jq.py: a lightweight and flexible JSON processor ================================================ This project contains Python bindings for `jq `_. Installation ------------ During installation, the source for jq 1.4 is downloaded over HTTPS and built. Therefore, installation requires any programs required to build jq. This includes: * Autoreconf * The normal C compiler toolchain, such as gcc and make. * Flex * Bison * libtool * Python headers. Debian, Ubuntu or relatives ~~~~~~~~~~~~~~~~~~~~~~~~~~~ If on Debian, Ubuntu or relatives, running the following command should be sufficient: .. code-block:: sh apt-get install autoconf build-essential flex bison libtool python-dev Red Hat, Fedora, CentOS or relatives ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ If on Red Hat, Fedora, CentOS, or relatives, running the following command should be sufficient: .. code-block:: sh yum groupinstall "Development Tools" yum install autoconf flex bison libtool python Mac OS X ~~~~~~~~ If on Mac OS X, you probably want to install `Xcode `_ and `Homebrew `_. Once Homebrew is installed, you can install the remaining dependencies with: .. code-block:: sh brew install autoconf bison libtool Note that at least some versions of Mac OS X ship a version of ``bison`` that is too old for ``jq`` to use. jq.py will try to use the version from Homebrew if available. Usage ----- A program can be compiled by passing it to ``jq.jq``. To apply the program to an input, call the ``transform`` method. jq.py expects the value to be valid JSON, such as values returned from ``json.load``. .. code-block:: python from jq import jq jq(".").transform("42") == "42" jq(".").transform({"a": 1}) == {"a": 1} If the value is unparsed JSON text, pass it in using the ``text`` argument: .. code-block:: python jq(".").transform(text="42") == 42 The ``text_output`` argument can be used to serialise the output into JSON text: .. code-block:: python jq(".").transform("42", text_output=True) == '"42"' If there are multiple output elements, each element is represented by a separate line, irrespective of the value of ``multiple_output``: .. code-block:: python jq(".[]").transform([1, 2, 3], text_output=True) == "1\n2\n3" If ``multiple_output`` is ``False`` (the default), then the first output is used: .. code-block:: python jq(".[]+1").transform([1, 2, 3]) == 2 If ``multiple_output`` is ``True``, all output elements are returned in an array: .. code-block:: python jq(".[]+1").transform([1, 2, 3], multiple_output=True) == [2, 3, 4]