|
| 1 | +/** |
| 2 | + * Provides classes modeling security-relevant aspects of the standard libraries. |
| 3 | + * Note: some modeling is done internally in the dataflow/taint tracking implementation. |
| 4 | + */ |
| 5 | + |
| 6 | +private import python |
| 7 | +private import semmle.python.dataflow.new.DataFlow |
| 8 | +private import semmle.python.dataflow.new.TaintTracking |
| 9 | +private import semmle.python.dataflow.new.RemoteFlowSources |
| 10 | +private import experimental.semmle.python.Concepts |
| 11 | +private import semmle.python.ApiGraphs |
| 12 | + |
| 13 | +private module NoSQL { |
| 14 | + // API Nodes returning `Mongo` instances. |
| 15 | + /** Gets a reference to `pymongo.MongoClient` */ |
| 16 | + private API::Node pyMongo() { |
| 17 | + result = API::moduleImport("pymongo").getMember("MongoClient").getReturn() |
| 18 | + } |
| 19 | + |
| 20 | + /** Gets a reference to `flask_pymongo.PyMongo` */ |
| 21 | + private API::Node flask_PyMongo() { |
| 22 | + result = API::moduleImport("flask_pymongo").getMember("PyMongo").getReturn() |
| 23 | + } |
| 24 | + |
| 25 | + /** Gets a reference to `mongoengine` */ |
| 26 | + private API::Node mongoEngine() { result = API::moduleImport("mongoengine") } |
| 27 | + |
| 28 | + /** Gets a reference to `flask_mongoengine.MongoEngine` */ |
| 29 | + private API::Node flask_MongoEngine() { |
| 30 | + result = API::moduleImport("flask_mongoengine").getMember("MongoEngine").getReturn() |
| 31 | + } |
| 32 | + |
| 33 | + /** |
| 34 | + * Gets a reference to an initialized `Mongo` instance. |
| 35 | + * See `pyMongo()`, `flask_PyMongo()` |
| 36 | + */ |
| 37 | + private API::Node mongoInstance() { |
| 38 | + result = pyMongo() or |
| 39 | + result = flask_PyMongo() |
| 40 | + } |
| 41 | + |
| 42 | + /** |
| 43 | + * Gets a reference to an initialized `Mongo` DB instance. |
| 44 | + * See `mongoEngine()`, `flask_MongoEngine()` |
| 45 | + */ |
| 46 | + private API::Node mongoDBInstance() { |
| 47 | + result = mongoEngine().getMember(["get_db", "connect"]).getReturn() or |
| 48 | + result = mongoEngine().getMember("connection").getMember(["get_db", "connect"]).getReturn() or |
| 49 | + result = flask_MongoEngine().getMember("get_db").getReturn() |
| 50 | + } |
| 51 | + |
| 52 | + /** |
| 53 | + * Gets a reference to a `Mongo` DB use. |
| 54 | + * |
| 55 | + * See `mongoInstance()`, `mongoDBInstance()`. |
| 56 | + */ |
| 57 | + private DataFlow::LocalSourceNode mongoDB(DataFlow::TypeTracker t) { |
| 58 | + t.start() and |
| 59 | + ( |
| 60 | + exists(SubscriptNode subscript | |
| 61 | + subscript.getObject() = mongoInstance().getAUse().asCfgNode() and |
| 62 | + result.asCfgNode() = subscript |
| 63 | + ) |
| 64 | + or |
| 65 | + result.(DataFlow::AttrRead).getObject() = mongoInstance().getAUse() |
| 66 | + or |
| 67 | + result = mongoDBInstance().getAUse() |
| 68 | + ) |
| 69 | + or |
| 70 | + exists(DataFlow::TypeTracker t2 | result = mongoDB(t2).track(t2, t)) |
| 71 | + } |
| 72 | + |
| 73 | + /** |
| 74 | + * Gets a reference to a `Mongo` DB use. |
| 75 | + * |
| 76 | + * ```py |
| 77 | + * from flask_pymongo import PyMongo |
| 78 | + * mongo = PyMongo(app) |
| 79 | + * mongo.db.user.find({'name': safe_search}) |
| 80 | + * ``` |
| 81 | + * |
| 82 | + * `mongo.db` would be a use of a `Mongo` instance, and so the result. |
| 83 | + */ |
| 84 | + private DataFlow::Node mongoDB() { mongoDB(DataFlow::TypeTracker::end()).flowsTo(result) } |
| 85 | + |
| 86 | + /** |
| 87 | + * Gets a reference to a `Mongo` collection use. |
| 88 | + * |
| 89 | + * See `mongoDB()`. |
| 90 | + */ |
| 91 | + private DataFlow::LocalSourceNode mongoCollection(DataFlow::TypeTracker t) { |
| 92 | + t.start() and |
| 93 | + ( |
| 94 | + exists(SubscriptNode subscript | result.asCfgNode() = subscript | |
| 95 | + subscript.getObject() = mongoDB().asCfgNode() |
| 96 | + ) |
| 97 | + or |
| 98 | + result.(DataFlow::AttrRead).getObject() = mongoDB() |
| 99 | + ) |
| 100 | + or |
| 101 | + exists(DataFlow::TypeTracker t2 | result = mongoCollection(t2).track(t2, t)) |
| 102 | + } |
| 103 | + |
| 104 | + /** |
| 105 | + * Gets a reference to a `Mongo` collection use. |
| 106 | + * |
| 107 | + * ```py |
| 108 | + * from flask_pymongo import PyMongo |
| 109 | + * mongo = PyMongo(app) |
| 110 | + * mongo.db.user.find({'name': safe_search}) |
| 111 | + * ``` |
| 112 | + * |
