Command-line reference

srcmetrics <COMMAND>

  analyze   Analyze a file or directory
  stats     Descriptive statistics, per-language baselines and correlations
  report    Self-contained HTML report
  model     Experimental readability model (train / predict)
  serve     HTTP API and web UI

Errors exit with status 1 and a message saying what to do next.

analyze

srcmetrics analyze <PATH> [--project NAME] [-o FILE] [--format json|csv]

• PATH is a file or a directory. Directories are walked respecting .gitignore; only files with supported extensions are analyzed. A file given explicitly with an unsupported extension is an error.
• --project sets the project name recorded in the result (default: the directory name).
• --format csv writes one row per project, file and function. See Output format.
• Files that fail to parse are kept in the result with "status": "error", and their count and errors are printed on stderr. The exit status is still 0.

stats

srcmetrics stats <RESULT.json>... [--scope file|function]

Computes, over the files or functions of one or more results:

• per metric: n (units with a value), missing, mean, sd (sample), min, q1, median, q3, max. Quantiles use linear interpolation (R type 7).
• the same per language (by_language), as baselines.
• Pearson and Spearman correlations for every pair of metrics, using the units where both values are present (n is reported). Fewer than 3 units or zero variance gives null.

report

srcmetrics report <RESULT.json> [-o FILE]

Writes a single HTML file with the run metadata, project metrics, per-language medians, histograms of every metric and a Spearman correlation heatmap. It loads no external resources, so it works offline and can be shared as one file.

model

srcmetrics model train --labels LABELS.csv [--features IDS] [--lambda L] [-o MODEL.json] <RESULT.json>...
srcmetrics model predict --model MODEL.json <RESULT.json>...

See Experimental readability model.

serve

srcmetrics serve [--bind 127.0.0.1] [--port 8080]

See HTTP API.