phuoc/html_cleanup
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

test upload ok

Browse Source
This commit is contained in:
Phuoc
2023-11-28 17:56:37 +01:00
parent 37655ad6f5
commit 8b15b7ab27
12 changed files with 715 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
development.ini
View File

@@ -14,9 +14,8 @@ pyramid.default_locale_name = en
pyramid.includes = pyramid.includes =
pyramid_debugtoolbar pyramid_debugtoolbar
# By default, the toolbar only appears for clients from IP addresses # temp folder location
# '127.0.0.1' and '::1'. temp_folder = /Users/phuoc/pyramid/html_cleanup/html_cleanup/static/temp/
# debugtoolbar.hosts = 127.0.0.1 ::1
### ###
# wsgi server configuration # wsgi server configuration

5
html_cleanup/static/temp/clean_Nantes_anthologie_cinema_viet.htm Normal file
View File

@@ -0,0 +1,5 @@
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<title>untitled</title>
</head>

5
html_cleanup/static/temp/clean_www.icloud.com.html Normal file
View File

@@ -0,0 +1,5 @@
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<title>untitled</title>
</head>

496
html_cleanup/static/temp/www.icloud.com.html Normal file
View File

File diff suppressed because one or more lines are too long

2
html_cleanup/templates/404.jinja2
View File

@@ -2,7 +2,7 @@
{% block content %} {% block content %}
<div class="content"> <div class="content">
<h1><span class="font-semi-bold">Pyramid</span> <span class="smaller">Starter project</span></h1> <h1>HTML Cleanup</h1>
<p class="lead"><span class="font-semi-bold">404</span> Page Not Found</p> <p class="lead"><span class="font-semi-bold">404</span> Page Not Found</p>
</div> </div>
{% endblock content %} {% endblock content %}

39
html_cleanup/templates/home.jinja2 Normal file
View File

@@ -0,0 +1,39 @@
{% extends "layout.jinja2" %}
{% block content %}
{% if message %}
<div class="alert alert-danger">
{{ message }}
</div>
{% endif %}
<form id="uploadfile-form" action="{{ request.route_url('home') }}" method="post"
accept-charset="utf-8" enctype="multipart/form-data">
<div class="form-group">
<label for="uploadfile">Veuillez séléctionner un fichier :</label>
<input class="file" id="uploadfile" name="uploadfile" type="file" />
</div>
<br />
<div class="form-group">
<button id="uploadButton" class="btn btn-success" type="submit" name="form.submitted">
<i class="glyphicon glyphicon-arrow-up"></i> Clean up</button>
</div>
</form>
<div class="panel-footer">
<ul>
<li>Seuls les fichiers au format <b>HTML</b> seront acceptés.</li>
<li>La taille du fichier ne doit <b>pas dépasser 10 Mo</b>.</li>
</ul>
</div>
{% if file_url %}
<h2 class="text-info font-semi-bold">CLEANUP réussi</h2>
Votre fichier nettoyé : <b>{{ file_name }}</b>: <a href="{{ file_url }}""></a>
<a class="btn btn-primary" href="{{ file_url }}" download="{{ file_name }}"><span class="glyphicon glyphicon-arrow-up"></span> Télécharger</a>
{% endif %}
{% endblock content %}

18
html_cleanup/templates/layout.jinja2
View File

@@ -24,18 +24,22 @@
</head> </head>
<body> <body>
<!-- Container (Above content Section) -->
<div class="container"> <div class="container">
<div class="row">
<!-- Display Page Title -->
{% if page_title %}
<h1 class="font-semi-bold">{{ page_title }}</h1>
<br />
{% endif %}
<!-- display page content-->
<div class="container row">
{% block content %} {% block content %}
<p>No content</p> <p>No content</p>
{% endblock content %} {% endblock content %}
</div> </div>
<div class="row"> <!-- Container (Above content Section) -->
<div class="copyright">
Copyright &copy; Pylons Project
</div>
</div>
</div> </div>

8
html_cleanup/templates/mytemplate.jinja2
View File

@@ -1,8 +0,0 @@
{% extends "layout.jinja2" %}
{% block content %}
<div class="content">
<h1><span class="font-semi-bold">Pyramid</span> <span class="smaller">Starter project</span></h1>
<p class="lead">Welcome to <span class="font-normal">{{project}}</span>, a&nbsp;Pyramid application generated&nbsp;by<br><span class="font-normal">Cookiecutter</span>.</p>
</div>
{% endblock content %}

