#!/usr/bin/env python """ $Id$ This file is part of the sqlmap project, http://sqlmap.sourceforge.net. Copyright (c) 2007-2009 Bernardo Damele A. G. Copyright (c) 2006 Daniele Bellucci sqlmap is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation version 2 of the License. sqlmap is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with sqlmap; if not, write to the Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA """ import os import re from lib.core.common import dataToDumpFile from lib.core.data import conf from lib.core.data import logger class Dump: """ This class defines methods used to parse and output the results of SQL injection actions """ def __init__(self): self.__outputFile = None self.__outputFP = None def __write(self, data, n=True): if n: print data self.__outputFP.write("%s\n" % data) else: print data, self.__outputFP.write("%s " % data) self.__outputFP.flush() conf.loggedToOut = True def setOutputFile(self): self.__outputFile = "%s%slog" % (conf.outputPath, os.sep) self.__outputFP = open(self.__outputFile, "a") def string(self, header, data, sort=True): if isinstance(data, (list, tuple, set)): self.lister(header, data, sort) return data = str(data) if data: data = data.replace("__NEWLINE__", "\n").replace("__TAB__", "\t") data = data.replace("__START__", "").replace("__STOP__", "") data = data.replace("__DEL__", ", ") if "\n" in data: self.__write("%s:\n---\n%s---\n" % (header, data)) else: self.__write("%s: '%s'\n" % (header, data)) else: self.__write("%s:\tNone\n" % header) def lister(self, header, elements, sort=True): if elements: self.__write("%s [%d]:" % (header, len(elements))) if sort: try: elements = set(elements) elements = list(elements) elements.sort(key=lambda x: x.lower()) except: pass for element in elements: if isinstance(element, str): self.__write("[*] %s" % element) elif isinstance(element, (list, tuple, set)): self.__write("[*] " + ", ".join(e for e in element)) if elements: self.__write("") def userSettings(self, header, userSettings, subHeader): self.__areAdmins = set() if userSettings: self.__write("%s:" % header) if isinstance(userSettings, (tuple, list, set)): self.__areAdmins = userSettings[1] userSettings = userSettings[0] users = userSettings.keys() users.sort(key=lambda x: x.lower()) for user in users: settings = userSettings[user] if user in self.__areAdmins: self.__write("[*] %s (administrator) [%d]:" % (user, len(settings))) else: self.__write("[*] %s [%d]:" % (user, len(settings))) settings.sort() for setting in settings: self.__write(" %s: %s" % (subHeader, setting)) print def dbTables(self, dbTables): if not isinstance(dbTables, dict): self.string("tables", dbTables) return maxlength = 0 for tables in dbTables.values(): for table in tables: maxlength = max(maxlength, len(table)) lines = "-" * (int(maxlength) + 2) for db, tables in dbTables.items(): tables.sort(key=lambda x: x.lower()) self.__write("Database: %s" % db) if len(tables) == 1: self.__write("[1 table]") else: self.__write("[%d tables]" % len(tables)) self.__write("+%s+" % lines) for table in tables: blank = " " * (maxlength - len(table)) self.__write("| %s%s |" % (table, blank)) self.__write("+%s+\n" % lines) def dbTableColumns(self, tableColumns): for db, tables in tableColumns.items(): if not db: db = "All" for table, columns in tables.items(): maxlength1 = 0 maxlength2 = 0 colList = columns.keys() colList.sort(key=lambda x: x.lower()) for column in colList: colType = columns[column] maxlength1 = max(maxlength1, len(column)) maxlength2 = max(maxlength2, len(colType)) maxlength1 = max(maxlength1, len("COLUMN")) maxlength2 = max(maxlength2, len("TYPE")) lines1 = "-" * (int(maxlength1) + 2) lines2 = "-" * (int(maxlength2) + 2) self.__write("Database: %s\nTable: %s" % (db, table)) if len(columns) == 1: self.__write("[1 column]") else: self.__write("[%d columns]" % len(columns)) self.__write("+%s+%s+" % (lines1, lines2)) blank1 = " " * (maxlength1 - len("COLUMN")) blank2 = " " * (maxlength2 - len("TYPE")) self.__write("| Column%s | Type%s |" % (blank1, blank2)) self.__write("+%s+%s+" % (lines1, lines2)) for column in colList: colType = columns[column] blank1 = " " * (maxlength1 - len(column)) blank2 = " " * (maxlength2 - len(colType)) self.__write("| %s%s | %s%s |" % (column, blank1, colType, blank2)) self.__write("+%s+%s+\n" % (lines1, lines2)) def dbTableValues(self, tableValues): db = tableValues["__infos__"]["db"] if not db: db = "All" table = tableValues["__infos__"]["table"] if not conf.multipleTargets: dumpDbPath = "%s%s%s" % (conf.dumpPath, os.sep, db) if not os.path.isdir(dumpDbPath): os.makedirs(dumpDbPath, 0755) dumpFileName = "%s%s%s.csv" % (dumpDbPath, os.sep, table) dumpFP = open(dumpFileName, "w") count = int(tableValues["__infos__"]["count"]) separator = "" field = 1 fields = len(tableValues) - 1 columns = tableValues.keys() columns.sort(key=lambda x: x.lower()) for column in columns: if column != "__infos__": info = tableValues[column] lines = "-" * (int(info["length"]) + 2) separator += "+%s" % lines separator += "+" self.__write("Database: %s\nTable: %s" % (db, table)) if count == 1: self.__write("[1 entry]") else: self.__write("[%d entries]" % count) self.__write(separator) for column in columns: if column != "__infos__": info = tableValues[column] maxlength = int(info["length"]) blank = " " * (maxlength - len(column)) self.__write("| %s%s" % (column, blank), n=False) if not conf.multipleTargets and field == fields: dataToDumpFile(dumpFP, "\"%s\"" % column) elif not conf.multipleTargets: dataToDumpFile(dumpFP, "\"%s\"," % column) field += 1 self.__write("|\n%s" % separator) if not conf.multipleTargets: dataToDumpFile(dumpFP, "\n") for i in range(count): field = 1 for column in columns: if column != "__infos__": info = tableValues[column] value = info["values"][i] if re.search("^[\ *]*$", value): value = "NULL" maxlength = int(info["length"]) blank = " " * (maxlength - len(value)) self.__write("| %s%s" % (value, blank), n=False) if not conf.multipleTargets and field == fields: dataToDumpFile(dumpFP, "\"%s\"" % value) elif not conf.multipleTargets: dataToDumpFile(dumpFP, "\"%s\"," % value) field += 1 self.__write("|") if not conf.multipleTargets: dataToDumpFile(dumpFP, "\n") self.__write("%s\n" % separator) if not conf.multipleTargets: dataToDumpFile(dumpFP, "\n") dumpFP.close() logger.info("Table '%s.%s' dumped to CSV file '%s'" % (db, table, dumpFileName)) # object to manage how to print the retrieved queries output to # standard output and sessions file dumper = Dump()