JSONPath Query

$ is the whole document. .name takes a field, [0] takes the first item of a list, [-1] the last, [*] all of them, [1:3] the 2nd and 3rd items, and ..name finds that field at any depth. For example: $.users[0].name or $..price.

Use [?(...)], where @ is each item being checked. For example $.products[?(@.price < 10)] takes the products priced under 10. You can compare with ==, !=, <, <=, > and >=, and combine conditions with && (and), || (or) and ! (not). Just write [?(@.isbn)] to keep the items that have that field.

Put the name in square brackets and double quotes, for example $["first name"] or $.users[0]["first-name"].

A path can match several values, so the result is always a list. If only one value matches, the list has exactly one item, and if nothing matches it is the empty list [].

Yes, in filters. length(@.name) counts the characters of text (or the items of a list), count(@.tags[*]) counts the values found, value(@..color) gives the single value found, match(@.name, "^An") checks that the whole text fits a pattern, and search(@.name, "An") checks that the text contains the pattern. For example: $.users[?(length(@.name) > 3 && match(@.name, "^B"))].

Basic regex syntax: plain characters, a dot (any one character), [abc] and [a-z], [^abc] (anything but), * + ? and exact repeat counts, | (or), groups (…), ^ and $ (start and end of the text), \d \w \s, \b (word edge) and Unicode groups such as \p for capital letters. There are no look-aheads or look-behinds (?=…) and no back-references (\1). In return, a pattern always runs fast — even a badly written one can't freeze the page. Note: in the query a backslash is written twice, for example \"\\d+\" for the pattern \d+.

No. Numbers are kept exactly as they are in the JSON, and when you filter by a number the comparison is exact to the last digit.

No. The JSON is processed right on your device.

It's completely free, no account needed.