entrygraph 0.1.7__tar.gz → 0.1.9__tar.gz

{entrygraph-0.1.7 → entrygraph-0.1.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: entrygraph
-Version: 0.1.7
+Version: 0.1.9
 Summary: Language-agnostic code graph: query symbols, entrypoints, and source-to-sink call paths from a SQLite index
 Project-URL: Repository, https://github.com/brettbergin/entrygraph
 Author-email: Brett Bergin <brettberginbc@yahoo.com>

{entrygraph-0.1.7 → entrygraph-0.1.9}/src/entrygraph/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '0.1.7'
-__version_tuple__ = version_tuple = (0, 1, 7)
+__version__ = version = '0.1.9'
+__version_tuple__ = version_tuple = (0, 1, 9)
 
 __commit_id__ = commit_id = None

{entrygraph-0.1.7 → entrygraph-0.1.9}/src/entrygraph/extract/javascript.py

@@ -238,12 +238,77 @@ class JavaScriptExtractor:
                         )
         for node in caps.get("export.from", []):
             self._reexport(node, ctx, out)
+        for node in caps.get("require", []):
+            self._commonjs_require(node, ctx, out)
 
         for imp in out.imports:
             out.framework_signals.append(
                 ("import", imp.module.split(".")[0] if "." in imp.module else imp.module)
             )
 
+    def _commonjs_require(self, call, ctx, out) -> None:
+        """CommonJS `require('m')` bindings -> RawImport, mirroring ESM handling.
+
+        ``const cp = require('m')``          -> whole module bound to ``cp``
+        ``const { x, y: z } = require('m')`` -> named imports ``x`` and ``y`` (as ``z``)
+        bare / chained / reassigned          -> module-only (side-effect) import,
+        which still yields the framework signal even when no alias can be bound.
+        """
+        args = call.child_by_field_name("arguments")
+        spec = None
+        if args is not None:
+            for child in args.named_children:
+                if child.type == "string":
+                    spec = node_text(child).strip("'\"`")
+                    break
+        if not spec:
+            return  # dynamic require (variable/template arg) — target unknowable
+        module = self._resolve_module(spec, ctx)
+        span = span_of(call)
+
+        parent = call.parent
+        name_node = None
+        if parent is not None and parent.type == "variable_declarator":
+            name_node = parent.child_by_field_name("name")
+        elif parent is not None and parent.type == "assignment_expression":
+            name_node = parent.child_by_field_name("left")
+
+        if name_node is not None and name_node.type == "identifier":
+            # Bind the alias to the module itself (same as an ESM default import),
+            # so `const cp = require('child_process'); cp.exec()` canonicalizes to
+            # child_process.exec.
+            out.imports.append(
+                RawImport(module=module, imported_name=None, alias=node_text(name_node), span=span)
+            )
+            return
+        if name_node is not None and name_node.type == "object_pattern":
+            for child in name_node.named_children:
+                if child.type == "shorthand_property_identifier_pattern":  # { exec }
+                    nm = node_text(child)
+                    out.imports.append(
+                        RawImport(module=module, imported_name=nm, alias=nm, span=span)
+                    )
+                elif child.type == "pair_pattern":  # { exec: run }
+                    key = child.child_by_field_name("key")
+                    value = child.child_by_field_name("value")
+                    if key is None:
+                        continue
+                    alias = (
+                        node_text(value)
+                        if value is not None and value.type == "identifier"
+                        else node_text(key)
+                    )
+                    out.imports.append(
+                        RawImport(
+                            module=module, imported_name=node_text(key), alias=alias, span=span
+                        )
+                    )
+            return
+        # bare `require('m')`, `require('m').foo`, `module.exports = require('m')`, etc.
+        out.imports.append(
+            RawImport(module=module, imported_name=None, alias=module.split(".")[-1], span=span)
+        )
+
     def _reexport(self, node, ctx, out) -> None:
         """`export { X as Y } from "./m"` / `export * from "./m"` -> RawReexport."""
         source = node.child_by_field_name("source")

