[feature] Complexity Score Combination#32
Open
seojeongm wants to merge 6 commits into
Open
Conversation
CREATE TABLE IF NOT EXISTS alone won't add the column to existing DBs, so initDb now checks via PRAGMA table_info and issues an ALTER TABLE ADD COLUMN for DBs created before this column existed.
Ten new weights/blends for the upcoming ComplexityScorer, following the same load/upsert pattern already used for alpha/beta/damping so they're user-adjustable via reading_sequence_config without a recompile.
📝 WalkthroughWalkthroughAdds configurable percentile-based file complexity scoring, persists scores in SQLite with schema migration support, integrates scoring into the algorithm runner, and adds normalization, scoring, database, and generated-file handling tests. ChangesComplexity scoring
Estimated code review effort: 4 (Complex) | ~45 minutes Possibly related issues
Possibly related PRs
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/algo/Scoring.cpp`:
- Around line 188-213: Update the SQLite SELECT loop in the scoring function to
capture each sqlite3_step result, process SQLITE_ROW rows, and after the loop
verify the final status is SQLITE_DONE. On any other status, handle and
propagate the read error before calling compute() or writing scores, preventing
partial scoring. Use the surrounding statement-loading function and its existing
error-handling conventions as the integration point.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: 57ecbdc7-c3f0-411c-9032-ba61e1013988
📒 Files selected for processing (9)
CMakeLists.txtsrc/algo/AlgoConfig.hsrc/algo/AlgoDbWriter.cppsrc/algo/AlgoRunner.cppsrc/algo/Scoring.cppsrc/algo/Scoring.hsrc/db/db.cpptests/algo/test_complexity_scorer.cpptests/algo/test_percentile_rank.cpp
Comment on lines
+188
to
+213
| while (sqlite3_step(stmt) == SQLITE_ROW) { | ||
| const char* fid = reinterpret_cast<const char*>(sqlite3_column_text(stmt, 0)); | ||
| if (!fid) continue; | ||
|
|
||
| ComplexityFileMetrics m; | ||
| m.file_id = fid; | ||
| m.max_cyclomatic_complexity = sqlite3_column_int(stmt, 1); | ||
| m.avg_cyclomatic_complexity = sqlite3_column_double(stmt, 2); | ||
| m.max_block_depth = sqlite3_column_int(stmt, 3); | ||
| m.avg_block_depth = sqlite3_column_double(stmt, 4); | ||
| m.logical_loc = sqlite3_column_int(stmt, 5); | ||
|
|
||
| // Isolated files with no CALLS/INHERITS/IMPORTS edges still get a node | ||
| // (GraphBuilder pre-populates every file_id), so a missing lookup here | ||
| // means the file wasn't in the file table when the graph was built -- | ||
| // fall back to 0/0 rather than dropping the file from scoring. | ||
| auto it = file_graph.id_to_node.find(m.file_id); | ||
| if (it != file_graph.id_to_node.end()) { | ||
| NodeId nid = it->second; | ||
| m.inbound = static_cast<int>(file_graph.radj[nid].size()); | ||
| m.outbound = static_cast<int>(file_graph.adj[nid].size()); | ||
| } | ||
|
|
||
| inputs.push_back(std::move(m)); | ||
| } | ||
| sqlite3_finalize(stmt); |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick win
SELECT loop swallows step errors, risking partial scoring.
The loop exits on any non-SQLITE_ROW return, so a mid-iteration error (e.g. SQLITE_BUSY, SQLITE_ERROR) is indistinguishable from SQLITE_DONE. Execution then proceeds to compute() and writes scores computed over a truncated population — silently corrupting the heatmap. Capture and check the final step code.
🛡️ Proposed fix to propagate read errors
std::vector<ComplexityFileMetrics> inputs;
- while (sqlite3_step(stmt) == SQLITE_ROW) {
+ int stepRc;
+ while ((stepRc = sqlite3_step(stmt)) == SQLITE_ROW) {
const char* fid = reinterpret_cast<const char*>(sqlite3_column_text(stmt, 0));
if (!fid) continue;
@@
inputs.push_back(std::move(m));
}
sqlite3_finalize(stmt);
+ if (stepRc != SQLITE_DONE) return stepRc;📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| while (sqlite3_step(stmt) == SQLITE_ROW) { | |
| const char* fid = reinterpret_cast<const char*>(sqlite3_column_text(stmt, 0)); | |
| if (!fid) continue; | |
| ComplexityFileMetrics m; | |
| m.file_id = fid; | |
| m.max_cyclomatic_complexity = sqlite3_column_int(stmt, 1); | |
| m.avg_cyclomatic_complexity = sqlite3_column_double(stmt, 2); | |
| m.max_block_depth = sqlite3_column_int(stmt, 3); | |
| m.avg_block_depth = sqlite3_column_double(stmt, 4); | |
| m.logical_loc = sqlite3_column_int(stmt, 5); | |
| // Isolated files with no CALLS/INHERITS/IMPORTS edges still get a node | |
| // (GraphBuilder pre-populates every file_id), so a missing lookup here | |
| // means the file wasn't in the file table when the graph was built -- | |
| // fall back to 0/0 rather than dropping the file from scoring. | |
| auto it = file_graph.id_to_node.find(m.file_id); | |
| if (it != file_graph.id_to_node.end()) { | |
| NodeId nid = it->second; | |
| m.inbound = static_cast<int>(file_graph.radj[nid].size()); | |
| m.outbound = static_cast<int>(file_graph.adj[nid].size()); | |
| } | |
| inputs.push_back(std::move(m)); | |
| } | |
| sqlite3_finalize(stmt); | |
| int stepRc; | |
| while ((stepRc = sqlite3_step(stmt)) == SQLITE_ROW) { | |
| const char* fid = reinterpret_cast<const char*>(sqlite3_column_text(stmt, 0)); | |
| if (!fid) continue; | |
| ComplexityFileMetrics m; | |
