--- name: "python-fedora" section: "1" platform: "linux" package: "python-fedora-doc" version: "1.1.1-5" release: "Debian 13 (Trixie)" platforms: ["linux"] url: "https://new.carta.tech/python-fedora" --- # python-fedora(1) python-fedora 1.1.1 The python-fedora module makes it easier for programmers to create both Fedora Services and clients that talk to the services. It's licensed under the GNU Lesser General Public License version 2 or later. ## NAME python-fedora - python-fedora 1.1.1 The python-fedora module makes it easier for programmers to create both Fedora Services and clients that talk to the services. It's licensed under the GNU Lesser General Public License version 2 or later. ## FEDORA CLIENT - **Authors** Toshio Kuratomi Luke Macken - **Date** 28 May 2008 - **For Version** 0\.3.x The client module allows you to easily code an application that talks to a *Fedora Service*. It handles the details of decoding the data sent from the Service into a python data structure and raises an *Exception* if an error is encountered. ### BaseClient The *BaseClient* class is the basis of all your interactions with the server. It is flexible enough to be used as is for talking with a service but is really meant to be subclassed and have methods written for it that do the things you specifically need to interact with the *Fedora Service* you care about. Authors of a *Fedora Service* are encouraged to provide their own subclasses of *BaseClient* that make it easier for other people to use a particular Service out of the box. ### Using Standalone If you don't want to subclass, you can use *BaseClient* as a utility class to talk to any *Fedora Service*. There's three steps to this. First you import the *BaseClient* and *Exceptions* from the **fedora.client** module. Then you create a new *BaseClient* with the URL that points to the root of the *Fedora Service* you're interacting with. Finally, you retrieve data from a method on the server. Here's some code that illustrates the process: ``` from fedora.client import BaseClient, AppError, ServerError client = BaseClient('https://admin.fedoraproject.org/pkgdb') try: collectionData = client.send_request('/collections', auth=False) except ServerError as e: print('%s' % e) except AppError as e: print('%s: %s' % (e.name, e.message)) for collection in collectionData['collections']: print('%s %s' % (collection['name'], collection['version']) ``` ### BaseClient Constructor In our example we only provide **BaseClient()** with the URL fragment it uses as the base of all requests. There are several more optional parameters that can be helpful. If you need to make an authenticated request you can specify the username and password to use when you construct your *BaseClient* using the **username** and **password** keyword arguments. If you do not use these, authenticated requests will try to connect via a cookie that was saved from previous runs of *BaseClient*. If that fails as well, *BaseClient* will throw an *Exception* which you can catch in order to prompt for a new username and password: ``` from fedora.client import BaseClient, AuthError import getpass MAX_RETRIES = 5 client = BaseClient('https://admin.fedoraproject.org/pkgdb', username='foo', password='bar') # Note this is simplistic. It only prompts once for another password. # Your application may want to loop through this several times. while (count < MAX_RETRIES): try: collectionData = client.send_request('/collections', auth=True) except AuthError as e: client.password = getpass.getpass('Retype password for %s: ' % username) else: # data retrieved or we had an error unrelated to username/password break count = count + 1 ``` **WARNING:** Note that although you can set the **username** and **password** as shown above you do have to be careful in cases where your application is multithreaded or simply processes requests for more than one user with the same *BaseClient*. In those cases, you can accidentally overwrite the **username** and **password** between two requests. To avoid this, make sure you instantiate a separate *BaseClient* for every thread of control or for every request you handle or use *ProxyClient* instead. The **useragent** parameter is useful for identifying in log files that your script is calling the server rather than another. The default value is **Fedora BaseClient/VERSION** where VERSION is the version of the *BaseClient* module. If you want to override this just give another string to this: ``` client = BaseClient('https://admin.fedoraproject.org/pkgdb', useragent='Package Database Client/1.0') ``` The **debug** parameter turns on a little extra output when running the program. Set it to true if you're having trouble and want to figure out what is happening inside of the *BaseClient* code. ### send\_request() **send\_request()** is what does the heavy lifting of making a request of the server, receiving the reply, and turning that into a python dictionary. The usage is pretty straightforward. The first argument to **send\_request()** is **method**. It contains the name of the method on the server. It also has any of the positional parameters that the method expects (extra path information interpreted by the server for those building non-*TurboGears* applications). The **auth** keyword argument is a boolean. If True, the session cookie for the user is sent to the server. If this fails, the **username** and **password** are sent. If that fails, an *Exception* is raised that you can handle in your code. **req\_params** contains a