All Projects → ejgallego → Coq Serapi

ejgallego / Coq Serapi

Licence: other
Coq Protocol Playground with Se(xp)rialization of Internal Structures.

Programming Languages

ocaml
1615 projects

Projects that are alternatives of or similar to Coq Serapi

Jstp
Fast RPC for browser and Node.js based on TCP, WebSocket, and MDSF
Stars: ✭ 132 (+51.72%)
Mutual labels:  json, serialization, protocol
Flatcc
FlatBuffers Compiler and Library in C for C
Stars: ✭ 434 (+398.85%)
Mutual labels:  json, serialization, protocol
Protobuf Convert
Macros for convenient serialization of Rust data structures into/from Protocol Buffers 3
Stars: ✭ 22 (-74.71%)
Mutual labels:  serialization, protocol
Play Json Extra
playframework2 json extra module. provide convenience functions for define Format, Reads, Writes
Stars: ✭ 20 (-77.01%)
Mutual labels:  json, serialization
Fhir.js
Node.JS library for serializing/deserializing FHIR resources between JS/JSON and XML using various node.js XML libraries
Stars: ✭ 61 (-29.89%)
Mutual labels:  json, serialization
Groot
From JSON to Core Data and back.
Stars: ✭ 533 (+512.64%)
Mutual labels:  json, serialization
Fastjson
A fast JSON parser/generator for Java.
Stars: ✭ 23,997 (+27482.76%)
Mutual labels:  json, serialization
Tech1 Benchmarks
Java JMH Benchmarks repository. No Longer Supported.
Stars: ✭ 50 (-42.53%)
Mutual labels:  json, serialization
Fastbinaryencoding
Fast Binary Encoding is ultra fast and universal serialization solution for C++, C#, Go, Java, JavaScript, Kotlin, Python, Ruby, Swift
Stars: ✭ 421 (+383.91%)
Mutual labels:  serialization, protocol
Devis
A microservices framework for Node.js
Stars: ✭ 72 (-17.24%)
Mutual labels:  json, protocol
Dictfier
Python library to convert/serialize class instances(Objects) both flat and nested into a dictionary data structure. It's very useful in converting Python Objects into JSON format
Stars: ✭ 67 (-22.99%)
Mutual labels:  json, serialization
Json Mobx
Simple undo/redo and persistence for MobX
Stars: ✭ 78 (-10.34%)
Mutual labels:  json, serialization
Iguana
universal serialization engine
Stars: ✭ 481 (+452.87%)
Mutual labels:  json, serialization
Airframe
Essential Building Blocks for Scala
Stars: ✭ 442 (+408.05%)
Mutual labels:  json, serialization
Jackson Module Kotlin
Module that adds support for serialization/deserialization of Kotlin (http://kotlinlang.org) classes and data classes.
Stars: ✭ 830 (+854.02%)
Mutual labels:  json, serialization
Eminim
JSON serialization framework for Nim, works from a Stream directly to any type and back. Depends only on stdlib.
Stars: ✭ 32 (-63.22%)
Mutual labels:  json, serialization
Jsonifier
Fast and simple JSON encoding toolkit
Stars: ✭ 79 (-9.2%)
Mutual labels:  json, serialization
Handyjson
A handy swift json-object serialization/deserialization library
Stars: ✭ 3,913 (+4397.7%)
Mutual labels:  json, serialization
Jsoniter Scala
Scala macros for compile-time generation of safe and ultra-fast JSON codecs
Stars: ✭ 410 (+371.26%)
Mutual labels:  json, serialization
Simplify Core
Simplify 为简化重复的JAVA代码而生,基于JDK8,无其它jar包依赖,提供序列化,json parse/generator,日期处理,asm && jdkproxy 实现动态代理功能 等常见操作。
Stars: ✭ 65 (-25.29%)
Mutual labels:  json, serialization

SerAPI: Machine-Friendly, Data-Centric Serialization for Coq

Build Status Gitter

$ opam install coq-serapi
$ sertop --help

SerAPI is a library for machine-to-machine interaction with the Coq proof assistant, with particular emphasis on applications in IDEs, code analysis tools, and machine learning. SerAPI provides automatic serialization of Coq's internal OCaml datatypes from/to JSON or S-expressions (sexps).

SerAPI is a proof-of-concept and should be considered alpha-quality. However, it is fully functional and supports, among other things, asynchronous proof checking, full-document parsing, and serialization of Coq's core datatypes. SerAPI can also be run as WebWorker thread, providing a self-contained Coq system inside the browser. Typical load times in Google Chrome are less than a second.

The main design philosophy of SerAPI is to make clients' lives easy, by providing a convenient, robust interface that hides most of the scary details involved in interacting with Coq.

Feedback from Coq users and developers is very welcome and intrinsic to the project. We are open to implementing new features and exploring new use cases.

Documentation and Help:

API WARNING: The protocol is experimental and may change often.

