fegger/obsidian_ollama
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Add list_vault_tags and get_vault_stats tools to all agent modes

Browse Source 
Extend the tool registry to support vault-wide tag listing and statistical
overview tools, adding them to READ, EDIT, ORGANIZE, and RESEARCH agent
modes. Includes implementations for tag aggregation, folder structure
reporting, and metadataCache integration in VaultIndexer.
This commit is contained in:
Florian Egger
2026-05-21 10:08:05 +02:00
parent 5350df92dd
commit af9f0d165a
10 changed files with 676 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files
main.js
+212 -27
View File
@@ -3124,10 +3124,17 @@ var ALL_AGENT_MODES = ["ask", "edit", "organize", "research", "workflow"];
function filterToolsByName(tools, allowed) {
return tools.filter((t) => allowed.has(t.function.name));
}
var READ_TOOLS = /* @__PURE__ */ new Set(["read_vault_file", "search_vault_files"]);
var READ_TOOLS = /* @__PURE__ */ new Set([
"read_vault_file",
"search_vault_files",
"list_vault_tags",
"get_vault_stats"
]);
var ORGANIZE_TOOLS = /* @__PURE__ */ new Set([
"read_vault_file",
"search_vault_files",
"list_vault_tags",
"get_vault_stats",
"update_frontmatter",
"rename_note",
"move_note",
@@ -3136,6 +3143,8 @@ var ORGANIZE_TOOLS = /* @__PURE__ */ new Set([
var EDIT_TOOLS = /* @__PURE__ */ new Set([
"read_vault_file",
"search_vault_files",
"list_vault_tags",
"get_vault_stats",
"create_note",
"append_to_note",
"replace_note_section",
@@ -3145,7 +3154,12 @@ var EDIT_TOOLS = /* @__PURE__ */ new Set([
"delete_note",
"insert_link"
]);
var RESEARCH_TOOLS = /* @__PURE__ */ new Set(["read_vault_file", "search_vault_files"]);
var RESEARCH_TOOLS = /* @__PURE__ */ new Set([
"read_vault_file",
"search_vault_files",
"list_vault_tags",
"get_vault_stats"
]);
var AGENT_MODE_CONFIGS = {
ask: {
label: "Ask",
@@ -3182,8 +3196,9 @@ When editing notes:
label: "Organize",
description: "Tag, rename, move, and link notes to keep the vault tidy.",
systemPrompt: `You are an assistant that helps organize the user's Obsidian vault.
You can search notes, read them, update frontmatter tags, rename files, move files to folders, and insert wiki-links.
IMPORTANT: When you need vault information, do NOT say you will search or read files. You MUST immediately emit a tool_call to the appropriate tool.
You can search notes, read them, list all vault tags, get vault structure stats, update frontmatter tags, rename files, move files to folders, and insert wiki-links.
IMPORTANT: When the user asks about tags or folder structure, use list_vault_tags or get_vault_stats FIRST instead of searching blindly.
When you need vault information, do NOT say you will search or read files. You MUST immediately emit a tool_call to the appropriate tool.
Only respond to the user after you have received and analyzed the tool results.
When organizing:
- Suggest consistent tag vocabularies.
@@ -3199,7 +3214,8 @@ When organizing:
description: "Deep vault search and synthesis across multiple notes.",
systemPrompt: `You are a research assistant that dives deep into the user's Obsidian vault.
Your job is to synthesize information across multiple notes, find connections, and produce comprehensive summaries.
IMPORTANT: When you need vault information, do NOT say you will search or read files. You MUST immediately emit a tool_call to search_vault_files or read_vault_file.
IMPORTANT: When the user asks about tags or vault structure, use list_vault_tags or get_vault_stats FIRST before searching.
When you need vault information, do NOT say you will search or read files. You MUST immediately emit a tool_call to search_vault_files or read_vault_file.
Only respond to the user after you have received and analyzed the tool results.
Search broadly, read key sources, and cross-reference information.
Cite specific notes and quotes where possible.
