simplehoster - use parseFileSize

8 files changed, 64 insertions, 61 deletions

diff --git a/module/plugins/hoster/CrockoCom.py b/module/plugins/hoster/CrockoCom.py
index 7eafa67ed..9598025ec 100644
--- a/module/plugins/hoster/CrockoCom.py
+++ b/module/plugins/hoster/CrockoCom.py
@@ -9,7 +9,7 @@ class CrockoCom(SimpleHoster):
     __name__ = "CrockoCom"
     __type__ = "hoster"
     __pattern__ = r"http://(www\.)?(crocko|easy-share).com/.*"
-    __version__ = "0.10"
+    __version__ = "0.11"
     __description__ = """Crocko Download Hoster"""
     __author_name__ = ("zoidberg")
     __author_mail__ = ("zoidberg@mujmail.cz")
@@ -23,7 +23,7 @@ class CrockoCom(SimpleHoster):
     FORM_PATTERN = r'<form  method="post" action="([^"]+)">(.*?)</form>'
     FORM_INPUT_PATTERN = r'<input[^>]* name="?([^" ]+)"? value="?([^" ]+)"?[^>]*>'
     
-    NAME_REPLACEMENTS = [(r'<[^>]*>', '')]
+    FILE_NAME_REPLACEMENTS = [(r'<[^>]*>', '')]
 
     def handleFree(self):
         if "You need Premium membership to download this file." in self.html:
diff --git a/module/plugins/hoster/CzshareCom.py b/module/plugins/hoster/CzshareCom.py
index 158fb0d1d..0ef9c267c 100644
--- a/module/plugins/hoster/CzshareCom.py
+++ b/module/plugins/hoster/CzshareCom.py
@@ -45,21 +45,20 @@ class CzshareCom(SimpleHoster):
     __name__ = "CzshareCom"
     __type__ = "hoster"
     __pattern__ = r"http://(\w*\.)*czshare\.(com|cz)/(\d+/|download.php\?).*"
-    __version__ = "0.85"
+    __version__ = "0.86"
     __description__ = """CZshare.com"""
     __author_name__ = ("zoidberg")
 
-    SIZE_REPLACEMENTS = [(',', '.'), (' ', '')]
+    FILE_NAME_PATTERN = r'<div class="tab" id="parameters">\s*<p>\s*Cel. n.zev: <a href=[^>]*>(?P<N>[^<]+)</a>'
+    FILE_SIZE_PATTERN = r'<div class="tab" id="category">(?:\s*<p>[^\n]*</p>)*\s*Velikost:\s*(?P<S>[0-9., ]+)(?P<U>[kKMG])i?B\s*</div>'
+    FILE_OFFLINE_PATTERN = r'<div class="header clearfix">\s*<h2 class="red">'
+    FILE_SIZE_REPLACEMENTS = [(' ', '')]
+
     FREE_URL_PATTERN = r'<a href="([^"]+)" class="page-download">[^>]*alt="([^"]+)" /></a>'
     FREE_FORM_PATTERN = r'<form action="download.php" method="post">\s*<img src="captcha.php" id="captcha" />(.*?)</form>'
     PREMIUM_FORM_PATTERN = r'<form action="/profi_down.php" method="post">(.*?)</form>'
     FORM_INPUT_PATTERN = r'<input[^>]* name="([^"]+)" value="([^"]+)"[^>]*/>'
-    #FILE_OFFLINE_PATTERN = r'<h2 class="red">[^<]*[Ss]oubor (nenalezen|expiroval|je po.kozen)[^<]*<span>&nbsp;</span></h2>'
-    FILE_OFFLINE_PATTERN = r'<div class="header clearfix">\s*<h2 class="red">'
     MULTIDL_PATTERN = r"<p><font color='red'>Z[^<]*PROFI.</font></p>"
-    #FILE_NAME_PATTERN = r'<h1>([^<]+)<span>&nbsp;</span></h1>'
-    FILE_NAME_PATTERN = r'<div class="tab" id="parameters">\s*<p>\s*Cel. n.zev: <a href=[^>]*>(?P<N>[^<]+)</a>'
-    FILE_SIZE_PATTERN = r'<div class="tab" id="category">(?:\s*<p>[^\n]*</p>)*\s*Velikost:\s*(?P<S>[0-9., ]+)(?P<U>[kKMG])i?B\s*</div>'
     USER_CREDIT_PATTERN = r'<div class="credit">\s*kredit: <strong>([0-9., ]+)([kKMG]i?B)</strong>\s*</div><!-- .credit -->'
 