150
html_cleanup/views/default.py
View File

@@ -1,6 +1,150 @@
from pyramid.view import view_config from pyramid.view import view_config
from pyramid.httpexceptions import HTTPFound
import os
import shutil
import magic
import sys
@view_config(route_name='home', renderer='html_cleanup:templates/home.jinja2')
def home(request):
@view_config(route_name='home', renderer='html_cleanup:templates/mytemplate.jinja2') message = ''
def my_view(request): file_name = ''
return {'project': 'html_cleanup'} file_url = ''
if 'form.submitted' in request.params:
if request.POST['uploadfile'] != b'':
input_file = request.POST['uploadfile'].file
input_name = request.POST['uploadfile'].filename
# récupère le fichier download, faire les controles et traiter
message = process_file(request, input_file, input_name)
if message == '':
file_name = "clean_" + input_name
file_url = request.static_url('html_cleanup:static/temp/') + file_name
return {
'page_title': "HTML cleanup",
'message': message,
'file_url': file_url,
'file_name': file_name,
}
def clean_html(input_file, output_file):
# cleanup undesirable tags in html file
encoding = "utf-8"
message = ""
import pdb;pdb.set_trace()
try:
with open(input_file, 'r', encoding='utf-8') as fi, open(output_file, 'w', encoding='utf-8') as fo:
fo.write("<html>\n<head>\n<meta http-equiv=\"Content-Type\" content=\"text/html; charset={}\">\n<title>untitled</title>\n</head>\n")
body = False
skip_tag = False
nb_lines = 0
for line_in in fi:
line_out = ""
if not body:
init = line_in.find("<body")
if init != -1 and init == 0:
body = True
if body:
nb_lines += 1
if skip_tag:
start_from = line_in.find(">")
if start_from != -1 and start_from >= 0:
skip_tag = False
start_from = start_from + 1 if start_from < len(line_in) else 0
else:
start_from = init if nb_lines == 1 else 0
if start_from >= 0:
done = False
while not done:
next_tag = line_in.find("<", start_from)
if next_tag == -1:
line_out += line_in[start_from:]
done = True
skip_tag = False
else:
if next_tag > start_from:
line_out += line_in[start_from:next_tag]
end_tag = line_in.find(">", next_tag + 1)
if end_tag == -1:
s = line_in[next_tag + 1:]
done = True
skip_tag = True
else:
s = line_in[next_tag + 1:end_tag]
if end_tag < len(line_in):
start_from = end_tag + 1
done = False
skip_tag = False
else:
done = True
skip_tag = False
tag = s.split(" ")
if tag:
tag[0] = tag[0].lower()
if tag[0] in ["table", "div", "img", "a"]:
line_out += "<{}>".format(s)
elif tag[0] in ["p"]:
line_out += "<p align=\"justify\">"
elif tag[0] in ["b", "/b", "em", "/em", "i", "/i", "li", "li/", "/li",
"ol", "/ol", "ul", "/ul", "strong", "/strong", "sub", "/sub",
"sup", "/sup", "u", "/u"]:
line_out += "<{}>".format(tag[0])
elif tag[0] in ["h1", "/h1", "h2", "/h2", "h3", "/h3", "h4", "/h4", "h5",
"/h5", "h6", "/h6", "/p", "body", "/body", "/html", "/div",
"/img", "/a"]:
line_out += "<{}>".format(tag[0])
elif tag[0] in ["table", "/table", "td", "/td", "th", "/th", "tr", "/tr"]:
line_out += "<{}>".format(tag[0])
elif tag[0] in ["br", "br/"]:
line_out += "<{}>".format(tag[0])
if line_out != "":
fo.write(line_out + "\n")
except Exception as e:
message = "Error: {}".format(str(e))
return message
def process_file(request, input_file, input_name):
# Check file mime type and size
# and if OK, process file
message = ''
ext_allowed = ['text/html', 'image/png', 'image/jpeg']
max_size = 10 * (1024 ** 2) # 10 Mb
temp_folder = request.registry.settings['temp_folder']
mime = magic.from_buffer(input_file.read(), mime=True)
# types de fichiers autorisés ?
if mime not in ext_allowed:
message = "ERREUR: Le format du fichier n'est pas valide. Téléchargement refusé."
else:
# lire la taille du fichier
pos = input_file.tell()
input_file.seek(0, 2) #seek to end
filesize = input_file.tell()
input_file.seek(pos) # back to original position
# controler la taille du fichier
if filesize > max_size:
message = "ERREUR: La taille du fichier dépasse la limite autorisée. Téléchargement refusé."
else:
# controle OK, traiter le fichier
output_name = "clean_" + input_name
output_file = os.path.join(temp_folder, output_name)
message = clean_html(input_file, output_file)
return message

1
html_cleanup/views/notfound.py
View File

@@ -1,6 +1,5 @@
from pyramid.view import notfound_view_config from pyramid.view import notfound_view_config
@notfound_view_config(renderer='html_cleanup:templates/404.jinja2') @notfound_view_config(renderer='html_cleanup:templates/404.jinja2')
def notfound_view(request): def notfound_view(request):
request.response.status = 404 request.response.status = 404

3
production.ini
View File

@@ -12,6 +12,9 @@ pyramid.debug_notfound = false
pyramid.debug_routematch = false pyramid.debug_routematch = false
pyramid.default_locale_name = en pyramid.default_locale_name = en
# temp folder location
temp_folder = /pyramid/html_cleanup/html_cleanup/static/temp/
### ###
# wsgi server configuration # wsgi server configuration
### ###

1
setup.py
View File

@@ -13,6 +13,7 @@ requires = [
'pyramid', 'pyramid',
'pyramid_jinja2', 'pyramid_jinja2',
'pyramid_debugtoolbar', 'pyramid_debugtoolbar',
'python-magic',
'waitress', 'waitress',
] ]