@@ -8565,11 +8581,14 @@ var VaultIndexer = class {
setVectorStore(vectorStore) {
this.vectorStore = vectorStore;
}
setApp(app) {
this.app = app;
}
tokenize(text) {
return text.toLowerCase().replace(/[^\w\s]/g, "").split(/\s+/).filter((token) => token.length > 1 && !STOP_WORDS.has(token));
}
tokenizeContent(content, file) {
const parsed = this.parseMarkdown(content);
const parsed = this.parseMarkdown(content, file);
const bodyWithoutFrontmatter = content.replace(/^---[\s\S]*?---\n/, "");
const paragraphs = bodyWithoutFrontmatter.split(/\n\n+/).map((p) => p.trim()).filter((p) => p && !p.startsWith("#"));
const firstParagraph = paragraphs[0] || "";
@@ -8629,7 +8648,7 @@ var VaultIndexer = class {
for (const file of files) {
try {
const content = typeof this.vault.cachedRead === "function" ? await this.vault.cachedRead(file) : await this.vault.read(file);
const parsed = this.parseMarkdown(content);
const parsed = this.parseMarkdown(content, file);
entries.push({
file,
title: parsed.frontmatter.title || file.basename,
@@ -8777,32 +8796,62 @@ var VaultIndexer = class {
const queryStem = this.stemToken(queryLower);
return textLower.includes(queryLower) || textLower.includes(queryStem);
}
parseMarkdown(content) {
const frontmatterRegex = /^---\n([\s\S]*?)\n---\n/;
const frontmatterMatch = content.match(frontmatterRegex);
const frontmatter = {};
if (frontmatterMatch) {
parseMarkdown(content, file) {
let frontmatter = {};
let title = "";
let headings = [];
if (this.app && file) {
try {
const lines = frontmatterMatch[1].trim().split("\n");
for (const line of lines) {
const [key, ...valueParts] = line.split(":");
if (!key) continue;
const value = valueParts.join(":").trim();
if (key.trim() === "title" && value) frontmatter.title = value;
else if (key.trim() === "tags" && value) frontmatter.tags = value;
const cache = this.app.metadataCache.getFileCache(file);
if (cache?.frontmatter) {
const fm = cache.frontmatter;
if (typeof fm.title === "string") frontmatter.title = fm.title;
const rawTags = fm.tags;
if (Array.isArray(rawTags)) {
frontmatter.tags = rawTags.map(String).join(", ");
} else if (typeof rawTags === "string") {
frontmatter.tags = rawTags;
}
}
if (cache?.headings) {
headings = cache.headings.map((h) => h.heading);
}
if (headings.length > 0) {
title = headings[0];
}
} catch {
Logger.warn("Failed to parse frontmatter", "vault-indexer");
}
}
const titleMatch = content.match(/^# (.+)$/m);
const title = titleMatch ? titleMatch[1] : "";
const headings = [];
const headingRegex = /^#{1,6} (.+)$/gm;
let headingMatch;
while ((headingMatch = headingRegex.exec(content)) !== null) {
headings.push(headingMatch[1]);
if (Object.keys(frontmatter).length === 0) {
const frontmatterRegex = /^---\n([\s\S]*?)\n---\n/;
const frontmatterMatch2 = content.match(frontmatterRegex);
if (frontmatterMatch2) {
try {
const lines = frontmatterMatch2[1].trim().split("\n");
for (const line of lines) {
const [key, ...valueParts] = line.split(":");
if (!key) continue;
const value = valueParts.join(":").trim();
if (key.trim() === "title" && value) frontmatter.title = value;
else if (key.trim() === "tags" && value) frontmatter.tags = value;
}
} catch {
Logger.warn("Failed to parse frontmatter", "vault-indexer");
}
}
}
if (!title) {
const titleMatch = content.match(/^# (.+)$/m);
title = titleMatch ? titleMatch[1] : "";
}
if (headings.length === 0) {
const headingRegex = /^#{1,6} (.+)$/gm;
let headingMatch;
while ((headingMatch = headingRegex.exec(content)) !== null) {
headings.push(headingMatch[1]);
}
}
const frontmatterMatch = content.match(/^---\n([\s\S]*?)\n---\n/);
const bodyWithoutFrontmatter = frontmatterMatch ? content.substring(frontmatterMatch[0].length) : content;
const bodyText = bodyWithoutFrontmatter.replace(/#{1,6} .+/g, "").replace(/^\s*[\r\n]/gm, "").trim();
return { frontmatter, title, headings, content: bodyText };
@@ -8923,6 +8972,12 @@ var ToolExecutor = class {
case "insert_link":
result = await this.handleInsertLink(parsedArgs);
break;
case "list_vault_tags":
result = await this.handleListVaultTags(parsedArgs);
break;
case "get_vault_stats":
result = await this.handleGetVaultStats(parsedArgs);
break;
default:
result = { success: false, message: `Unknown tool: ${toolName}` };
}
