File size: 8,479 Bytes
1e588f4
 
 
 
 
 
 
 
 
 
 
33df71b
 
4e90ce6
33df71b
 
 
 
 
4e90ce6
33df71b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4e90ce6
33df71b
 
 
 
4e90ce6
33df71b
 
 
 
 
4e90ce6
 
33df71b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1e588f4
 
 
 
 
 
 
 
bfef0ca
1e588f4
 
 
 
 
 
 
 
 
 
4e90ce6
 
 
 
 
 
 
 
 
 
1e588f4
 
 
 
4e90ce6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d6046c5
4e90ce6
 
 
 
 
 
 
 
 
 
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
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
# AUTOGENERATED! DO NOT EDIT! File to edit: ../nbs/02_app_gradio.ipynb.

# %% auto 0
__all__ = ['categories', 'k', 'min_words', 'max_words', 'ignore_text', 'ignore_common', 'learn', 'text', 'label', 'examples',
           'intf', 'predict']

# %% ../nbs/02_app_gradio.ipynb 4
import warnings
warnings.filterwarnings('ignore')
from fastai.text.all import *
import gradio as gr
import requests
from bs4 import BeautifulSoup
import enchant
import re
import random
from collections import Counter
import hashlib
import pickle
from wordcloud import WordCloud


# %% ../nbs/01_data.ipynb 8
class Webpage:
    def __init__(self, url):
        self.url = url
        self.hash = self.get_hash_str()
        self.requested = False
        self.page_text = ""
        self.html = ""
        self.links = []
        self.text = []
        self.cleaned_text = []
        self.most_common_words = []
    
    def get_page(self, headers, min_size, max_size):
        r = requests.get(self.url, stream=True, headers=headers)
        content_length = int(r.headers.get('Content-Length', 0))
        data = []
        length = 0

        if content_length > max_size:
            return None

        for chunk in r.iter_content(1024):
            data.append(chunk)
            length += len(chunk)
            if length > max_size:
                return None
        r._content = b''.join(data)
        if len(r.text) < min_size: return None
        return r.text

    def get_page_html(self, min_size=1000, max_size=2000000):
        user_agents = [ 
            'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36', 
            'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/92.0.4515.107 Safari/537.36', 
            'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.212 Safari/537.36', 
            'Mozilla/5.0 (iPhone; CPU iPhone OS 12_2 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Mobile/15E148', 
            'Mozilla/5.0 (Linux; Android 11; SM-G960U) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.72 Mobile Safari/537.36' 
        ] 
        user_agent = random.choice(user_agents) 
        headers = {'User-Agent': user_agent} 
        self.page_text = self.get_page(headers, min_size, max_size)
        self.html = BeautifulSoup(self.page_text, "html.parser")
        self.requested = True

    def get_hash_str(self, inp=""):
        return hashlib.sha3_256((self.url+inp).encode()).hexdigest()

    def get_html_anchors(self, keyword="http"):
        for anchor in self.html.findAll('a'):
            link = anchor.get('href')
            if link == None or link == "":
                continue
            if keyword in link:
                self.links.append(link)
                
    def get_html_text(self, tags=["p"]):
        for tag in tags:
            for p in self.html.findAll(tag):
                p_text = p.getText().strip()
                if p_text == None or p_text == '':
                    continue
                self.text.append(p_text)

    def clean_html_text(self, max_words, enchant_dict="en_US", ignore=[], rx="[^a-zA-Z ]+", min_word_len=2):
        all_text = ' '.join(self.text).lower()
        regex_text = re.sub(rx,'',all_text).strip()
        split = regex_text.split()
        split = [word for word in split if word not in ignore]
        if enchant_dict != "": d = enchant.Dict(enchant_dict)
        for word in split:
            if len(self.cleaned_text) >= max_words: break
            if len(word) >= min_word_len:
                if enchant_dict == "":
                    self.cleaned_text.append(word)
                elif d.check(word): 
                    self.cleaned_text.append(word)

    def k_common_words(self, k=10, ignore=[]):
        if self.cleaned_text == "":
            text = self.text
        else:
            text = self.cleaned_text
        all_text = ' '.join(text).lower()
        split = all_text.split()
        split_ignore = [word for word in split if word not in ignore]
        counts = Counter(split_ignore)
        k_most_common = counts.most_common(k)
        self.most_common_words = k_most_common

    def save_text(self, path, fname):
        file = open(path+fname, 'wb')
        pickle.dump(self.text, file)
        file.close()

    def load_text(self, path, fname):
        file = open(path+fname, 'rb')
        self.text = pickle.load(file)
        file.close()

    def save_links(self, path, fname):
        file = open(path+fname, 'wb')
        pickle.dump(self.links, file)
        file.close()

    def load_links(self, path, fname):
        file = open(path+fname, 'rb')
        self.links = pickle.load(file)
        file.close()

