dictmaker is a stand-alone, single-binary server application for building and publishing dictionary websites. Alar (Kannada-English dictionary) is an example in production.

• Generic entry-relation (many)-entry structure for dictionary data in a Postgres database with just two tables (entries, relations)
• Entirely built on top of Postgres full text search using tsvector tokens
• Works with any language. Plug in external tokenizers or use a built in tokenizer supported by Postgres for search
• Possible to have entries and definitions in any number of languages in a single database
• HTTP/JSON REST like APIs
• Themes and templates to publish dictionary websites
• Paginated A-Z (or any alphabet for any language) glossary generation for dictionary words

dictmaker has no concept of language or semantics. To make a universal dictionary interface possible, it treats all data as unicode strings that can be searched with Postgres DB's fulltext capabilities, where search tokens are generated and stored along with the dictionary entries. There are several built-in fulltext dictionaries and tokenizers that Postgres supports out of the box, mostly European languages (\dFd lists installed dictionaries). For languages that do not have fulltext dictionaries, it is possible to generate search tokens manually using an external algorithm. For instance, to make a phonetic English dictionary, Metaphone tokens can be generated manually and search queries issued against them.

• There can be any number of languages defined in the dictionary. eg: 'english', 'malayalam', 'kannada' etc.
• All content, the entry words and their definitions, are stored in the entries table
• Entry-definition many-to-many relationships are stored in the relations table, represented by from_id (entry word) -> to_id (definition word), where both IDs refer to the entries table.

1. Download the latest release release of dictmaker
2. Run ./dictmaker --new to generate a sample config.toml and DB schema.sql
3. Create a Postgres DB and execute schema.sql on it to create the tables
4. Define your dictionary's languages and properties along with other configuration in config.toml
5. Populate the entries and relations tables with your dictionary data. See the "Sample dictionary" section below
6. Run the binary: ./dictmaker

# /dictionary/from_lang/to_lang/word
# Optional query params: ?type=noun&type=noun2&tag=a&tag=b ...
curl localhost:8080/dictionary/english/english/apple

The sample/sample.sql shows how to setup an English-English and English-Italian dictionary. Retaining the English-Italian config in the generated sample config file, execute sample/sample.sql on your database.

Then try:

curl localhost:8080/dictionary/english/english/apple
curl localhost:8080/dictionary/english/italian/apple

See the dictmaker-alar repository that powers the Alar dictionary. A theme is a directory with a collection of Go HTML templates. Run a theme by passing ./dictmaker --site=theme_dir.

For languages that do not have Postgres fulltext dictionaries and tokenizers, dictmaker supports loading compiled Go tokenizer plugins that implement the search.Tokenizer interface. See tokenizers/kannada (and Makefile for compilation help).

Licensed under the AGPL v3 license.