refactor: Refactor fulltext and other search code into explain/index

kgilpin · kgilpin · commit 4545b121bf3e · 2024-11-23T11:52:52.000-05:00
diff --git a/packages/cli/src/cmds/search/search.ts b/packages/cli/src/cmds/search/search.ts
@@ -9,14 +9,14 @@ import { FileIndex } from '@appland/search';
 import { handleWorkingDirectory } from '../../lib/handleWorkingDirectory';
 import { verbose } from '../../utils';
 import searchSingleAppMap, { SearchOptions as SingleSearchOptions } from './searchSingleAppMap';
-import { SearchResponse as DiagramsSearchResponse } from '../../fulltext/appmap-match';
+import { SearchResponse as DiagramsSearchResponse } from '../../rpc/explain/index/appmap-match';
 import {
   SearchResult as EventSearchResult,
   SearchResponse as EventSearchResponse,
 } from '../../fulltext/FindEvents';
 import { openInBrowser } from '../open/openers';
-import { buildAppMapIndex, search } from '../../fulltext/appmap-index';
-import buildIndexInTempDir from '../../rpc/explain/build-index-in-temp-dir';
+import { buildAppMapIndex, search } from '../../rpc/explain/index/appmap-index';
+import buildIndexInTempDir from '../../rpc/explain/index/build-index-in-temp-dir';
 
 export const command = 'search <query>';
 export const describe =
diff --git a/packages/cli/src/cmds/search/searchSingleAppMap.ts b/packages/cli/src/cmds/search/searchSingleAppMap.ts
@@ -12,6 +12,7 @@ export default async function searchSingleAppMap(
   query: string,
   options: SearchOptions = {}
 ): Promise<SearchResponse> {
+  // eslint-disable-next-line no-param-reassign
   if (appmap.endsWith('.appmap.json')) appmap = appmap.slice(0, -'.appmap.json'.length);
 
   const findEvents = new FindEvents(appmap);
diff --git a/packages/cli/src/fulltext/FindEvents.ts b/packages/cli/src/fulltext/FindEvents.ts
@@ -7,7 +7,7 @@ import assert from 'assert';
 
 import { verbose } from '../utils';
 import { collectParameters } from './collectParameters';
-import { fileNameMatchesFilterPatterns } from './filter-patterns';
+import { fileNameMatchesFilterPatterns } from '../rpc/explain/index/filter-patterns';
 
 type IndexItem = {
   fqid: string;
diff --git a/packages/cli/src/rpc/explain/EventCollector.ts b/packages/cli/src/rpc/explain/EventCollector.ts
@@ -1,7 +1,7 @@
 import { isAbsolute, join } from 'path';
 import { ContextV2 } from '@appland/navie';
 import { SearchRpc } from '@appland/rpc';
-import { SearchResponse as AppMapSearchResponse } from '../../fulltext/appmap-match';
+import { SearchResponse as AppMapSearchResponse } from './index/appmap-match';
 import FindEvents, {
   SearchResponse as EventSearchResponse,
   SearchOptions as EventsSearchOptions,
diff --git a/packages/cli/src/rpc/explain/SearchContextCollector.ts b/packages/cli/src/rpc/explain/SearchContextCollector.ts
@@ -5,14 +5,14 @@ import { ContextV2, applyContext } from '@appland/navie';
 import { SearchRpc } from '@appland/rpc';
 import { FileIndex, FileSearchResult } from '@appland/search';
 
-import { SearchResponse as AppMapSearchResponse } from '../../fulltext/appmap-match';
+import { SearchResponse as AppMapSearchResponse } from './index/appmap-match';
 import { DEFAULT_MAX_DIAGRAMS } from '../search/search';
 import EventCollector from './EventCollector';
 import indexFiles from './index-files';
 import indexSnippets from './index-snippets';
 import collectSnippets from './collect-snippets';
-import buildIndexInTempDir from './build-index-in-temp-dir';
-import { buildAppMapIndex, search } from '../../fulltext/appmap-index';
+import buildIndexInTempDir from './index/build-index-in-temp-dir';
+import { buildAppMapIndex, search } from './index/appmap-index';
 
