benoit@41: # -*- coding: utf-8 -*-
benoit@41: # Description:
benoit@41: # Methods for parsing CPEs
benoit@41: #
benoit@41: # Authors:
benoit@41: # Benoît Allard <benoit.allard@greenbone.net>
benoit@41: #
benoit@41: # Copyright:
benoit@41: # Copyright (C) 2014 Greenbone Networks GmbH
benoit@41: #
benoit@41: # This program is free software; you can redistribute it and/or
benoit@41: # modify it under the terms of the GNU General Public License
benoit@41: # as published by the Free Software Foundation; either version 2
benoit@41: # of the License, or (at your option) any later version.
benoit@41: #
benoit@41: # This program is distributed in the hope that it will be useful,
benoit@41: # but WITHOUT ANY WARRANTY; without even the implied warranty of
benoit@41: # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
benoit@41: # GNU General Public License for more details.
benoit@41: #
benoit@41: # You should have received a copy of the GNU General Public License
benoit@41: # along with this program; if not, write to the Free Software
benoit@41: # Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
benoit@41: 
benoit@40: """\
benoit@40: a cpe class to ease the creation of a producttree based on cpe
benoit@40: 
benoit@40: This is based on:
benoit@40: 
benoit@40:     NIST Interagency Report 7695
benoit@40:     Common Platform Enumeration: Naming Specification Version 2.3
benoit@40: 
benoit@40: CPE is a trademark of The MITRE Corporation.
benoit@40: 
benoit@40: """
benoit@40: 
benoit@40: import re
benoit@40: 
benoit@40: PCT_MAP ={'!': "%21", '"': "%22", '#': "%23", '$': "%24", '%': "%25", '&': "%26",
benoit@40:           "'": "%27", '(': "%28", ')': "%29", '*': "%2a", '+': "%2b", ',': "%2c",
benoit@40:           '/': "%2f", ':': "%3a", ';': "%3b", '<': "%3c", "=": "%3d", '>': "%3e",
benoit@40:           '?': "%3f", '@': "%40", '[': "%5b", '\\': "%5c","]": "%5d", '^': "%5e",
benoit@40:           '`': "%60", '{': "%7b", '|': "%7c", '}': "%7d", "~": "%7e"}
benoit@40: 
benoit@40: PCT_MAP_i = dict((v, k) for k, v in PCT_MAP.iteritems())
benoit@40: 
benoit@40: def pct_encode(c):
benoit@40:     """ Returns the right percent-encoding of c """
benoit@40:     if c in "-.":
benoit@40:         return c
benoit@40:     return PCT_MAP[c]
benoit@40:     return {'!': "%21", '"': "%22", '#': "%23", '$': "%24", '%': "%25", '&': "%26",
benoit@40:             "'": "%27", '(': "%28", ')': "%29", '*': "%2a", '+': "%2b", ',': "%2c",
benoit@40:             "-": c, '.': c, '/': "%2f", ':': "%3a", ';': "%3b", '<': "%3c",
benoit@40:             "=": "%3d", '>': "%3e", '?': "%3f", '@': "%40", '[': "%5b", '\\': "%5c",
benoit@40:             "]": "%5d", '^': "%5e", '`': "%60", '{': "%7b", '|': "%7c", '}': "%7d",
benoit@40:             "~": "%7e"}[c]
benoit@40: 
benoit@40: def decode(s):
benoit@40:     if s == '':
benoit@40:         return ANY
benoit@40:     if s == '-':
benoit@40:         return NA
benoit@40:     s = s.lower()
benoit@40:     res = ""
benoit@40:     idx = 0
benoit@40:     embedded = False
benoit@40:     while idx < len(s):
benoit@40:         c = s[idx]
benoit@40:         if c in ".-~":
benoit@40:             res += "\\" + c
benoit@40:             embedded = True
benoit@40:         elif c != '%':
benoit@40:             res += c
benoit@40:             embedded = True
benoit@40:         else:
benoit@40:             form = s[idx:idx+3]
benoit@40:             if form == "%01":
benoit@40:                 if (((idx == 0) or (idx == (len(s) - 3))) or
benoit@40:                     ( not embedded and (s[idx-4:idx-1] == "%01")) or
benoit@40:                     (embedded and (len(s) > idx + 6) and (s[idx+3:idx+6] == "%01"))):
benoit@40:                     res += '?'
