizers/punkt/english.pickle')

    for filename in os.listdir(ref_in_path):

        content = open(ref_in_path+'/'+filename, 'r').read()
        code = filename.split(".")

        cond = code[0]+code[3] if year==2006 else (code[0]+code[3]).lower()

        if cond in test_refs:

            new_name = (code[0]+code[3]+"_"+code[4]).lower()
            out_file = open(ref_out_path+new_name, "w")

            for s in sent_detector.tokenize(content):
                out_file.write(s+"\n")

            out_file.close()


# other automatic summaries
if rivals:

    path_in = "./rivals/peers/"
    path_out = "./rivals2/peers/"
    count = 0
    test_refs_low = [x.lower() for x in test_refs]

    for filename in os.listdir(path_in):

        content = open(path_in+filename, 'r').read()
        code = filename.split(".")

        if (code[0]+code[3]).lower() in test_refs_low:

            new_name = (code[0]+code[3]+"_"+code[4]).lower()
            out_file = open(path_out+new_name, "w")

            out_file.write(content)
            out_file.close()


# generate single doc for training language models
if LM:

    path_dict = "./pickled_counts.pkl"

    path_in = "./data/collections/"
    path_out_train = "../neural-language-modelling/duc-train.txt"
    path_out_dev = "../neural-language-modelling/duc-dev.txt"
    path_out_test = "../neural-language-modelling/duc-test.txt"

    out_file_train = open(path_out_train, "w")
    out_file_dev = open(path_out_dev, "w")
    out_file_test = open(path_out_test, "w")

    dict = pk.load(open(path_dict,"r"))

    for year in [2005, 2006]:

        for foldername in os.listdir(path_in+str(year)):

            if foldername in ["duc2005_topics.sgml", "duc2006_topics.sgml"]:
                continue

            count = 0

            for filename in os.listdir(path_in+str(year)+"/"+foldername):

                count += 1

                if count <= 24:

                    root = ET.parse(path_in+str(year)+"/"+foldername+"/"+filename).getroot()
                    node = root.find('TEXT')

                    if node!=None:

                        content = root.find('TEXT').text
                        content = clean(content)
                        new = ""
                        for sent in sent_detector.tokenize(content):
                            if random_UKN:
                                new += ' '.join(["<UKN>" if (random.random()>0.98 and dict[word]>20 and dict[word]<484) else word for word in nltk.tokenize.word_tokenize(sent)])+"\n"
                        content = new

                        if count > 20 and count <=22 and random.random() < 0.7:
                            out_file_dev.write(content)
                        elif count > 22 and count <=24 and random.random() < 0.74:
                            out_file_test.write(content)
                        else:
                            out_file_train.write(content)
                else:
                    continue

    out_file_train.close()
    out_file_dev.close()
    out_file_test.close()


if learnDic:

    dict = defaultdict(int)
    path_in = "./data/collections/"

    for year in [2005, 2006]:

        for foldername in os.listdir(path_in+str(year)):

            if foldername in ["duc2005_topics.sgml", "duc2006_topics.sgml"]:
                continue

            for filename in os.listdir(path_in+str(year)+"/"+foldername):

                root = ET.parse(path_in+str(year)+"/"+foldername+"/"+filename).getroot()
                node = root.find('TEXT')

                if node!=None:

                    content = root.find('TEXT').text

                    for word in nltk.tokenize.word_tokenize(clean(content)):
                        dict[word] += 1

    pk.dump(dict, open("./pickled_counts.pkl","wb"))