{entrygraph-0.1.7 → entrygraph-0.1.9}/src/entrygraph/pipeline/scanner.py

@@ -432,7 +432,9 @@ def _write_edges_and_entrypoints(
     entrypoint_rows: list[dict] = []
 
     for path, x, is_package in extractions:
-        resolver = FileResolver(x, module_ids[path], table, externals, is_package)
+        resolver = FileResolver(
+            x, module_ids[path], table, externals, is_package, sink_registry=sink_registry
+        )
         file_id = file_id_by_path[path]
         for edge in resolver.resolve():
             is_call = edge.kind is EdgeKind.CALLS

{entrygraph-0.1.7 → entrygraph-0.1.9}/src/entrygraph/resolve/resolver.py

@@ -17,6 +17,7 @@ incremental re-resolution can heal them when the target appears.
 from __future__ import annotations
 
 from dataclasses import dataclass
+from typing import TYPE_CHECKING
 
 from entrygraph.extract.ir import FileExtraction, RawReference
 from entrygraph.kinds import Confidence, EdgeKind, SymbolKind
@@ -29,6 +30,9 @@ from entrygraph.resolve.hierarchy import (
 )
 from entrygraph.resolve.symbol_table import SymbolTable
 
+if TYPE_CHECKING:
+    from entrygraph.detect.taint import SinkRegistry
+
 # expand_relative re-exported for backward compatibility with callers/tests.
 __all__ = ["FileResolver", "ResolvedEdge", "expand_relative"]
 
@@ -66,12 +70,14 @@ class FileResolver:
         table: SymbolTable,
         externals: ExternalRegistry,
         is_package: bool = False,
+        sink_registry: SinkRegistry | None = None,
     ) -> None:
         self.x = extraction
         self.module_symbol_id = module_symbol_id
         self.table = table
         self.externals = externals
         self.is_package = is_package
+        self.sink_registry = sink_registry
         self.prefix = LANG_PREFIX.get(extraction.language, extraction.language)
         self.import_map, self.wildcard_modules = build_import_map(extraction, is_package)
 
@@ -217,6 +223,14 @@ class FileResolver:
                 return self.table.by_fqn[target]
         return self.table.unique_by_name(name, (SymbolKind.FUNCTION, SymbolKind.METHOD))
 
+    def _guess_is_sink(self, ref: RawReference) -> bool:
+        """Whether the receiver-agnostic guess for this attribute call matches a
+        registered sink — checked with the same call the index-time stamper uses."""
+        if self.sink_registry is None:
+            return False
+        guess = f"{self.prefix}:*.{ref.callee_name}"
+        return self.sink_registry.match(guess, ref.arg_preview) is not None
+
     def _bind(self, ref: RawReference) -> tuple[int | None, str, Confidence, str | None]:
         # 1. import-map expansion (chase re-export chains for project targets)
         first_seg = ref.callee_text.split(".", 1)[0].split("(", 1)[0]
@@ -270,14 +284,21 @@ class FileResolver:
         # where runner is a local variable) this is the only way to recover the
         # call-graph edge without local type inference — it's the documented
         # FUZZY tradeoff.
-        kinds = (
-            (SymbolKind.METHOD,)
-            if ref.receiver_text is not None
-            else (SymbolKind.FUNCTION, SymbolKind.CLASS)
-        )
-        dst_id = self.table.unique_by_name(ref.callee_name, kinds)
-        if dst_id is not None:
-            return dst_id, self.table.qname_of[dst_id], Confidence.FUZZY, None
+        #
+        # But if the receiver-agnostic guess (`py:*.execute`) is itself a
+        # registered sink, do NOT fuzzy-bind: rewriting dst_qname to a unique
+        # project method named `execute` would erase the sink stamp and hide the
+        # vulnerability. Prefer the guess so the sink survives (step 5); CHA still
+        # adds the project method as a via="cha" candidate, so no edge is lost.
+        if not (ref.receiver_text is not None and self._guess_is_sink(ref)):
+            kinds = (
+                (SymbolKind.METHOD,)
+                if ref.receiver_text is not None
+                else (SymbolKind.FUNCTION, SymbolKind.CLASS)
+            )
+            dst_id = self.table.unique_by_name(ref.callee_name, kinds)
+            if dst_id is not None:
+                return dst_id, self.table.qname_of[dst_id], Confidence.FUZZY, None
 