@@ -9217,6 +9272,105 @@ ${lines.join("\n")}
await this.vault.delete(file);
return { success: true, message: `Note ${path} deleted successfully` };
}
async handleListVaultTags(args) {
const sortBy = args.sortBy === "count" ? "count" : "name";
const files = this.vault.getMarkdownFiles();
const tagMap = /* @__PURE__ */ new Map();
for (const file of files) {
try {
const cache = this.app.metadataCache.getFileCache(file);
const rawTags = cache?.frontmatter?.tags;
const tagList = [];
if (Array.isArray(rawTags)) {
tagList.push(...rawTags.map(String));
} else if (typeof rawTags === "string") {
tagList.push(
...rawTags.split(/[,\n]+/).map((t) => t.trim()).filter((t) => t.length > 0)
);
}
for (const tag of tagList) {
const existing = tagMap.get(tag);
if (existing) {
existing.count++;
if (existing.notes.length < 5) existing.notes.push(file.path);
} else {
tagMap.set(tag, { count: 1, notes: [file.path] });
}
}
} catch {
}
}
const entries = Array.from(tagMap.entries()).map(([tag, data]) => ({
tag,
count: data.count,
sampleNotes: data.notes.slice(0, 3)
}));
if (sortBy === "count") {
entries.sort((a, b) => b.count - a.count);
} else {
entries.sort((a, b) => a.tag.localeCompare(b.tag));
}
return {
success: true,
message: `Found ${entries.length} unique tags across ${files.length} notes`,
data: entries
};
}
async handleGetVaultStats(args) {
const files = this.vault.getMarkdownFiles();
const folderSet = /* @__PURE__ */ new Set();
let totalLength = 0;
let taggedCount = 0;
let untaggedCount = 0;
const tagMap = /* @__PURE__ */ new Map();
const recentFiles = [];
for (const file of files) {
try {
const folder = file.path.split("/").slice(0, -1).join("/") || "(root)";
folderSet.add(folder);
const content = await this.vault.cachedRead(file);
totalLength += content.length;
const cache = this.app.metadataCache.getFileCache(file);
const rawTags = cache?.frontmatter?.tags;
let hasTags = false;
if (Array.isArray(rawTags) && rawTags.length > 0) {
hasTags = true;
for (const tag of rawTags.map(String)) {
tagMap.set(tag, (tagMap.get(tag) ?? 0) + 1);
}
} else if (typeof rawTags === "string" && rawTags.trim().length > 0 && rawTags.trim() !== "[]") {
hasTags = true;
for (const tag of rawTags.split(/[,\n]+/).map((t) => t.trim()).filter((t) => t.length > 0)) {
tagMap.set(tag, (tagMap.get(tag) ?? 0) + 1);
}
}
if (hasTags) {
taggedCount++;
} else {
untaggedCount++;
}
if (file.stat?.mtime) {
recentFiles.push({ path: file.path, mtime: file.stat.mtime });
}
} catch {
}
}
recentFiles.sort((a, b) => b.mtime - a.mtime);
return {
success: true,
message: `Vault has ${files.length} notes in ${folderSet.size} folders`,
data: {
totalNotes: files.length,
totalFolders: folderSet.size,
folders: Array.from(folderSet).sort(),
taggedNotes: taggedCount,
untaggedNotes: untaggedCount,
topTags: Array.from(tagMap.entries()).sort((a, b) => b[1] - a[1]).slice(0, 20).map(([tag, count]) => ({ tag, count })),
avgNoteLength: files.length > 0 ? Math.round(totalLength / files.length) : 0,
recentFiles: recentFiles.slice(0, 10).map((f) => f.path)
}
};
}
async handleInsertLink(args) {
const sourcePath = args.sourcePath;
const targetPath = args.targetPath;
@@ -9700,6 +9854,7 @@ var VARIABLE_PATTERN = /\{\{([\w.]+)\}\}/g;
var WorkflowEngine = class _WorkflowEngine {
constructor(vault, app, ollamaUrl, model, options) {
this.vaultIndexer = new VaultIndexer(vault);
this.vaultIndexer.setApp(app);
this.toolExecutor = new ToolExecutor(vault, app, void 0, this.vaultIndexer);
this.ollamaClient = new OllamaClient(ollamaUrl, model, void 0, options?.cacheConfig);
this.conversationStateManager = new ConversationStateManager();
@@ -10748,6 +10903,7 @@ var ChatView = class extends import_obsidian5.ItemView {
this.ollamaClient = this.createOllamaClient(settings.chatModel ?? settings.model, settings);
this.agentOllamaClient = (settings.agentModel ?? settings.model) === (settings.chatModel ?? settings.model) ? this.ollamaClient : this.createOllamaClient(settings.agentModel ?? settings.model, settings);
this.vaultIndexer = new VaultIndexer(this.app.vault, void 0, vectorStore);
this.vaultIndexer.setApp(this.app);
this.toolExecutor = new ToolExecutor(
this.app.vault,
this.app,
@@ -11272,6 +11428,34 @@ var ChatView = class extends import_obsidian5.ItemView {
}
}
},
