copy_from

The copy_from template derives a new dictionary from an existing one named by FROM, inheriting its template and all of its options, and lets you override just the ones you want to change. This avoids repeating a long definition when you need close variants — for instance copying an English text dictionary and overriding only STEMMING or the stop-word list.

Any option accepted by the source dictionary's template can be given here to override the inherited value; everything you do not mention is carried over unchanged. Because the source's template is inherited too, prefixed options reach into composed sources — STEP2_CASE overrides a pipeline step, TOKENIZER_STOPWORDS overrides a minhash nested analyzer.

Options

Option	Type	Default	Description
FROM	string	required	Name of the source dictionary to copy
any option	—	—	Override any option accepted by the source's template; unmentioned options are inherited

Tokenization

A copy behaves exactly like its source except where you override. Starting from an English text dictionary that lowercases and stems — so running flies indexes as {run,fli} — a copy that overrides only STEMMING = false keeps the inherited locale and lower-casing but emits the full words {running,flies}. The same input through the two dictionaries shows precisely what the override changed and what it left alone.

Input	Dictionary	Tokens
running flies	source text (CASE = 'lower', STEMMING = true)	{run,fli}
running flies	copy overriding STEMMING = false	{running,flies}

Define a stemming English dictionary, then copy it and turn stemming off:

Query

CREATE TEXT SEARCH DICTIONARY english_dict (    template = 'text',    locale = 'en_US.UTF-8',    case = 'lower',    stemming = true);

Query

CREATE TEXT SEARCH DICTIONARY english_no_stem (    template = 'copy_from',    from = 'english_dict',    -- inherit locale and case, override only stemming    stemming = false);

The source stems each word to its root:

Query

SELECT ts_lexize('english_dict', 'running flies');

Result

 ts_lexize----------- {run,fli}

The copy inherits locale and case but keeps full words, because only STEMMING was overridden:

Query

SELECT ts_lexize('english_no_stem', 'running flies');

Result

 ts_lexize----------------- {running,flies}

Examples

Override a pipeline step option

A prefixed option overrides the matching step of a copied pipeline, leaving the other steps intact:

Query

CREATE TEXT SEARCH DICTIONARY pipe_variant (    template = 'copy_from',    from = 'pipe_dict',    STEP2_CASE = 'upper');

Extend a pipeline with additional steps

Naming a step number beyond the source's last step appends a new step to the copied pipeline:

Query

CREATE TEXT SEARCH DICTIONARY pipe_extended (    template = 'copy_from',    from = 'pipe_dict',    STEP3_TEMPLATE = 'stopwords',    STEP3_STOPWORDS = '"bar","foo"');

Override a nested minhash analyzer option

A TOKENIZER_-prefixed option reaches into the nested analyzer of a copied minhash dictionary:

Query

CREATE TEXT SEARCH DICTIONARY minhash_variant (    template = 'copy_from',    from = 'minhash_text',    TOKENIZER_STOPWORDS = '"fox"');

See also

• pipeline — composed source whose steps can be overridden
• minhash — composed source whose nested analyzer can be overridden
• CREATE TEXT SEARCH DICTIONARY