Quick Overview and Install:

SerAPI can be installed as the OPAM package coq-serapi. See build instructions for manual installation. The experimental in-browser version is also online.

SerAPI provides an interactive "Read-Print-Eval-Loop" sertop, a batch-oriented compiler sercomp, and a batch-oriented tokenizer sertok. See the manual pages and --help pages of each command for more details.

To get familiar with SerAPI we recommend launching the sertop REPL, as it provides a reasonably human-friendly experience:

$ rlwrap sertop --printer=human

You can then input commands. Ctrl-C will interrupt a busy Coq process in the same way it interrupts coqtop.

The program sercomp provides a command-line interface to some key functionality of SerAPI and can be used for batch processing of Coq documents, e.g., to serialize Coq source files from/to lists of S-expressions of Coq vernacular sentences. See sercomp --help for some usage examples and an overview of the main options. The program sertok provides similar functionality at the level of Coq source file tokens.

Protocol Commands:

Interaction with sertop is done using commands, which can be optionally tagged in the form of (tag cmd); otherwise, an automatic tag will be assigned. For every command, SerAPI will always reply with (Answer tag Ack) to indicate that the command was successfully parsed and delivered to Coq, or with a SexpError if parsing failed.

There are three categories of commands:

  • Document manipulation: Add, Cancel, Exec, ...: these commands instruct Coq to perform some action on the current document. Every command will produce zero or more different tagged answers, and a final answer (Answer tag Completed), indicating that there won't be more output.

    SerAPI document commands are an evolution of the OCaml STM API, here and here you can find a few informal notes on how it works. We are working on a more detailed specification, for now you can get some more details in the issue tracker.

  • Queries: (Query ((opt value) ...) kind):

    Queries stream Coq objects of type kind. This can range from options, goals and hypotheses, tactics, etc. The first argument is a list of options: preds is a list of conjunctive filters, limit specifies how many values the query may return. pp controls the output format: PpSer for full serialization, or PpStr for "pretty printing". For instance:

    (tag (Query ((preds (Prefix "Debug")) (limit 10) (pp PpSexp)) Option))
    

    will stream all Coq options that start with "Debug", limiting to the first 10 and printing the full internal Coq datatype:

    (CoqOption (Default Goal Selector)
       ((opt_sync true) (opt_depr false) (opt_name "default goal selector")
       (opt_value (StringValue 1))))
    ...
    

    Options can be omitted, as in: (tag (Query ((limit 10)) Option)), and currently supported queries can be seen here

  • Printing: (Print opts obj): The Print command provides access to the Coq pretty printers. Its intended use is for printing (maybe IDE manipulated) objects returned by Query.

Roadmap:

SerAPI 0.13.x is based on Coq 8.13. These days, most work related to SerAPI is directly happening over Coq's upstream itself. The main objective is to improve the proof-document model; building a rich query language will be next.

Clients and Users:

SerAPI has been used in a few contexts already, we provide a few pointers here, feel free to add your own!

  • jsCoq allows you to run Coq in your browser. jsCoq is the predecessor of SerAPI and will shortly be fully based on it.

  • mCoq is a tool for mutation analysis of Coq projects, based on serializing and deserializing Coq code via SerAPI. See the accompanying tool paper, and the research paper which describes and evaluates the technique.

  • elcoq, an emacs technology demo based on SerAPI by Clément Pit-Claudel. elcoq is not fully functional but illustrates some noteworthy features of SerAPI.

  • PeaCoq, a Coq IDE for the browser, has an experimental branch that uses SerAPI.

  • GrammaTech's Software Evolution Library (SEL) provides tools for programmatically modifying and evaluating software. SEL operates over multiple software representations including source code in several languages and compiled machine code. Its Coq module uses SerAPI to serialize Coq source code into ASTs, which are parsed into Common Lisp objects for further manipulation. GrammaTech uses this library to synthesize modifications to software so that it satisfies an objective function, e.g., a suite of unit tests. SerAPI support was added to SEL by Rebecca Swords.

  • CoqGym is a Coq-based learning environment for theorem proving. It uses SerAPI to interact with Coq and perform feature-extraction. Its author notes:

    CoqGym relies heavily on SerAPI for serializing the internal structures of Coq. I tried to use Coq's native printing functions when I started with this project, but soon I found SerAPI could save a lot of the headaches with parsing Coq's output. Thanks to SerAPI authors, this project wouldn't be possible (or at least in its current form) without SerAPI.

    See also the paper describing CoqGym.

  • Proverbot9001 is a proof search system based on machine learning techniques, and uses SerAPI to interface with Coq. See also the paper describing the system.

  • Roosterize is a tool for suggesting lemma names in Coq projects based on machine learning. See the paper describing the technique and tool.

  • MathComp corpus is a machine learning dataset based on the Mathematical Components family of Coq projects, and includes several machine-readable representations of Coq code generated via SerAPI. The dataset was used to train and evaluate the Roosterize tool.