{
type: "function",
function: {
name: "list_vault_tags",
description: "Lists all unique tags used across the vault with usage counts. Use this when the user asks about tags, tag organization, or wants to see all tags. Do NOT use search_vault_files for tag listing.",
parameters: {
type: "object",
properties: {
sortBy: {
type: "string",
description: 'Sort by "name" (alphabetical) or "count" (most used first). Default: "name"',
enum: ["name", "count"]
}
}
}
}
},
{
type: "function",
function: {
name: "get_vault_stats",
description: "Returns an overview of the vault: total notes, folder structure, tag distribution, average note length, and recent files. Use this when the user asks about vault structure, folder organization, or wants a high-level overview before making organizational suggestions.",
parameters: {
type: "object",
properties: {}
}
}
},
{
type: "function",
function: {
@@ -13182,6 +13366,7 @@ var OllamaPlugin = class extends import_obsidian7.Plugin {
);
if (!this.autoLinker) {
const vaultIndexer = new VaultIndexer(this.app.vault, void 0, this.vaultVectorStore);
vaultIndexer.setApp(this.app);
this.autoLinker = new AutoLinker(
this.app.vault,
vaultIndexer,
src/agent-modes.ts
+21 -5
View File
@@ -25,11 +25,18 @@ function filterToolsByName(tools: OllamaTool[], allowed: Set<string>): OllamaToo
return tools.filter((t) => allowed.has(t.function.name));
}
const READ_TOOLS = new Set(['read_vault_file', 'search_vault_files']);
const READ_TOOLS = new Set([
'read_vault_file',
'search_vault_files',
'list_vault_tags',
'get_vault_stats',
]);
const ORGANIZE_TOOLS = new Set([
'read_vault_file',
'search_vault_files',
'list_vault_tags',
'get_vault_stats',
'update_frontmatter',
'rename_note',
'move_note',
@@ -39,6 +46,8 @@ const ORGANIZE_TOOLS = new Set([
const EDIT_TOOLS = new Set([
'read_vault_file',
'search_vault_files',
'list_vault_tags',
'get_vault_stats',
'create_note',
'append_to_note',
'replace_note_section',
@@ -49,7 +58,12 @@ const EDIT_TOOLS = new Set([
'insert_link',
]);
const RESEARCH_TOOLS = new Set(['read_vault_file', 'search_vault_files']);
const RESEARCH_TOOLS = new Set([
'read_vault_file',
'search_vault_files',
'list_vault_tags',
'get_vault_stats',
]);
export const AGENT_MODE_CONFIGS: Record<AgentMode, AgentModeConfig> = {
ask: {
@@ -89,8 +103,9 @@ When editing notes:
label: 'Organize',
description: 'Tag, rename, move, and link notes to keep the vault tidy.',
systemPrompt: `You are an assistant that helps organize the user's Obsidian vault.
You can search notes, read them, update frontmatter tags, rename files, move files to folders, and insert wiki-links.
IMPORTANT: When you need vault information, do NOT say you will search or read files. You MUST immediately emit a tool_call to the appropriate tool.
You can search notes, read them, list all vault tags, get vault structure stats, update frontmatter tags, rename files, move files to folders, and insert wiki-links.
IMPORTANT: When the user asks about tags or folder structure, use list_vault_tags or get_vault_stats FIRST instead of searching blindly.
When you need vault information, do NOT say you will search or read files. You MUST immediately emit a tool_call to the appropriate tool.
Only respond to the user after you have received and analyzed the tool results.
When organizing:
- Suggest consistent tag vocabularies.
@@ -107,7 +122,8 @@ When organizing:
description: 'Deep vault search and synthesis across multiple notes.',
systemPrompt: `You are a research assistant that dives deep into the user's Obsidian vault.
Your job is to synthesize information across multiple notes, find connections, and produce comprehensive summaries.
IMPORTANT: When you need vault information, do NOT say you will search or read files. You MUST immediately emit a tool_call to search_vault_files or read_vault_file.
IMPORTANT: When the user asks about tags or vault structure, use list_vault_tags or get_vault_stats FIRST before searching.
When you need vault information, do NOT say you will search or read files. You MUST immediately emit a tool_call to search_vault_files or read_vault_file.
Only respond to the user after you have received and analyzed the tool results.
Search broadly, read key sources, and cross-reference information.
