A download link for an archive cooked through the vault (directory or
revision_gitfast) does not remain alive ad vitam aeternam.

As the history of previously cooked objects remains stored in the browser
local storage, we need to handle the case where one wants to download a
previously cooked archive from a dead link.

In that case, check the link availability first before proceeding to the
real download. If it is dead, display a modal asking the user if he wants
to recook the archive.

Closes T1082

Related T1081
Closes D341

Closes T1081

This commit enables the rendering of pdf files in the content view
through the use of the pdf.js library from Mozilla (dynamically loaded
when required).

  - do not expand it to its original width when hovering the mouse on it

  - update brand image

Promote the use of following urls /browse/origin/(origin_url)/...
instead of /browse/origin/(origin_type)/url/(origin_url)/... in browse.

In a same manner, to provide origin context through query parameters,
simply use <url>?origin=(origin_url) instead of
<url>?origin_type=(origin_type)&origin_url=(origin_url)

Nicolas Dandrimont authored 6 years ago and Antoine Lambert committed 6 years ago

Summary:
Some bits of tests were never executed, being after the exception has been
raised in an assertRaises block. Fix all such instances

Test Plan: the unit tests pass

Reviewers: #reviewers, anlambert

Reviewed By: #reviewers, anlambert

Differential Revision: https://forge.softwareheritage.org/D338

Summary:
We used to call the database twice for directory listings. We now
special-case the empty directory, which allows us to only call the database
once, saving one round trip on the most used query.

Test Plan: Unit tests updated

Reviewers: anlambert, #reviewers!

Differential Revision: https://forge.softwareheritage.org/D337

This commits adds a popover panel reachable from the 'Show identifiers'
entry in the 'Actions' menu for displaying swh persistent identifiers
related to the currently browsed objects (content/directory/release/
revision + snapshot context if any).

The panel offers easy copy to clipboard of the swh ids and related
urls but also the possibility to add optional parts (origin info
for instance).

Closes T1004

Closes T1050

When clicking on the 'Browse' menu entry, redirect to the search page
if no swh objects were previously browsed in the current browser
session instead of displaying an info text.

This commits adds support for optional parts of persistent identifiers
in swh-web in order to add origin context when resolving the swh object
to browse. It is also possible to set lines to highlight for swh content
object.

Also to be noted, persistent id urls are now rooted to / instead of
/browse. Nevertheless backward compatibility is maintained for HAL.

Closes T1042

This commit improves the overall design of swh web applications by
making it more compact in order to gain more space for displaying the
archive content.

The following changes have been applied:

  - use a collapsible sidebar for switching between the different
    features (currently Web API, Browse, Search, Vault, Help) instead
    of putting them in the top navbar

  - remove the use of panels (cards in Boostrap 4.x) as the main layout
    for swh-web pages, instead move their headers in the top navbar and
    use their content as the main page layout

Closes T1052
Closes T1032

Closes T1067

Closes T1065

When one wants to display a content, HTTP requests are sent to the indexer
storage in order to retrieve the associated mime type, language and license.
Those requests may fail so properly handle errors in order to avoid
content display errors.

Closes T1060

This commit removes the collapsible panel, located in the top
part of numerous views, used to display the metadata associated
to a swh object in favor of a popover reachable from the Actions
menu located on the right of the view.

Related T1052

When a README is written in reStructuredText (rst), convert it to html server
side through the use of pypandoc in order to get a proper display of it.

Indeed, the client side library used to render markdown (showdown.js) does
not handle rst properly and there is no viable solution to convert rst
to html directly in the browser.