summaryrefslogtreecommitdiffstats
path: root/src/fetch/scripts/fr.allocine.py
diff options
context:
space:
mode:
Diffstat (limited to 'src/fetch/scripts/fr.allocine.py')
-rw-r--r--src/fetch/scripts/fr.allocine.py60
1 files changed, 30 insertions, 30 deletions
diff --git a/src/fetch/scripts/fr.allocine.py b/src/fetch/scripts/fr.allocine.py
index 97a2247..83e41d0 100644
--- a/src/fetch/scripts/fr.allocine.py
+++ b/src/fetch/scripts/fr.allocine.py
@@ -15,20 +15,20 @@
# ***************************************************************************
# Version 0.4: 2007-08-27
-# * Fixed parsing errors: some fields in allocine's HTML pages have changed recently. Multiple actors and genres
+# * Fixed parsing errors: some fields in allocine's HTML pages have changed recently. Multiple actors and genres
# could not be retrieved. Fixed bad http request error due to some changes in HTML code.
#
# Version 0.3:
# * Fixed parsing: some fields in allocine's HTML pages have changed. Movie's image could not be fetched anymore. Fixed.
-#
+#
# Version 0.2:
# * Fixed parsing: allocine's HTML pages have changed. Movie's image could not be fetched anymore.
-#
+#
# Version 0.1:
# * Initial release.
import sys, os, re, md5, random
-import urllib, urllib2, time, base64
+import urllib.request, urllib.parse, urllib.error, time, base64
import xml.dom.minidom
XML_HEADER = """<?xml version="1.0" encoding="UTF-8"?>"""
@@ -48,16 +48,16 @@ class BasicTellicoDOM:
self.__root = self.__doc.createElement('tellico')
self.__root.setAttribute('xmlns', 'http://periapsis.org/tellico/')
self.__root.setAttribute('syntaxVersion', '9')
-
+
self.__collection = self.__doc.createElement('collection')
self.__collection.setAttribute('title', 'My Movies')
self.__collection.setAttribute('type', '3')
-
+
self.__fields = self.__doc.createElement('fields')
# Add all default (standard) fields
self.__dfltField = self.__doc.createElement('field')
self.__dfltField.setAttribute('name', '_default')
-
+
# Add a custom 'Collection' field
self.__customField = self.__doc.createElement('field')
self.__customField.setAttribute('name', 'titre-original')
@@ -67,7 +67,7 @@ class BasicTellicoDOM:
self.__customField.setAttribute('format', '1')
self.__customField.setAttribute('type', '1')
self.__customField.setAttribute('i18n', 'yes')
-
+
self.__fields.appendChild(self.__dfltField)
self.__fields.appendChild(self.__customField)
self.__collection.appendChild(self.__fields)
@@ -90,23 +90,23 @@ class BasicTellicoDOM:
entryNode.setAttribute('id', str(self.__currentId))
titleNode = self.__doc.createElement('title')
- titleNode.appendChild(self.__doc.createTextNode(unicode(d['title'], 'latin-1').encode('utf-8')))
+ titleNode.appendChild(self.__doc.createTextNode(str(d['title'], 'latin-1').encode('utf-8')))
otitleNode = self.__doc.createElement('titre-original')
- otitleNode.appendChild(self.__doc.createTextNode(unicode(d['otitle'], 'latin-1').encode('utf-8')))
+ otitleNode.appendChild(self.__doc.createTextNode(str(d['otitle'], 'latin-1').encode('utf-8')))
yearNode = self.__doc.createElement('year')
- yearNode.appendChild(self.__doc.createTextNode(unicode(d['year'], 'latin-1').encode('utf-8')))
+ yearNode.appendChild(self.__doc.createTextNode(str(d['year'], 'latin-1').encode('utf-8')))
genresNode = self.__doc.createElement('genres')
for g in d['genres']:
genreNode = self.__doc.createElement('genre')
- genreNode.appendChild(self.__doc.createTextNode(unicode(g, 'latin-1').encode('utf-8')))
+ genreNode.appendChild(self.__doc.createTextNode(str(g, 'latin-1').encode('utf-8')))
genresNode.appendChild(genreNode)
natsNode = self.__doc.createElement('nationalitys')
natNode = self.__doc.createElement('nat')
- natNode.appendChild(self.__doc.createTextNode(unicode(d['nat'], 'latin-1').encode('utf-8')))
+ natNode.appendChild(self.__doc.createTextNode(str(d['nat'], 'latin-1').encode('utf-8')))
natsNode.appendChild(natNode)
castsNode = self.__doc.createElement('casts')
@@ -114,7 +114,7 @@ class BasicTellicoDOM:
castNode = self.__doc.createElement('cast')
col1Node = self.__doc.createElement('column')
col2Node = self.__doc.createElement('column')
- col1Node.appendChild(self.__doc.createTextNode(unicode(g, 'latin-1').encode('utf-8')))
