sqlmap/lib/core/dump.py
2023-11-16 11:09:48 +02:00

726 lines
28 KiB
Python

#!/usr/bin/env python
"""
Copyright (c) 2006-2023 sqlmap developers (https://sqlmap.org/)
See the file 'LICENSE' for copying permission
"""
import hashlib
import os
import re
import shutil
import tempfile
import threading
from lib.core.common import Backend
from lib.core.common import checkFile
from lib.core.common import dataToDumpFile
from lib.core.common import dataToJsonFile
from lib.core.common import dataToStdout
from lib.core.common import filterNone
from lib.core.common import getSafeExString
from lib.core.common import isListLike
from lib.core.common import isNoneValue
from lib.core.common import normalizeUnicode
from lib.core.common import openFile
from lib.core.common import prioritySortColumns
from lib.core.common import randomInt
from lib.core.common import safeCSValue
from lib.core.common import unArrayizeValue
from lib.core.common import unsafeSQLIdentificatorNaming
from lib.core.compat import xrange
from lib.core.convert import getBytes
from lib.core.convert import getConsoleLength
from lib.core.convert import getText
from lib.core.convert import getUnicode
from lib.core.convert import htmlEscape
from lib.core.data import conf
from lib.core.data import kb
from lib.core.data import logger
from lib.core.dicts import DUMP_REPLACEMENTS
from lib.core.enums import CONTENT_STATUS
from lib.core.enums import CONTENT_TYPE
from lib.core.enums import DBMS
from lib.core.enums import DUMP_FORMAT
from lib.core.exception import SqlmapGenericException
from lib.core.exception import SqlmapSystemException
from lib.core.exception import SqlmapValueException
from lib.core.replication import Replication
from lib.core.settings import DUMP_FILE_BUFFER_SIZE
from lib.core.settings import HTML_DUMP_CSS_STYLE
from lib.core.settings import IS_WIN
from lib.core.settings import METADB_SUFFIX
from lib.core.settings import MIN_BINARY_DISK_DUMP_SIZE
from lib.core.settings import TRIM_STDOUT_DUMP_SIZE
from lib.core.settings import UNICODE_ENCODING
from lib.core.settings import UNSAFE_DUMP_FILEPATH_REPLACEMENT
from lib.core.settings import VERSION_STRING
from lib.core.settings import WINDOWS_RESERVED_NAMES
from lib.utils.safe2bin import safechardecode
from thirdparty import six
from thirdparty.magic import magic
class Dump(object):
"""
This class defines methods used to parse and output the results
of SQL injection actions
"""
def __init__(self):
self._outputFile = None
self._outputFP = None
self._lock = threading.Lock()
def _write(self, data, newline=True, console=True, content_type=None):
text = "%s%s" % (data, "\n" if newline else " ")
if conf.api:
dataToStdout(data, contentType=content_type, status=CONTENT_STATUS.COMPLETE)
elif console:
dataToStdout(text)
if self._outputFP:
multiThreadMode = kb.multiThreadMode
if multiThreadMode:
self._lock.acquire()
try:
self._outputFP.write(text)
except IOError as ex:
errMsg = "error occurred while writing to log file ('%s')" % getSafeExString(ex)
raise SqlmapGenericException(errMsg)
if multiThreadMode:
self._lock.release()
kb.dataOutputFlag = True
def flush(self):
if self._outputFP:
try:
self._outputFP.flush()
except IOError:
pass
def setOutputFile(self):
if conf.noLogging:
self._outputFP = None
return
self._outputFile = os.path.join(conf.outputPath, "log")
try:
self._outputFP = openFile(self._outputFile, "ab" if not conf.flushSession else "wb")
except IOError as ex:
