doci/AGENTS.md

AGENTS.md - XQuery Code Style and Build Guide

Build & Test Commands

Running Tests

basex -t /path/to/test/file.xqm    # Run all tests in a file
basex -t /path/to/test/dir         # Run all tests in directory
basex -t test/simple.xqm           # Run doci unit tests (6 tests)
basex -t test/smoke.xq             # Run smoke tests (no assertions)

Query Execution

basex query.xq                     # Execute XQuery file
basex -V query.xq                  # Verbose mode with timing info (includes compile/execute times)
basex -q "query string"            # Execute inline query
basex -Q test/file.xqm             # Execute XQuery file (explicit flag, for query files without test functions)
basex -c "commands"               # Execute BaseX commands from string
basex -C script.bxs              # Execute BaseX command script
basex  # Start interactive BaseX shell

Database Operations

(Already included in Query Execution section above)

Environment

• XQuery Engine: BaseX 12.2
• XQuery Version: 3.1+
• Module System: BaseX module system with module namespace
• Test Framework: http://basex.org/modules/unit

Code Style Guidelines

File Naming

• Module files: lowercase.xqm (e.g., doci.xqm)
• Test files: snake_case.xqm or snake_case.xq
• Use hyphens for multi-word names (e.g., pdfbox.xqm)
• Resource files: lowercase with hyphens (e.g., test-data.txt)

Namespace Declarations

module namespace module-name = 'urn:project:namespace';

• Use reverse domain name notation for consistency
• Match namespace URI to file path structure

Records & Types

• Use declare record for structured data
• Use declare type for custom types
• Record fields use ?field syntax for access
• Mark optional fields with ? after field name

declare record doci:doci(
  lines as xs:string+,
  separator? as xs:string
);

Function Naming

• Module functions: prefix with module namespace (e.g., doci:build)
• Helper functions: use camelCase internally
• Test functions: prefix with test: namespace and test: prefix
• Use descriptive names that clarify purpose

Import/Module Syntax

import module namespace mod = 'namespace' at "path/to/file.xqm";

• Use relative paths for local modules
• Keep imports at top of file, grouped logically

Module Organization

• Group related functionality in modules
• Export public API functions
• Keep private functions marked with %private if needed
• Document exported functions with (:~ doc comments

Line Separators

• Detect automatically using doci:separator()
• Default: file:line-separator()
• Handle all common separators: \n, \r\n, \r

Module Function Implementation Notes

• Use => for function chaining/piping
• Prefer let over for for single items
• Use if-then-else expressions
• Leverage XQuery 3.1+ functions: tokenize(), string-join()
• Use ! for mapping single items
• Use * for mapping sequences: expr!func()
• Prefer switch() case when over nested if-then-else

Error Handling

try {
  ...
} catch * {
  error(xs:QName("mod:code"), "descriptive message")
}

• Use try/catch for external operations (files, PDFs, network)
• Define custom error codes as xs:QName
• Validate parameters and ranges before processing
• Provide meaningful error messages with context
• Use #mod:code for errorQName in error() calls

XQuery Best Practices

• Use => for function chaining/piping
• Prefer let over for for single items
• Use if-then-else expressions
• Leverage XQuery 3.1+ functions: tokenize(), string-join()
• Use ! for mapping single items
• Use * for mapping sequences: expr!func()
• Prefer switch() case when over nested if-then-else

Type Checking

declare function foo($param as xs:string) as xs:integer { ... }

• Specify parameter types using as xs:type
• Use item(), item()* for flexible/untyped sequences
• Return type declarations are mandatory for exported functions
• Use doci:num for union types (e.g., integer|double)

Record Field Access

• Use ?field syntax for record field access (e.g., $doci?lines)
• Optional fields marked with ? after field name in record declaration

Documentation

(:~ 
 * Documentation comment
 * @param $name Description
 * @return Description
 * @see Reference URL
 :)

• Use (:~ for module and function documentation
• Document parameters, return values, and side effects
• Include @see references to external docs when relevant

Test Structure

Test Module Pattern

module namespace test = 'test:module-name';
import module namespace mod = 'namespace' at "../src/file.xqm";

declare %unit:test function test:name() {
  let $result := mod:function(...)
  return unit:assert-equals(expected, $result)
};

Test Assertions

unit:assert-equals(expected, actual)
unit:assert(condition)

• Use unit:assert-equals for value comparison
• Use unit:assert for boolean conditions
• Provide meaningful test names: test:functionality

Running Single Tests

basex -t test/file.xqm  # Runs all tests in the file
# To run a single test, comment out others or use test: namespaces

Test Helper Functions

declare function test:read($path as xs:string) as xs:string {
  file:resolve-path($path, file:base-dir()) => file:read-text()
};

Test Data

• Place test resources in test/resources/
• Use descriptive filenames (e.g., sample.txt, empty.txt)
• Use test:read("resources/filename.txt") helper

Test Organization

• Group tests by functionality
• Test edge cases: empty input, null values, boundaries
• Test error cases where applicable
• Keep tests independent and order-independent

Commit Standards

Commit Message Format

verb: description

Examples

fix: handle empty input
add: support多元 line separators
refactor: improve error handling
test: add edge case for range updates
docs: update module documentation

Commit Guidelines

• Add test cases for bug fixes
• Update documentation for public API changes
• Keep commits focused on single changes
• Verify tests pass before committing

Notes

• BaseX 12.2 supports XQuery 3.1 syntax
• The fn:do-until function may not be available; use hof:until as alternative
• Java interoperability available through Q{namespace} syntax (e.g., Q{java:java.io.File}new())
• For PDF processing (pdfbox module), external JAR dependencies required