pg_trgm Similarity Calculator

The pg_trgm extension powers fuzzy search in PostgreSQL. It splits text into trigrams (groups of three characters) and scores how many two strings share. The scores are easy to misread. similarity('postgres', 'postgresql') is 0.6666667, not 0.9, because the extra letters add trigrams. 'MySQL' and 'my sql' differ because the space starts a new word. A search term that matches one word inside a long value scores low with similarity() and high with word_similarity(). This calculator runs the same algorithm as pg_trgm. It lowercases the text, splits it into words of letters and digits, pads each word with two spaces in front and one behind, and hashes trigrams that contain non-ASCII characters. Enter a search term and the values you expect it to find. You get show_trgm() for each string, the shared trigrams and the exact similarity, word_similarity and strict_word_similarity values, printed the way PostgreSQL prints a real. It also shows the <->, <<-> and <<<-> distances and whether %, <% and <<% match at your thresholds. Use it to choose pg_trgm.similarity_threshold before you create a GIN index. Results were checked against PostgreSQL 18.6 with 12,000 random string pairs (English, accented, Greek, Cyrillic, CJK and emoji text) and matched every value. Everything runs in your browser.