Firefox/Projects/PlacesQueryAPIRedesign

From MozillaWiki
< Firefox‎ | Projects
Revision as of 20:37, 11 May 2010 by Mak77 (talk | contribs)
Jump to navigation Jump to search

Places Query API

Create a sensible, easy to use Query API for Places for Firefox.next & Jetpack. This API should make it possible to do targeted queries against history and bookmarks with a minimum of code.

Goals/Use Cases

  • An elegant and easy to use API. Fx devs, Jetpack and extension developers will all benefit.
  • allows us to prototype the new ui for Fx.next
  • 'Pluggable' results handling Easy wrappable querying object and results
  • Focus on JS usage
  • Paging Support - Sounds like a great use of Generators! dropped, there is no perf gain, can be a follow-up
  • Fetch individual record by Id or other property

(After talking to the Labs guys about Weave, we should try to provide a plugin-interface whereby you can allow a developer to keep track of and index random JSON objects.

Dan Mills has example code that was created for the people store. Link coming soon.)

Non Goals

  • Tailoring the API to tree views
  • Creating the perfect API (lets iterate)
  • Create an abstract datastore
  • Intended as a snap-in replacement of the current API Future follow-up through a live-updating wrapper
  • Wrap current sync and XPCOM API

Status

  • Old prototype landed in Jetpack 0.8, currently being revised
  • Team
    • API: ddahl, mak
    • UX/UI: not required so far, future users could though
  • Implementation bugs: bug 522572 bug 545700 bug 531940 bug 543888
  • Next Steps
    • Iterate the feel of the API with the Places team and other interested developers. Ideas and feedback are encouraged. Build working examples to encourage feedback.

Timeline / Milestones

  • 2010/02 Landed in Jetpack 0.8
  • 2010/03 Work on "skeleton" API design with Marco
  • 2010/05 New redesign, collecting feedback and proceeding.

Delivery Requirements

  • This is purely additive.
  • Coordination with Jetpack as this should be coded once, checked into places and imported into Jetpack's implementation.

Constraints

  • Final implementation should use only Async Storage APIs

Dependencies

Testing

  • Comprehensive xpcshell tests
  • Example/Documentation style Browser Chrome tests Not for now, when treeviews will use this.
  • TDD

Related Projects

  • Jetpack

Additional Details

The initial API "sketches" are here: [1]

Faaborg has been mocking up designs and putting a lot pf thought into the UX: http://blog.mozilla.com/faaborg/2009/10/13/browsing-your-personal-web

Inspiration

Links to code we should study...

Gloda

Gloda: [2]

facet.js: [3]

Gloda Fundamental Attribute provider: [4]

Explicit Attribute Provider: [5]

FacetView: [6]

Current Sketches

function callback(results) {
  // results is a simple array of ResultItem, it's pretty easy to wrap it
  // if more complex management is needed.
  if (results.length == 0)
    dump("En empty results array indicates last results push\n");
  else
    dump("Collected " + results.length + " results this turn\n");
}

// Results are pushed to the callback as soon as they are available.
new PlacesQuery([object]QueryConf, [function]callback, [scope]thisObject);

// It's possible to pass multiple QueryConf objects, results will be merged
// based on the .merge attribute of each config.
// First query results are always unioned.
// In case of multiple queries, sort, group and limit of the first query will
// be used for the global result.
new PlacesQuery([QueryConf1, QueryConf2], callback, thisObject);

// It's possible to create the query, but run it later.
let query = new PlacesQuery(QueryConf);
query.execute(callback, thisObject);

//The query can be so configured:
QueryConf = {

  phrase: string.
          Show only pages containing this string in either title, uri or
          tags.  Case insensitive.  Can use ^ and $ to match beginning or
          end.

  host: string.
        Show only pages containing this string in the host.  Case
        insensitive.  Can use ^ and $ to match beginning or end.

  uri: string.
       Show only pages containing this string in the uri.  Case
       insensitive.  Can use ^ and $ to match beginning or end.

  annotated: array of strings.
             Show only pages with these annotations (Either page or item).

  bookmarked: object {
    tags: array of strings.
          Show only pages tagged with these tags.
    folders: array of numbers.
             Show contents of these folders. (non-recursive)
    items: array of numbers.
           Show only these items.
    when: array of 2 Date objects.
          Show only bookmarks created between these times.
          Can use null beginning or end time to match till epoch
          or now.
    modified: array of 2 Date objects.
              Show only bookmarks modified between these times.
              Can use null beginning or end time to match till
              epoch or now.
    excludeNonContainers: boolean.
                          Removes any non-container from results.
                          Default is false.
    excludeReadOnlyContainers: boolean.
                               Removes read only containers from
                               results.  Default is false.
  }

  visited: object {
         howMany: array of numbers.
                  Show only pages with these so many visits.
                  Can use null minimum or maximum to match anything.
         transitions: array fo transition types.
                      Show only pages with at least one visit with these
                      transitions.
         when: array of 2 Date objects.
               Show only pages with visits between these times.
               Can use null beginning or end time to match till epoch
               or now.
         excludeRedirectsSources: boolean.
                                  Removes redirects sources from results.
                                  Default is false.
         excludeRedirectsTargets: boolean.
                                  Removes redirects targets from results.
                                  Default is false.
         includeHiddenPages: boolean.
                             Returns also pages marked as hidden.
                             Default is false.
         includeVisits: boolean.
                        Returns all visits.
                        Default is false, that means visits are grouped
                        by uri, and no duplicates are returned.
  }

  sort: object {
    by: string.
        Either "none", "title", "time", "uri", "accessCount" or
        "lastModified", "frecency".  Defaults to "none".
    dir: string.
         Either "asc" or "desc".  Defaults to "asc".
  }

  group: string.
         Either "tags", "folders", "day", "month", "year" or "domain".
         Defaults to "none".

  limit: number.
         Maximum umber of results to return.  Defaults to all results.

  merge: string.
         How to merge this query's results with others in the same request.
         Either "union", "intersect" or "except".
}