| 113 | + * `mongo.db.user` would be a use of a `Mongo` collection, and so the result. |
| 114 | + */ |
| 115 | + private DataFlow::Node mongoCollection() { |
| 116 | + mongoCollection(DataFlow::TypeTracker::end()).flowsTo(result) |
| 117 | + } |
| 118 | + |
| 119 | + /** This class represents names of find_* relevant `Mongo` collection-level operation methods. */ |
| 120 | + private class MongoCollectionMethodNames extends string { |
| 121 | + MongoCollectionMethodNames() { |
| 122 | + this in [ |
| 123 | + "find", "find_raw_batches", "find_one", "find_one_and_delete", "find_and_modify", |
| 124 | + "find_one_and_replace", "find_one_and_update", "find_one_or_404" |
| 125 | + ] |
| 126 | + } |
| 127 | + } |
| 128 | + |
| 129 | + /** |
| 130 | + * Gets a reference to a `Mongo` collection method. |
| 131 | + * |
| 132 | + * ```py |
| 133 | + * from flask_pymongo import PyMongo |
| 134 | + * mongo = PyMongo(app) |
| 135 | + * mongo.db.user.find({'name': safe_search}) |
| 136 | + * ``` |
| 137 | + * |
| 138 | + * `mongo.db.user.find` would be a collection method, and so the result. |
| 139 | + */ |
| 140 | + private DataFlow::Node mongoCollectionMethod() { |
| 141 | + mongoCollection() = result.(DataFlow::AttrRead).getObject() and |
| 142 | + result.(DataFlow::AttrRead).getAttributeName() instanceof MongoCollectionMethodNames |
| 143 | + } |
| 144 | + |
| 145 | + /** |
| 146 | + * Gets a reference to a `Mongo` collection method call |
| 147 | + * |
| 148 | + * ```py |
| 149 | + * from flask_pymongo import PyMongo |
| 150 | + * mongo = PyMongo(app) |
| 151 | + * mongo.db.user.find({'name': safe_search}) |
| 152 | + * ``` |
| 153 | + * |
| 154 | + * `mongo.db.user.find({'name': safe_search})` would be a collection method call, and so the result. |
| 155 | + */ |
| 156 | + private class MongoCollectionCall extends DataFlow::CallCfgNode, NoSQLQuery::Range { |
| 157 | + MongoCollectionCall() { this.getFunction() = mongoCollectionMethod() } |
| 158 | + |
| 159 | + override DataFlow::Node getQuery() { result = this.getArg(0) } |
| 160 | + } |
| 161 | + |
| 162 | + /** |
| 163 | + * Gets a reference to a call from a class whose base is a reference to `mongoEngine()` or `flask_MongoEngine()`'s |
| 164 | + * `Document` or `EmbeddedDocument` objects and its attribute is `objects`. |
| 165 | + * |
| 166 | + * ```py |
| 167 | + * from flask_mongoengine import MongoEngine |
| 168 | + * db = MongoEngine(app) |
| 169 | + * class Movie(db.Document): |
| 170 | + * title = db.StringField(required=True) |
| 171 | + * |
| 172 | + * Movie.objects(__raw__=json_search) |
| 173 | + * ``` |
| 174 | + * |
| 175 | + * `Movie.objects(__raw__=json_search)` would be the result. |
| 176 | + */ |
| 177 | + private class MongoEngineObjectsCall extends DataFlow::CallCfgNode, NoSQLQuery::Range { |
| 178 | + MongoEngineObjectsCall() { |
| 179 | + this = |
| 180 | + [mongoEngine(), flask_MongoEngine()] |
| 181 | + .getMember(["Document", "EmbeddedDocument"]) |
| 182 | + .getASubclass() |
| 183 | + .getMember("objects") |
| 184 | + .getACall() |
| 185 | + } |
| 186 | + |
| 187 | + override DataFlow::Node getQuery() { result = this.getArgByName(_) } |
| 188 | + } |
| 189 | + |
| 190 | + /** Gets a reference to `mongosanitizer.sanitizer.sanitize` */ |
| 191 | + private class MongoSanitizerCall extends DataFlow::CallCfgNode, NoSQLSanitizer::Range { |
| 192 | + MongoSanitizerCall() { |
| 193 | + this = |
| 194 | + API::moduleImport("mongosanitizer").getMember("sanitizer").getMember("sanitize").getACall() |
| 195 | + } |
| 196 | + |
| 197 | + override DataFlow::Node getAnInput() { result = this.getArg(0) } |
| 198 | + } |
| 199 | + |
| 200 | + /** |
| 201 | + * ObjectId returns a string representing an id. |
| 202 | + * If at any time ObjectId can't parse it's input (like when a tainted dict in passed in), |
| 203 | + * then ObjectId will throw an error preventing the query from running. |
| 204 | + */ |
| 205 | + private class BsonObjectIdCall extends DataFlow::CallCfgNode, NoSQLSanitizer::Range { |
| 206 | + BsonObjectIdCall() { |
| 207 | + this = |
| 208 | + API::moduleImport(["bson", "bson.objectid", "bson.json_util"]) |
| 209 | + .getMember("ObjectId") |
| 210 | + .getACall() |
| 211 | + } |
| 212 | + |
| 213 | + override DataFlow::Node getAnInput() { result = this.getArg(0) } |
| 214 | + } |
| 215 | +} |
0 commit comments