summaryrefslogtreecommitdiffstats
path: root/pyload/plugin/extractor
diff options
context:
space:
mode:
Diffstat (limited to 'pyload/plugin/extractor')
-rw-r--r--pyload/plugin/extractor/SevenZip.py155
-rw-r--r--pyload/plugin/extractor/UnRar.py248
-rw-r--r--pyload/plugin/extractor/UnZip.py68
-rw-r--r--pyload/plugin/extractor/__init__.py1
4 files changed, 472 insertions, 0 deletions
diff --git a/pyload/plugin/extractor/SevenZip.py b/pyload/plugin/extractor/SevenZip.py
new file mode 100644
index 000000000..4b23c6ff2
--- /dev/null
+++ b/pyload/plugin/extractor/SevenZip.py
@@ -0,0 +1,155 @@
+# -*- coding: utf-8 -*-
+
+import os
+import re
+
+from subprocess import Popen, PIPE
+
+from pyload.plugin.internal.UnRar import ArchiveError, CRCError, PasswordError, UnRar, renice
+from pyload.utils import fs_encode, safe_join
+
+
+class SevenZip(UnRar):
+ __name__ = "SevenZip"
+ __version__ = "0.08"
+
+ __description__ = """7-Zip extractor plugin"""
+ __license__ = "GPLv3"
+ __authors__ = [("Michael Nowak", ""),
+ ("Walter Purcaro", "vuolter@gmail.com")]
+
+
+ CMD = "7z"
+ VERSION = ""
+
+ EXTENSIONS = [".7z", ".xz", ".zip", ".gz", ".gzip", ".tgz", ".bz2", ".bzip2",
+ ".tbz2", ".tbz", ".tar", ".wim", ".swm", ".lzma", ".rar", ".cab",
+ ".arj", ".z", ".taz", ".cpio", ".rpm", ".deb", ".lzh", ".lha",
+ ".chm", ".chw", ".hxs", ".iso", ".msi", ".doc", ".xls", ".ppt",
+ ".dmg", ".xar", ".hfs", ".exe", ".ntfs", ".fat", ".vhd", ".mbr",
+ ".squashfs", ".cramfs", ".scap"]
+
+
+ #@NOTE: there are some more uncovered 7z formats
+ re_filelist = re.compile(r'([\d\:]+)\s+([\d\:]+)\s+([\w\.]+)\s+(\d+)\s+(\d+)\s+(.+)')
+ re_wrongpwd = re.compile(r'(Can not open encrypted archive|Wrong password)', re.I)
+ re_wrongcrc = re.compile(r'Encrypted\s+\=\s+\+', re.I)
+ re_version = re.compile(r'7-Zip\s(?:\[64\]\s)?(\d+\.\d+)', re.I)
+
+
+ @classmethod
+ def isUsable(cls):
+ if os.name == "nt":
+ cls.CMD = os.path.join(pypath, "7z.exe")
+ p = Popen([cls.CMD], stdout=PIPE, stderr=PIPE)
+ out,err = p.communicate()
+ else:
+ p = Popen([cls.CMD], stdout=PIPE, stderr=PIPE)
+ out, err = p.communicate()
+
+ cls.VERSION = cls.re_version.search(out).group(1)
+
+ return True
+
+
+ def check(self):
+ file = fs_encode(self.filename)
+
+ p = self.call_cmd("t", file)
+ out, err = p.communicate()
+
+ if p.returncode > 1:
+ raise CRCError(err)
+
+ p = self.call_cmd("l", "-slt", file)
+ out, err = p.communicate()
+
+ if p.returncode > 1:
+ raise ArchiveError(_("Process return code: %d") % p.returncode)
+
+ # check if output or error macthes the 'wrong password'-Regexp
+ if self.re_wrongpwd.search(out):
+ raise PasswordError
+
+ # check if output matches 'Encrypted = +'
+ if self.re_wrongcrc.search(out):
+ raise CRCError(_("Header protected"))
+
+
+ def isPassword(self, password):
+ p = self.call_cmd("l", fs_encode(self.filename), password=password)
+ p.communicate()
+ return p.returncode == 0
+
+
+ def repair(self):
+ return False
+
+
+ def extract(self, password=None):
+ command = "x" if self.fullpath else "e"
+
+ p = self.call_cmd(command, '-o' + self.out, fs_encode(self.filename), password=password)
+
+ renice(p.pid, self.renice)
+
+ # communicate and retrieve stderr
+ self._progress(p)
