@codragraph/cli 2.1.1 → 2.1.5

package/dist/_shared/lbug/schema-constants.js

@@ -1,67 +0,0 @@
-/**
- * LadybugDB schema constants — single source of truth.
- *
- * NODE_TABLES and REL_TYPES define what the knowledge graph can contain.
- * Both CLI and web must agree on these for data compatibility.
- *
- * Full DDL schemas remain in each package's own schema.ts because
- * the CLI uses native LadybugDB and the web uses WASM.
- */
-export const NODE_TABLES = [
-    'File',
-    'Folder',
-    'Function',
-    'Class',
-    'Interface',
-    'Method',
-    'CodeElement',
-    'Community',
-    'Process',
-    'Section',
-    'Struct',
-    'Enum',
-    'Macro',
-    'Typedef',
-    'Union',
-    'Namespace',
-    'Trait',
-    'Impl',
-    'TypeAlias',
-    'Const',
-    'Static',
-    'Variable',
-    'Property',
-    'Record',
-    'Delegate',
-    'Annotation',
-    'Constructor',
-    'Template',
-    'Module',
-    'Route',
-    'Tool',
-];
-export const REL_TABLE_NAME = 'CodeRelation';
-export const REL_TYPES = [
-    'CONTAINS',
-    'DEFINES',
-    'IMPORTS',
-    'CALLS',
-    'EXTENDS',
-    'IMPLEMENTS',
-    'HAS_METHOD',
-    'HAS_PROPERTY',
-    'ACCESSES',
-    'METHOD_OVERRIDES',
-    'OVERRIDES', // Legacy compat alias — kept until all stored indexes are migrated
-    'METHOD_IMPLEMENTS',
-    'MEMBER_OF',
-    'STEP_IN_PROCESS',
-    'HANDLES_ROUTE',
-    'FETCHES',
-    'HANDLES_TOOL',
-    'ENTRY_POINT_OF',
-    'WRAPS',
-    'QUERIES',
-];
-export const EMBEDDING_TABLE_NAME = 'CodeEmbedding';
-//# sourceMappingURL=schema-constants.js.map

package/dist/_shared/lbug/schema-constants.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"schema-constants.js","sourceRoot":"","sources":["../../src/lbug/schema-constants.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,MAAM,CAAC,MAAM,WAAW,GAAG;IACzB,MAAM;IACN,QAAQ;IACR,UAAU;IACV,OAAO;IACP,WAAW;IACX,QAAQ;IACR,aAAa;IACb,WAAW;IACX,SAAS;IACT,SAAS;IACT,QAAQ;IACR,MAAM;IACN,OAAO;IACP,SAAS;IACT,OAAO;IACP,WAAW;IACX,OAAO;IACP,MAAM;IACN,WAAW;IACX,OAAO;IACP,QAAQ;IACR,UAAU;IACV,UAAU;IACV,QAAQ;IACR,UAAU;IACV,YAAY;IACZ,aAAa;IACb,UAAU;IACV,QAAQ;IACR,OAAO;IACP,MAAM;CACE,CAAC;AAIX,MAAM,CAAC,MAAM,cAAc,GAAG,cAAc,CAAC;AAE7C,MAAM,CAAC,MAAM,SAAS,GAAG;IACvB,UAAU;IACV,SAAS;IACT,SAAS;IACT,OAAO;IACP,SAAS;IACT,YAAY;IACZ,YAAY;IACZ,cAAc;IACd,UAAU;IACV,kBAAkB;IAClB,WAAW,EAAE,mEAAmE;IAChF,mBAAmB;IACnB,WAAW;IACX,iBAAiB;IACjB,eAAe;IACf,SAAS;IACT,cAAc;IACd,gBAAgB;IAChB,OAAO;IACP,SAAS;CACD,CAAC;AAIX,MAAM,CAAC,MAAM,oBAAoB,GAAG,eAAe,CAAC"}

package/dist/core/graphstore/lbug-row-source.d.ts

@@ -1,19 +0,0 @@
-/**
- * Adapter exposing a live LadybugDB instance as a `@codragraph/graphstore`
- * `RowSource`. Used by the analyze pipeline (Phase 4) to snapshot the
- * loaded graph into the content-addressed store.
- *
- * Best-effort by design: any table that errors at query time is skipped
- * (with the failure surfaced through the optional `onSkip` callback) so
- * the surrounding analyze flow never breaks because the versioning hook
- * misbehaves.
- */
-import type { RowSource } from '@codragraph/graphstore';
-import { type NodeTableName } from '../../_shared/index.js';
-export interface LbugRowSourceOptions {
-    /** Filter the node tables enumerated by `listNodeTables` — defaults to every NODE_TABLE. */
-    readonly nodeTables?: readonly NodeTableName[];
-    /** Called with `(table, error)` when a table query fails — defaults to a no-op. */
-    readonly onSkip?: (tableName: string, error: unknown) => void;
-}
-export declare const createLbugRowSource: (opts?: LbugRowSourceOptions) => RowSource;

