This project offers a high level python library to perform code analysis with CPG and Joern server. Several API methods including integration with NetworkX and PyTorch Geometric are offered to perform code analysis and research on complex code bases in a pythonic manner from cli and from notebooks.
pip install joern-lib
# To install the optional science pack, clone this repo and use poetry > 1.5 to install the science group
poetry install --with science # cpu
poetry install --with science-cu117 # cuda 11.7
poetry install --with science-cu118 # cuda 11.8
The repository includes docker compose configuration to interactively query the joern server with polynote notebooks.
Run joern server and polynote locally.
git clone https://github.com/appthreat/joern-lib.git
# Edit docker-compose.yml to set sources directory
docker compose up -d
# podman-compose up --build
Navigate to http://localhost:8192 for an interactive polynote notebook. You could open one of the sample notebooks from the contrib directory to learn about Joern server and this library.
Refer to the API documentation for programmatic usage.
python -m asyncio
Execute single query
from joern_lib import client, workspace, utils
from joern_lib.detectors import common as cpg
connection = await client.get("http://localhost:9000", "http://localhost:7072", "admin", "admin")
# connection = await client.get("http://localhost:9000")
res = await client.q(connection, "val a=1");
# {'response': 'a: Int = 1\n'}
Execute bulk query
res = await client.bulk_query(connection, ["val a=1", "val b=2", "val c=a+b"]);
# [{'response': 'a: Int = 1\n'}, {'response': 'b: Int = 2\n'}, {'response': 'c: Int = 3\n'}]
List workspaces
res = await workspace.ls(connection)
Get workspace path
res = await workspace.get_path(connection)
# /workspace (Response would be parsed)
Check if cpg exists
await workspace.cpg_exists(connection, "NodeGoat")
Import code for analysis
res = await workspace.import_code(connection, "/app", "NodeGoat")
# True
Import an existing CPG for analysis
res = await workspace.import_cpg(connection, "/app/sandbox/crAPI/cpg_out/crAPI-python-cpg.bin.zip", "crAPI-python")
Create a CPG with a remote cpggen server
res = await workspace.create_cpg(connection, "/app/sandbox/crAPI", out_dir="/app/sandbox/crAPI/cpg_out", languages="python", project_name="crAPI-python")
List files
res = await cpg.list_files(connection)
# list of files
Print call tree
res = await cpg.get_call_tree(connection, "com.example.vulnspring.WebController.issue:java.lang.String(org.springframework.ui.Model,java.lang.String)")
utils.print_tree(res)
from joern_lib.detectors import java
List http routes
await java.list_http_routes(connection)
from joern_lib.detectors import js
List http routes
await js.list_http_routes(connection)
Name of the variable containing express()
await js.get_express_appvar(connection)
List of require statements
await js.list_requires(connection)
List of import statements
await js.list_imports(connection)
List of NoSQL DB collection names
await js.list_nosql_collections(connection)
Get HTTP sources
await js.get_http_sources(connection)
await js.get_http_sinks(connection)
Requires TypeScript project
await js.list_aws_modules(connection)
If Joern server stops responding after a while restart docker.
docker compose down
docker compose up -d
Adding asyncio.sleep(0) seems to fix such errors.
# Workaround to fix websockets.exceptions.ConnectionClosedError
await asyncio.sleep(0)
Alternatively, use the sync api.
pygraphviz/graphviz_wrap.c:2711:10: fatal error: graphviz/cgraph.h: No such file or directory
2711 | #include "graphviz/cgraph.h"
| ^~~~~~~~~~~~~~~~~~~
compilation terminated.
error: command '/usr/bin/gcc' failed with exit code 1
Install graphviz-devel
or graphviz-dev
package for your OS. See here