         # 5. unresolved: language-prefixed guess, still a real node for sinks
         if ref.receiver_text is not None:

entrygraph-0.1.9/tests/fixtures/javascript/commonjs_app/server.js

@@ -0,0 +1,15 @@
+const cp = require('child_process');
+const { exec } = require('child_process');
+const express = require('express');
+
+function runDefault(input) {
+  cp.execSync(input);
+}
+
+function runNamed(input) {
+  exec(input);
+}
+
+const app = express();
+app.get('/a', (req, res) => runDefault(req.query.cmd));
+app.get('/b', (req, res) => runNamed(req.query.cmd));

entrygraph-0.1.9/tests/fixtures/python/fuzzy_sink/app.py

@@ -0,0 +1,14 @@
+import sqlite3
+
+
+class Runner:
+    # the ONLY project method named "execute" — the collision that used to
+    # swallow cursor.execute via unique-name fuzzy binding
+    def execute(self):
+        return 1
+
+
+def query(db_path, user_id):
+    conn = sqlite3.connect(db_path)
+    cur = conn.cursor()
+    cur.execute("SELECT * FROM users WHERE id = " + user_id)

entrygraph-0.1.9/tests/test_commonjs.py

@@ -0,0 +1,41 @@
+"""End-to-end CommonJS `require()` support (Phase 1.3).
+
+Regression: the JS extractor dropped the `require()` captures entirely, so a
+pure-CommonJS app produced no imports — no framework detection and no
+canonicalization of `cp.execSync` to a `child_process` sink.
+"""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+import pytest
+
+from entrygraph import CodeGraph
+
+COMMONJS_APP = Path(__file__).parent / "fixtures" / "javascript" / "commonjs_app"
+
+
+@pytest.fixture(scope="module")
+def graph(tmp_path_factory) -> CodeGraph:
+    db = tmp_path_factory.mktemp("db") / "graph.db"
+    g = CodeGraph.index(COMMONJS_APP, db=db)
+    yield g
+    g.close()
+
+
+def test_commonjs_require_enables_framework_detection(graph):
+    names = {f.name for f in graph.detect().frameworks}
+    assert "express" in names
+
+
+def test_commonjs_require_canonicalizes_command_sinks(graph):
+    # both the default-bound (cp.execSync) and destructured (exec) forms must
+    # resolve to child_process command_exec sinks
+    assert graph.stats().sink_edges >= 2
+    sinks = {
+        row["dst_qname"]
+        for row in graph.sql("SELECT dst_qname FROM edges WHERE sink_id IS NOT NULL")
+    }
+    assert "js:child_process.execSync" in sinks
+    assert "js:child_process.exec" in sinks

{entrygraph-0.1.7 → entrygraph-0.1.9}/tests/test_extract_javascript.py

@@ -72,6 +72,32 @@ def test_relative_import_resolution():
     assert any(i.module == "db.conn" for i in x.imports)
 
 
+def test_commonjs_require_bindings():
+    x = extract(
+        "const cp = require('child_process');\n"
+        "const { exec, spawn: launch } = require('child_process');\n"
+        "const express = require('express');\n"
+        "require('./side-effect');\n"
+        "const rel = require('../db/conn');\n",
+        path="src/routes/users.js",
+    )
+    imports = {(i.module, i.imported_name, i.alias) for i in x.imports}
+    assert ("child_process", None, "cp") in imports  # default -> module binding
+    assert ("child_process", "exec", "exec") in imports  # destructured
+    assert ("child_process", "spawn", "launch") in imports  # destructured + rename
+    assert ("express", None, "express") in imports
+    assert ("db.conn", None, "rel") in imports  # relative resolved to dotted, bound to `rel`
+    # side-effect require still registers the module (framework signal source)
+    assert any(m == "routes.side-effect" and n is None for m, n, _ in imports)
+
+
+def test_commonjs_require_emits_framework_signal():
+    # regression: CommonJS `require('express')` used to yield no import at all,
+    # so express was undetectable in pure-CJS apps.
+    x = extract("const express = require('express');\n")
+    assert ("import", "express") in x.framework_signals
+
+
 def test_calls_with_receivers():
     x = extract(
         "import cp from 'child_process';\nfunction run() { cp.execSync('ls'); helper(); }\n"