+ err = p.stderr.read().strip()
+
+ if err:
+ if self.re_wrongpwd.search(err):
+ raise PasswordError
+
+ elif self.re_wrongcrc.search(err):
+ raise CRCError(err)
+
+ else: #: raise error if anything is on stderr
+ raise ArchiveError(err)
+
+ if p.returncode > 1:
+ raise ArchiveError(_("Process return code: %d") % p.returncode)
+
+ self.files = self.list(password)
+
+
+ def list(self, password=None):
+ command = "l" if self.fullpath else "l"
+
+ p = self.call_cmd(command, fs_encode(self.filename), password=password)
+ out, err = p.communicate()
+
+ if "Can not open" in err:
+ raise ArchiveError(_("Cannot open file"))
+
+ if p.returncode > 1:
+ raise ArchiveError(_("Process return code: %d") % p.returncode)
+
+ result = set()
+ for groups in self.re_filelist.findall(out):
+ f = groups[-1].strip()
+ result.add(safe_join(self.out, f))
+
+ return list(result)
+
+
+ def call_cmd(self, command, *xargs, **kwargs):
+ args = []
+
+ #overwrite flag
+ if self.overwrite:
+ args.append("-y")
+
+ #set a password
+ if "password" in kwargs and kwargs["password"]:
+ args.append("-p'%s'" % kwargs["password"])
+ else:
+ args.append("-p-")
+
+ #@NOTE: return codes are not reliable, some kind of threading, cleanup whatever issue
+ call = [self.CMD, command] + args + list(xargs)
+
+ self.manager.logDebug(" ".join(call))
+
+ p = Popen(call, stdout=PIPE, stderr=PIPE)
+ return p
diff --git a/pyload/plugin/extractor/UnRar.py b/pyload/plugin/extractor/UnRar.py
new file mode 100644
index 000000000..8a3985678
--- /dev/null
+++ b/pyload/plugin/extractor/UnRar.py
@@ -0,0 +1,248 @@
+# -*- coding: utf-8 -*-
+
+import os
+import re
+
+from glob import glob
+from string import digits
+from subprocess import Popen, PIPE
+
+from pyload.plugin.internal.Extractor import Extractor, ArchiveError, CRCError, PasswordError
+from pyload.utils import decode, fs_encode, safe_join
+
+
+def renice(pid, value):
+ if value and os.name != "nt":
+ try:
+ Popen(["renice", str(value), str(pid)], stdout=PIPE, stderr=PIPE, bufsize=-1)
+
+ except Exception:
+ pass
+
+
+class UnRar(Extractor):
+ __name__ = "UnRar"
+ __version__ = "1.13"
+
+ __description__ = """Rar extractor plugin"""
+ __license__ = "GPLv3"
+ __authors__ = [("RaNaN", "RaNaN@pyload.org"),
+ ("Walter Purcaro", "vuolter@gmail.com"),
+ ("Immenz", "immenz@gmx.net"),]
+
+
+ CMD = "unrar"
+ VERSION = ""
+
+ EXTENSIONS = [".rar"]
+
+
+ re_multipart = re.compile(r'\.(part|r)(\d+)(?:\.rar)?',re.I)
+
+ re_filefixed = re.compile(r'Building (.+)')
+ re_filelist = re.compile(r'^(.)(\s*[\w\.\-]+)\s+(\d+\s+)+(?:\d+\%\s+)?[\d\-]{8}\s+[\d\:]{5}', re.M|re.I)
+
+ re_wrongpwd = re.compile(r'password', re.I)
+ re_wrongcrc = re.compile(r'encrypted|damaged|CRC failed|checksum error', re.I)
+
+ re_version = re.compile(r'UNRAR\s(\d+\.\d+)', re.I)
+
+
+ @classmethod
+ def isUsable(cls):
+ if os.name == "nt":
+ cls.CMD = os.path.join(pypath, "UnRAR.exe")
+ p = Popen([cls.CMD], stdout=PIPE, stderr=PIPE)
+ out, err = p.communicate()
+ else:
+ try:
+ p = Popen([cls.CMD], stdout=PIPE, stderr=PIPE)
+ out, err = p.communicate()
+
+ except OSError: #: fallback to rar
+ cls.CMD = "rar"
+ p = Popen([cls.CMD], stdout=PIPE, stderr=PIPE)
+ out, err = p.communicate()
+