 export default class SearchContextCollector {
   public excludePatterns: RegExp[] | undefined;
diff --git a/packages/cli/src/rpc/explain/collect-context.ts b/packages/cli/src/rpc/explain/collect-context.ts
diff --git a/packages/cli/src/rpc/explain/collect-snippets.ts b/packages/cli/src/rpc/explain/collect-snippets.ts
@@ -1,6 +1,6 @@
 import { ContextV2 } from '@appland/navie';
 import { parseFileChunkSnippetId, SnippetIndex, SnippetSearchResult } from '@appland/search';
-import { CHARS_PER_SNIPPET } from './collectContext';
+import { CHARS_PER_SNIPPET } from './collect-context';
 
 export default function collectSnippets(
   snippetIndex: SnippetIndex,
diff --git a/packages/cli/src/rpc/explain/explain.ts b/packages/cli/src/rpc/explain/explain.ts
@@ -18,7 +18,7 @@ import collectProjectInfos from '../../cmds/navie/projectInfo';
 import configuration, { AppMapDirectory } from '../configuration';
 import { getLLMConfiguration } from '../llmConfiguration';
 import { RpcError, RpcHandler } from '../rpc';
-import collectContext from './collectContext';
+import collectContext from './collect-context';
 import { initializeHistory } from './navie/historyHelper';
 import { ThreadAccessError } from './navie/ihistory';
 import INavie, { INavieProvider } from './navie/inavie';
diff --git a/packages/cli/src/rpc/explain/fileFilter.ts b/packages/cli/src/rpc/explain/fileFilter.ts
@@ -1,6 +1,6 @@
 import { FilterFn, isBinaryFile, isDataFile, isLargeFile } from '@appland/search';
 import makeDebug from 'debug';
-import { fileNameMatchesFilterPatterns } from '../../fulltext/filter-patterns';
+import { fileNameMatchesFilterPatterns } from './index/filter-patterns';
 
 const debug = makeDebug('appmap:rpc:explain:file-filter');
 
diff --git a/packages/cli/src/rpc/explain/index-files.ts b/packages/cli/src/rpc/explain/index-files.ts
@@ -1,5 +1,4 @@
 import sqlite3 from 'better-sqlite3';
-import makeDebug from 'debug';
 
 import {
   buildFileIndex,
@@ -10,8 +9,6 @@ import {
 } from '@appland/search';
 import fileFilter from './fileFilter';
 
-const debug = makeDebug('appmap:rpc:explain:index-files');
-
 export default async function indexFiles(
   db: sqlite3.Database,
   directories: string[],
diff --git a/packages/cli/src/rpc/explain/index/appmap-file-index.ts b/packages/cli/src/rpc/explain/index/appmap-file-index.ts
@@ -0,0 +1,31 @@
+import sqlite3 from 'better-sqlite3';
+
+import { FileIndex } from '@appland/search';
+
+import buildIndexInTempDir, { CloseableIndex } from './build-index-in-temp-dir';
+import { buildAppMapIndex, search } from './appmap-index';
+import { SearchResponse } from './appmap-match';
+
+export async function buildAppMapFileIndex(
+  appmapDirectories: string[]
+): Promise<CloseableIndex<FileIndex>> {
+  return await buildIndexInTempDir<FileIndex>('appmaps', async (indexFile) => {
+    const db = new sqlite3(indexFile);
+    const fileIndex = new FileIndex(db);
+    await buildAppMapIndex(fileIndex, appmapDirectories);
+    return fileIndex;
+  });
+}
+
+export async function searchAppMapFiles(
+  appmapDirectories: string[],
+  vectorTerms: string[],
+  maxDiagrams: number
+): Promise<SearchResponse> {
+  const index = await buildAppMapFileIndex(appmapDirectories);
+  try {
+    return await search(index.index, vectorTerms.join(' OR '), maxDiagrams);
+  } finally {
+    index.close();
+  }
+}
diff --git a/packages/cli/src/rpc/explain/index/appmap-index.ts b/packages/cli/src/rpc/explain/index/appmap-index.ts
@@ -5,7 +5,7 @@ import { readFile } from 'fs/promises';
 import { Metadata } from '@appland/models';
 import { buildFileIndex, FileIndex, fileTokens } from '@appland/search';
 
