PbIdea Framework API Query

This skill full-text indexes all exported source code of the PbIdea framework (

code/

, 566 PB objects: user objects/datawindows/windows/functions/structures) into an SQLite database

references/pbidea.db

, enabling millisecond-level retrieval via scripts. Core Principle: Do not fabricate PbIdea API details from memory. For any specific function signatures, property names, constant values (e.g., HttpGet=1), or component existence, always check the index. The indexed content directly comes from the framework's source code itself (external function declaration blocks

type prototypes

.sru

files, Chinese comments, constants, event scripts), making it the most authoritative and complete API reference.

Prerequisites (Copy and Use Directly)

Only need
```
references/pbidea.db
```
(pre-built) in this skill directory.
The query script only uses Python's built-in
```
sqlite3
```
(FTS5 trigram tokenizer), no third-party packages required; however, Python 3.11+ runtime is needed (it comes with SQLite ≥3.37 which supports trigram).
Check if Python is available on the current machine:
```
python --version
```
(try
```
py --version
```
on Windows). If
python
/
py
is not found: Install Python 3.11+ first (https://www.python.org/downloads/), otherwise the retrieval script cannot run.
Add
```
PYTHONIOENCODING=utf-8
```
when running the script on Windows to avoid encoding errors when printing Chinese in the GBK console.
The entire skill directory can be copied to any machine with Python 3.11+ for use, no internet connection or additional models required.

Component Map (Know What's Available First, Then Search)

PbIdea is a "PowerBuilder interface extension library based on JSON and WebAPI", with its core implementation compiled in

PbIdea.dll

, and the PB side calls it via external function declarations. The

catalog

table in

references/pbidea.db

stores the names, affiliated libraries, and a Chinese description of all 566 objects; you can browse them using

list_objects.py

. The following is a quick overview of main components:

Category	Objects (websuite library)	Description
JSON	`uo_json`	Core JSON object: Parse/ToString/Set/Get, cascading paths like `/0/sub/2/dj` (array indices start from 0)
JSON	`vo_json_datawindow`	Mutual conversion between DataWindow/DataStore and JSON
JSON	`uo_json_s` , `uo_bson` , `uo_csv` , `uo_config`	Standard library functions, BSON for binary storage, csv2json, JSON format configuration files
HTTP	`uo_httpclient` + `uo_response`	WinHttp client: `Request(nHttpType,url,...)` , HttpGet=1/HttpPost=2/HttpPut=3..., upload and download, BasicAuth/JWT
HTTP	`uo_curl` , `uo_ftp` , `uo_websocket_client` / `uo_websocket_server` , `uo_aliyun`	curl-based client, FTP, WebSocket, Alibaba Cloud
Encryption & Encoding	`uo_crypto` , `uo_rsa` , `uo_hmac` , `sm2utils` / `sm3utils` / `sm4utils`	openssl algorithm library, RSA encryption/decryption/signature verification, SM2/SM3/SM4
Database	(sql library) `uo_database` , `uo_database_pool` , `uo_recordset` , `uo_field`	Database access, connection pool, result set
Data Processing	`uo_datawindowex` , `uo_dbf` , `uo_xlsx` , `uo_xml` , `uo_sqlite3`	DW enhancement, DBF/xlsx/XML reading and writing
Compression	`uo_compress` (zlib), `uo_7z` , `uo_zip`	Compression and decompression
System Tools	`uo_thread` , `uo_timer` , `uo_process` , `uo_serial` , `uo_string` , `uo_map` / `uo_kv` , `uo_buffer` / `uo_blob` , `uo_biginteger` , `uo_datetime` , `uo_file` , `uo_scintilla`	Thread/timer/process/serial port/string/key-value pair/blob/big integer/datetime/file/syntax editor
UI	`uo_chart` (painter library), `uo_led` , `uo_cover` , `uo_datawindowex`	Charts, LED, mask window
Other Libraries	sciter library `uo_sciter` (HTML UI), web library `uo_datastore` / `uo_html` / `nvo_webreponse_` , web_client library `uo_datawindow_client` , haikang library `uo_fingerprint` (fingerprint device)/ `uo_hcusb` (ID card), pbjson library for demonstration programs	—

To get the complete list, run

PYTHONIOENCODING=utf-8 python scripts/list_objects.py

(you can add pbl name,

--type

--name

for filtering).