+ cls.VERSION = cls.re_version.search(out).group(1)
+
+ return True
+
+
+ @classmethod
+ def isMultipart(cls,filename):
+ multipart = cls.re_multipart.search(filename)
+ if multipart:
+ # First Multipart file (part1.rar for *.part1-9.rar format or *.rar for .r1-9 format) handled as normal Archive
+ return False if (multipart.group(1) == "part" and int(multipart.group(2)) == 1) else True
+
+ return False
+
+
+ def check(self):
+ p = self.call_cmd("l", "-v", fs_encode(self.filename))
+ out, err = p.communicate()
+
+ if self.re_wrongpwd.search(err):
+ raise PasswordError
+
+ if self.re_wrongcrc.search(err):
+ raise CRCError(err)
+
+ # output only used to check if passworded files are present
+ for attr in self.re_filelist.findall(out):
+ if attr[0].startswith("*"):
+ raise PasswordError
+
+
+ def isPassword(self, password):
+ # at this point we can only verify header protected files
+ p = self.call_cmd("l", "-v", fs_encode(self.filename), password=password)
+ out, err = p.communicate()
+ return False if self.re_wrongpwd.search(err) else True
+
+
+ def repair(self):
+ p = self.call_cmd("rc", fs_encode(self.filename))
+
+ # communicate and retrieve stderr
+ self._progress(p)
+ err = p.stderr.read().strip()
+
+ if err or p.returncode:
+ p = self.call_cmd("r", fs_encode(self.filename))
+
+ # communicate and retrieve stderr
+ self._progress(p)
+ err = p.stderr.read().strip()
+
+ if err or p.returncode:
+ return False
+ else:
+ dir = os.path.dirname(filename)
+ name = re_filefixed.search(out).group(1)
+
+ self.filename = os.path.join(dir, name)
+
+ return True
+
+
+ def _progress(self, process):
+ s = ""
+ while True:
+ c = process.stdout.read(1)
+ # quit loop on eof
+ if not c:
+ break
+ # reading a percentage sign -> set progress and restart
+ if c == '%':
+ self.notifyProgress(int(s))
+ s = ""
+ # not reading a digit -> therefore restart
+ elif c not in digits:
+ s = ""
+ # add digit to progressstring
+ else:
+ s += c
+
+
+ def extract(self, password=None):
+ command = "x" if self.fullpath else "e"
+
+ p = self.call_cmd(command, fs_encode(self.filename), self.out, password=password)
+
+ renice(p.pid, self.renice)
+
+ # communicate and retrieve stderr
+ self._progress(p)
+ err = p.stderr.read().strip()
+
+ if err:
+ if self.re_wrongpwd.search(err):
+ raise PasswordError
+
+ elif self.re_wrongcrc.search(err):
+ raise CRCError(err)
+
+ else: #: raise error if anything is on stderr
+ raise ArchiveError(err)
+
+ if p.returncode:
+ raise ArchiveError(_("Process return code: %d") % p.returncode)
+
+ self.files = self.list(password)
+
+
+ def getDeleteFiles(self):
+ dir, name = os.path.split(self.filename)
+
+ # actually extracted file
+ files = [self.filename]
+
+ # eventually Multipart Files
+ files.extend(safe_join(dir, os.path.basename(file)) for file in filter(self.isMultipart, os.listdir(dir))
+ if re.sub(self.re_multipart,".rar",name) == re.sub(self.re_multipart,".rar",file))
+
+ return files
+
+
+ def list(self, password=None):
+ command = "vb" if self.fullpath else "lb"
+
+ p = self.call_cmd(command, "-v", fs_encode(self.filename), password=password)
+ out, err = p.communicate()
+
+ if "Cannot open" in err:
+ raise ArchiveError(_("Cannot open file"))
+
+ if err.strip(): #: only log error at this point