-import { findFiles, isNodeError, verbose } from '../utils';
+import { findFiles, isNodeError, verbose } from '../../../utils';
 import {
   downscoreOutOfDateMatches,
   Match,
@@ -14,12 +14,12 @@ import {
   scoreMatches,
   SearchResponse,
 } from './appmap-match';
-import loadAppMapConfig from '../lib/loadAppMapConfig';
+import loadAppMapConfig from '../../../lib/loadAppMapConfig';
 
-type ClassMapEntry = {
+export type ClassMapEntry = {
   name: string;
   type: string;
-  labels: string[];
+  labels?: string[];
   children: ClassMapEntry[];
   static?: boolean;
   sourceLocation?: string;
@@ -48,43 +48,46 @@ export async function listAppMaps(directory: string): Promise<string[]> {
   return appmapFiles.map(relativeToPath);
 }
 
+export async function readIndexFile<T>(
+  appmapName: string,
+  indexName: string
+): Promise<T | undefined> {
+  const indexFile = join(appmapName, [indexName, '.json'].join(''));
+  let indexStr: string;
+  try {
+    indexStr = await readFile(indexFile, 'utf-8');
+  } catch (e) {
+    if (isNativeError(e) && !isNodeError(e, 'ENOENT')) {
+      warn(`Error reading metadata file ${indexFile}: ${e.message}`);
+    }
+    return undefined;
+  }
+
+  try {
+    return JSON.parse(indexStr) as T;
+  } catch (e) {
+    const errorMessage = isNativeError(e) ? e.message : String(e);
+    warn(`Error parsing metadata file ${indexFile}: ${errorMessage}`);
+  }
+}
+
 /**
  * Read all content for an AppMap. For efficiency, utilizes the AppMap index files, rather
  * than reading the entire AppMap file directly.
  */
 export async function readAppMapContent(appmapFile: string): Promise<string> {
   const appmapName = appmapFile.replace(/\.appmap\.json$/, '');
 
-  async function readIndexFile<T>(name: string): Promise<T | undefined> {
-    const indexFile = join(appmapName, [name, '.json'].join(''));
-    let indexStr: string;
-    try {
-      indexStr = await readFile(indexFile, 'utf-8');
-    } catch (e) {
-      if (isNativeError(e) && !isNodeError(e, 'ENOENT')) {
-        warn(`Error reading metadata file ${indexFile}: ${e.message}`);
-      }
-      return undefined;
-    }
-
-    try {
-      return JSON.parse(indexStr) as T;
-    } catch (e) {
-      const errorMessage = isNativeError(e) ? e.message : String(e);
-      warn(`Error parsing metadata file ${indexFile}: ${errorMessage}`);
-    }
-  }
-
   const appmapWords = new Array<string>();
 
-  const metadata = await readIndexFile<Metadata>('metadata');
+  const metadata = await readIndexFile<Metadata>(appmapName, 'metadata');
   if (metadata) {
     appmapWords.push(metadata.name);
     if (metadata.labels) appmapWords.push(...metadata.labels);
     if (metadata.exception) appmapWords.push(metadata.exception.message);
   }
 
-  const classMap = (await readIndexFile<ClassMapEntry[]>('classMap')) ?? [];
+  const classMap = (await readIndexFile<ClassMapEntry[]>(appmapName, 'classMap')) ?? [];
 
   const queries = new Array<string>();
   const codeObjects = new Array<string>();
@@ -119,7 +122,7 @@ export async function readAppMapContent(appmapFile: string): Promise<string> {
   classMap.forEach((co) => collectClassMapEntry(co));
   appmapWords.push(...queries, ...codeObjects, ...routes, ...externalRoutes);
 
