@pugi/cli 0.1.0-beta.99 → 1.0.0-alpha.10

package/dist/core/edits/format-detector.js

@@ -1,260 +0,0 @@
-/**
- * Edit-format auto-select (PUGI-79).
- *
- * The 4-layer diff dispatcher (Layer A single-block, B ordered batch,
- * C sha256-gated rewrite, D AST-aware) is layer-agnostic at the wire
- * — the model emits whichever marker family it was prompted with, and
- * `dispatch.ts` routes per envelope. Empirically though, different
- * model families have markedly different success rates per layer:
- *
- *   - Anthropic Claude family — native function-calling shape;
- *     produces clean single-block Layer A edits with high recall
- *     against `+++ NEW / --- OLD / ===` markers.
- *   - Open-weight Qwen3 / Kimi / DeepSeek — historically stronger
- *     with whole-file Layer C rewrites OR sha256-gated patches; the
- *     conflict-marker (Gemini-style) Layer A surface has higher
- *     ambiguity rates on these models.
- *   - DeepSeek-coder family — anchors well on Layer C primary with
- *     Layer A as fallback (their RLHF set leaned hard on patch-style
- *     emit).
- *   - Reasoning models (o-series, gpt-5) — handle Layer D AST-aware
- *     operations when an LSP bridge is available; Layer C primary
- *     otherwise.
- *
- * The existing `format-matrix.ts` keyed exact slugs only — fine for
- * the canonical set but blind to vendor-prefixed slugs that come back
- * from the Anvil gateway (e.g. `anthropic/claude-sonnet-4-20250514`,
- * `qwen/qwen3-coder`, `deepseek/deepseek-chat-v3.1`). This detector
- * layers ON TOP of the matrix:
- *
- *   1. exact-key match against `EDIT_FORMAT_MATRIX` — wins outright.
- *   2. vendor-prefix strip → second exact-key probe (so
- *      `anthropic/claude-sonnet-4-6` resolves the same as `claude-sonnet-4-6`).
- *   3. family inference from the (possibly prefixed) slug — returns
- *      the family-default chain (see `FAMILY_DEFAULTS`).
- *   4. unknown — wildcard `*` from the matrix.
- *
- * The output is the same `EditFormatChain` shape the matrix already
- * exposes so callers (dispatcher hint, persona prompt hint) consume a
- * single contract.
- *
- * NOT a routing decision: the dispatcher's per-layer routing is still
- * driven by what the model actually emits. The detector's output is a
- * HINT — surfaced to the persona prompt so the model has the right
- * marker template loaded, and surfaced to the dispatcher's fuzzy ladder
- * / fallback ordering so an ambiguous response gets retried in an order
- * that matches the model's empirical strengths.
- *
- * Spec: this is the PUGI-79 implementation; the issue ships the
- * detector + dispatcher hint + persona hint together.
- */
-import { EDIT_FORMAT_MATRIX, } from './format-matrix.js';
-/**
- * Family-level default chains. These are deliberately CONSERVATIVE —
- * each family's primary is the layer with the highest observed apply-
- * rate across the 2026-05 dogfood corpus; fallbacks rank in observed
- * success order. The exact-slug entries in `EDIT_FORMAT_MATRIX` can
- * override per-model; family defaults are the floor.
- *
- * Anthropic:
- *   The Anvil corpus shows Claude family hitting ~95% Layer A apply on
- *   first try; Layer C is the rescue when the file is large enough
- *   that the operator's region selection ambiguates. Matrix entries
- *   for `claude-opus-4-7` + `claude-sonnet-4-6` override to Layer C
- *   primary because the larger context windows let those models hold
- *   the whole file comfortably.
- *
- * Open-weight (qwen / deepseek / kimi):
- *   These ship whole-file rewrites more reliably than partial diffs —
- *   the patch-style envelopes their fine-tune sets prefer match Layer
- *   C semantics. Layer A is the fallback because they can still emit
- *   sensible search/replace pairs when prompted explicitly.
- *
- * Reasoning (gpt-5 / o-series):
- *   Layer C primary because the longer reasoning trace burns through
- *   the partial-diff format's positional constraints. Layer D rides
- *   the wire when an LSP is bridged; the matrix entries override per-
- *   model when that's available (see `qwen3-coder-480b → D`).
- *
- * Unknown:
- *   Defers to the wildcard chain from the matrix (Layer A primary).
- */
-const FAMILY_DEFAULTS = {
-    anthropic: { primary: 'A', fallback: ['C', 'B'] },
-    openai: { primary: 'C', fallback: ['A', 'B'] },
-    gemini: { primary: 'A', fallback: ['C', 'B'] },
-    qwen: { primary: 'A', fallback: ['C', 'B'] },
-    deepseek: { primary: 'C', fallback: ['A', 'B'] },
-    kimi: { primary: 'C', fallback: ['A', 'B'] },
-    unknown: { primary: 'A', fallback: ['B', 'C'] },
-};
-/**
- * Detect the model family from a slug. Handles vendor-prefixed forms
- * (`<vendor>/<model>` and `<provider>:<model>`) plus bare-name forms.
