gnu-recutils（gnu 斜角）

GNU Recutils

GNU recutils is a set of tools for managing human-readable, plain-text databases. Records are stored in .rec files with a simple field: value syntax.

Quick Reference

Core Commands

Basic File Format

# Comment line
Field_name: field value
Another_field: another value

Field_name: second record
Another_field: its value

Records are separated by blank lines. Field names are case-sensitive and conventionally capitalized.

Multi-line Values

Use + continuation for multi-line content:

Description: This is the first line
+ and this continues on the second line
+ and a third line too.

Or use backslash at end of line:

Description: This is a long value that \
continues on the next line.

Schema Definitions (%rec Descriptors)

Place schema declarations before records:

%rec: Entity
%key: Id
%mandatory: Id Name
%allowed: Id Name Prototype Description OnLook OnTouch
%type: Prototype rec Entity
%type: Count int
%type: Active bool
%unique: Name

Id: sword
Name: Iron Sword
Count: 1
Active: yes

Common Descriptors

Field Types

Foreign Key References

%rec: Room
%key: Id

Id: tavern
Name: The Rusty Tavern

%rec: Entity
%key: Id
%type: Room rec Room

Id: mug
Name: Beer Mug
Room: tavern

Entity Containment (MUDD)

Entities can be nested using a Container field:

%rec: Entity
%key: Id
%type: Prototype rec Entity
%type: Container rec Entity

Id: table
Name: Wooden Table
Prototype: furniture

Id: lamp
Name: Brass Lamp
Prototype: object
Container: table

The lamp is "on" the table. Recfix validates that Container references exist.

Common Operations

Query Records

# All records
recsel data.rec

# Filter by field value
recsel -e 'Name = "sword"' data.rec

# Pattern matching
recsel -e 'Name ~ "Iron"' data.rec

# Numeric comparison
recsel -e 'Count > 5' data.rec

# Multiple conditions
recsel -e 'Type = "weapon" && Count > 0' data.rec

# Select specific fields
recsel -p Id,Name data.rec

# Count matching records
recsel -c -e 'Active = yes' data.rec

# Select by record type
recsel -t Entity data.rec

Insert Records

# Insert from stdin
echo -e "Id: axe\nName: Battle Axe" | recins data.rec

# Insert with field values
recins -f Id -v sword -f Name -v "Iron Sword" data.rec

Modify Records

# Update field value
recset -e 'Id = "sword"' -f Count -s 5 data.rec

# Delete field from record
recset -e 'Id = "sword"' -f Obsolete -d data.rec

Delete Records

# Delete matching records
recdel -e 'Count = 0' data.rec

# Delete by record number
recdel -n 3 data.rec

Validate

# Check file integrity
recfix data.rec

# Check and report all errors
recfix --check data.rec

Convert Formats

# To CSV
rec2csv data.rec > data.csv

# From CSV
csv2rec data.csv > data.rec

Format Output

# Custom output format
recsel data.rec | recfmt '{{Id}}: {{Name}}\n'

# Template-based formatting (recfmt reads from stdin)
recsel data.rec | recfmt -f template.fmt

Expression Syntax

Used with -e flag in recsel, recdel, recset:

Tips

• Use recfix --check before loading data to catch schema violations
• Foreign keys (%type: Field rec OtherType) validate references exist
• Field names start with a letter and contain only [a-zA-Z0-9_]; special fields start with %
• Empty lines separate records; use + continuation for multi-line values
• Comments start with # at the beginning of a line
• For JSON output, pipe through recsel -p and parse with a script

MUDD Naming Convention

In this project, entity fields use PascalCase (e.g., DescriptionShort, OnAttack). This avoids visual noise from underscores. See data/entities.rec for examples.