Since ogCP multi ogServer support, g.server is not used anymore. The
list of available servers is stored in the global variable "servers" at
og_server.py
Commit 648a1a315e1d35f67c5301833bddd3753cc51793 changes do not work well
with large scopes trees. It takes several minutes to check a scope.
Improve javascript code to speed up scope selection.
From MDN:
Technically, the value for an id attribute may contain any character,
except whitespace characters. However, to avoid inadvertent errors, only
ASCII letters, digits, '_', and '-' should be used and the value for an
id attribute should start with a letter.
https://developer.mozilla.org/en-US/docs/Web/HTML/Global_attributes/id
Clients pills show MACs regardless of the ogServer they belong to. Flask
GET /client/mac API uses "get_server_from_clients()" function to send
the command to the correct ogServer.
When users select an image, its ogServer is sent too. Then, the back-end
(flask) processes the form to obtain this ogServer and send it the
pertinent requests.
It also makes "Image details" view use this new parameter, and adds
javascript code to ensure that users can only work with several images
at the same time if they belong to the same ogServer.
Images view fetch images from all ogServers configured and show them in
the left tree as a nested list. ogServers are represented as the parents
of their images.
This commit adds a new select input with all available ogServers to "Add
center" view. The ogCP creates the center in the ogServer selected by
the user.
With this commit when users select a scope, the ogServer to
which it belongs is also sent. Then when processing the form we obtain
this ogServer and we can send it the pertinent requests.
This commit also makes action "Add client" to use the ogServer sent in
the form.
This commit adapt how "get_scopes()" accesses the name of the ogServer
because commit eae64dd changes how "multi_request()" returns it.
Fixes: eae64dd ("Initial support of ogServer requests routing")
This commit adds a helper function ("get_server_from_clients()") to get
the adequate ogServer according to the targeted clients.
It also makes "Client details" command use this function.
'multi_request()' function make requests to ogServers defined in the
configuration and aggregates it responses.
This commit also makes 'get_scopes()' use it.
Update the configuration file format to support several ogServers and
retrieve scopes from those ogServers.
The old config file format is still supported.
On image creation use the id to identify repositories instead of the IP.
ogServer related commit:
* 52a38d3 ("#915 Use the repository id on image creation")
This patch moves login password hashing from the frontend/javascript to
the backend/flask.
This patch moves password hashing of login and user management forms.
Related commits:
* c7b0024 ("Add password hashing")
* 661254b ("Add 'Add user' to Users section")
Creates "Edit user" form with the following inputs: password, password
confirmation, role (administrator or regular), allowed scopes. It does
no allow to change/edit the username.
The front-end now hashes passwords before sending them to the back-end.
It uses SHA-512.
This commit adds a hidden input which sends the password hash to not
interfere with browsers' save password functionality.
Also change passwords of the template configuration file for their
hashed/digested versions.
Add import clients form with required inputs: room and dhcpd.conf.
This permits users to rapidly add large amounts of clients to a room
using dhcpd.conf's syntax. Users can copy full dhcpd.conf files to the
text area and the parser only matches lines with the following format as
clients:
host dummy {hardware ethernet 12:34:56:78:90:ab; fixed-address 192.168.1.55; }
Allows updating the image file for an existing image.
This action is related to /image/create in ogServer API. If ogServer
receives an POST /image/create without the parameter "description", it
does not create a new image and only updates.
Each checkbox may have child checkboxes. If all those children are
checked, it be checked. If none are checked, it is unchecked. If some of
them are checked, then it’s in an indeterminate state (in this case
symbolically meaning “partially” checked).
Old id format example: id_1-1_2-4_3-2
New id format example: scope-1-4-2
It represents the node and its ancestors' position with respect to
theirs siblings. The rightmost value is the current node position.
Commit f70d90ba32 introduces the old format.
Disk selection uses "scopesForm" to submit the change, but setup/show
view receives new parameters not provided by this form.
Commit 17757baa4727 adds those parameters.
This commit gives disk selection its own form with all the required
parameters.
Otherwise, users can not switch between actions without going back to
the parent view.
This makes "scopesForm" work again in all views and not only in
scopes.html and commands.html.
Commit feed135554 introduces this regression.
get_client_setup() takes an IP address that identify the client,
instead of passing an iterator with one single IP address, pass
directly such IP address.
Otherwise, scopes with whitespaces in their names breaks the javascript
code.
From
https://developer.mozilla.org/en-US/docs/Web/HTML/Global_attributes/id:
---
Note: Technically, in HTML5, the value for an id attribute may contain
any character, except whitespace characters. However, to avoid
inadvertent errors, only ASCII letters, digits, '_', and '-' should be
used and the value for an id attribute should start with a letter. For
example, . has a special meaning in CSS (it acts as a class selector).
Unless you are careful to escape it in the CSS, it won't be recognized
as part of the value of an id attribute. It is easy to forget to do
this, resulting in bugs in your code that could be hard to detect.
---
* Rename "ogLives" to "ogLive images".
* Move "Number of images" table to take less space.
* Show storage data as "Storage size | Used (%) | Available (%)".