     def setup(self):
diff --git a/module/plugins/hoster/FourSharedCom.py b/module/plugins/hoster/FourSharedCom.py
index 5d10204a7..b1cc252e2 100644
--- a/module/plugins/hoster/FourSharedCom.py
+++ b/module/plugins/hoster/FourSharedCom.py
@@ -8,7 +8,7 @@ class FourSharedCom(SimpleHoster):
     __name__ = "FourSharedCom"
     __type__ = "hoster"
     __pattern__ = r"http://[\w\.]*?4shared(-china)?\.com/(account/)?(download|get|file|document|photo|video|audio)/.+?/.*"
-    __version__ = "0.23"
+    __version__ = "0.24"
     __description__ = """4Shared Download Hoster"""
     __author_name__ = ("jeix", "zoidberg")
     __author_mail__ = ("jeix@hasnomail.de", "zoidberg@mujmail.cz")
@@ -16,10 +16,10 @@ class FourSharedCom(SimpleHoster):
     FILE_NAME_PATTERN = '<meta name="title" content="(?P<N>[^"]+)" />'
     FILE_SIZE_PATTERN = '<span title="Size: (?P<S>[0-9,.]+) (?P<U>[kKMG])i?B">'
     FILE_OFFLINE_PATTERN = 'The file link that you requested is not valid\.|This file was deleted.'
+    FILE_NAME_REPLACEMENTS = [(r"&#(\d+).", lambda m: unichr(int(m.group(1))))]
+    
     DOWNLOAD_BUTTON_PATTERN = '<a href="([^"]+)"\s*class="dbtn'
     DOWNLOAD_URL_PATTERN = "<div class=\"(?:dl|xxlarge bold)\">\s*<a href='([^']+)'"
-    
-    NAME_REPLACEMENTS = [(r"&#(\d+).", lambda m: unichr(int(m.group(1))))]
 
     def handleFree(self):
         found = re.search(self.DOWNLOAD_BUTTON_PATTERN, self.html)
diff --git a/module/plugins/hoster/HellspyCz.py b/module/plugins/hoster/HellspyCz.py
index 9a8817c54..a03e2bf21 100644
--- a/module/plugins/hoster/HellspyCz.py
+++ b/module/plugins/hoster/HellspyCz.py
@@ -23,14 +23,14 @@ class HellspyCz(SimpleHoster):
     __name__ = "HellspyCz"
     __type__ = "hoster"
     __pattern__ = r"http://(?:\w*\.)*hellspy\.(?:cz|com|sk|hu)(/\S+/\d+)/?.*"
-    __version__ = "0.23"
+    __version__ = "0.24"
     __description__ = """HellSpy.cz"""
     __author_name__ = ("zoidberg")
     __author_mail__ = ("zoidberg@mujmail.cz")
 
     FILE_INFO_PATTERN = '<span class="filesize right">(?P<S>[0-9.]+) <span>(?P<U>[kKMG])i?B</span></span>\s*<h1>(?P<N>[^<]+)</h1>'
     FILE_OFFLINE_PATTERN = r'<h2>(404 - Page|File) not found</h2>'
-    URL_REPLACEMENTS = [(r"http://(?:\w*\.)*hellspy\.(?:cz|com|sk|hu)(/\S+/\d+)/?.*", r"http://www.hellspy.com\1")]
+    FILE_URL_REPLACEMENTS = [(r"http://(?:\w*\.)*hellspy\.(?:cz|com|sk|hu)(/\S+/\d+)/?.*", r"http://www.hellspy.com\1")]
     