Cite specific notes and quotes where possible.
src/chat-view.ts
+32
View File
@@ -74,6 +74,7 @@ export class ChatView extends ItemView {
? this.ollamaClient
: this.createOllamaClient(settings.agentModel ?? settings.model, settings);
this.vaultIndexer = new VaultIndexer(this.app.vault, undefined, vectorStore);
this.vaultIndexer.setApp(this.app);
this.toolExecutor = new ToolExecutor(
this.app.vault,
this.app,
@@ -666,6 +667,37 @@ export class ChatView extends ItemView {
},
},
},
{
type: 'function',
function: {
name: 'list_vault_tags',
description:
'Lists all unique tags used across the vault with usage counts. Use this when the user asks about tags, tag organization, or wants to see all tags. Do NOT use search_vault_files for tag listing.',
parameters: {
type: 'object',
properties: {
sortBy: {
type: 'string',
description:
'Sort by "name" (alphabetical) or "count" (most used first). Default: "name"',
enum: ['name', 'count'],
},
},
},
},
},
{
type: 'function',
function: {
name: 'get_vault_stats',
description:
'Returns an overview of the vault: total notes, folder structure, tag distribution, average note length, and recent files. Use this when the user asks about vault structure, folder organization, or wants a high-level overview before making organizational suggestions.',
parameters: {
type: 'object',
properties: {},
},
},
},
{
type: 'function',
function: {
src/main.ts
+1
View File
@@ -211,6 +211,7 @@ export default class OllamaPlugin extends Plugin {
if (!this.autoLinker) {
const vaultIndexer = new VaultIndexer(this.app.vault, undefined, this.vaultVectorStore);
vaultIndexer.setApp(this.app);
this.autoLinker = new AutoLinker(
this.app.vault,
vaultIndexer,
src/tool-executor.ts
+133
View File
@@ -157,6 +157,12 @@ export class ToolExecutor {
case 'insert_link':
result = await this.handleInsertLink(parsedArgs);
break;
case 'list_vault_tags':
result = await this.handleListVaultTags(parsedArgs);
break;
case 'get_vault_stats':
result = await this.handleGetVaultStats(parsedArgs);
break;
default:
result = { success: false, message: `Unknown tool: ${toolName}` };
}
@@ -535,6 +541,133 @@ export class ToolExecutor {
return { success: true, message: `Note ${path} deleted successfully` };
}
private async handleListVaultTags(args: Record<string, unknown>): Promise<ToolResult> {
const sortBy = args.sortBy === 'count' ? 'count' : 'name';
const files = this.vault.getMarkdownFiles();
const tagMap = new Map<string, { count: number; notes: string[] }>();
for (const file of files) {
try {
const cache = this.app.metadataCache.getFileCache(file);
const rawTags: unknown = cache?.frontmatter?.tags;
const tagList: string[] = [];
if (Array.isArray(rawTags)) {
tagList.push(...rawTags.map(String));
} else if (typeof rawTags === 'string') {
tagList.push(
...rawTags
.split(/[,\n]+/)
.map((t) => t.trim())
.filter((t) => t.length > 0)
);
}
for (const tag of tagList) {
const existing = tagMap.get(tag);
if (existing) {
existing.count++;
if (existing.notes.length < 5) existing.notes.push(file.path);
} else {
tagMap.set(tag, { count: 1, notes: [file.path] });
}
}
} catch {
// skip unreadable files
}
}
const entries = Array.from(tagMap.entries()).map(([tag, data]) => ({
tag,
count: data.count,
sampleNotes: data.notes.slice(0, 3),
}));
if (sortBy === 'count') {
entries.sort((a, b) => b.count - a.count);
} else {
entries.sort((a, b) => a.tag.localeCompare(b.tag));
}
return {
success: true,
message: `Found ${entries.length} unique tags across ${files.length} notes`,
data: entries,
};
}
private async handleGetVaultStats(args: Record<string, unknown>): Promise<ToolResult> {
const files = this.vault.getMarkdownFiles();
const folderSet = new Set<string>();
let totalLength = 0;
let taggedCount = 0;
let untaggedCount = 0;
const tagMap = new Map<string, number>();
const recentFiles: { path: string; mtime: number }[] = [];
for (const file of files) {
try {
const folder = file.path.split('/').slice(0, -1).join('/') || '(root)';
folderSet.add(folder);
const content = await this.vault.cachedRead(file);
totalLength += content.length;
const cache = this.app.metadataCache.getFileCache(file);
const rawTags: unknown = cache?.frontmatter?.tags;
let hasTags = false;
if (Array.isArray(rawTags) && rawTags.length > 0) {
hasTags = true;
for (const tag of rawTags.map(String)) {
