Summary-based information flow analysis for javascript

[nettrino]

Hello all,

I am trying to replicate the steps in the (rather old) javascript doc for information flow analysis. In particular I have done the following:

1. Checked out node-mkdir
2. Created a database from within the root of the repo using codeql database create ~/.local/codeql/databases/js-mkdir --language=javascript --overwrite
3. Run an analysis using codeql query run --database ~/.local/codeql/databases/js-mkdir --threads=8 --ram=10000 -- qlstarter/ql/javascript/ql/src/experimental/Summaries/ExtractSourceSummaries.ql

I cannot see any of the results outlined in the documentation, which has references to older Semmle tooling. How can I achieve equivalent results as described in the docs using the latest codeql?

P.S. I also added some comment directives in the repo's /lib/mkdirp-native.js without much luck:

 const mkdirpNative = (path, opts) => {
   opts.recursive = true
-  const parent = dirname(path)
+  const parent = dirname(
+    path /* Semmle: sink: taint, TaintedPath */
+  )

I do see how the intent of the doc is to examine/analyze applications actually using mkdirp and not the package itself, however I would expect there would be source summaries just by analysing mkdirp as a standalone pkg?

[nettrino]

I think I understood better what the issue is after going through https://codeql.github.com/docs/codeql-cli/manual/database-analyze/.

The results of the query are generated if I run
codeql database run-queries --threads=8 --ram=10000 -v -- ~/.local/codeql/databases/js-mkdir qlstarter/ql/javascript/ql/src/experimental/Summaries/ExtractSourceSummaries.ql, and stored locally in the database folder, however they are still empty if inspected via

codeql bqrs decode ~/.local/codeql/databases/js-mkdir/results/codeql/javascript-queries/experimental/Summaries/ExtractSourceSummaries.bqrs

Am I missing something?

[asgerf]

Hi @nettrino,

Sorry but the document is so far out of date that I'd say it's not worth trying to make it work again. I am going to remove the document and the queries, as it describes a feature that is unsupported and unmaintained. Apologies for any inconvenience this has caused you.

[nettrino]

No problem - and thank you for the reply! I am interested in the functionality overall so if there is a way to achieve the equivalent of extracting such summaries with the current APIs that would be great!