Projets-2016-2017-GeoDiff

Preambule

The project subject GeoDiff

This project is handled by three Polytech students: AMAURIN Alexandre, BECHER Hervé, BROCHIER Aymeric

Project Presentation

The goal of this project is to identify the differences between two spatial data sets and visualize them.

Team

• Supervisors: Nicolas Palix

• Members: AMAURIN Alexandre, BECHER Hervé, BROCHIER Aymeric

• Department : RICM 4, Polytech Grenoble

Progress of the project

The project started January 14th, 2015.

Week 1 (January 09th - January 15th)

• Project selection

Week 2 (January 16th - January 22th)

• Project discovery
• AIR page and GitHub project setup
• Team organization
• Software Requirement Specification

SRS

Week 4 (January 30th - February 05th)

• UML diagrams (Use Case and Sequence)

UML

Week 5 (February 6th - February 12th)

• First look at Javascript and Leaflet
• Finished first SRS version

Week 6 (February 13th - February 19th)

• Progress at the level of the delta and at the level of the graphic display.

Week 7 (February 20th - February 26th)

• Holiday break

Week 8 (February 27th - March 5th)

• Rough interface (using Leaflet) for showing delta using layers
• Files input and unique identifier selection

Week 9 (March 6th - March 12th)

• Command-line tool (written in Java) development started
  • Pretty rough (prototype)
  • Done:
    • Delta computing module
    • Command-line argument parser (using jopt-simple) that currently takes two files and an optional unique identifier for each element
    • Support for GeoJSON as input files
    • Output delta as GeoJSON file, with each feature having additional properties used by GeoDiff tools (currently 'geodiff-type' which identifies the type of modification, i.e 'add', 'del', 'mod', or 'id')
  • To do:
    • Support more formats
    • Add abstraction layers for internal manipulation of data sets (adapters, etc.)
    • Allow the user to add his own module(s) for parsing input files
  • Technologies used:
    • Java
    • Gson (from Google) for manipulating JSON objects (under Apache License 2.0, GitHub repo)
    • GeoJSON addon for Gson (from user filosganga) that adds convenient structures and methods for manipulating GeoJSON objects (under Apache License 2.0, GitHub repo)
    • JOpt Simple for parsing command line options (under The MIT License, website)
• Browser tool (JavaScript)
  • Done: (...)
  • To do: (...)
• To do:
  • Merge the map view and data inputs (web interface)

Week 10 (March 20th - March 26th)

• Command-line tool
  • Made internal data structures all use filosganga's geogson classes
  • Added parsers for reading and converting their input into our internal Element structure
    • Allows for simple modules for other formats
    • Currently GeoJSON and CSV supported
  • Added Javadoc
  • To do:
    • Improve the modularity of the program (=> better API)
    • Maybe write our own (simplified) data structures
• Created test sets to check if the Java and JavaScript tools are consistent in regard to the GeoJSON standard
  • Done: tests on JavaScript
  • To do: tests on Java
• Browser tool
  • Done: (...)
  • To do: (...)
• To do:
  • Merge the map view and data inputs (web interface)

Week 11 (March 27th - April 2nd)

• Command-line tool
  • Done:
    • Tests done and successful
    • Flaw detected: the parser doesn't allow empty (possibly null too?) JSON objects (like empty geometry objects)
  • To do:
    • Add options to filter output data
    • Javadoc
  • Done:
    • Tests (see section below) with a UI for easier testing
    • Squashed bugs in the comparison functions
• Browser tool
  • Done:
    • Option to download a zip archive containing four files (additions, deletions, modifications and all three) in GeoJSON
    • Use of PHP to create the files
    • Add geodiff-type property to all features
  • To do:
    • Clean up code and improve security (for ex. prevent the modification of critical variables)
    • Merge the map view and data inputs - almost there
• Tests
  • Test sets in GeoJSON to facilitate the checks in Java
  • All the geometrical structures have been tested: Point, MultiPoint, LineString, MultiLineString, Polygon and MultiPolygon (with consideration of the height for JavaScript)