-  const parameters = (await readIndexFile<string[]>('canonical.parameters')) ?? [];
+  const parameters = (await readIndexFile<string[]>(appmapName, 'canonical.parameters')) ?? [];
   appmapWords.push(...parameters);
   appmapWords.push(...types);
 
diff --git a/packages/cli/src/rpc/explain/index/appmap-match.ts b/packages/cli/src/rpc/explain/index/appmap-match.ts
@@ -1,5 +1,5 @@
-import UpToDate from '../lib/UpToDate';
-import { exists } from '../utils';
+import UpToDate from '../../../lib/UpToDate';
+import { exists } from '../../../utils';
 
 import makeDebug from 'debug';
 
diff --git a/packages/cli/src/rpc/explain/index/build-index-in-temp-dir.ts b/packages/cli/src/rpc/explain/index/build-index-in-temp-dir.ts
@@ -10,7 +10,7 @@ export interface Closeable {
   close(): void;
 }
 
-type CloseableIndex<T> = {
+export type CloseableIndex<T> = {
   index: T;
   close: () => void;
 };
diff --git a/packages/cli/src/rpc/explain/index/filter-patterns.ts b/packages/cli/src/rpc/explain/index/filter-patterns.ts
diff --git a/packages/cli/src/rpc/explain/index/project-file-index.ts b/packages/cli/src/rpc/explain/index/project-file-index.ts
@@ -0,0 +1,83 @@
+import sqlite3 from 'better-sqlite3';
+import makeDebug from 'debug';
+
+import {
+  buildFileIndex,
+  FileIndex,
+  FileSearchResult,
+  fileTokens,
+  FilterFn,
+  isBinaryFile,
+  isDataFile,
+  isLargeFile,
+  listProjectFiles,
+  readFileSafe,
+} from '@appland/search';
+import { fileNameMatchesFilterPatterns } from './filter-patterns';
+
+import buildIndexInTempDir, { CloseableIndex } from './build-index-in-temp-dir';
+
+const debug = makeDebug('appmap:index:project-files');
+
+function fileFilter(
+  includePatterns: RegExp[] | undefined,
+  excludePatterns: RegExp[] | undefined
+): FilterFn {
+  return async (path: string) => {
+    debug('Filtering: %s', path);
+    if (isBinaryFile(path)) {
+      debug('Skipping binary file: %s', path);
+      return false;
+    }
+
+    const includeFile = fileNameMatchesFilterPatterns(path, includePatterns, excludePatterns);
+    if (!includeFile) return false;
+
+    const isData = isDataFile(path);
+    if (isData && (await isLargeFile(path))) {
+      debug('Skipping large data file: %s', path);
+      return false;
+    }
+
+    return true;
+  };
+}
+
+async function indexFiles(
+  db: sqlite3.Database,
+  directories: string[],
+  includePatterns: RegExp[] | undefined,
+  excludePatterns: RegExp[] | undefined
+): Promise<FileIndex> {
+  const fileIndex = new FileIndex(db);
+
+  const filter = fileFilter(includePatterns, excludePatterns);
+  await buildFileIndex(fileIndex, directories, listProjectFiles, filter, readFileSafe, fileTokens);
+
+  return fileIndex;
+}
+
+export async function buildProjectFileIndex(
+  sourceDirectories: string[],
+  includePatterns: RegExp[] | undefined,
+  excludePatterns: RegExp[] | undefined
+): Promise<CloseableIndex<FileIndex>> {
+  return await buildIndexInTempDir('files', async (indexFile) => {
+    const db = new sqlite3(indexFile);
+    return await indexFiles(db, sourceDirectories, includePatterns, excludePatterns);
+  });
+}
+
+export async function searchProjectFiles(
+  sourceDirectories: string[],
+  includePatterns: RegExp[] | undefined,
+  excludePatterns: RegExp[] | undefined,
+  vectorTerms: string[]
+): Promise<FileSearchResult[]> {
+  const index = await buildProjectFileIndex(sourceDirectories, includePatterns, excludePatterns);
+  try {
+    return index.index.search(vectorTerms.join(' OR '));
+  } finally {
+    index.close();
+  }
+}
diff --git a/packages/cli/src/rpc/search/search.ts b/packages/cli/src/rpc/search/search.ts
@@ -4,12 +4,12 @@ import { FileIndex } from '@appland/search';
 import { SearchRpc } from '@appland/rpc';
 