benoit@40:                 else:
benoit@40:                     raise ValueError
benoit@40:             elif form == "%02":
benoit@40:                 if (idx == 0) or (idx == len(s) - 3):
benoit@40:                     res += '*'
benoit@40:                 else:
benoit@40:                     raise ValueError
benoit@40:             else:
benoit@40:                 res += '\\' + PCT_MAP_i[form]
benoit@40:             embedded = True
benoit@40:             idx += 2
benoit@40:         idx += 1
benoit@40:     return CPEAttribute(res)
benoit@40: 
benoit@40: def unbind_value_fs(s):
benoit@40:     if s == '*':
benoit@40:         return ANY
benoit@40:     if s == '-':
benoit@40:         return NA
benoit@40:     res = ""
benoit@40:     idx = 0
benoit@40:     embedded = False
benoit@40:     while idx < len(s):
benoit@40:         c = s[idx]
benoit@40:         if re.match("[a-zA-Z0-9_]", c) is not None:
benoit@40:             res += c
benoit@40:             embedded = True
benoit@40:         elif c == "\\":
benoit@40:             res += s[idx:idx+2]
benoit@40:             embedded = True
benoit@40:             idx += 1
benoit@40:         elif c == "*":
benoit@40:             if (idx == 0) or (idx == (len(s) - 1)):
benoit@40:                 res += c
benoit@40:                 embedded = True
benoit@40:             else:
benoit@40:                 raise ValueError
benoit@40:         elif c == "?":
benoit@40:             if (((idx == 0) or (idx == (len(s) - 1))) or
benoit@40:                 (not embedded and (s[idx - 1] == "?")) or
benoit@40:                 (embedded and (s[idx + 1] == "?"))):
benoit@40:                 res += c
benoit@40:                 embedded = False
benoit@40:             else:
benoit@40:                 raise ValueError
benoit@40:         else:
benoit@40:             res += "\\" + c
benoit@40:             embedded = True
benoit@40:         idx += 1
benoit@40:     return CPEAttribute(res)
benoit@40: 
benoit@40: class CPEAttribute(object):
benoit@40:     """ We need a special class to deal with ANY / NA / "string" """
benoit@40: 
benoit@40:     def __init__(self, value=None, any=False, na=False):
benoit@40:         self.any = any
benoit@40:         self.na = na
benoit@40:         self.value = value
benoit@40: 
benoit@40:     def bind_for_URI(self):
benoit@40: #        print self.any, self.na, self.value
benoit@40:         if self.any:
benoit@40:             return ""
benoit@40:         if self.na:
benoit@40:             return '-'
benoit@40:         return self.transform_for_uri()
benoit@40: 
benoit@40:     def transform_for_uri(self):
benoit@40:         res = ""
benoit@40:         idx = 0
benoit@40:         while idx < len(self.value):
benoit@40:             c = self.value[idx]
benoit@40:             if re.match("[a-zA-Z0-9_]", c) is not None:
benoit@40:                 res += c
benoit@40:             elif c == '\\':
benoit@40:                 idx += 1
benoit@40:                 c = self.value[idx]
benoit@40:                 res += pct_encode(c)
benoit@40:             elif c == '?':
benoit@40:                 res += "%01"
benoit@40:             elif c == '*':
benoit@40:                 res += "%02"
benoit@40:             idx += 1
benoit@40:         return res
benoit@40: 
benoit@40:     def bind_for_fs(self):
benoit@40:         if self.any:
benoit@40:             return "*"
benoit@40:         if self.na:
benoit@40:             return "-"
benoit@40:         return self.process_quoted_chars()
benoit@40: 
benoit@40:     def process_quoted_chars(self):
benoit@40:         res = ""
benoit@40:         idx = 0
benoit@40:         while idx < len(self.value):
benoit@40:             c = self.value[idx]
benoit@40:             if c != '\\':
benoit@40:                 res += c
benoit@40:             else:
benoit@40:                 idx += 1
benoit@40:                 c = self.value[idx]
benoit@40:                 if c in ".-_":
benoit@40:                     res += c
benoit@40:                 else:
benoit@40:                     res += '\\' + c
benoit@40:             idx += 1
benoit@40:         return res
benoit@40: 
benoit@40: ANY = CPEAttribute(any=True)
benoit@40: NA = CPEAttribute(na=True)
benoit@40: 
benoit@40: class CPE(object):
benoit@40: 
benoit@40:     def __init__(self, part=None, vendor=None, product=None, version=None, update=None, edition=None, language=None, sw_edition=None, target_sw=None, target_hw=None, other=None):
benoit@40:         self.part = part or CPEAttribute(any=True)
benoit@40:         self.vendor = vendor or CPEAttribute(any=True)