| m.file_id = fid; | |
| m.max_cyclomatic_complexity = sqlite3_column_int(stmt, 1); | |
| m.avg_cyclomatic_complexity = sqlite3_column_double(stmt, 2); | |
| m.max_block_depth = sqlite3_column_int(stmt, 3); | |
| m.avg_block_depth = sqlite3_column_double(stmt, 4); | |
| m.logical_loc = sqlite3_column_int(stmt, 5); | |
| // Isolated files with no CALLS/INHERITS/IMPORTS edges still get a node | |
| // (GraphBuilder pre-populates every file_id), so a missing lookup here | |
| // means the file wasn't in the file table when the graph was built -- | |
| // fall back to 0/0 rather than dropping the file from scoring. | |
| auto it = file_graph.id_to_node.find(m.file_id); | |
| if (it != file_graph.id_to_node.end()) { | |
| NodeId nid = it->second; | |
| m.inbound = static_cast<int>(file_graph.radj[nid].size()); | |
| m.outbound = static_cast<int>(file_graph.adj[nid].size()); | |
| } | |
| inputs.push_back(std::move(m)); | |
| } | |
| sqlite3_finalize(stmt); | |
| if (stepRc != SQLITE_DONE) return stepRc; |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/algo/Scoring.cpp` around lines 188 - 213, Update the SQLite SELECT loop
in the scoring function to capture each sqlite3_step result, process SQLITE_ROW
rows, and after the loop verify the final status is SQLITE_DONE. On any other
status, handle and propagate the read error before calling compute() or writing
scores, preventing partial scoring. Use the surrounding statement-loading
function and its existing error-handling conventions as the integration point.
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
1 participant
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.
Summary
Implements #31 — combines the per-file complexity signals from #30 (cyclomatic complexity, block nesting depth, logical LOC) with inbound/outbound link counts into a single
complexity_scoreper file, using percentile-rank normalization instead of min-max so a single outlier file doesn't compress everyone else's score toward 0.complexity_score REAL NOT NULL DEFAULT 0.0to thefiletable, with a migration path for existing DBs (CREATE TABLE IF NOT EXISTSalone won't touch already-created tables).percentile_rank_normalizealongside the existingminmax_normalizeinScoring.cpp— PageRank/BC's min-max path is untouched.ComplexityScorer(compute+computeAndWrite), following the same pure-compute/DB-IO split already used byPageRank/AlgoDbWriter.AlgoConfigkeys (5 top-level weights, 2 pre-blend pairs, 1 include-generated flag), read/written viareading_sequence_configthe same wayalpha/beta/dampingalready are — no recompile needed to retune.ComplexityScorer::computeAndWriteintoAlgoRunner::run, soTelescodeAlgo <db_path>computes it end-to-end alongside the existing reading-sequence pipeline.Design notes
rank(i) = #{values strictly less than v[i]} / (N-1)), so it only cares about relative ordering.0.8·max + 0.2·avgblend before percentile-ranking (configurable);logical_loc, inbound, and outbound feed in unblended.WHERE is_generated = 0); includable viacomplexity_include_generated. Excluded files stay at the column default (0.0) — this is unambiguous sinceis_generatedis already its own column callers can filter on.Graph.radj/Graph.adjat scoring time (same patternPageRank::computealready uses forout_deg), not persisted asfiletable columns — avoids a second source of truth that would need invalidation on incremental updates.Commits
feat(db): add complexity_score column to file table with migrationfeat(algo): add complexity scoring config keys to AlgoConfigfeat(algo): add percentile_rank_normalize helper with unit testsfeat(algo): add ComplexityScorer::compute (blend, percentile-rank, weighted sum)feat(algo): add ComplexityScorer::computeAndWrite (DB read/write + graph join)feat(algo): wire ComplexityScorer into AlgoRunner::runTest plan
Unit tests (15 new cases: percentile-rank edge cases, blend/weighted-sum correctness against a hand-computed fixture, weight auto-normalization, DB round-trip via
computeAndWrite):Schema migration (verifies existing DBs get the column added without losing data):
Config seeding (verifies all 10 new keys get seeded with correct defaults on first run):
cmake --build build --target TelescodeScanner TelescodeAlgo --config Debug ./build/Debug/TelescodeScanner.exe repos/sherlock cfg_test.db ./build/Debug/TelescodeAlgo.exe cfg_test.db sqlite3 cfg_test.db "SELECT config_key, config_value FROM reading_sequence_config WHERE config_key LIKE 'complexity_%' ORDER BY config_key;" rm cfg_test.dbEnd-to-end (real repo, sanity-checks the resulting distribution):
./build/Debug/TelescodeScanner.exe repos/sherlock e2e_test.db ./build/Debug/TelescodeAlgo.exe e2e_test.db sqlite3 e2e_test.db "SELECT file_id, complexity_score FROM file WHERE is_generated=0 ORDER BY complexity_score DESC LIMIT 10;" rm e2e_test.dbExpect the largest/most-complex file in the repo (sherlock_project/sherlock.py, 714 loc, CC 48) at the top of the list with a score near the high end (~0.95), scores spread across the range rather than clustered.
Out of scope
Color mapping / heatmap visualization (separate issue)
UI for weight adjustment (separate issue — config keys defined here are what it will read/write)
Per-function complexity score (file-level only)