     CREDIT_LEFT_PATTERN = r'<strong>Credits: </strong>\s*(\d+)'
     DOWNLOAD_AGAIN_PATTERN = r'<a id="button-download-start"[^>]*title="You can download the file without deducting your credit.">'
diff --git a/module/plugins/hoster/IfolderRu.py b/module/plugins/hoster/IfolderRu.py
index 3177271c4..83b98ecc9 100644
--- a/module/plugins/hoster/IfolderRu.py
+++ b/module/plugins/hoster/IfolderRu.py
@@ -30,9 +30,11 @@ class IfolderRu(SimpleHoster):
     __author_name__ = ("zoidberg")
     __author_mail__ = ("zoidberg@mujmail.cz")
 
-    SIZE_UNITS = {u'Кб': 1, u'Мб': 2, u'Гб': 3}
+    FILE_SIZE_REPLACEMENTS = [(u'Кб', 'KB'), (u'Мб', 'MB'), (u'Гб', 'GB')]
     FILE_NAME_PATTERN = ur'(?:<div><span>)?Название:(?:</span>)? <b>(?P<N>[^<]+)</b><(?:/div|br)>'
-    FILE_SIZE_PATTERN = ur'(?:<div><span>)?Размер:(?:</span>)? <b>(?P<S>[0-9.]+) (?P<U>[^<]+)</b><(?:/div|br)>'
+    FILE_SIZE_PATTERN = ur'(?:<div><span>)?Размер:(?:</span>)? <b>(?P<S>[^<]+)</b><(?:/div|br)>'
+    FILE_OFFLINE_PATTERN = ur'<p>Файл номер <b>[^<]*</b> не найден !!!</p>'
+    
     SESSION_ID_PATTERN = r'<a href=(http://ints.ifolder.ru/ints/sponsor/\?bi=\d*&session=([^&]+)&u=[^>]+)>'
     FORM1_PATTERN = r'<form method=post name="form1" ID="Form1" style="margin-bottom:200px">(.*?)</form>'
     FORM_INPUT_PATTERN = r'<input[^>]* name="?([^" ]+)"? value="?([^" ]+)"?[^>]*>'
@@ -40,7 +42,6 @@ class IfolderRu(SimpleHoster):
     HIDDEN_INPUT_PATTERN = r"var v = .*?name='([^']+)' value='1'"
     DOWNLOAD_LINK_PATTERN = r'<a id="download_file_href" href="([^"]+)"'
     WRONG_CAPTCHA_PATTERN = ur'<font color=Red>неверный код,<br>введите еще раз</font><br>'
-    FILE_OFFLINE_PATTERN = ur'<p>Файл номер <b>[^<]*</b> не найден !!!</p>'
 
     def setup(self):
         self.resumeDownload = self.multiDL = True if self.account else False
diff --git a/module/plugins/hoster/UploadboxCom.py b/module/plugins/hoster/UploadboxCom.py
index 584c64e77..0eb023cb2 100644
--- a/module/plugins/hoster/UploadboxCom.py
+++ b/module/plugins/hoster/UploadboxCom.py
@@ -31,7 +31,7 @@ class UploadboxCom(SimpleHoster):
     __name__ = "Uploadbox"
     __type__ = "hoster"
     __pattern__ = r"http://(?:www\.)?uploadbox\.com/files/([^/]+).*"
-    __version__ = "0.03"
+    __version__ = "0.04"
     __description__ = """UploadBox.com plugin - free only"""
     __author_name__ = ("zoidberg")
     __author_mail__ = ("zoidberg@mujmail.cz")
@@ -39,7 +39,7 @@ class UploadboxCom(SimpleHoster):
     FILE_NAME_PATTERN = r'<p><span>File name:</span>\s*(?P<N>[^<]+)</p>'
     FILE_SIZE_PATTERN = r'<span>Size:</span>\s*(?P<S>[0-9.]+) (?P<U>[kKMG])i?B <span>'
     FILE_OFFLINE_PATTERN = r'<strong>File deleted from service</strong>'
-    NAME_REPLACEMENTS = [(r"(.*)", lambda m: unicode(m.group(1), 'koi8_r'))]
+    FILE_NAME_REPLACEMENTS = [(r"(.*)", lambda m: unicode(m.group(1), 'koi8_r'))]
     
