93 lines
4.4 KiB
Go
93 lines
4.4 KiB
Go
package config
|
|
|
|
import (
|
|
"bytes"
|
|
"fmt"
|
|
"os"
|
|
"path/filepath"
|
|
|
|
"gopkg.in/yaml.v3"
|
|
)
|
|
|
|
// pair.go: the LANGUAGE-PAIR configuration layer (D39.23, pack-15).
|
|
//
|
|
// A run config answers "how is THIS run wired" — stages, models, temperatures, budgets, gate
|
|
// switches. A pair config answers "what is true of zh→ru", and stays true across every run of every
|
|
// zh→ru book: the segmentation calibration (draft/edit budgets in OUTPUT tokens and the fertility
|
|
// coefficients that convert source characters into them) and the excision corridor for that pair.
|
|
// Before this split those numbers were copy-pasted into every shipping run config, so adding an arm
|
|
// meant re-pasting a pair calibration and a drift between two copies was invisible.
|
|
//
|
|
// It lives next to the run configs (`<config dir>/pairs/<pair>.yaml`) and is OPTIONAL: a project with
|
|
// no pair file falls back to the ratified generic numbers in LoadPipeline, exactly as before. The
|
|
// PROMPTS of a pair are resolved by convention rather than listed here — see promptsRoot.
|
|
|
|
// PairConfig is the per-language-pair layer of the configuration.
|
|
type PairConfig struct {
|
|
// Pair is the "<src>-<tgt>" key, asserted against the file name so a mis-filed calibration
|
|
// (zh-ru numbers saved as ja-ru.yaml) fails loud instead of silently mis-chunking a book.
|
|
Pair string `yaml:"pair"`
|
|
// PromptsRoot is the directory holding the per-pair prompt packs, as `<root>/<pair>/<role>.md`.
|
|
// Relative paths resolve against the pair file. Empty → "../../prompts" (the repo layout:
|
|
// backend/configs/pairs/<pair>.yaml → backend/prompts/<pair>/<role>.md).
|
|
PromptsRoot string `yaml:"prompts_root"`
|
|
// Segmentation is the pair's chunking calibration. A field left at zero keeps the generic
|
|
// fallback, so a pair file may carry only the corridor.
|
|
Segmentation Segmentation `yaml:"segmentation"`
|
|
// Coverage carries the pair's excision corridor (the lower bound is the one the gate checks).
|
|
Coverage PairCoverage `yaml:"coverage"`
|
|
}
|
|
|
|
// PairCoverage is the pair's slice of the coverage gate: the len_ratio corridor [low, high] for this
|
|
// language pair. The gate itself (on/off, sentence coverage, minimum chunk) stays run-scoped — whether
|
|
// to gate is a decision about a RUN, while what "too short for zh→ru" means is a fact about the pair.
|
|
type PairCoverage struct {
|
|
LenRatioBounds []float64 `yaml:"len_ratio_bounds"` // [low, high]
|
|
}
|
|
|
|
// pairsDirName is the directory, relative to the run config, that holds the pair layer.
|
|
const pairsDirName = "pairs"
|
|
|
|
// defaultPromptsRoot is where a pair's prompt pack lives when the pair file does not say otherwise —
|
|
// relative to the pair file itself (backend/configs/pairs → backend/prompts).
|
|
const defaultPromptsRoot = "../../prompts"
|
|
|
|
// LoadPair reads `<configDir>/pairs/<pair>.yaml`. A MISSING file is not an error (nil, nil): the pair
|
|
// layer is optional and its absence means "use the generic fallbacks" — the same behaviour a project
|
|
// had before the layer existed. A file that exists but is malformed, or whose `pair` disagrees with
|
|
// its name, fails LOUD: a calibration that silently does not apply is the failure mode this layer is
|
|
// supposed to remove.
|
|
func LoadPair(configDir, pair string) (*PairConfig, error) {
|
|
path := filepath.Join(configDir, pairsDirName, pair+".yaml")
|
|
raw, err := os.ReadFile(path)
|
|
if os.IsNotExist(err) {
|
|
return nil, nil
|
|
}
|
|
if err != nil {
|
|
return nil, fmt.Errorf("config: read pair config %s: %w", path, err)
|
|
}
|
|
var pc PairConfig
|
|
// STRICT decode, same class as the pipeline/book loaders: a mistyped key in a pair file means the pair
|
|
// calibration silently does not apply, which is precisely the failure this layer was created to remove.
|
|
dec := yaml.NewDecoder(bytes.NewReader(raw))
|
|
dec.KnownFields(true)
|
|
if err := dec.Decode(&pc); err != nil {
|
|
return nil, fmt.Errorf("config: parse pair config %s: %w", path, err)
|
|
}
|
|
if pc.Pair != "" && pc.Pair != pair {
|
|
return nil, fmt.Errorf("config: pair config %s declares pair %q — the file name is the pair key, so the calibration would apply to %q; rename the file or fix the field", path, pc.Pair, pair)
|
|
}
|
|
pc.Pair = pair
|
|
root := pc.PromptsRoot
|
|
if root == "" {
|
|
root = defaultPromptsRoot
|
|
}
|
|
if !filepath.IsAbs(root) {
|
|
root = filepath.Join(filepath.Dir(path), root)
|
|
}
|
|
pc.PromptsRoot = root
|
|
if b := pc.Coverage.LenRatioBounds; len(b) != 0 && len(b) != 2 {
|
|
return nil, fmt.Errorf("config: pair config %s: coverage.len_ratio_bounds must be [low, high] (2 values), got %d", path, len(b))
|
|
}
|
|
return &pc, nil
|
|
}
|