The file-reading and table-rendering behavior the adapters share, kept out of ccusage-core so that changing it does not touch the report, pricing, or config layers.

Owns #

file discovery by extension, including the JSONL helper most adapters use.
chunk_file_indexes_by_size — size-balanced work splitting, so one large file does not leave workers idle.
read_files_parallel — ordered parallel reads: results come back in file order, so output does not depend on worker scheduling, and --single-thread collapses to a plain sequential read.
filter_loaded_entries_by_date — the shared --since/--until filter.
jsonl — the streaming JSONL parsing helpers.
report::print_table_for_agent — the usage table for adapters whose rows are plain UsageSummary values, parameterized by agent name.

Public surface #

report::print_table_for_agent
collect_usage_files
collect_files_with_extension
filter_loaded_entries_by_date
chunk_file_indexes_by_size
read_files_parallel

Depends on #

ccusage-core
serde
serde_json

Build layer #

Built in the foundation Crane artifact layer, so a change here recompiles every adapter.