 import { RpcHandler } from '../rpc';
-import { SearchResponse } from '../../fulltext/appmap-match';
-import { search as searchAppMaps } from '../../fulltext/appmap-index';
+import { SearchResponse } from '../explain/index/appmap-match';
+import { search as searchAppMaps } from '../explain/index/appmap-index';
 import searchSingleAppMap from '../../cmds/search/searchSingleAppMap';
 import configuration, { AppMapDirectory } from '../configuration';
-import buildIndexInTempDir from '../explain/build-index-in-temp-dir';
-import { buildAppMapIndex } from '../../fulltext/appmap-index';
+import buildIndexInTempDir from '../explain/index/build-index-in-temp-dir';
+import { buildAppMapIndex } from '../explain/index/appmap-index';
 
 export const DEFAULT_MAX_DIAGRAMS = 10;
 export const DEFAULT_MAX_EVENTS_PER_DIAGRAM = 100;
diff --git a/packages/cli/tests/unit/fulltext/appmap-index.readAppMapContent.spec.ts b/packages/cli/tests/unit/fulltext/appmap-index.readAppMapContent.spec.ts
@@ -1,5 +1,5 @@
 import { vol } from 'memfs';
-import { readAppMapContent } from '../../../src/fulltext/appmap-index';
+import { readAppMapContent } from '../../../src/rpc/explain/index/appmap-index';
 import { Metadata } from '@appland/models';
 
 jest.mock('fs/promises', () => require('memfs').promises);
diff --git a/packages/cli/tests/unit/fulltext/appmap-index.search.spec.ts b/packages/cli/tests/unit/fulltext/appmap-index.search.spec.ts
@@ -3,8 +3,8 @@ import UpToDate, { AppMapIndex } from '../../../src/lib/UpToDate';
 import { PathLike } from 'fs';
 import { join } from 'path';
 import { FileIndex, FileSearchResult } from '@appland/search';
-import { search } from '../../../src/fulltext/appmap-index';
-import { SearchStats } from '../../../src/fulltext/appmap-match';
+import { search } from '../../../src/rpc/explain/index/appmap-index';
+import { SearchStats } from '../../../src/rpc/explain/index/appmap-match';
 
 jest.mock('../../../src/utils');
 jest.mock('../../../src/lib/UpToDate');
diff --git a/packages/cli/tests/unit/rpc/explain/ContextCollector.spec.ts b/packages/cli/tests/unit/rpc/explain/ContextCollector.spec.ts
@@ -1,4 +1,4 @@
-import { ContextCollector } from '../../../../src/rpc/explain/collectContext';
+import { ContextCollector } from '../../../../src/rpc/explain/collect-context';
 import * as SearchContextCollector from '../../../../src/rpc/explain/SearchContextCollector';
 import * as LocationContextCollector from '../../../../src/rpc/explain/LocationContextCollector';
 import * as navie from '@appland/navie';
diff --git a/packages/cli/tests/unit/rpc/explain/EventCollector.spec.ts b/packages/cli/tests/unit/rpc/explain/EventCollector.spec.ts
@@ -3,7 +3,7 @@ import { join } from 'path';
 
 import { textSearchResultToRpcSearchResult } from '../../../../src/rpc/explain/textSearchResultToRpcSearchResult';
 import buildContext from '../../../../src/rpc/explain/buildContext';
-import { SearchResponse as AppMapSearchResponse } from '../../../../src/fulltext/appmap-match';
+import { SearchResponse as AppMapSearchResponse } from '../../../../src/rpc/explain/index/appmap-match';
 import FindEvents, {
   SearchResponse as EventSearchResponse,
 } from '../../../../src/fulltext/FindEvents';
diff --git a/packages/cli/tests/unit/rpc/explain/index/appmap-index.search.spec.ts b/packages/cli/tests/unit/rpc/explain/index/appmap-index.search.spec.ts
diff --git a/packages/cli/tests/unit/rpc/explain/index/appmap-index.spec.ts b/packages/cli/tests/unit/rpc/explain/index/appmap-index.spec.ts
diff --git a/packages/cli/tests/unit/rpc/explain/pattern.spec.ts b/packages/cli/tests/unit/rpc/explain/pattern.spec.ts