tagMap.set(tag, (tagMap.get(tag) ?? 0) + 1);
}
} else if (
typeof rawTags === 'string' &&
rawTags.trim().length > 0 &&
rawTags.trim() !== '[]'
) {
hasTags = true;
for (const tag of rawTags
.split(/[,\n]+/)
.map((t) => t.trim())
.filter((t) => t.length > 0)) {
tagMap.set(tag, (tagMap.get(tag) ?? 0) + 1);
}
}
if (hasTags) {
taggedCount++;
} else {
untaggedCount++;
}
if (file.stat?.mtime) {
recentFiles.push({ path: file.path, mtime: file.stat.mtime });
}
} catch {
// skip unreadable files
}
}
recentFiles.sort((a, b) => b.mtime - a.mtime);
return {
success: true,
message: `Vault has ${files.length} notes in ${folderSet.size} folders`,
data: {
totalNotes: files.length,
totalFolders: folderSet.size,
folders: Array.from(folderSet).sort(),
taggedNotes: taggedCount,
untaggedNotes: untaggedCount,
topTags: Array.from(tagMap.entries())
.sort((a, b) => b[1] - a[1])
.slice(0, 20)
.map(([tag, count]) => ({ tag, count })),
avgNoteLength: files.length > 0 ? Math.round(totalLength / files.length) : 0,
recentFiles: recentFiles.slice(0, 10).map((f) => f.path),
},
};
}
private async handleInsertLink(args: Record<string, unknown>): Promise<ToolResult> {
const sourcePath = args.sourcePath;
const targetPath = args.targetPath;
src/types.ts
+1
View File
@@ -114,6 +114,7 @@ export interface OllamaTool {
[key: string]: {
type: string;
description?: string;
enum?: string[];
};
};
required?: string[];
src/vault-indexer.ts
+66 -24
View File
@@ -1,6 +1,6 @@
// src/vault-indexer.ts
import { Vault, TFile } from 'obsidian';
import { Vault, TFile, App } from 'obsidian';
import { Logger } from './utils';
import { Cache } from './cache';
import { VaultVectorStore } from './vault-vector-store';
@@ -110,6 +110,7 @@ const DEFAULT_RECENCY_HALF_LIFE = 30; // 30 days
export class VaultIndexer {
private vault: Vault;
private app?: App;
private cache?: Cache;
private vectorStore?: VaultVectorStore;
private readonly SCORING_WEIGHTS = {
@@ -134,6 +135,10 @@ export class VaultIndexer {
this.vectorStore = vectorStore;
}
setApp(app: App | undefined): void {
this.app = app;
}
tokenize(text: string): string[] {
return text
.toLowerCase()
@@ -143,7 +148,7 @@ export class VaultIndexer {
}
tokenizeContent(content: string, file: TFile): TokenizedContent {
const parsed = this.parseMarkdown(content);
const parsed = this.parseMarkdown(content, file);
const bodyWithoutFrontmatter = content.replace(/^---[\s\S]*?---\n/, '');
const paragraphs = bodyWithoutFrontmatter
.split(/\n\n+/)
@@ -221,7 +226,7 @@ export class VaultIndexer {
typeof this.vault.cachedRead === 'function'
? await this.vault.cachedRead(file)
: await this.vault.read(file);
const parsed = this.parseMarkdown(content);
const parsed = this.parseMarkdown(content, file);
entries.push({
file,
title: parsed.frontmatter.title || file.basename,
@@ -404,35 +409,72 @@ export class VaultIndexer {
return textLower.includes(queryLower) || textLower.includes(queryStem);
}
private parseMarkdown(content: string) {
const frontmatterRegex = /^---\n([\s\S]*?)\n---\n/;
const frontmatterMatch = content.match(frontmatterRegex);
const frontmatter: ParsedFrontmatter = {};
if (frontmatterMatch) {
private parseMarkdown(content: string, file?: TFile) {
let frontmatter: ParsedFrontmatter = {};
let title = '';
let headings: string[] = [];
// Use metadataCache when available for accurate frontmatter and headings parsing
if (this.app && file) {
try {
const lines = frontmatterMatch[1].trim().split('\n');
for (const line of lines) {
const [key, ...valueParts] = line.split(':');
if (!key) continue;
const value = valueParts.join(':').trim();
if (key.trim() === 'title' && value) frontmatter.title = value;
else if (key.trim() === 'tags' && value) frontmatter.tags = value;
const cache = this.app.metadataCache.getFileCache(file);
if (cache?.frontmatter) {
const fm = cache.frontmatter as Record<string, unknown>;
if (typeof fm.title === 'string') frontmatter.title = fm.title;
const rawTags = fm.tags;
if (Array.isArray(rawTags)) {
frontmatter.tags = rawTags.map(String).join(', ');
} else if (typeof rawTags === 'string') {
frontmatter.tags = rawTags;
}
}
if (cache?.headings) {
headings = cache.headings.map((h) => h.heading);
}
if (headings.length > 0) {
title = headings[0];
}
} catch {
Logger.warn('Failed to parse frontmatter', 'vault-indexer');