     FREE_FORM_PATTERN = r'<form action="([^"]+)" method="post" id="free" name="free">(.*?)</form>'
     FORM_INPUT_PATTERN = r'<input[^>]* name="([^"]+)" value="([^"]+)" />'
diff --git a/module/plugins/hoster/UploadedTo.py b/module/plugins/hoster/UploadedTo.py
index 174c386a8..39483cf86 100644
--- a/module/plugins/hoster/UploadedTo.py
+++ b/module/plugins/hoster/UploadedTo.py
@@ -59,7 +59,7 @@ def parseFileInfo(self, url = '', html = ''):
         found = re.search(self.FILE_INFO_PATTERN, html)
         if found:
             name, fileid = html_unescape(found.group('N')), found.group('ID')
-            size = parseFileSize(found.group('S'), found.group('U'))
+            size = parseFileSize(found.group('S'))
             status = 2
 
     return name, size, status, fileid
@@ -84,12 +84,12 @@ class UploadedTo(Hoster):
     __name__ = "UploadedTo"
     __type__ = "hoster"
     __pattern__ = r"(http://[\w\.-]*?uploaded\.to/.*?(file/|\?id=|&id=)[\w]+/?)|(http://[\w\.]*?ul\.to/(\?id=|&id=)?[\w\-]+/.+)|(http://[\w\.]*?ul\.to/(\?id=|&id=)?[\w\-]+/?)"
-    __version__ = "0.52"
+    __version__ = "0.53"
     __description__ = """Uploaded.to Download Hoster"""
     __author_name__ = ("spoob", "mkaay")
     __author_mail__ = ("spoob@pyload.org", "mkaay@mkaay.de")
 
-    FILE_INFO_PATTERN = r'<a href="file/(?P<ID>\w+)" id="filename">(?P<N>[^<]+)</a> &nbsp;\s*<small[^>]*>(?P<S>[0-9,]+) (?P<U>[KMG])B</small>'
+    FILE_INFO_PATTERN = r'<a href="file/(?P<ID>\w+)" id="filename">(?P<N>[^<]+)</a> &nbsp;\s*<small[^>]*>(?P<S>[^<]+)</small>'
     FILE_OFFLINE_PATTERN = r'<small class="cL">Error: 404</small>'
 
     def setup(self):
diff --git a/module/plugins/internal/SimpleHoster.py b/module/plugins/internal/SimpleHoster.py
index c101cbf6d..4a03ec60a 100644
--- a/module/plugins/internal/SimpleHoster.py
+++ b/module/plugins/internal/SimpleHoster.py
@@ -20,7 +20,7 @@ from urlparse import urlparse
 from re import search, sub
 
 from module.plugins.Hoster import Hoster
-from module.utils import html_unescape
+from module.utils import html_unescape, parseFileSize
 from module.network.RequestFactory import getURL
 
 def reSub(string, ruleslist):
@@ -31,40 +31,37 @@ def reSub(string, ruleslist):
 
 def parseFileInfo(self, url = '', html = ''):     
     if not html and hasattr(self, "html"): html = self.html
-    name, size, status, found = '', 0, 3, 0
+    info = {"name" : url, "size" : 0, "status" : 3}
     
     if hasattr(self, "FILE_OFFLINE_PATTERN") and search(self.FILE_OFFLINE_PATTERN, html):
         # File offline
-        status = 1
-    elif hasattr(self, "FILE_INFO_PATTERN"):
-        found = search(self.FILE_INFO_PATTERN, html)
-        if found:
-            name, size, units = found.group('N'), found.group('S'), found.group('U')
+        info['status'] = 1
     else:
-        if hasattr(self, "FILE_NAME_PATTERN"):
-            found = search(self.FILE_NAME_PATTERN, html)
-            if found:
-                name = found.group('N')
-        
-        if hasattr(self, "FILE_SIZE_PATTERN"):
-            found = search(self.FILE_SIZE_PATTERN, html)    
-            if found:
-                size, units = found.group('S'), found.group('U')
+        for pattern in ("FILE_INFO_PATTERN", "FILE_NAME_PATTERN", "FILE_SIZE_PATTERN"):
+            try:
+                info = dict(info, **search(getattr(self, pattern), html).groupdict())
+            except AttributeError:
+                continue
                 