package/dist/core/graphstore/lbug-row-source.js

@@ -1,141 +0,0 @@
-/**
- * Adapter exposing a live LadybugDB instance as a `@codragraph/graphstore`
- * `RowSource`. Used by the analyze pipeline (Phase 4) to snapshot the
- * loaded graph into the content-addressed store.
- *
- * Best-effort by design: any table that errors at query time is skipped
- * (with the failure surfaced through the optional `onSkip` callback) so
- * the surrounding analyze flow never breaks because the versioning hook
- * misbehaves.
- */
-import { NODE_TABLES, REL_TABLE_NAME } from '../../_shared/index.js';
-import { executeQuery } from '../lbug/lbug-adapter.js';
-export const createLbugRowSource = (opts = {}) => {
-    const onSkip = opts.onSkip ?? (() => { });
-    const tables = opts.nodeTables ?? NODE_TABLES;
-    const listNodeTables = async () => {
-        return [...tables];
-    };
-    const streamNodeTable = async function* (tableName) {
-        let rows;
-        try {
-            // `MATCH (n:T) RETURN n` returns one row per node. The node value
-            // is reachable as either `row.n` (named-column form) or `row[0]`
-            // (positional form) depending on the LadybugDB result-shape mode;
-            // we accept both, mirroring the resilient pattern used by
-            // `core/search/bm25-index.ts` for FTS results. Tables that do not
-            // exist on disk for a given repo throw here — we treat that as
-            // "no rows" via the onSkip callback rather than a hard failure.
-            rows = await executeQuery(`MATCH (n:${tableName}) RETURN n`);
-        }
-        catch (err) {
-            onSkip(tableName, err);
-            return;
-        }
-        let yielded = 0;
-        for (const raw of rows) {
-            const node = unwrapNode(raw);
-            if (!node)
-                continue;
-            yield normalizeNodeRow(node);
-            yielded++;
-        }
-        // If the query reported rows but none unwrapped, surface that as a
-        // skip so the analyze log makes the silent-empty failure mode
-        // visible instead of producing a 0-row snapshot for the table.
-        if (rows.length > 0 && yielded === 0) {
-            onSkip(tableName, new Error(`lbug-row-source: query returned ${rows.length} row(s) for "${tableName}" but none had an unwrappable node — ` +
-                `result shape changed? expected row.n or row[0] to be the node`));
-        }
-    };
-    const streamEdges = async function* () {
-        let rows;
-        try {
-            // Project `from`/`to`/`type` as scalar columns and the full rel as
-            // `rel`. Scalars give us a deterministic edge id even if the rel
-            // payload's shape changes; `rel` carries any extra properties for
-            // hashing.
-            rows = await executeQuery(`MATCH (a)-[r:${REL_TABLE_NAME}]->(b) RETURN a.id AS \`from\`, b.id AS \`to\`, r.type AS type, r AS rel`);
-        }
-        catch (err) {
-            onSkip(REL_TABLE_NAME, err);
-            return;
-        }
-        let yielded = 0;
-        for (const raw of rows) {
-            const r = raw;
-            const from = pickField(r, 'from', 0);
-            const to = pickField(r, 'to', 1);
-            const type = pickField(r, 'type', 2);
-            const rel = pickField(r, 'rel', 3);
-            if (typeof from !== 'string' || typeof to !== 'string')
-                continue;
-            yield normalizeEdgeRow({ from, to, type, rel: isPlainObject(rel) ? rel : null });
-            yielded++;
-        }
-        if (rows.length > 0 && yielded === 0) {
-            onSkip(REL_TABLE_NAME, new Error(`lbug-row-source: edges query returned ${rows.length} row(s) but none had a string from/to — ` +
-                `result shape changed?`));
-        }
-    };
-    return { listNodeTables, streamNodeTable, streamEdges };
-};
-/**
- * Pull the node out of an executeQuery result row, accepting either the
- * named-column form (`row.n`) or the positional form (`row[0]`). Returns
- * null when the row is missing or the node value isn't an object — the
- * caller treats that as "skip and surface".
- */
-const unwrapNode = (raw) => {
-    if (!raw || typeof raw !== 'object')
-        return null;
-    const r = raw;
-    const candidate = r['n'] ?? r[0];
-    return isPlainObject(candidate) ? candidate : null;
-};
-/** Read a field from an executeQuery row, falling back to the positional index. */
-const pickField = (row, named, positional) => {
-    if (!row)
-        return undefined;
-    return row[named] ?? row[positional];
-};
-const isPlainObject = (v) => typeof v === 'object' && v !== null && !Array.isArray(v);
-/**
- * Sanitize a node row for canonical hashing:
- *   - Drop LadybugDB-specific internal fields (`_id`, `_label`) that are
- *     not content-bearing — including them would make the hash sensitive
- *     to internal storage offsets and break dedup across snapshots.
- *   - Sort keys deterministically (canonical JSON in the serializer
- *     already does this, but doing it once here keeps the row payload
- *     stable when we ever swap engines).
- */
-const normalizeNodeRow = (node) => {
-    const out = {};
-    for (const key of Object.keys(node).sort()) {
-        if (key === '_id' || key === '_label')
-            continue;
-        out[key] = node[key];
-    }
-    return out;
-};
-const normalizeEdgeRow = (r) => {
-    const props = {};
-    if (r.rel && typeof r.rel === 'object') {
-        for (const key of Object.keys(r.rel).sort()) {
-            // Skip the synthetic from/to/type that show up under `rel` too —
-            // we already project them as top-level columns and don't want
-            // duplication in the canonical row.
-            if (key === 'from' || key === 'to' || key === 'type')
-                continue;
-            if (key.startsWith('_'))
-                continue;
-            props[key] = r.rel[key];
-        }
-    }
-    return {
-        from: String(r.from),
-        to: String(r.to),
-        type: typeof r.type === 'string' ? r.type : String(r.type ?? ''),
-        ...props,
-    };
-};

