Skip to main content
Ayase Quart provides powerful search capabilities for exploring archived content. Two search modes are available: SQL search for exact matching and full-text search (FTS) for faster, more flexible queries.

​
Search modes

SQL search

Exact pattern matching directly against the database. Slower but always current.

Full-text search

Fast indexed search using LNX, Meilisearch, or Typesense. May have slight data lag.
SQL search queries the archive database directly for exact matches:

​
Features

  • Always current: Searches live database with no indexing delay
  • Exact matching: Finds precise text patterns in posts
  • Multi-board support: Configurable single or multi-board searches
  • Per-board results: When searching multiple boards, results are paginated per board

​
Search fields

  • Comment: Search within post text
  • Subject: Search within post subjects/titles
  • OP Comment: Find posts in threads where the opening post matches
  • OP Subject: Find posts in threads with matching OP subjects
  • Tripcode: Search by poster tripcode
  • Filename: Search by original media filename
  • Post Number: Find specific post by ID
  • Comment Length: Filter by comment character count (>, <, =, >=, <=)
  • Subject Length: Filter by subject character count
  • Media Dimensions: Filter by image/video width and height
  • Has File / No File: Filter by media attachment
  • Is OP / Not OP: Filter opening posts vs replies
  • Is Deleted / Not Deleted: Filter deleted posts
  • Is Sticky / Not Sticky: Filter pinned threads
  • Date After: Posts on or after this date
  • Date Before: Posts on or before this date

​
Usage

1

Navigate to SQL search

Visit the SQL search page at /search (if enabled in configuration)
2

Select boards

Choose one or more boards to search depending on your configuration
3

Enter search terms

Fill in any combination of search fields. All specified filters are combined with AND logic.
4

Submit search

Click the Search button to view results
SQL search has configurable result limits per page. Default maximum is defined in config.toml under vanilla_search.hits_per_page

​
Full-text search (FTS)

FTS provides much faster searches using a dedicated search engine:

​
Supported engines

LNX

Fully supported and tested. Based on Tantivy. Best performance.

Meilisearch

Partial support. Easy setup with good performance.

Typesense

Partial support. Fast with typo tolerance.

​
Query syntax

FTS supports advanced query operators: 
"exact phrase"              # Exact phrase matching
+devices +usb-c -adapter     # Require devices AND usb-c, exclude adapter
x AND y OR z                 # Boolean operators
Full syntax support depends on the search engine used. LNX provides:
  • Phrase queries with quotes
  • Boolean operators: AND, OR, NOT
  • Required/excluded terms: +term, -term
  • Field-specific searches
  • Wildcard patterns
See Tantivy query documentation for complete details.

​
Search features

All the same search fields from SQL search are available, plus:
  • Highlighting: Search terms are highlighted in results (configurable)
  • Faster results: Pre-indexed data returns results in milliseconds
  • Typo tolerance: Some engines support fuzzy matching
  • Relevance ranking: Results are scored by relevance

​
Setting up FTS

1

Choose a search engine

Select LNX, Meilisearch, or Typesense based on your needs
2

Run the search engine

Use the provided Docker Compose configuration:
cd index_search/lnx/
docker compose up -d
3

Configure Ayase Quart

Edit config.toml to enable index search and set connection details:
[index_search]
enabled = true
engine = "lnx"
host = "localhost"
port = 8000
4

Create the index

Initialize the search index schema:
ayaseq search index create
5

Load data

Populate the index with archived posts:
ayaseq search load full a b c g
The initial index load can take time depending on archive size. Progress bars show loading status.
Both search modes support gallery mode for media-focused browsing:
1

Enable gallery mode

Check the “Gallery Mode” checkbox in the search form
2

Search as normal

Enter your search criteria (automatically filters to posts with media)
3

View results

Results display as a grid of images and videos without post text
Gallery mode automatically enables the “Has File” filter

​
Search plugins

Ayase Quart supports custom search plugins to extend functionality:

​
Plugin capabilities

  • Add custom search fields to the form
  • Pre-filter results before native search
  • Integrate external data sources
  • Implement custom search logic

​
Using plugins

Plugins are automatically detected and loaded from:
  • src/ayase_quart/plugins/search/
See search_example.py for a template plugin implementation.
Search plugins face a “post-filtering” challenge:Since plugin filtering happens before native search filtering (for reported posts, deleted posts, etc.), final page sizes can be unpredictable. Plugin results must go through a second round of filtering, which may reduce the number of results below the requested page size.Plugin developers should:
  • Return more results than hits_per_page when combined with native filters
  • Document this behavior for users
  • Consider returning 2-3x the page size for complex filters

​
Result display

Search results show:

​
Standard mode

  • Post metadata: Number, timestamp, board, thread
  • Post content: Subject and comment with formatting
  • Media: Thumbnails with click-to-view full size
  • Thread context: Link to view post in original thread
  • Quotelinks: Clickable references preserved
  • Highlighting: Search terms highlighted in text (FTS only, if enabled)
  • Grid of media thumbnails
  • Click to view full size
  • Link to source thread
  • Media metadata on hover

​
Pagination

Search results are paginated for performance:
  • Configurable results per page
  • Page navigation controls
  • Total hit count displayed
  • Current page indicator
  • Direct page number links
Page size limits are configured per search mode in config.toml

​
Performance considerations

​
SQL search performance

  • Slower for large archives (millions of posts)
  • Performance depends on database indexes
  • Consider for small archives or when FTS isn’t available
  • Results are always current with no lag

​
FTS performance

  • Very fast even on large archives
  • Requires separate search engine infrastructure
  • Index updates may lag behind scraper by minutes
  • Memory requirements depend on index size
For production deployments with large archives, FTS is strongly recommended

​
Rate limiting

Search endpoints are rate limited to prevent abuse:
  • FTS: 5 requests per minute per IP
  • SQL: Configurable limits
Exceeding limits returns a 429 Too Many Requests error.

​
API access

Search is also available via API (if enabled):
  • Same search parameters as web interface
  • JSON response format
  • Requires API authentication token
  • See API documentation for endpoint details
References:
  • Search handler: src/ayase_quart/blueprints/web/bp_search.py:24
  • FTS endpoint: src/ayase_quart/blueprints/web/bp_search_fts.py:18
  • Query builder: src/ayase_quart/search/query.py:12

Build docs developers (and LLMs) love

Get started for freeTalk to us