errMsg = "error occurred while opening log file ('%s')" % getSafeExString(ex)
raise SqlmapGenericException(errMsg)
def singleString(self, data, content_type=None):
self._write(data, content_type=content_type)
def string(self, header, data, content_type=None, sort=True):
if conf.api:
self._write(data, content_type=content_type)
if isListLike(data) and len(data) == 1:
data = unArrayizeValue(data)
if isListLike(data):
self.lister(header, data, content_type, sort)
elif data is not None:
_ = getUnicode(data)
if _.endswith("\r\n"):
_ = _[:-2]
elif _.endswith("\n"):
_ = _[:-1]
if _.strip(' '):
_ = _.strip(' ')
if "\n" in _:
self._write("%s:\n---\n%s\n---" % (header, _))
else:
self._write("%s: %s" % (header, ("'%s'" % _) if isinstance(data, six.string_types) else _))
def json(self, jsonFile, data):
dataToJsonFile(jsonFile, data)
def lister(self, header, elements, content_type=None, sort=True):
if elements and sort:
try:
elements = set(elements)
elements = list(elements)
elements.sort(key=lambda _: _.lower() if hasattr(_, "lower") else _)
except:
pass
if conf.api:
self._write(elements, content_type=content_type)
if elements:
self._write("%s [%d]:" % (header, len(elements)))
for element in elements:
if isinstance(element, six.string_types):
self._write("[*] %s" % element)
elif isListLike(element):
self._write("[*] " + ", ".join(getUnicode(e) for e in element))
if elements:
self._write("")
def banner(self, data):
self.string("banner", data, content_type=CONTENT_TYPE.BANNER)
def currentUser(self, data):
self.string("current user", data, content_type=CONTENT_TYPE.CURRENT_USER)
def currentDb(self, data):
if Backend.getIdentifiedDbms() in (DBMS.ORACLE, DBMS.PGSQL, DBMS.HSQLDB, DBMS.H2, DBMS.MONETDB, DBMS.VERTICA, DBMS.CRATEDB, DBMS.CACHE, DBMS.FRONTBASE):
self.string("current database (equivalent to schema on %s)" % Backend.getIdentifiedDbms(), data, content_type=CONTENT_TYPE.CURRENT_DB)
elif Backend.getIdentifiedDbms() in (DBMS.ALTIBASE, DBMS.DB2, DBMS.MIMERSQL, DBMS.MAXDB, DBMS.VIRTUOSO):
self.string("current database (equivalent to owner on %s)" % Backend.getIdentifiedDbms(), data, content_type=CONTENT_TYPE.CURRENT_DB)
else:
self.string("current database", data, content_type=CONTENT_TYPE.CURRENT_DB)
def hostname(self, data):
self.string("hostname", data, content_type=CONTENT_TYPE.HOSTNAME)
def dba(self, data):
self.string("current user is DBA", data, content_type=CONTENT_TYPE.IS_DBA)
def users(self, users):
self.lister("database management system users", users, content_type=CONTENT_TYPE.USERS)
def statements(self, statements):
self.lister("SQL statements", statements, content_type=CONTENT_TYPE.STATEMENTS)
def userSettings(self, header, userSettings, subHeader, content_type=None):
self._areAdmins = set()
if isinstance(userSettings, (tuple, list, set)):
self._areAdmins = userSettings[1]
userSettings = userSettings[0]
users = [_ for _ in userSettings.keys() if _ is not None]
users.sort(key=lambda _: _.lower() if hasattr(_, "lower") else _)
if conf.api:
self._write(userSettings, content_type=content_type)
if userSettings:
self._write("%s:" % header)
for user in users:
settings = filterNone(userSettings[user])
if isNoneValue(settings):
stringSettings = ""
else:
stringSettings = " [%d]:" % len(settings)
if user in self._areAdmins:
self._write("[*] %s (administrator)%s" % (user, stringSettings))
else:
self._write("[*] %s%s" % (user, stringSettings))
if settings:
settings.sort()
for setting in settings:
self._write(" %s: %s" % (subHeader, setting))
if userSettings:
self.singleString("")
def dbs(self, dbs):
self.lister("available databases", dbs, content_type=CONTENT_TYPE.DBS)
def dbTables(self, dbTables):
if isinstance(dbTables, dict) and len(dbTables) > 0:
if conf.api:
self._write(dbTables, content_type=CONTENT_TYPE.TABLES)
maxlength = 0
for tables in dbTables.values():
for table in tables:
if table and isListLike(table):
table = table[0]
maxlength = max(maxlength, getConsoleLength(unsafeSQLIdentificatorNaming(getUnicode(table))))
lines = "-" * (int(maxlength) + 2)
for db, tables in dbTables.items():
tables = sorted(filter(None, tables))
self._write("Database: %s" % unsafeSQLIdentificatorNaming(db) if db and METADB_SUFFIX not in db else "<current>")
if len(tables) == 1:
self._write("[1 table]")
else:
self._write("[%d tables]" % len(tables))
self._write("+%s+" % lines)
for table in tables:
if table and isListLike(table):
table = table[0]
table = unsafeSQLIdentificatorNaming(table)
blank = " " * (maxlength - getConsoleLength(getUnicode(table)))
self._write("| %s%s |" % (table, blank))
self._write("+%s+\n" % lines)
elif dbTables is None or len(dbTables) == 0:
self.singleString("No tables found", content_type=CONTENT_TYPE.TABLES)
else:
self.string("tables", dbTables, content_type=CONTENT_TYPE.TABLES)
def dbTableColumns(self, tableColumns, content_type=None):
if isinstance(tableColumns, dict) and len(tableColumns) > 0:
if conf.api:
self._write(tableColumns, content_type=content_type)
for db, tables in tableColumns.items():
if not db:
db = "All"
for table, columns in tables.items():
maxlength1 = 0
maxlength2 = 0
colType = None
colList = list(columns.keys())
colList.sort(key=lambda _: _.lower() if hasattr(_, "lower") else _)
for column in colList:
colType = columns[column]
column = unsafeSQLIdentificatorNaming(column)
maxlength1 = max(maxlength1, len(column or ""))
maxlength2 = max(maxlength2, len(colType or ""))
maxlength1 = max(maxlength1, len("COLUMN"))
lines1 = "-" * (maxlength1 + 2)
if colType is not None:
maxlength2 = max(maxlength2, len("TYPE"))
lines2 = "-" * (maxlength2 + 2)
self._write("Database: %s\nTable: %s" % (unsafeSQLIdentificatorNaming(db) if db and METADB_SUFFIX not in db else "<current>", unsafeSQLIdentificatorNaming(table)))
if len(columns) == 1:
self._write("[1 column]")
else:
self._write("[%d columns]" % len(columns))
if colType is not None:
self._write("+%s+%s+" % (lines1, lines2))
else:
self._write("+%s+" % lines1)
blank1 = " " * (maxlength1 - len("COLUMN"))
if colType is not None:
blank2 = " " * (maxlength2 - len("TYPE"))
if colType is not None:
self._write("| Column%s | Type%s |" % (blank1, blank2))
self._write("+%s+%s+" % (lines1, lines2))
else:
self._write("| Column%s |" % blank1)
self._write("+%s+" % lines1)
for column in colList:
colType = columns[column]
column = unsafeSQLIdentificatorNaming(column)
blank1 = " " * (maxlength1 - len(column))
if colType is not None:
blank2 = " " * (maxlength2 - len(colType))
self._write("| %s%s | %s%s |" % (column, blank1, colType, blank2))
else:
self._write("| %s%s |" % (column, blank1))
if colType is not None:
self._write("+%s+%s+\n" % (lines1, lines2))
else:
self._write("+%s+\n" % lines1)
def dbTablesCount(self, dbTables):
if isinstance(dbTables, dict) and len(dbTables) > 0:
if conf.api:
self._write(dbTables, content_type=CONTENT_TYPE.COUNT)
maxlength1 = len("Table")
maxlength2 = len("Entries")
for ctables in dbTables.values():