// Fall back to regex parsing below
}
}
const titleMatch = content.match(/^# (.+)$/m);
const title = titleMatch ? titleMatch[1] : '';
const headings: string[] = [];
const headingRegex = /^#{1,6} (.+)$/gm;
let headingMatch;
while ((headingMatch = headingRegex.exec(content)) !== null) {
headings.push(headingMatch[1]);
// Fallback regex parsing for frontmatter when metadataCache is unavailable
if (Object.keys(frontmatter).length === 0) {
const frontmatterRegex = /^---\n([\s\S]*?)\n---\n/;
const frontmatterMatch = content.match(frontmatterRegex);
if (frontmatterMatch) {
try {
const lines = frontmatterMatch[1].trim().split('\n');
for (const line of lines) {
const [key, ...valueParts] = line.split(':');
if (!key) continue;
const value = valueParts.join(':').trim();
if (key.trim() === 'title' && value) frontmatter.title = value;
else if (key.trim() === 'tags' && value) frontmatter.tags = value;
}
} catch {
Logger.warn('Failed to parse frontmatter', 'vault-indexer');
}
}
}
// Fallback regex parsing for title
if (!title) {
const titleMatch = content.match(/^# (.+)$/m);
title = titleMatch ? titleMatch[1] : '';
}
// Fallback regex parsing for headings
if (headings.length === 0) {
const headingRegex = /^#{1,6} (.+)$/gm;
let headingMatch;
while ((headingMatch = headingRegex.exec(content)) !== null) {
headings.push(headingMatch[1]);
}
}
const frontmatterMatch = content.match(/^---\n([\s\S]*?)\n---\n/);
const bodyWithoutFrontmatter = frontmatterMatch
? content.substring(frontmatterMatch[0].length)
: content;
src/workflow-engine/workflow-engine.ts
+1
View File
@@ -58,6 +58,7 @@ export class WorkflowEngine {
}
) {
this.vaultIndexer = new VaultIndexer(vault);
this.vaultIndexer.setApp(app);
this.toolExecutor = new ToolExecutor(vault, app, undefined, this.vaultIndexer);
this.ollamaClient = new OllamaClient(ollamaUrl, model, undefined, options?.cacheConfig);
this.conversationStateManager = new ConversationStateManager();
tests/tool-executor.test.ts
+135
View File
@@ -1295,6 +1295,141 @@ describe('ToolExecutor', () => {
});
});
describe('list_vault_tags tool', () => {
it('should list all tags sorted by name', async () => {
class MockTFile extends TFile {
path: string;
basename: string;
extension: string;
name: string;
constructor(path: string) {
super();
this.path = path;
this.basename = path.split('/').pop() || path;
this.extension = this.basename.split('.').pop() || '';
this.name = this.basename;
}
}
const files = [new MockTFile('a.md'), new MockTFile('b.md'), new MockTFile('c.md')];
mockVault.getMarkdownFiles = jest.fn().mockReturnValue(files);
mockApp.metadataCache.getFileCache = jest.fn().mockImplementation((f: MockTFile) => {
if (f.path === 'a.md') return { frontmatter: { tags: ['project', 'alpha'] } };
if (f.path === 'b.md') return { frontmatter: { tags: 'project, beta' } };
return { frontmatter: {} };
});
const call: ToolCall = {
id: 'call_lt1',
type: 'function',
function: {
name: 'list_vault_tags',
arguments: JSON.stringify({ sortBy: 'name' }),
},
};
const result = await executor.handleToolCall(call);
expect(result.success).toBe(true);
expect((result.data as any[]).length).toBe(3);
expect((result.data as any[])[0].tag).toBe('alpha');
expect((result.data as any[])[1].tag).toBe('beta');
expect((result.data as any[])[2].tag).toBe('project');
expect((result.data as any[])[2].count).toBe(2);
});
it('should sort tags by count', async () => {
class MockTFile extends TFile {
path: string;
basename: string;
extension: string;
name: string;
constructor(path: string) {
super();
this.path = path;
this.basename = path.split('/').pop() || path;
this.extension = this.basename.split('.').pop() || '';
this.name = this.basename;
}
}
const files = [new MockTFile('a.md'), new MockTFile('b.md')];
mockVault.getMarkdownFiles = jest.fn().mockReturnValue(files);
mockApp.metadataCache.getFileCache = jest.fn().mockImplementation((f: MockTFile) => {
if (f.path === 'a.md') return { frontmatter: { tags: ['common', 'rare'] } };
if (f.path === 'b.md') return { frontmatter: { tags: ['common'] } };
return { frontmatter: {} };
});
const call: ToolCall = {
id: 'call_lt2',
type: 'function',
function: {
name: 'list_vault_tags',
arguments: JSON.stringify({ sortBy: 'count' }),
},
};
const result = await executor.handleToolCall(call);