  • SerAPI is being used to improve the Coq regression proof selection tool iCoq, see the paper.

  • SerAPI is being used in additional software testing and machine learning projects; we will update this list as papers get out of embargo.

Quick Demo (not always up to date):

$ rlwrap sertop --printer=human
(Add () "Lemma addn0 n : n + 0 = n. Proof. now induction n. Qed.")
  > (Answer 0 Ack)
  > (Answer 0 (Added 2 ((fname "") (line_nb 1) (bol_pos 0) (line_nb_last 1) (bol_pos_last 0) (bp 0) (ep 26))
  >            NewTip))
  > ...
  > (Answer 0 (Added 5 ... NewTip))
  > (Answer 0 Completed)

(Exec 5)
  > (Answer 1 Ack)
  > (Feedback ((id 5) (route 0) (contents (ProcessingIn master))))
  > ...
  > (Feedback ((id 5) (route 0) (contents Processed)))
  > (Answer 1 Completed)

(Query ((sid 3)) Goals)
  > (Answer 2 Ack)
  > (Answer 2
  >  (ObjList ((CoqGoal ((fg_goals (((name 5) (ty (App (Ind ...))))
                         (bg_goals ()) (shelved_goals ()) (given_up_goals ()))))))
  > (Answer 2 Completed)

(Query ((sid 3) (pp ((pp_format PpStr)))) Goals)
  > (Answer 3 Ack)
  > (Answer 3 (ObjList ((CoqString
  >   "\
  >    \n  n : nat\
  >    \n============================\
  >    \nn + 0 = n"))))
  > (Answer 3 Completed)

(Query ((sid 4)) Ast)
  > (Answer 4 Ack)
  > (Answer 4 (ObjList ((CoqAst ((((fname "") (line_nb 1) (bol_pos 0) (line_nb_last 1)
  >                                (bol_pos_last 0) (bp 34) (ep 50)))
  > ...
  >            ((Tacexp
  >              (TacAtom
  >                (TacInductionDestruct true false
  > ...
  > (Answer 4 Completed)

(pp_ex (Print ((sid 4) (pp ((pp_format PpStr)))) (CoqConstr (App (Rel 0) ((Rel 0))))))
  > (Answer pp_ex Ack)
  > (Answer pp_ex(ObjList((CoqString"(_UNBOUND_REL_0 _UNBOUND_REL_0)"))))

(Query () (Vernac "Print nat. "))
  > (Answer 6 Ack)
  > (Feedback ((id 5) (route 0) (contents
  >    (Message Notice ()
  >    ((Pp_box (Pp_hovbox 0) ...)
  > (Answer 6 (ObjList ()))
  > (Answer 6 Completed)

(Query () (Definition nat))
  > (Answer 7 Ack)
  > (Answer 7 (ObjList ((CoqMInd (Mutind ....)))))
  > (Answer 7 Completed)

Technical Report:

There is a brief technical report describing the motivation, design, and implementation of SerAPI. If you are using SerAPI in a project, please cite the technical report in any related publications:

@techreport{GallegoArias2016SerAPI,
  title = {{SerAPI: Machine-Friendly, Data-Centric Serialization for Coq}},
  author = {Gallego Arias, Emilio Jes{\'u}s},
  url = {https://hal-mines-paristech.archives-ouvertes.fr/hal-01384408},
  institution = {MINES ParisTech},
  year = {2016},
  month = Oct,
}

Developer Information

Technical Details

SerAPI has four main components:

  • serapi, an extended version of the current IDE protocol;
  • serlib, a library providing automatic de/serialization of most Coq data structures using ppx_conv_sexp. This should be eventually incorporated into Coq itself. Support for ppx_deriving_yojson is work in progress;
  • sertop, sertop_js, toplevels offering implementations of the protocol;
  • sercomp, sertok, command-line tools providing access to key features of serlib.

Building your own toplevels using serlib and serapi is encouraged.

Advanced Use Cases

With a bit more development effort, you can also:

  • Use SerAPI as an OCaml library. The low-level serialization library serlib/ and the higher-level SerAPI protocol in serapi/serapi_protocol.mli can be linked standalone.

  • Use SerAPI's web worker JavaScript Worker from your web/node application. In this model, you communicate with SerAPI using the typical onmessage/postMessage worker API. Ready-to-use builds may be found at here, we provide an example REPL at: https://x80.org/rhino-hawk

We would also like to provide a Jupyter/IPython kernel.

Developer/Users Mailing List

SerAPI development is mainly discussed on GitHub and in the Gitter channel. You can also use the jsCoq mailing list by subscribing at: https://x80.org/cgi-bin/mailman/listinfo/jscoq

The mailing list archives should also be available at the Gmane group: gmane.science.mathematics.logic.coq.jscoq. You can post to the list using nntp.

Acknowledgments

SerAPI has been developed at the Centre de Recherche en Informatique of MINES ParisTech (former École de Mines de Paris) and was partially supported by the FEEVER project.

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].