for tables in ctables.values():
for table in tables:
maxlength1 = max(maxlength1, getConsoleLength(getUnicode(table)))
for db, counts in dbTables.items():
self._write("Database: %s" % unsafeSQLIdentificatorNaming(db) if db and METADB_SUFFIX not in db else "<current>")
lines1 = "-" * (maxlength1 + 2)
blank1 = " " * (maxlength1 - len("Table"))
lines2 = "-" * (maxlength2 + 2)
blank2 = " " * (maxlength2 - len("Entries"))
self._write("+%s+%s+" % (lines1, lines2))
self._write("| Table%s | Entries%s |" % (blank1, blank2))
self._write("+%s+%s+" % (lines1, lines2))
sortedCounts = list(counts.keys())
sortedCounts.sort(reverse=True)
for count in sortedCounts:
tables = counts[count]
if count is None:
count = "Unknown"
tables.sort(key=lambda _: _.lower() if hasattr(_, "lower") else _)
for table in tables:
blank1 = " " * (maxlength1 - getConsoleLength(getUnicode(table)))
blank2 = " " * (maxlength2 - len(str(count)))
self._write("| %s%s | %d%s |" % (table, blank1, count, blank2))
self._write("+%s+%s+\n" % (lines1, lines2))
else:
logger.error("unable to retrieve the number of entries for any table")
def dbTableValues(self, tableValues):
replication = None
rtable = None
dumpFP = None
appendToFile = False
warnFile = False
if tableValues is None:
return
db = tableValues["__infos__"]["db"]
if not db:
db = "All"
table = tableValues["__infos__"]["table"]
if conf.api:
self._write(tableValues, content_type=CONTENT_TYPE.DUMP_TABLE)
try:
dumpDbPath = os.path.join(conf.dumpPath, unsafeSQLIdentificatorNaming(db))
except UnicodeError:
try:
dumpDbPath = os.path.join(conf.dumpPath, normalizeUnicode(unsafeSQLIdentificatorNaming(db)))
except (UnicodeError, OSError):
tempDir = tempfile.mkdtemp(prefix="sqlmapdb")
warnMsg = "currently unable to use regular dump directory. "
warnMsg += "Using temporary directory '%s' instead" % tempDir
logger.warning(warnMsg)
dumpDbPath = tempDir
if conf.dumpFormat == DUMP_FORMAT.SQLITE:
replication = Replication(os.path.join(conf.dumpPath, "%s.sqlite3" % unsafeSQLIdentificatorNaming(db)))
elif conf.dumpFormat in (DUMP_FORMAT.CSV, DUMP_FORMAT.HTML):
if not os.path.isdir(dumpDbPath):
try:
os.makedirs(dumpDbPath)
except:
warnFile = True
_ = re.sub(r"[^\w]", UNSAFE_DUMP_FILEPATH_REPLACEMENT, unsafeSQLIdentificatorNaming(db))
dumpDbPath = os.path.join(conf.dumpPath, "%s-%s" % (_, hashlib.md5(getBytes(db)).hexdigest()[:8]))
if not os.path.isdir(dumpDbPath):
try:
os.makedirs(dumpDbPath)
except Exception as ex:
tempDir = tempfile.mkdtemp(prefix="sqlmapdb")
warnMsg = "unable to create dump directory "
warnMsg += "'%s' (%s). " % (dumpDbPath, getSafeExString(ex))
warnMsg += "Using temporary directory '%s' instead" % tempDir
logger.warning(warnMsg)
dumpDbPath = tempDir
dumpFileName = conf.dumpFile or os.path.join(dumpDbPath, re.sub(r'[\\/]', UNSAFE_DUMP_FILEPATH_REPLACEMENT, "%s.%s" % (unsafeSQLIdentificatorNaming(table), conf.dumpFormat.lower())))
if not checkFile(dumpFileName, False):
try:
openFile(dumpFileName, "w+b").close()
except SqlmapSystemException:
raise
except:
warnFile = True
_ = re.sub(r"[^\w]", UNSAFE_DUMP_FILEPATH_REPLACEMENT, normalizeUnicode(unsafeSQLIdentificatorNaming(table)))
if len(_) < len(table) or IS_WIN and table.upper() in WINDOWS_RESERVED_NAMES:
_ = re.sub(r"[^\w]", UNSAFE_DUMP_FILEPATH_REPLACEMENT, unsafeSQLIdentificatorNaming(table))
dumpFileName = os.path.join(dumpDbPath, "%s-%s.%s" % (_, hashlib.md5(getBytes(table)).hexdigest()[:8], conf.dumpFormat.lower()))