# %% ../nbs/01_data.ipynb 14
def get_page_all(url, k, max_words, ignore_text, ignore_common, path = None):
    page = Webpage(url)
    fname_text = page.hash+'.text'
    fname_links = page.hash+'.links'
    if path == None:
        page.get_page_html()
        page.get_html_text(tags=["p","h1","h2","h3","span"])
        page.get_html_anchors()
    else:
        if os.path.isfile(path+fname_text): 
            page.load_text(path, fname_text)
        else:
            page.get_page_html()
            page.get_html_text(tags=["p","h1","h2","h3","span"])
            page.save_text(path, fname_text)

        if os.path.isfile(path+fname_links): 
            page.load_links(path, fname_links)
        else:
            if page.html == "": page.get_page_html()
            page.get_html_anchors()
            page.save_links(path, fname_links)

    if page.text is not None:
        page.clean_html_text(max_words, ignore=ignore_text, rx="[^a-zA-Z ]+")
        page.k_common_words(k=k, ignore=ignore_common)
    return page

# %% ../nbs/02_app_gradio.ipynb 6
categories = ('pseudoscience','science')
k = 30
min_words = 20
max_words = 450
ignore_text = ['the', 'of', 'to', 'and', 'a', 'in', 'it', 'that', 'for', 'on'] 
ignore_common = ignore_text
learn = load_learner('model.pkl', cpu=True)

def predict(url):
    page = get_page_all(url, k, max_words, ignore_text, ignore_common)
    length = len(page.cleaned_text)
    if  length < min_words:
        return "ERROR: Returned "+str(length)+" words"
    else:
        text = ' '.join(page.cleaned_text)
        with learn.no_bar(), learn.no_logging():
            pred,idx,probs = learn.predict(text)
        wordcloud = WordCloud(width = 800, height = 800,
                background_color ='white',
                min_font_size = 10).generate(text)
                
        # plot the WordCloud image                      
        fig = plt.figure(figsize = (8, 8), facecolor = None)
        plt.imshow(wordcloud)
        plt.axis("off")
        plt.tight_layout(pad = 0)
        return (dict(zip(categories, map(float,probs))), fig)

# %% ../nbs/02_app_gradio.ipynb 8
examples = ['https://www.theskepticsguide.org/about','https://www.foxnews.com/opinion']

pseudo_sources = ["http://www.ageofautism.com/",
 "http://www.naturalnews.com", 
 "https://foodbabe.com/starthere/",
 "http://www.chopra.com",
 "https://www.mercola.com/",
 "https://www.history.com/",
 "https://doctoroz.com/",
 "https://www.disclose.tv/",
 "https://nationalreport.net/",
 "https://heartland.org/",
 "https://www.dailymail.co.uk/",
 "https://www.motherjones.com/"]

science_sources = ["https://sciencebasedmedicine.org/",
 "https://www.hopkinsmedicine.org/gim/research/method/ebm.html",
 "https://www.bbc.com/news/science_and_environment",
 "https://www.nature.com/",
 "https://www.science.org/",
 "https://www.snopes.com/top/",
 "https://quackwatch.org/",
 "https://www.skepdic.com/",
 "http://scibabe.com/",
 "http://pandasthumb.org/",
 "https://skepticalscience.com/",
 "https://www.cdc.gov/",
 "https://apnews.com/"]

with gr.Blocks() as blocks:
    gr.Markdown("# Pseudometer")
    gr.Markdown("Prototype machine learning pseudoscience detector for websites!")
    text = gr.Textbox(label="Input URL (http format):")
    label = gr.Label()
    btn = gr.Button("Analyze!")
    with gr.Accordion("Pseudoscience Primary Training Sources"):
        gr.Markdown(', '.join(pseudo_sources))
    with gr.Accordion("Science Primary Training Sources"):
        gr.Markdown(', '.join(science_sources))
    example = gr.Examples(examples=examples, inputs=text)

    btn.click(fn=predict, inputs=text, outputs=[label, gr.Plot(label="Wordcloud")])
    
blocks.launch()