Retrieval Workflow

Full-text Retrieval Localization (Main Path). Extract English keywords (object names, function names, constant names) and Chinese keywords (function descriptions) from user questions, then use
```
search_db.py
```
to locate hit objects in milliseconds:
bash
```
PYTHONIOENCODING=utf-8 python scripts/search_db.py Request
PYTHONIOENCODING=utf-8 python scripts/search_db.py "uo_json Get"
PYTHONIOENCODING=utf-8 python scripts/search_db.py 加密 签名
```
Multiple keywords default to AND to narrow down results (quotes are optional). Note: The index uses a trigram tokenizer, Boolean operators like AND/OR are not supported. To widen or narrow the search, directly change keywords or adjust the number of words. The hit output shows "Library / Object Name (Type) + Hit Fragments".
Retrieve Complete Context at Once. Add
```
--pages N
```
during retrieval to make
```
search_db.py
```
output the complete source code of the top N hit objects along with the hit fragments (function signatures, Chinese comments, constants, event scripts are in the same file for complementation). Usually this step is sufficient for answering:
bash
```
PYTHONIOENCODING=utf-8 python scripts/search_db.py "uo_httpclient Request" --pages 2
```
If you know the object name and want to view the complete source code directly, use
```
get_object.py
```
:
bash
```
PYTHONIOENCODING=utf-8 python scripts/get_object.py uo_json
PYTHONIOENCODING=utf-8 python scripts/get_object.py uo_json websuite   # Specify library for duplicate names
```
Localize by Component Map (Auxiliary). If you don't know which keyword to start with, first check the "Component Map" to determine the object's affiliation, then use
```
get_object.py
```
to directly retrieve the object's source code, or
```
list_objects.py --name uo_
```
to browse similar objects.
Answer Naturally. Answer based on the extracted original content, but do not copy the code wording or list references. The answer should be like a PowerBuilder developer who has used PbIdea for many years answering face-to-face: directly provide usage, examples, and calling conventions (e.g.,
```
uo_json.Get("/0/sub/2/dj", ref ld_dj)
```
where array indices start from 0). The final answer must not contain any library names, object file names, script commands, or source identifiers like "source" — the retrieval script uses file names internally, but that's an internal workflow detail and should not be disclosed to users. Also, do not repeatedly emphasize version numbers in the text (e.g., do not add "in PbIdea" to every sentence). Version numbers should only be mentioned when the user explicitly asks about version features/differences. If the extracted content is insufficient to answer, expand keywords or try other objects.

Why Query Instead of Extrapolate

PbIdea has an extremely wide API surface (566 objects, with different function signatures, constant values, and cascading path rules), and its habits differ from mainstream languages (e.g., JSON array indices start from 0,

system library "PbIdea.dll"

external function declaration,

uo_response

response object). It's easy to give wrong parameters, misremember constants, or fictionalize non-existent components based on impressions.

pbidea.db

is the authoritative source. Following the two steps of "search keywords → retrieve complete object source code", most questions can be locked in milliseconds.

Notes

```
pbidea.db
```
is pre-built (constructed from exported source code of the PbIdea framework, with automatic GBK/UTF-8 recognition and decoding), delivered with the skill, and can be used directly.

After updating the source code, you can rebuild the index with one click using

scripts/build_db.py

bash

PYTHONIOENCODING=utf-8 python scripts/build_db.py --source <PbIdea source code directory>

The index stores text content. Layout coordinates, fonts, and other attributes in PB exported files are fully retained but have no actual rendering value; function signatures, comments, event scripts, and constant declarations are fully readable.
Query dependencies: Only Python's built-in
```
sqlite3
```
(including FTS5 trigram), no third-party packages.

pbidea

NPX Install

Tags

SKILL.md Content (Chinese)