benoit@40:         self.product = product or CPEAttribute(any=True)
benoit@40:         self.version = version or CPEAttribute(any=True)
benoit@40:         self.update = update or CPEAttribute(any=True)
benoit@40:         self.edition = edition or CPEAttribute(any=True)
benoit@40:         self.language = language or CPEAttribute(any=True)
benoit@40:         # Extended attributes:
benoit@40:         self.sw_edition = sw_edition or CPEAttribute(any=True)
benoit@40:         self.target_sw = target_sw or CPEAttribute(any=True)
benoit@40:         self.target_hw = target_hw or CPEAttribute(any=True)
benoit@40:         self.other = other or CPEAttribute(any=True)
benoit@40: 
benoit@40:     def bind_to_URI(self):
benoit@40:         uri = 'cpe:/'
benoit@40:         uri += ':'.join(a.bind_for_URI() for a in (self.part, self.vendor, self.product, self.version, self.update))
benoit@40:         # Special handling for edition
benoit@40:         ed = self.edition.bind_for_URI()
benoit@40:         sw_ed = self.sw_edition.bind_for_URI()
benoit@40:         t_sw = self.target_sw.bind_for_URI()
benoit@40:         t_hw = self.target_hw.bind_for_URI()
benoit@40:         oth = self.other.bind_for_URI()
benoit@40:         if sw_ed == "" and t_sw == "" and t_hw == "" and oth == "":
benoit@40:             uri += ":" + ed
benoit@40:         else:
benoit@40:             uri += ":~" + '~'.join([ed, sw_ed, t_sw, t_hw, oth])
benoit@40:         uri += ':' + self.language.bind_for_URI()
benoit@40:         return uri.rstrip(':')
benoit@40: 
benoit@40:     def unbind_URI(self, uri):
benoit@40:         for idx, comp in enumerate(uri.split(':')):
benoit@40:             if idx == 0:
benoit@40:                 continue
benoit@40:             elif idx == 1:
benoit@40:                 self.part = decode(comp[1:])
benoit@40:             elif idx == 2:
benoit@40:                 self.vendor = decode(comp)
benoit@40:             elif idx == 3:
benoit@40:                 self.product = decode(comp)
benoit@40:             elif idx == 4:
benoit@40:                 self.version = decode(comp)
benoit@40:             elif idx == 5:
benoit@40:                 self.update = decode(comp)
benoit@40:             elif idx == 6:
benoit@40:                 if comp == "" or comp[0] != '~':
benoit@40:                     self.edition = decode(comp)
benoit@40:                 else:
benoit@40:                     ed, sw_ed, t_sw, t_hw, oth = comp[1:].split('~')
benoit@40:                     self.edition = decode(ed)
benoit@40:                     self.sw_edition = decode(sw_ed)
benoit@40:                     self.target_sw = decode(t_sw)
benoit@40:                     self.target_hw = decode(t_hw)
benoit@40:                     self.other = decode(oth)
benoit@40:             elif idx == 7:
benoit@40:                 self.language = decode(comp)
benoit@40: 
benoit@40:     def bind_to_fs(self):
benoit@40:         fs = 'cpe:2.3:'
benoit@40:         fs += ':'.join(a.bind_for_fs() for a in (self.part, self.vendor, self.product, self.version, self.update, self.edition, self.language, self.sw_edition, self.target_sw, self.target_hw, self.other))
benoit@40:         return fs
benoit@40: 
benoit@40:     def unbind_fs(self, fs):
benoit@40:         for idx, v in enumerate(fs.split(':')):
benoit@40:             v = unbind_value_fs(v)
benoit@40:             if idx == 2:
benoit@40:                 self.part = v
benoit@40:             elif idx == 3:
benoit@40:                 self.vendor = v
benoit@40:             elif idx == 4:
benoit@40:                 self.product = v
benoit@40:             elif idx == 5:
benoit@40:                 self.version = v
benoit@40:             elif idx == 6:
benoit@40:                 self.update = v
benoit@40:             elif idx == 7:
benoit@40:                 self.edition = v
benoit@40:             elif idx == 8:
benoit@40:                 self.language = v
benoit@40:             elif idx == 9:
benoit@40:                 self.sw_edition = v
benoit@40:             elif idx == 10:
benoit@40:                 self.target_sw = v
benoit@40:             elif idx == 11:
benoit@40:                 self.target_hw = v
benoit@40:             elif idx == 12:
benoit@40:                 self.other = v
benoit@41: 
benoit@41: def parse(s):
benoit@41:     cpe = CPE()
benoit@41:     if s[:5] == 'cpe:/':
benoit@41:         cpe.unbind_URI(s)
benoit@41:     elif s[:8] == 'cpe:2.3:':
benoit@41:         cpe.unbind_fs(s)
benoit@41:     else:
benoit@41:         raise ValueError(s)
benoit@41:     return cpe