entrygraph-0.1.9/tests/test_fuzzy_sink.py

@@ -0,0 +1,36 @@
+"""Sink preservation vs unique-name fuzzy binding (Phase 1.4).
+
+Regression: an unknown-receiver call like `cursor.execute(sql)` fuzzy-bound to a
+unique project method named `execute`, rewriting dst_qname so the `py:*.execute`
+SQL sink never got stamped — the vulnerability silently vanished from the graph.
+"""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+import pytest
+
+from entrygraph import CodeGraph
+
+FUZZY_SINK_APP = Path(__file__).parent / "fixtures" / "python" / "fuzzy_sink"
+
+
+@pytest.fixture(scope="module")
+def graph(tmp_path_factory) -> CodeGraph:
+    db = tmp_path_factory.mktemp("db") / "graph.db"
+    g = CodeGraph.index(FUZZY_SINK_APP, db=db)
+    yield g
+    g.close()
+
+
+def test_sql_sink_survives_unique_name_collision(graph):
+    sinks = graph.sql("SELECT dst_qname, sink_id FROM edges WHERE sink_id IS NOT NULL")
+    by_qname = {r["dst_qname"]: r["sink_id"] for r in sinks}
+    assert by_qname.get("py:*.execute") == "py.sql-execute"
+
+
+def test_sql_path_reachable_despite_collision(graph):
+    paths = graph.paths(source="app.query", sink_category="sql", include_unresolved=True)
+    assert paths
+    assert paths[0].symbols[-1].qname == "py:*.execute"

{entrygraph-0.1.7 → entrygraph-0.1.9}/tests/test_resolver.py

@@ -20,7 +20,7 @@ def make_table() -> SymbolTable:
     return table
 
 
-def make_resolver(table, refs=(), imports=(), module="app.routes", module_id=2):
+def make_resolver(table, refs=(), imports=(), module="app.routes", module_id=2, sink_registry=None):
     x = FileExtraction(
         path="app/routes.py",
         language="python",
@@ -31,7 +31,7 @@ def make_resolver(table, refs=(), imports=(), module="app.routes", module_id=2):
         references=list(refs),
     )
     externals = ExternalRegistry(iter(range(100, 200)).__next__)
-    return FileResolver(x, module_id, table, externals), externals
+    return FileResolver(x, module_id, table, externals, sink_registry=sink_registry), externals
 
 
 def ref(callee_text, callee_name=None, receiver=None, caller=None, kind="call"):
@@ -180,6 +180,36 @@ def test_unresolved_gets_prefixed_placeholder():
     assert by_qname["py:eval"].confidence is Confidence.UNRESOLVED
 
 
+def test_sink_named_attribute_call_is_not_fuzzy_bound():
+    # `cursor.execute(...)` guesses to the sink `py:*.execute`. Even though a
+    # unique project method `Runner.execute` (id 12) exists, we must NOT fuzzy-bind
+    # to it — that would rewrite dst_qname and erase the SQL sink. Keep the guess.
+    from entrygraph.detect.taint import builtin_registry
+
+    table = make_table()
+    resolver, _ = make_resolver(
+        table, [ref("cursor.execute", receiver="cursor")], sink_registry=builtin_registry()
+    )
+    call = next(e for e in resolver.resolve() if e.kind is EdgeKind.CALLS and e.via != "cha")
+    assert call.dst_qname == "py:*.execute"
+    assert call.confidence is Confidence.UNRESOLVED
+    assert call.dst_symbol_id != 12
+
+
+def test_non_sink_attribute_call_still_fuzzy_binds_with_registry():
+    # `obj.render` is not a sink, so unique-name fuzzy binding still applies even
+    # with a sink registry present — the fix is targeted to sink names only.
+    from entrygraph.detect.taint import builtin_registry
+
+    table = make_table()
+    resolver, _ = make_resolver(
+        table, [ref("obj.render", receiver="obj")], sink_registry=builtin_registry()
+    )
+    call = next(e for e in resolver.resolve() if e.kind is EdgeKind.CALLS)
+    assert call.dst_symbol_id == 13  # Runner.render
+    assert call.confidence is Confidence.FUZZY
+
+
 def test_relative_import_expansion():
     table = make_table()
     table.add_symbol(40, "app.utils", 40 and "utils", SymbolKind.MODULE)