expect(result.success).toBe(true);
const data = result.data as any[];
expect(data[0].tag).toBe('common');
expect(data[0].count).toBe(2);
expect(data[1].tag).toBe('rare');
expect(data[1].count).toBe(1);
});
});
describe('get_vault_stats tool', () => {
it('should return vault overview stats', async () => {
class MockTFile extends TFile {
path: string;
basename: string;
extension: string;
name: string;
constructor(path: string, mtime?: number) {
super();
this.path = path;
this.basename = path.split('/').pop() || path;
this.extension = this.basename.split('.').pop() || '';
this.name = this.basename;
if (mtime) {
(this as any).stat = { mtime, ctime: mtime, size: 100 };
}
}
}
const files = [
new MockTFile('Projects/alpha.md', 1000),
new MockTFile('Projects/beta.md', 2000),
new MockTFile('notes/daily.md', 1500),
];
mockVault.getMarkdownFiles = jest.fn().mockReturnValue(files);
mockVault.cachedRead = jest.fn().mockResolvedValue('content');
mockApp.metadataCache.getFileCache = jest.fn().mockImplementation((f: MockTFile) => {
if (f.path === 'Projects/alpha.md') return { frontmatter: { tags: ['project'] } };
if (f.path === 'Projects/beta.md') return { frontmatter: { tags: ['project', 'done'] } };
return { frontmatter: {} };
});
const call: ToolCall = {
id: 'call_vs1',
type: 'function',
function: {
name: 'get_vault_stats',
arguments: JSON.stringify({}),
},
};
const result = await executor.handleToolCall(call);
expect(result.success).toBe(true);
const data = result.data as any;
expect(data.totalNotes).toBe(3);
expect(data.totalFolders).toBe(2);
expect(data.folders).toContain('Projects');
expect(data.folders).toContain('notes');
expect(data.taggedNotes).toBe(2);
expect(data.untaggedNotes).toBe(1);
expect(data.topTags).toHaveLength(2);
expect(data.topTags[0].tag).toBe('project');
expect(data.topTags[0].count).toBe(2);
expect(data.recentFiles[0]).toBe('Projects/beta.md');
});
});
describe('telemetry integration', () => {
let telemetryManager: TelemetryManager;
let telemetryExecutor: ToolExecutor;
tests/vault-indexer.test.ts
+74
View File
@@ -489,4 +489,78 @@ describe('VaultIndexer', () => {
expect(key2).toContain('norecency');
});
});
describe('with metadataCache', () => {
it('should parse YAML array tags from metadataCache', async () => {
const file = {
basename: 'Note A',
path: 'projects/note-a.md',
};
const mockApp = {
metadataCache: {
getFileCache: jest.fn().mockReturnValue({
frontmatter: {
title: 'Project Alpha',
tags: ['project', 'alpha', 'urgent'],
},
headings: [{ heading: 'Project Alpha' }, { heading: 'Overview' }],
}),
},
};
const indexedWithApp = new VaultIndexer(mockVault as unknown as any);
indexedWithApp.setApp(mockApp as unknown as any);
mockVault.getMarkdownFiles = jest.fn().mockReturnValue([file]);
mockVault.read = jest
.fn()
.mockResolvedValue('# Project Alpha\n\nSome content here.\n\n## Overview\n\nMore text.');
const results = await indexedWithApp.searchVault('alpha', 5);
expect(results.length).toBe(1);
expect(results[0].title).toBe('Project Alpha');
expect(results[0].tags).toBe('project, alpha, urgent');
});
it('should fall back to regex parsing when metadataCache is unavailable', async () => {
const file = {
basename: 'Note B',
path: 'note-b.md',
};
mockVault.getMarkdownFiles = jest.fn().mockReturnValue([file]);
mockVault.read = jest
.fn()
.mockResolvedValue(
'---\ntitle: Legacy Note\ntags: legacy, old\n---\n\n# Legacy Note\n\nContent here.'
);
const results = await indexer.searchVault('legacy', 5);
expect(results.length).toBe(1);
expect(results[0].title).toBe('Legacy Note');
expect(results[0].tags).toBe('legacy, old');
});
it('should use metadataCache headings when available', () => {
const file = {
basename: 'Note C',
path: 'note-c.md',
};
const mockApp = {
metadataCache: {
getFileCache: jest.fn().mockReturnValue({
frontmatter: {},
headings: [{ heading: 'First Heading' }, { heading: 'Second Heading' }],
}),
},
};
const indexedWithApp = new VaultIndexer(mockVault as unknown as any);
indexedWithApp.setApp(mockApp as unknown as any);
const tokenized = (indexedWithApp as any).tokenizeContent(
'Some content\n\n# First Heading\n\n# Second Heading\n\nBody.',
file
);
expect(tokenized.headings).toEqual(['First Heading', 'Second Heading']);
expect(tokenized.title).toBe('First Heading');
});
});
});