SQL schema-prefix rename

Rename a schema (qualified-name prefix) such as ra02. or rtmp. to a new prefix inside a folder of Python-wrapped Spark SQL scripts. These files look like Python but embed SQL in Sql = [ ... ] triple-quoted strings; do not be confused by the Python scaffolding. The SQL statements use -- comments, while the file header uses # Python comments.

Workflow

1. Locate and inspect. List the target folder (including subfolders) for .txt / .py / .sql scripts. Read one file to confirm the structure: a # -*- coding: utf-8 -*- header block (# 目标表:, # 依赖表:, …), then Sql = [ ... ] holding triple-quoted SQL with -- comments.

2. Enumerate occurrences. Grep for the source prefixes case-insensitively (to also catch uppercase RA02. / RTMP.). Classify each hit by context:
   - INSERT INTO <prefix>.<table> and FROM / JOIN <prefix>.<table> inside an INSERT…SELECT (active or ---commented).
   - CREATE TABLE / ALTER TABLE / REFRESH TABLE / DROP TABLE <prefix>.<table> (active or ---commented).
   - # 目标表: <prefix>.<table> Python header comments.

3. Confirm scope before editing. The scope is genuinely ambiguous and the edit is hard to reverse by hand, so ask the user (offer a recommended default):
   - INSERT statements only — target table plus FROM/JOIN sources inside INSERT…SELECT, including --INSERT INTO blocks; leave CREATE/ALTER/REFRESH/DROP and # comments unchanged.
   - All SQL statements — everything inside the SQL strings (INSERT/CREATE/ALTER/REFRESH/DROP, active and ---commented), keep # header comments.
   - Everything — also the # header comments.
   Also confirm whether uppercase variants (RA02. / RTMP.) count as the same schema (they normally do in Hive/Spark, where schema names are case-insensitive).

4. Back up first. Copy the originals to a sibling _backup_before_rename/ before any in-place edit.

5. Run the rename. Use scripts/rename_schema.py (resolve relative to this skill directory):

text python scripts/rename_schema.py --dir <dir> --map "ra02.=rcdm_temp.,rtmp.=rcdm_temp." \ --scope insert|all --case-insensitive [--keep-hash-comments] [--ext .txt,.py,.sql]

• --map is a comma-separated list of from.=to. pairs; the trailing . is significant (it matches the schema-qualified-name prefix).
• --scope insert uses a line state machine: an active INSERT INTO spans to its terminating ;; a --INSERT INTO spans its contiguous -- block until a blank line, a dashes-only line, or a non-comment line.
• --scope all replaces on every line, except #-comment lines when --keep-hash-comments is set.
• Use the bundled Python (load_workspace_dependencies) if a system Python is unavailable.

6. Verify. Grep for the source prefixes again (case-insensitive). The only remaining hits must be the deliberately-preserved # header comments (or none, for full scope). Reconcile counts: original_total = replaced + remaining. Confirm UTF-8 and line endings are preserved.

Notes

• Preserve uppercase table names; only the schema prefix changes.
• The script preserves UTF-8 encoding and CRLF/LF line endings exactly.
• Always report the backup path so the user can roll back.