- * Case-insensitive on the family name; case-preserving on the rest.
- *
- * Order matters: vendor prefix wins when present (the gateway is
- * authoritative about which family routed). Otherwise the bare slug
- * gets prefix-matched against well-known family stems. Unknown slugs
- * surface as `unknown` — the caller falls back to the matrix wildcard.
- */
-export function detectModelFamily(modelSlug) {
-    if (!modelSlug)
-        return 'unknown';
-    const raw = modelSlug.trim();
-    if (raw.length === 0)
-        return 'unknown';
-    const lower = raw.toLowerCase();
-    // Vendor-prefix form: `<vendor>/<model>`. The vendor wins outright
-    // because the gateway is authoritative — `anthropic/<anything>`
-    // routes Claude family regardless of the suffix.
-    const slashIdx = lower.indexOf('/');
-    if (slashIdx > 0) {
-        const vendor = lower.slice(0, slashIdx);
-        switch (vendor) {
-            case 'anthropic':
-                return 'anthropic';
-            case 'openai':
-                return 'openai';
-            case 'google':
-            case 'gemini':
-            case 'xai':
-                return 'gemini';
-            case 'qwen':
-            case 'alibaba':
-                return 'qwen';
-            case 'deepseek':
-                return 'deepseek';
-            case 'moonshot':
-            case 'kimi':
-                return 'kimi';
-            default:
-                // Unknown vendor — fall through to bare-name detection on the
-                // suffix. Some gateways (OpenRouter style) put the family
-                // name AFTER the vendor (e.g. `together/qwen-coder`), so the
-                // suffix still carries signal.
-                return detectFamilyFromBareName(lower.slice(slashIdx + 1));
-        }
-    }
-    // Provider-prefix form: `ollama:<model>` (engine-VM local runtime).
-    // The prefix is informational; strip and continue.
-    const colonIdx = lower.indexOf(':');
-    if (colonIdx > 0) {
-        const prefix = lower.slice(0, colonIdx);
-        if (prefix === 'ollama' || prefix === 'lmstudio' || prefix === 'llama-cpp') {
-            return detectFamilyFromBareName(lower.slice(colonIdx + 1));
-        }
-    }
-    return detectFamilyFromBareName(lower);
-}
-/**
- * Bare-name detector. Lower-case input. Prefix-match against family
- * stems; longest match wins so `claude-opus-4-7` doesn't accidentally
- * route through the generic `c*` lane.
- */
-function detectFamilyFromBareName(name) {
-    if (name.startsWith('claude'))
-        return 'anthropic';
-    if (name.startsWith('gpt') || name.startsWith('o1') || name.startsWith('o3') || name.startsWith('o4')) {
-        return 'openai';
-    }
-    if (name.startsWith('gemini') || name.startsWith('grok'))
-        return 'gemini';
-    if (name.startsWith('qwen'))
-        return 'qwen';
-    if (name.startsWith('deepseek'))
-        return 'deepseek';
-    if (name.startsWith('kimi') || name.startsWith('moonshot'))
-        return 'kimi';
-    return 'unknown';
-}
-/**
- * Family-default chain. Exported for callers that want the family-
- * level recommendation without running the full detector pipeline
- * (e.g. observability surfaces that already resolved the family).
- */
-export function familyDefaultChain(family) {
-    // Defensive clone: the FAMILY_DEFAULTS table is readonly conceptually
-    // but the EditFormatChain type allows fallback array mutation. Clone
-    // so callers cannot accidentally mutate the table.
-    const base = FAMILY_DEFAULTS[family];
-    return { primary: base.primary, fallback: [...base.fallback] };
-}
-/**
- * Resolve the preferred edit-format chain for a model slug.
- *
- * Resolution order:
- *   1. Exact slug match in EDIT_FORMAT_MATRIX — overrides everything.
- *      This lets per-model tuning (e.g. `qwen3-coder-480b → D`) win
- *      over the family default.
- *   2. Vendor-prefix strip + exact match — so `anthropic/claude-...`
- *      resolves to the bare `claude-...` entry when present.
- *   3. Family inference — `FAMILY_DEFAULTS[detectModelFamily(slug)]`.
- *   4. Wildcard fallback — `EDIT_FORMAT_MATRIX['*']`.
- *
- * The returned chain is always a fresh object so caller mutations
- * cannot leak back into the matrix.
- */
-export function detectEditFormat(modelSlug) {
-    const fallbackWildcard = EDIT_FORMAT_MATRIX['*'] ?? { primary: 'A', fallback: [] };
-    if (!modelSlug || modelSlug.trim().length === 0) {
-        return { primary: fallbackWildcard.primary, fallback: [...fallbackWildcard.fallback] };
-    }
-    const raw = modelSlug.trim();
-    // Step 1 — exact match wins.
