| English | 中文 |
Use gcam_inspect_output_structure for its fixed R inspection sequence. It accepts max_memory (default 1g) and runner-supplied output_dir; it has no action/scenario/region/sector selector. The database is ${GCAM_RELEASE_DIR}/output/database_basexdb, with database_R/Prov_code_china.csv. This differs from the per-scenario Python extractor layout.
Follow setup, invocation and artifact checks. Check all four CSVs and their actual query contents; individual query failures can be caught inside the script, so process completion alone is insufficient.
Read only the needed detail: XML paths, custom query templates, units, or province mapping. Custom queries require separately reviewed task-copy adaptation; these references are not extra runner arguments. Interpret results using the GCAM method.