package/dist/core/lbug/content-read.d.ts

@@ -1,46 +0,0 @@
-/**
- * Read-side decoder for `content` columns in lbug node rows.
- *
- * RFC 0001 Phase 2 introduces an optional `contentEncoding` column on
- * every node table that has `content`. Default is `'none'` (passthrough)
- * so existing reads keep working unchanged. When a writer opts into
- * `--compress brotli|zstd`, the column carries the encoding tag and the
- * `content` column carries base64-encoded compressed bytes — readers
- * MUST run those bytes back through `decodeContent` before handing them
- * to a consumer (MCP tool result, HTTP API response, embedding model,
- * LLM input).
- *
- * Centralizing the decode in one helper has two benefits:
- *   1. Shim sites are 2-line changes: add `, n.contentEncoding AS
- *      contentEncoding` to the Cypher RETURN, and pipe the row through
- *      `decodeContentField` (or `decodeContentRow`) at the boundary.
- *   2. Anyone hunting for "where does the read path decode compressed
- *      bytes" greps for `decodeContentField` and gets every site in one
- *      shot — no per-table feature detection scattered across files.
- */
-/**
- * Decode a single (content, contentEncoding) pair from a Cypher row.
- *
- * Returns the input content unchanged when:
- *   - the encoding is missing / empty / `'none'` (the common case for
- *     1.6.x – 1.7.x indexes, plus any 1.8+ index written without
- *     `--compress`);
- *   - content is null/undefined (caller decides whether that's an error);
- *   - content is not a string (pre-Phase-2 indexes never wrote non-string
- *     content, but defensive: don't crash a read path on a malformed row).
- *
- * Throws (via `decodeContent`) only when the row claims an encoding this
- * CLI build can't decode — that's a forward-compat error and the right
- * behavior is to fail loudly rather than return wrong content.
- */
-export declare function decodeContentField(content: unknown, encoding: unknown): string | undefined;
-/**
- * Apply `decodeContentField` to a row that carries `content` and
- * `contentEncoding` keys (or their numeric column-index aliases).
- *
- * The numeric-fallback shape (`r[N]`) mirrors LadybugDB's row format —
- * driver versions vary on whether named keys are populated, so existing
- * read sites do `r.content ?? r[N]`. This helper accepts the same
- * pattern. Returns a NEW object (does not mutate input).
- */
-export declare function decodeContentRow<T extends Record<string, unknown>>(row: T, contentKey?: keyof T, encodingKey?: keyof T): T;

