You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
ENH: Packages and their methods are not present as nodes #62
Still we dont have any connection between the 2 nodes, seemingly because the node is a function defined in a package, which isnt being added to the graph DB.
This issue reports the symptom — a call resolves (use_random → randint) but no edge appears, because the callee lives in a package that was never added to the graph. #53 requests the mechanism that fixes it: an index_dependencies flag on add_code_to_graph that indexes imported packages so those nodes exist to link to.
Keeping #53 as canonical: it is older, has the actionable design, and carries the live thread with claimants.
One distinction worth carrying over, since your example is stdlib (random) while #53 says "all the Python packages": package_resolver._is_stdlib_module currently exists to classify and exclude stdlib, so stdlib and third-party may need different handling. I've noted that on #53.
Also relevant to whoever picks it up: #1528 — add_package_to_graph on a flat single-module package currently resolves to the entire site-packages directory, which would make naive dependency indexing catastrophic. That should be fixed first.
Activity
DhineshPonnarasan commentedon Feb 8, 2026
HI @Shashankss1205 , Bro shall I work on this
Shashankss1205 commentedon Aug 2, 2026
Closing as a duplicate of #53.
This issue reports the symptom — a call resolves (
use_random→randint) but no edge appears, because the callee lives in a package that was never added to the graph. #53 requests the mechanism that fixes it: anindex_dependenciesflag onadd_code_to_graphthat indexes imported packages so those nodes exist to link to.Keeping #53 as canonical: it is older, has the actionable design, and carries the live thread with claimants.
One distinction worth carrying over, since your example is stdlib (
random) while #53 says "all the Python packages":package_resolver._is_stdlib_modulecurrently exists to classify and exclude stdlib, so stdlib and third-party may need different handling. I've noted that on #53.Also relevant to whoever picks it up: #1528 —
add_package_to_graphon a flat single-module package currently resolves to the entiresite-packagesdirectory, which would make naive dependency indexing catastrophic. That should be fixed first.