-    if size:
+    if len(info) > 3:
         # File online, return name and size
-        size = float(reSub(size, self.SIZE_REPLACEMENTS)) * 1024 ** self.SIZE_UNITS[units]
-        status = 2
+        info['status'] = 2
+        if 'N' in info: info['name'] = reSub(info['N'], self.FILE_NAME_REPLACEMENTS)
+        if 'S' in info:
+            size = info['S'] + info['U'] if 'U' in info else info['S']
+            print repr(size)
+            size = parseFileSize(reSub(size, self.FILE_SIZE_REPLACEMENTS))
+            print repr(self.FILE_SIZE_REPLACEMENTS), repr(size)
+            info['size'] = size
     
-    name = reSub(name, self.NAME_REPLACEMENTS) if name else url
-                    
-    return name, size, status, url
+    print info              
+    return info
 
 def create_getInfo(plugin):
     def getInfo(urls):
         for url in urls:
-            file_info = parseFileInfo(plugin, url, getURL(reSub(url, plugin.URL_REPLACEMENTS), decode=True))
-            yield file_info
+            file_info = parseFileInfo(plugin, url, getURL(reSub(url, plugin.FILE_URL_REPLACEMENTS), decode=True))
+            yield file_info['name'], file_info['size'], file_info['status'], url
     return getInfo
 
 class PluginParseError(Exception):
@@ -90,20 +87,18 @@ class SimpleHoster(Hoster):
     FILE_OFFLINE_PATTERN = r'File (deleted|not found)'
     TEMP_OFFLINE_PATTERN = r'Server maintainance'
     """
-    #TODO: could be replaced when using utils.parseFileSize ?
-    #some plugins need to override these
-    SIZE_UNITS = {'k': 1, 'K': 1, 'M': 2, 'G': 3}
-    SIZE_REPLACEMENTS = [(',', ''), (' ', '')]
-    NAME_REPLACEMENTS = []
-    URL_REPLACEMENTS = []
+
+    FILE_SIZE_REPLACEMENTS = []
+    FILE_NAME_REPLACEMENTS = []
+    FILE_URL_REPLACEMENTS = []
 
     def setup(self):
         self.resumeDownload = self.multiDL = True if self.account else False   
 
     def process(self, pyfile):
-        pyfile.url = reSub(pyfile.url, self.URL_REPLACEMENTS) 
+        pyfile.url = reSub(pyfile.url, self.FILE_URL_REPLACEMENTS) 
         self.html = self.load(pyfile.url, decode = True)
-        self.getFileInfo()    
+        self.file_info = self.getFileInfo()    
         if self.account:
             self.handlePremium()
         else:
@@ -114,17 +109,25 @@ class SimpleHoster(Hoster):
         if hasattr(self, "TEMP_OFFLINE_PATTERN") and search(self.TEMP_OFFLINE_PATTERN, html):
             self.tempOffline()
           
-        name, size, status, url = parseFileInfo(self)           
-        if status == 1: 
+        file_info = parseFileInfo(self)           
+        if file_info['status'] == 1: 
             self.offline()
-        elif status != 2: 
+        elif file_info['status'] != 2:
+            self.logDebug(file_info) 
             self.parseError('File info')
             
-        if not name:
-            name = html_unescape(urlparse(self.pyfile.url).path.split("/")[-1])
+        if file_info['name']:
+            self.pyfile.name = file_info['name']
+        else:
+            self.pyfile.name = html_unescape(urlparse(self.pyfile.url).path.split("/")[-1])
+            
+        if file_info['size']:
+            self.pyfile.size = file_info['size']
+        else:
+            self.logError("File size not parsed")
 
-        self.logDebug("FILE NAME: %s FILE SIZE: %s" % (name, size))        
-        self.pyfile.name, self.pyfile.size = name, size
+        self.logDebug("FILE NAME: %s FILE SIZE: %s" % (self.pyfile.name, self.pyfile.size))
+        return file_info               
     
     def handleFree(self):
         self.fail("Free download not implemented")