else:
dumpFileName = os.path.join(dumpDbPath, "%s.%s" % (_, conf.dumpFormat.lower()))
else:
appendToFile = any((conf.limitStart, conf.limitStop))
if not appendToFile:
count = 1
while True:
candidate = "%s.%d" % (dumpFileName, count)
if not checkFile(candidate, False):
try:
shutil.copyfile(dumpFileName, candidate)
except IOError:
pass
break
else:
count += 1
dumpFP = openFile(dumpFileName, "wb" if not appendToFile else "ab", buffering=DUMP_FILE_BUFFER_SIZE)
count = int(tableValues["__infos__"]["count"])
separator = str()
field = 1
fields = len(tableValues) - 1
columns = prioritySortColumns(list(tableValues.keys()))
if conf.col:
cols = conf.col.split(',')
columns = sorted(columns, key=lambda _: cols.index(_) if _ in cols else 0)
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" % (unsafeSQLIdentificatorNaming(db) if db and METADB_SUFFIX not in db else "<current>", unsafeSQLIdentificatorNaming(table)))
if conf.dumpFormat == DUMP_FORMAT.SQLITE:
cols = []
for column in columns:
if column != "__infos__":
colType = Replication.INTEGER
for value in tableValues[column]['values']:
try:
if not value or value == " ": # NULL
continue
int(value)
except ValueError:
colType = None
break
if colType is None:
colType = Replication.REAL
for value in tableValues[column]['values']:
try:
if not value or value == " ": # NULL
continue
float(value)
except ValueError:
colType = None
break
cols.append((unsafeSQLIdentificatorNaming(column), colType if colType else Replication.TEXT))
rtable = replication.createTable(table, cols)
elif conf.dumpFormat == DUMP_FORMAT.HTML:
dataToDumpFile(dumpFP, "<!DOCTYPE html>\n<html>\n<head>\n")
dataToDumpFile(dumpFP, "<meta http-equiv=\"Content-type\" content=\"text/html;charset=%s\">\n" % UNICODE_ENCODING)
dataToDumpFile(dumpFP, "<meta name=\"generator\" content=\"%s\" />\n" % VERSION_STRING)
dataToDumpFile(dumpFP, "<title>%s</title>\n" % ("%s%s" % ("%s." % db if METADB_SUFFIX not in db else "", table)))
dataToDumpFile(dumpFP, HTML_DUMP_CSS_STYLE)
dataToDumpFile(dumpFP, "\n</head>\n<body>\n<table>\n<thead>\n<tr>\n")
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]
column = unsafeSQLIdentificatorNaming(column)
maxlength = int(info["length"])
blank = " " * (maxlength - getConsoleLength(column))
self._write("| %s%s" % (column, blank), newline=False)
if not appendToFile:
if conf.dumpFormat == DUMP_FORMAT.CSV:
if field == fields:
dataToDumpFile(dumpFP, "%s" % safeCSValue(column))
else:
dataToDumpFile(dumpFP, "%s%s" % (safeCSValue(column), conf.csvDel))
elif conf.dumpFormat == DUMP_FORMAT.HTML:
dataToDumpFile(dumpFP, "<th>%s</th>" % getUnicode(htmlEscape(column).encode("ascii", "xmlcharrefreplace")))
field += 1
if conf.dumpFormat == DUMP_FORMAT.HTML:
dataToDumpFile(dumpFP, "\n</tr>\n</thead>\n<tbody>\n")
self._write("|\n%s" % separator)
if conf.dumpFormat == DUMP_FORMAT.CSV:
dataToDumpFile(dumpFP, "\n" if not appendToFile else "")
elif conf.dumpFormat == DUMP_FORMAT.SQLITE:
rtable.beginTransaction()
if count > TRIM_STDOUT_DUMP_SIZE:
warnMsg = "console output will be trimmed to "
warnMsg += "last %d rows due to " % TRIM_STDOUT_DUMP_SIZE
warnMsg += "large table size"
logger.warning(warnMsg)
for i in xrange(count):
console = (i >= count - TRIM_STDOUT_DUMP_SIZE)
field = 1
values = []
if conf.dumpFormat == DUMP_FORMAT.HTML:
dataToDumpFile(dumpFP, "<tr>")
for column in columns:
if column != "__infos__":
info = tableValues[column]
if len(info["values"]) <= i:
continue
if info["values"][i] is None:
value = u''
else:
value = getUnicode(info["values"][i])
value = DUMP_REPLACEMENTS.get(value, value)
values.append(value)
maxlength = int(info["length"])
blank = " " * (maxlength - getConsoleLength(value))
self._write("| %s%s" % (value, blank), newline=False, console=console)
if len(value) > MIN_BINARY_DISK_DUMP_SIZE and r'\x' in value:
try:
mimetype = getText(magic.from_buffer(value, mime=True))
if any(mimetype.startswith(_) for _ in ("application", "image")):
if not os.path.isdir(dumpDbPath):
os.makedirs(dumpDbPath)
_ = re.sub(r"[^\w]", UNSAFE_DUMP_FILEPATH_REPLACEMENT, normalizeUnicode(unsafeSQLIdentificatorNaming(column)))
filepath = os.path.join(dumpDbPath, "%s-%d.bin" % (_, randomInt(8)))
warnMsg = "writing binary ('%s') content to file '%s' " % (mimetype, filepath)
logger.warning(warnMsg)
with openFile(filepath, "w+b", None) as f:
_ = safechardecode(value, True)
f.write(_)
except Exception as ex:
logger.debug(getSafeExString(ex))
if conf.dumpFormat == DUMP_FORMAT.CSV:
if field == fields:
dataToDumpFile(dumpFP, "%s" % safeCSValue(value))
else:
dataToDumpFile(dumpFP, "%s%s" % (safeCSValue(value), conf.csvDel))
elif conf.dumpFormat == DUMP_FORMAT.HTML:
dataToDumpFile(dumpFP, "<td>%s</td>" % getUnicode(htmlEscape(value).encode("ascii", "xmlcharrefreplace")))
field += 1
if conf.dumpFormat == DUMP_FORMAT.SQLITE:
try:
rtable.insert(values)
except SqlmapValueException:
pass
elif conf.dumpFormat == DUMP_FORMAT.CSV:
dataToDumpFile(dumpFP, "\n")
elif conf.dumpFormat == DUMP_FORMAT.HTML:
dataToDumpFile(dumpFP, "</tr>\n")
self._write("|", console=console)
self._write("%s\n" % separator)
if conf.dumpFormat == DUMP_FORMAT.SQLITE:
rtable.endTransaction()
logger.info("table '%s.%s' dumped to SQLITE database '%s'" % (db, table, replication.dbpath))
elif conf.dumpFormat in (DUMP_FORMAT.CSV, DUMP_FORMAT.HTML):
if conf.dumpFormat == DUMP_FORMAT.HTML:
dataToDumpFile(dumpFP, "</tbody>\n</table>\n</body>\n</html>")
else:
dataToDumpFile(dumpFP, "\n")
dumpFP.close()
msg = "table '%s.%s' dumped to %s file '%s'" % (db, table, conf.dumpFormat, dumpFileName)
if not warnFile:
logger.info(msg)
else:
logger.warning(msg)
def dbColumns(self, dbColumnsDict, colConsider, dbs):
if conf.api:
self._write(dbColumnsDict, content_type=CONTENT_TYPE.COLUMNS)
for column in dbColumnsDict.keys():
if colConsider == "1":
colConsiderStr = "s LIKE '%s' were" % unsafeSQLIdentificatorNaming(column)
else:
colConsiderStr = " '%s' was" % unsafeSQLIdentificatorNaming(column)
found = {}
for db, tblData in dbs.items():
for tbl, colData in tblData.items():
for col, dataType in colData.items():
if column.lower() in col.lower():
if db in found:
if tbl in found[db]:
found[db][tbl][col] = dataType
else:
found[db][tbl] = {col: dataType}
else:
found[db] = {}
found[db][tbl] = {col: dataType}
continue
if found:
msg = "column%s found in the " % colConsiderStr
msg += "following databases:"
self._write(msg)
self.dbTableColumns(found)
def sqlQuery(self, query, queryRes):
self.string(query, queryRes, content_type=CONTENT_TYPE.SQL_QUERY)
def rFile(self, fileData):
self.lister("files saved to", fileData, sort=False, content_type=CONTENT_TYPE.FILE_READ)
def registerValue(self, registerData):
self.string("Registry key value data", registerData, content_type=CONTENT_TYPE.REG_READ, sort=False)
# object to manage how to print the retrieved queries output to
# standard output and sessions file
dumper = Dump()