Official PowerBuilder Documentation Query

references/pb_docs.db

Runtime Prerequisites (Ready to Copy and Use)

• The only required file is

  references/pb_docs.db

  in this skill directory, which has already been prebuilt.
• The query scripts use only Python's built-in sqlite3 module (FTS5), so no third-party packages need to be installed; however, a Python 3.11+ runtime is required.
• Check whether Python is available on the current machine:

  python --version

  (on Windows, you can also try

  py --version

  ).
• If

  python

  /

  py

  is not found: Python is not installed on the machine. Install Python 3.11+ first (https://www.python.org/downloads/), or

  search_db.py

  /

  get_pages.py

  will not run.
• The entire skill directory can be copied as a whole to any machine with Python 3.11+ and used without an internet connection or an additional model.

Document Inventory (Topics)

Retrieval Workflow

1. Locate content through full-text search (primary path). Extract English keywords from the user's question, such as function names, property names, and syntax keywords, and use

   search_db.py

   to locate matching pages in milliseconds:
   bash

   python scripts/search_db.py Modify
   python scripts/search_db.py "SetTransObject AND Retrieve" 20

   Results show the "file name + page number + snippet". Multi-term queries use AND by default; use OR to broaden the search and enclose phrases in quotation marks. On Windows, use

   PYTHONIOENCODING=utf-8

   .
2. Retrieve the complete context in one pass. Add

   --pages N

   to the search command so that

   search_db.py

   outputs the full text of the first N matching pages together with the matching snippets. Function signatures and code examples often span multiple pages, so this step is usually sufficient to answer the question:
   bash

   python scripts/search_db.py "SetTransObject" 3 --pages 1

   If additional pages adjacent to a match are needed, use

   get_pages.py

   separately:
   bash

   python scripts/get_pages.py 08_datawindow_reference 760 765

3. Browse by topic (supporting method). If you do not know which keywords to search for or want to browse by topic, use the "Document Inventory" to understand the scope of each topic, and then combine browsing with a precise search. Search results directly provide the file name and page number.
4. Answer naturally. Base the response on the extracted source content, but do not copy the documentation's wording verbatim or list citations. Respond as a senior PowerBuilder developer would when answering directly: practical, natural, and concise rather than mechanical or academic. The final answer must not contain any file names, document names, page numbers, or source identifiers such as "Source: xx". File names are used internally by the retrieval scripts, but those implementation details should not be exposed to the user. Do not repeatedly emphasize the version number in the response either—for example, do not add "in PowerBuilder" to every sentence. Mention the version only when the user explicitly asks about version-specific features or differences. If the extracted content is insufficient to answer the question, broaden the keywords or page range and try again.

Why Query Instead of Extrapolating

dw_1.Modify()

GetItemStatus()

PowerString

Notes

• pb_docs.db

  is prebuilt. The full text was extracted from the official documentation and indexed with SQLite FTS5. It is included with the skill, ready to use, and can be copied as a whole to another machine.
• The index stores textual content. A small amount of table alignment and images is lost, but the body text, code, and signatures remain complete and readable.
• Dependencies: only Python's built-in

  sqlite3

  module (with FTS5); no third-party packages are required.