update from fork#1
Open
harm1999 wants to merge 36 commits into
Open
Conversation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
ScriptDescription, StructureDescription, ComponentDescription are now Pydantic BaseModels. Tool dicts for the legacy tool-calling path are auto-generated via model_json_schema(), eliminating the risk of schema drift. TOOL_MODELS maps tool name → model class for the structured-output client path. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Primary path uses OpenAI json_schema strict mode and validates responses with Pydantic model_validate_json(). Falls back to the legacy tool-calling path if the provider does not support structured outputs (controlled by [llm] use_structured_output). Both paths retry up to 3 times with exponential backoff (2s/4s/8s) before giving up. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
load_checkpoint() reads a JSONL checkpoint file and applies node property updates and edges back into a graph, so a resumed run skips already-processed nodes via the existing per-node description guards in processors. configure_writer() sets a stable checkpoint path and open mode: 'w' for fresh runs (overwrite), 'a' for resume runs (append). Writer calls are now lock-protected for concurrent processor access. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…g context Execution model: NodeExecutor submits all nodes to a ThreadPoolExecutor immediately. Each worker waits only on the futures of its direct dependencies before making its LLM call, giving finer-grained parallelism than level-by-level topological batching. Workers is configurable via [llm] workers (default 8). Override/overload context: ScriptProcessor collects all methods sharing the same simpleName in the enclosing class (overloads) and ancestor classes (overrides/hides). When present, these are injected into the prompt with an explicit instruction to differentiate implementations while keeping terminology consistent. Sibling-aware class processing: StructureProcessor collects already-processed sibling classes (sharing the same parent) and passes them as context with an instruction to maintain terminological consistency while differentiating responsibilities. Also strengthens the parent-class context instruction and fixes a "Inhertis" typo. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Reads [llm] workers, checkpoint_file, and resume from config; calls configure_writer() so processors share the right checkpoint file/mode. - Calls load_checkpoint() before processing when resume = true. - Builds a type_ancestor_ids map (BFS over specializes edges) and passes it to ScriptProcessor for same-name family / override context lookups. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Adds two new classification dimensions based on the Dragan/Al-Ramadan Stereocode taxonomy (17 method stereotypes across five categories: structural accessors, structural mutators, creational, collaborational, degenerate; and 14 class stereotypes). Each becomes a Dimension node with Category nodes in the graph, and elements receive 'implements' edges to their assigned categories. Method stereotypes are multi-label (e.g. 'get collaborator'); class stereotypes are single-label. Both are gated behind [stereocode] enabled in config so existing runs are unaffected. Pydantic output models gain optional stereocodeStereotype / stereocodeClassStereotype fields (default empty) so they parse cleanly regardless of whether stereocode is enabled. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude refactor
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.