+ self.manager.logError(err.strip())
+
+ result = set()
+ if not self.fullpath and self.VERSION.startswith('5'):
+ # NOTE: Unrar 5 always list full path
+ for f in decode(out).splitlines():
+ f = safe_join(self.out, os.path.basename(f.strip()))
+ if os.path.isfile(f):
+ result.add(safe_join(self.out, os.path.basename(f)))
+ else:
+ for f in decode(out).splitlines():
+ f = f.strip()
+ result.add(safe_join(self.out, f))
+
+ return list(result)
+
+
+ def call_cmd(self, command, *xargs, **kwargs):
+ args = []
+
+ # overwrite flag
+ if self.overwrite:
+ args.append("-o+")
+ else:
+ args.append("-o-")
+ if self.delete:
+ args.append("-or")
+
+ for word in self.excludefiles:
+ args.append("-x'%s'" % word.strip())
+
+ # assume yes on all queries
+ args.append("-y")
+
+ # set a password
+ if "password" in kwargs and kwargs['password']:
+ args.append("-p%s" % kwargs['password'])
+ else:
+ args.append("-p-")
+
+ if self.keepbroken:
+ args.append("-kb")
+
+ # NOTE: return codes are not reliable, some kind of threading, cleanup whatever issue
+ call = [self.CMD, command] + args + list(xargs)
+
+ self.manager.logDebug(" ".join(call))
+
+ p = Popen(call, stdout=PIPE, stderr=PIPE)
+ return p
diff --git a/pyload/plugin/extractor/UnZip.py b/pyload/plugin/extractor/UnZip.py
new file mode 100644
index 000000000..cb6621f99
--- /dev/null
+++ b/pyload/plugin/extractor/UnZip.py
@@ -0,0 +1,68 @@
+# -*- coding: utf-8 -*-
+
+from __future__ import with_statement
+
+import os
+import sys
+import zipfile
+
+from pyload.plugin.internal.Extractor import Extractor, ArchiveError, CRCError, PasswordError
+from pyload.utils import fs_encode
+
+
+class UnZip(Extractor):
+ __name__ = "UnZip"
+ __version__ = "1.10"
+
+ __description__ = """Zip extractor plugin"""
+ __license__ = "GPLv3"
+ __authors__ = [("Walter Purcaro", "vuolter@gmail.com")]
+
+
+ EXTENSIONS = [".zip", ".zip64"]
+ VERSION ="(python %s.%s.%s)" % (sys.version_info[0], sys.version_info[1], sys.version_info[2])
+
+
+ @classmethod
+ def isUsable(cls):
+ return sys.version_info[:2] >= (2, 6)
+
+
+ def list(self, password=None):
+ with zipfile.ZipFile(fs_encode(self.filename), 'r', allowZip64=True) as z:
+ z.setpassword(password)
+ return z.namelist()
+
+
+ def check(self):
+ with zipfile.ZipFile(fs_encode(self.filename), 'r', allowZip64=True) as z:
+ badfile = z.testzip()
+
+ if badfile:
+ raise CRCError(badfile)
+ else:
+ raise PasswordError
+
+
+ def extract(self, password=None):
+ try:
+ with zipfile.ZipFile(fs_encode(self.filename), 'r', allowZip64=True) as z:
+ z.setpassword(password)
+
+ badfile = z.testzip()
+
+ if badfile:
+ raise CRCError(badfile)
+ else:
+ z.extractall(self.out)
+
+ except (zipfile.BadZipfile, zipfile.LargeZipFile), e:
+ raise ArchiveError(e)
+
+ except RuntimeError, e:
+ if "encrypted" in e:
+ raise PasswordError
+ else:
+ raise ArchiveError(e)
+ else:
+ self.files = z.namelist()
diff --git a/pyload/plugin/extractor/__init__.py b/pyload/plugin/extractor/__init__.py
new file mode 100644
index 000000000..40a96afc6
--- /dev/null
+++ b/pyload/plugin/extractor/__init__.py
@@ -0,0 +1 @@
+# -*- coding: utf-8 -*-
generated by cgit v1.2.3 (git 2.25.1) at 2025-01-24 03:36:05 +0100