-    const exact = EDIT_FORMAT_MATRIX[raw];
-    if (exact)
-        return { primary: exact.primary, fallback: [...exact.fallback] };
-    // Step 2 — vendor-prefix strip (e.g. `anthropic/claude-sonnet-4-6`).
-    const slashIdx = raw.indexOf('/');
-    if (slashIdx > 0) {
-        const suffix = raw.slice(slashIdx + 1);
-        const stripped = EDIT_FORMAT_MATRIX[suffix];
-        if (stripped)
-            return { primary: stripped.primary, fallback: [...stripped.fallback] };
-    }
-    // Step 3 — family inference.
-    const family = detectModelFamily(raw);
-    if (family !== 'unknown')
-        return familyDefaultChain(family);
-    // Step 4 — wildcard.
-    return { primary: fallbackWildcard.primary, fallback: [...fallbackWildcard.fallback] };
-}
-/**
- * Render a one-line marker-format hint for the persona prompt. The
- * dispatcher already accepts every marker family (anthropic / gemini /
- * openai), so the hint is preference, not gating — surface the
- * preferred template + alternative so the model knows which envelope
- * has the highest apply-rate for its family.
- *
- * Returns `null` when the slug is unknown / unset; the caller drops
- * the hint module on null so the prompt is not polluted with empty
- * sections.
- */
-export function renderEditFormatHint(modelSlug) {
-    if (!modelSlug || modelSlug.trim().length === 0)
-        return null;
-    const family = detectModelFamily(modelSlug);
-    if (family === 'unknown')
-        return null;
-    const chain = detectEditFormat(modelSlug);
-    const markerStyle = FAMILY_MARKER_TEMPLATE[family];
-    return `# Edit-format hint (auto-selected for ${family})
-Preferred layer: Layer ${chain.primary}${chain.fallback.length > 0 ? ` (fallback: ${chain.fallback.map((l) => `Layer ${l}`).join(' -> ')})` : ''}.
-Marker template: ${markerStyle}
-The dispatcher accepts any marker family; this hint optimises apply-rate for the resolved model.`;
-}
-/**
- * Marker template literal per family. Mirrors the three envelopes the
- * marker-parser accepts; used by `renderEditFormatHint` so the model
- * sees a concrete shape next to the layer recommendation.
- */
-const FAMILY_MARKER_TEMPLATE = {
-    anthropic: '+++ NEW <file> / <new contents> / --- OLD <file> / <old contents> / ===',
-    openai: '@@@ REWRITE <file> sha256=<hex> / <full new contents> / @@@ END',
-    gemini: '<<<<<<< SEARCH <file> / <old> / ======= / <new> / >>>>>>> REPLACE',
-    qwen: '<<<<<<< SEARCH <file> / <old> / ======= / <new> / >>>>>>> REPLACE',
-    deepseek: '@@@ REWRITE <file> sha256=<hex> / <full new contents> / @@@ END',
-    kimi: '@@@ REWRITE <file> sha256=<hex> / <full new contents> / @@@ END',
-};
-//# sourceMappingURL=format-detector.js.map

package/dist/core/edits/format-matrix.js

@@ -1,26 +0,0 @@
-export const EDIT_FORMAT_MATRIX = {
-    'claude-opus-4-7': { primary: 'C', fallback: ['A', 'B'] },
-    'claude-sonnet-4-6': { primary: 'C', fallback: ['A', 'B'] },
-    'claude-haiku-4-5': { primary: 'A', fallback: ['B'] },
-    'gpt-5': { primary: 'C', fallback: ['A'] },
-    'gpt-5-mini': { primary: 'A', fallback: ['B'] },
-    'deepseek-v4-pro': { primary: 'A', fallback: ['B', 'C'] },
-    'deepseek-coder-v2': { primary: 'A', fallback: ['B'] },
-    'qwen3-coder-480b': { primary: 'D', fallback: ['A', 'B'] },
-    'qwen3:0.6b': { primary: 'A', fallback: [] },
-    'gemini-2.5-pro': { primary: 'C', fallback: ['A'] },
-    'gemini-2.0-flash': { primary: 'A', fallback: ['B'] },
-    'ollama:qwen3-coder:30b': { primary: 'A', fallback: ['B'] },
-    '*': { primary: 'A', fallback: ['B', 'C'] },
-};
-export function pickEditFormat(model, fileSize, isLSPAvailable = () => false) {
-    const chain = EDIT_FORMAT_MATRIX[model] ?? EDIT_FORMAT_MATRIX['*'] ?? { primary: 'A', fallback: [] };
-    if (chain.primary === 'C' && fileSize > 4000) {
-        return { primary: 'A', fallback: ['B'] };
-    }
-    if (chain.primary === 'D' && !isLSPAvailable()) {
-        return { primary: chain.fallback[0] ?? 'A', fallback: chain.fallback.slice(1) };
-    }
-    return chain;
-}
-//# sourceMappingURL=format-matrix.js.map