package/dist/core/lbug/content-read.js

@@ -1,64 +0,0 @@
-/**
- * Read-side decoder for `content` columns in lbug node rows.
- *
- * RFC 0001 Phase 2 introduces an optional `contentEncoding` column on
- * every node table that has `content`. Default is `'none'` (passthrough)
- * so existing reads keep working unchanged. When a writer opts into
- * `--compress brotli|zstd`, the column carries the encoding tag and the
- * `content` column carries base64-encoded compressed bytes — readers
- * MUST run those bytes back through `decodeContent` before handing them
- * to a consumer (MCP tool result, HTTP API response, embedding model,
- * LLM input).
- *
- * Centralizing the decode in one helper has two benefits:
- *   1. Shim sites are 2-line changes: add `, n.contentEncoding AS
- *      contentEncoding` to the Cypher RETURN, and pipe the row through
- *      `decodeContentField` (or `decodeContentRow`) at the boundary.
- *   2. Anyone hunting for "where does the read path decode compressed
- *      bytes" greps for `decodeContentField` and gets every site in one
- *      shot — no per-table feature detection scattered across files.
- */
-import { decodeContent } from '@codragraph/graphstore';
-/**
- * Decode a single (content, contentEncoding) pair from a Cypher row.
- *
- * Returns the input content unchanged when:
- *   - the encoding is missing / empty / `'none'` (the common case for
- *     1.6.x – 1.7.x indexes, plus any 1.8+ index written without
- *     `--compress`);
- *   - content is null/undefined (caller decides whether that's an error);
- *   - content is not a string (pre-Phase-2 indexes never wrote non-string
- *     content, but defensive: don't crash a read path on a malformed row).
- *
- * Throws (via `decodeContent`) only when the row claims an encoding this
- * CLI build can't decode — that's a forward-compat error and the right
- * behavior is to fail loudly rather than return wrong content.
- */
-export function decodeContentField(content, encoding) {
-    if (content === undefined || content === null)
-        return undefined;
-    if (typeof content !== 'string')
-        return content;
-    if (typeof encoding !== 'string' || encoding === '' || encoding === 'none') {
-        return content;
-    }
-    return decodeContent(content, encoding);
-}
-/**
- * Apply `decodeContentField` to a row that carries `content` and
- * `contentEncoding` keys (or their numeric column-index aliases).
- *
- * The numeric-fallback shape (`r[N]`) mirrors LadybugDB's row format —
- * driver versions vary on whether named keys are populated, so existing
- * read sites do `r.content ?? r[N]`. This helper accepts the same
- * pattern. Returns a NEW object (does not mutate input).
- */
-export function decodeContentRow(row, contentKey = 'content', encodingKey = 'contentEncoding') {
-    const content = row[contentKey];
-    if (content === undefined || content === null)
-        return row;
-    const encoding = row[encodingKey];
-    if (typeof encoding !== 'string' || encoding === '' || encoding === 'none')
-        return row;
-    return { ...row, [contentKey]: decodeContentField(content, encoding) };
-}

package/dist/core/lbug/csv-generator.d.ts

@@ -1,29 +0,0 @@
-/**
- * CSV Generator for LadybugDB Hybrid Schema
- *
- * Streams CSV rows directly to disk files in a single pass over graph nodes.
- * File contents are lazy-read from disk per-node to avoid holding the entire
- * repo in RAM. Rows are buffered (FLUSH_EVERY) before writing to minimize
- * per-row Promise overhead.
- *
- * RFC 4180 Compliant:
- * - Fields containing commas, double quotes, or newlines are enclosed in double quotes
- * - Double quotes within fields are escaped by doubling them ("")
- * - All fields are consistently quoted for safety with code content
- */
-import { KnowledgeGraph } from '../graph/types.js';
-import { NodeTableName } from './schema.js';
-import { type ContentEncoding } from '@codragraph/graphstore';
-export declare const sanitizeUTF8: (str: string) => string;
-export declare const escapeCSVField: (value: string | number | undefined | null) => string;
-export declare const escapeCSVNumber: (value: number | undefined | null, defaultValue?: number) => string;
-export declare const isBinaryContent: (content: string) => boolean;
-export interface StreamedCSVResult {
-    nodeFiles: Map<NodeTableName, {
-        csvPath: string;
-        rows: number;
-    }>;
-    relCsvPath: string;
-    relRows: number;
-}
-export declare const streamAllCSVsToDisk: (graph: KnowledgeGraph, repoPath: string, csvDir: string, compress?: ContentEncoding) => Promise<StreamedCSVResult>;