+ col1Node.appendChild(self.__doc.createTextNode(str(g, 'latin-1').encode('utf-8')))
castNode.appendChild(col1Node)
castNode.appendChild(col2Node)
castsNode.appendChild(castNode)
@@ -122,17 +122,17 @@ class BasicTellicoDOM:
dirsNode = self.__doc.createElement('directors')
for g in d['dirs']:
dirNode = self.__doc.createElement('director')
- dirNode.appendChild(self.__doc.createTextNode(unicode(g, 'latin-1').encode('utf-8')))
+ dirNode.appendChild(self.__doc.createTextNode(str(g, 'latin-1').encode('utf-8')))
dirsNode.appendChild(dirNode)
timeNode = self.__doc.createElement('running-time')
- timeNode.appendChild(self.__doc.createTextNode(unicode(d['time'], 'latin-1').encode('utf-8')))
+ timeNode.appendChild(self.__doc.createTextNode(str(d['time'], 'latin-1').encode('utf-8')))
- allocineNode = self.__doc.createElement(unicode('allocin�-link', 'latin-1').encode('utf-8'))
- allocineNode.appendChild(self.__doc.createTextNode(unicode(d['allocine'], 'latin-1').encode('utf-8')))
+ allocineNode = self.__doc.createElement(str('allocin�-link', 'latin-1').encode('utf-8'))
+ allocineNode.appendChild(self.__doc.createTextNode(str(d['allocine'], 'latin-1').encode('utf-8')))
plotNode = self.__doc.createElement('plot')
- plotNode.appendChild(self.__doc.createTextNode(unicode(d['plot'], 'latin-1').encode('utf-8')))
+ plotNode.appendChild(self.__doc.createTextNode(str(d['plot'], 'latin-1').encode('utf-8')))
if d['image']:
imageNode = self.__doc.createElement('image')
@@ -140,12 +140,12 @@ class BasicTellicoDOM:
imageNode.setAttribute('id', d['image'][0])
imageNode.setAttribute('width', '120')
imageNode.setAttribute('height', '160')
- imageNode.appendChild(self.__doc.createTextNode(unicode(d['image'][1], 'latin-1').encode('utf-8')))
+ imageNode.appendChild(self.__doc.createTextNode(str(d['image'][1], 'latin-1').encode('utf-8')))
coverNode = self.__doc.createElement('cover')
coverNode.appendChild(self.__doc.createTextNode(d['image'][0]))
- for name in ( 'titleNode', 'otitleNode', 'yearNode', 'genresNode', 'natsNode',
+ for name in ( 'titleNode', 'otitleNode', 'yearNode', 'genresNode', 'natsNode',
'castsNode', 'dirsNode', 'timeNode', 'allocineNode', 'plotNode' ):
entryNode.appendChild(eval(name))
@@ -154,7 +154,7 @@ class BasicTellicoDOM:
self.__images.appendChild(imageNode)
self.__collection.appendChild(entryNode)
-
+
self.__currentId += 1
def printXML(self):
@@ -162,8 +162,8 @@ class BasicTellicoDOM:
Outputs XML content to stdout
"""
self.__collection.appendChild(self.__images)
- print XML_HEADER; print DOCTYPE
- print self.__root.toxml()
+ print(XML_HEADER); print(DOCTYPE)
+ print(self.__root.toxml())
class AlloCineParser:
@@ -185,7 +185,7 @@ class AlloCineParser:
'otitle' : 'Titre original *: *<i>(?P<otitle>.+?)</i>',
'plot' : """(?s)<td valign="top" style="padding:10 0 0 0"><div align="justify"><h4> *(?P<plot>.+?) *</h4>""",
'image' : """<td valign="top" width="120".*?<img src="(?P<image>.+?)" border"""}
-
+
self.__domTree = BasicTellicoDOM()
@@ -203,7 +203,7 @@ class AlloCineParser:
Fetch HTML data from url
"""
- u = urllib2.urlopen(url)
+ u = urllib.request.urlopen(url)
self.__data = u.read()
u.close()
@@ -224,7 +224,7 @@ class AlloCineParser:
matches = data = {}
- for name, regexp in self.__regExps.iteritems():
+ for name, regexp in self.__regExps.items():
if name == 'image':
matches[name] = re.findall(self.__regExps[name], self.__data, re.S | re.I)
else:
@@ -272,7 +272,7 @@ class AlloCineParser:
elif name == 'image':
# Save image to a temporary folder
md5 = genMD5()
- imObj = urllib2.urlopen(matches[name][0].strip())
+ imObj = urllib.request.urlopen(matches[name][0].strip())
img = imObj.read()
imObj.close()
imgPath = "/tmp/%s.jpeg" % md5
@@ -303,7 +303,7 @@ class AlloCineParser:
if not len(title): return
self.__title = title
- self.__getHTMLContent(self.__searchURL % urllib.quote(self.__title))
+ self.__getHTMLContent(self.__searchURL % urllib.parse.quote(self.__title))
# Get all links
links = self.__fetchMovieLinks()
@@ -321,7 +321,7 @@ class AlloCineParser:
def showUsage():
- print "Usage: %s movietitle" % sys.argv[0]
+ print("Usage: %s movietitle" % sys.argv[0])
sys.exit(1)
def main():
generated by cgit v1.2.1 (git 2.18.0) at 2024-12-18 20:13:07 +0000