* Move "Latest images" an "ogLive images" to the bottom.
Some css classes are common to all pills statuses, so we do not need to
add or delete those classes on status change.
It also remove "badge-info" class because is never used.
Both "Add client" and "Client details" views use client_details.html
template. With this commit, "Add client" extends scopes.html and "Client
details" extend commands.hmlt.
Otherwise, scopes with dots in their names breaks the javascript code.
From
https://developer.mozilla.org/en-US/docs/Web/HTML/Global_attributes/id:
---
Note: Technically, in HTML5, the value for an id attribute may contain
any character, except whitespace characters. However, to avoid
inadvertent errors, only ASCII letters, digits, '_', and '-' should be
used and the value for an id attribute should start with a letter. For
example, . has a special meaning in CSS (it acts as a class selector).
Unless you are careful to escape it in the CSS, it won't be recognized
as part of the value of an id attribute. It is easy to forget to do
this, resulting in bugs in your code that could be hard to detect.
---
Extend commands.html template to keep scope tree, and action buttons
state.
When in the software inventory view, there exists two options: 'update'
or 'view'. Do not print raw json response from ogserver about the software
inventory/profile.
When vieweing the software list, print a html table using the new
template: software_list.html.
Do not print blank pages with plain text html error codes when something
goes wrong. Instead, print an error message after redirecting to the
commands view.
On scopes and commands views, draw clients as users selects them in the
scopes tree.
Trigger client drawing on two events:
1."change" event, occurs when the user clicks a client checkbox. This
event is standard [1].
2. "show-client" event, fires when ogcp get selected clients from
localStorage and when an user checks a parent checkbox. This event is
custom.
Dot characters (".") in clients names are replaced by underscore("_")
when used as id to avoid errors.
1. https://developer.mozilla.org/en-US/docs/Web/API/HTMLElement/change_event
If something goes wrong, the application shows an error message after
redirecting to the commands view.
Do not show blank pages with plain text error codes.
Extending from command.html is required for action views to mantain
the scope context on the left side column.
Enables setting dropdown button and action button as active when loading
the hardware inventory view.
Group related actions in the "commands" section inside a
dropdown button.
Declutters the panel in which the buttons are displayed.
Dropdown button jinja block name is used as prefix for the block names
of inside buttons. For example, for the image dropwdown button:
<button class="...{% block nav_image %}{% endblock %}...">
{{ _('Image') }}
</button>
Actions such as restore and create will show a jinja block inside class
attr like:
<input class="... dropdown-item{% block nav_image_restore %}{% endblock %}" ...
<input class="... dropdown-item{% block nav_image_create %}{% endblock %}" ...
Adds empty jinja blocks for the class attribute of all action buttons.
Child templates from command (actions) can set their nav button as
active overriding the corresponding {% block %}.
Currently, not all action templates extends the commands.html template.
This is not a problem, as the blocks are just empty.
This is a preparation for changes on each action template to extend
commands.html template.
{% block %} defined in nav template cannot be overriden by child
templates from base. This is a limitation in jinja.
Merge nav into base template so jinja blocks so no {% include %} is used
and these blocks can be overriden by child templates (commands, images,
dashboard...)
Avoid using request.endpoint to determine active nav item, decoupling
endpoint names from navigation. Instead use child templates to override
jinja blocks.
[1] https://stackoverflow.com/q/40537752https://stackoverflow.com/a/40562662https://github.com/pallets/jinja/issues/243
Otherwise, users have to click clients checkboxes every time they
switch view.
This commit adds one hook to capture when clients checkboxes changes.
When they are checked, ogcp (front-end) stores them in browser's local
storage. When they are unchecked, ogcp removes them from local storage.
Every time users load a page with the scopes tree, ogcp checks local
storage to restore selected clients.
Otherwise, users have to expand the scopes tree every time they switch
view.
This commit adds two hooks to capture when an element of the tree is
shown or is hidden[1]. When they are shown, ogcp (front-end) stores them
in browser's local storage[2]. When they are hidden, ogcp removes them
from local storage.
Every time users load a page with the scopes tree, ogcp checks local
storage to restore the scopes tree state.
Store and remove functions use stopPropagation() to prevent ancestors of
clicked elements to also be stored or removed from local storage.[3]
1. https://getbootstrap.com/docs/4.1/components/collapse/#events
2. https://developer.mozilla.org/en-US/docs/Web/API/Web_Storage_API
3. https://javascript.info/bubbling-and-capturing
User can confirm when deleting an image by double checking the
information displayed on this page.
Adds a new optional field in GenericForm: 'ids'. Used when dealing with
ids instead of ips.
This commit adds an extra view to ensure users do not reboot clients
accidentally.
ogcp GET /reboot returns the confirmation page and POST /reboot
builds and sends the request to ogServer.
It also includes Spanish translation of the new strings.
This commit adds an extra view to ensure users do not power off clients
accidentally.
ogcp GET /poweroff returns the confirmation page and POST /poweroff
builds and sends the request to ogServer.
It also includes Spanish translation of the new strings.
Javier Sánchez Parra
Daniel García Moreno
Jose M. Guisado