dictionary of additional keyword arguments for the server method. These would be the names and values returned via a form if it was a CGI. Note that parameters passed as extra path information should be added to the **method** argument instead. An example: ``` import BaseClient client = BaseClient('https://admin.fedoraproject.org/pkgdb/') client.send_request('/package/name/python-fedora', auth=False, req_params={'collectionVersion': '9', 'collectionName': 'Fedora'}) ``` In this particular example, knowing how the server works, **/packages/name/** defines the method that the server is going to invoke. **python-fedora** is a positional parameter for the name of the package we're looking up. **auth=False** means that we'll try to look at this method without having to authenticate. The **req\_params** sends two additional keyword arguments: **collectionName** which specifies whether to filter on a single distro or include Fedora, Fedora EPEL, Fedora OLPC, and Red Hat Linux in the output and **collectionVersion** which specifies which version of the distribution to output for. The URL constructed by *BaseClient* to the server could be expressed as\[#\]\_: ``` https://admin.fedoraproject.org/pkgdb/package/name/python-fedora/?collectionName=Fedora&collectionVersion=9 ``` In previous releases of python-fedora, there would be one further query parameter: **tg\_format=json**. That parameter instructed the server to return the information as JSON data instead of HTML. Although this is usually still supported in the server, *BaseClient* has deprecated this method. Servers should be configured to use an **Accept** header to get this information instead. See the *JSON output* section of the *Fedora Service* documentation for more information about the server side. ### Subclassing Building a client using subclassing builds on the information you've already seen inside of *BaseClient*. You might want to use this if you want to provide a module for third parties to access a particular *Fedora Service*. A subclass can provide a set of standard methods for calling the server instead of forcing the user to remember the URLs used to access the server directly. Here's an example that turns the previous calls into the basis of a python API to the *Fedora Package Database*: ``` import getpass import sys from fedora.client import BaseClient, AuthError class MyClient(BaseClient): def __init__(self, baseURL='https://admin.fedoraproject.org/pkgdb', username=None, password=None, useragent='Package Database Client/1.0', debug=None): super(BaseClient, self).__init__(baseURL, username, password, useragent, debug) def collection_list(self): '''Return a list of collections.''' return client.send_request('/collection') def package_owners(self, package, collectionName=None, collectionVersion=None): '''Return a mapping of release to owner for this package.''' pkgData = client.send_request('/packages/name/%s' % (package), {'collectionName': collectionName, 'collectionVersion': collectionVersion}) ownerMap = {} for listing in pkgData['packageListings']: ownerMap['-'.join(listing['collection']['name'], listing['collection']['version'])] = \ listing['owneruser'] return ownerMap ``` A few things to note: - 1\. In our constructor we list a default **baseURL** and **useragent**. This is usually a good idea as we know the URL of the *Fedora Service* we're connecting to and we want to know that people are using our specific API. - 2\. Sometimes we'll want methods that are thin shells around the server methods like **collection\_list()**. Other times we'll want to do more post processing to get specific results as **package\_owners()** does. Both types of methods are valid if they fit the needs of your API. If you find yourself writing more of the latter, though, you may want to consider getting a new method implemented in the server that can return results more appropriate to your needs as it could save processing on the server and bandwidth downloading the data to get information that more closely matches what you need. See **pydoc fedora.client.fas2** for a module that implements a standard client API for the *Fedora Account System* ### Handling Errors *BaseClient* will throw a variety of errors that can be caught to tell you what kind of error was generated. ### Exceptions - **FedoraServiceError** The base of all exceptions raised by *BaseClient*. If your code needs to catch any of the listed errors then you can catch that to do so. - **ServerError** Raised if there's a problem communicating with the service. For instance, if we receive an HTML response instead of JSON. - **AuthError** If something happens during authentication, like an invalid usernsme or password, **AuthError** will be raised. You can catch this to prompt the user for a new usernsme. - **AppError** If there is a *server side error* when processing a request, the *Fedora Service* can alert the client of this by setting certain flags in the response. *BaseClient* will see these flags and raise an AppError. The name of the error will be stored in AppError's **name** field. The error's message will be stored in **message**. ### Example Here's an example of the exceptions in action: ``` from fedora.client import ServerError, AuthError, AppError, BaseClient import getpass MAXRETRIES = 5 client = BaseClient('https://admin.fedoraproject.org/pkgdb') for retry in range(0, MAXRETRIES): try: collectionData = client.send_request('/collections', auth=True) except AuthError as e: from six.moves import input client.username = input('Username: ').strip() client.password = getpass.getpass('Password: ') continue except ServerError as e: print('Error talking to the server: %s' % e) break except AppError as e: print('The server issued the following exception: %s: %s' % ( e.name, e.message)) for collection in collectionData['collections']: print('%s %s' % (collection[0]['name'], collection[0]['version'])) ``` ### OpenIdBaseClient Applications that use OpenId to authenticate are not able to use the standard BaseClient because the pattern of authenticating is very different. We've written a separate client object called *OpenIdBaseClient* to do this. ## EXISTING SERVICES There are many Services in Fedora. Many of these have an interface that we can query and get back information as *JSON* data. There is documentation here about both the services and the client modules that can access them. ### Fedora Account System FAS is the Fedora Account System. It holds the account data for all of our contributors. - **class fedora.client.AccountSystem(base\_url='https://admin.fedoraproject.org/accounts/', \*args, \*\*kwargs)** An object for querying the Fedora Account System. The Account System object provides a python API for talking to the Fedora Account System. It abstracts the http requests, cookie handling, and other details so you can concentrate on the methods that are important to your program. **WARNING:** If your code is trying to use the AccountSystem object to connect to fas for multiple users you probably want to use *FasProxyClient* instead. If your code is trying to reuse a single instance of AccountSystem for multiple users you *definitely* want to use *FasProxyClient* instead. Using AccountSystem in these cases may result in a user being logged in as a different user. (This may be the case even if you instantiate a new AccountSystem object for each user if :attr:cache\_session: is True since that creates a file on the file system that can end up loading session credentials for the wrong person. Changed in version 0.3.26: Added *gravatar\_url()* that returns a url to a gravatar for a user. Changed in version 0.3.33: Renamed *gravatar\_url()* to *avatar\_url()*. - **avatar\_url(username, size=64, default=None, lookup\_email=True, service=None)** Returns a URL to an avatar for a given username. Avatars are drawn from third party services. - **Parameters** - • **username** -- FAS username to construct a avatar url for - • **size** -- size of the avatar. Allowed sizes are 32, 64, 140. Default: 64 - • **default** -- If the service does not have a avatar image for the email address, this url is returned instead. Default: the fedora logo at the specified size. - • **lookup\_email** -- If true, use the email from FAS for gravatar.com lookups, otherwise just append @fedoraproject.org to the username. For libravatar.org lookups, this is ignored. The openid identifier of the user is used instead. Note that gravatar.com lookups will be much slower if lookup\_email is set to True since we'd have to make a query against FAS itself. - • **service** -- One of 'libravatar' or 'gravatar'. Default: 'libravatar'. - **Raises** **ValueError** -- if the size parameter is not allowed or if the service is not one of 'libravatar' or 'gravatar' - **Return type** **str** - **Returns** url of a avatar for the user If that user has no avatar entry, instruct the remote service to redirect us to the Fedora logo. If that user has no email attribute, then make a fake request to the third party service. Added in version 0.3.26. - **create\_group(name, display\_name, owner, group\_type, invite\_only=0, needs\_sponsor=0, user\_can\_remove=1, prerequisite='', joinmsg='', apply\_rules='None')** Creates a FAS group. - **Parameters** - • **name** -- The short group name (alphanumeric only). - • **display\_name** -- A longer version of the group's name. - • **owner** -- The username of the FAS account which owns the new group. - • **group\_type** -- The kind of group being created. Current valid options are git, svn, hg, shell, and tracking. - • **invite\_only** -- Users must be invited to the group, they cannot join on their own. - • **needs\_sponsor** -- Users must be sponsored into the group. - • **user\_can\_remove** -- Users can remove themselves from the group. - • **prerequisite** -- Users must be in the given group (string) before they can join the new group. - • **joinmsg** -- A message shown to users when they apply to the group. - • **apply\_rules** -- Rules for applying to the group, shown to users before they apply. - **Return type** **munch.Munch** - **Returns** A Munch containing information about the group that was created. Added in version 0.3.29. - **get\_config(username, application, attribute)** Return the config entry for the key values. - **Parameters** - • **username** -- Username of the person - • **application** -- Application for which the config is set - • **attribute** -- Attribute key to lookup - **Raises** *AppError* -- if the server returns an exception - **Returns** The unicode string that describes the value. If no entry matched the username, application, and attribute then None is returned. - **get\_configs\_like(username, application, pattern='\*')** Return the config entries that match the keys and the pattern. Note: authentication on the server will prevent anyone but the user or a fas admin from viewing or changing their configs. - **Parameters** - • **username** -- Username of the person - • **application** -- Application for which the config is set - • **pattern** -- A pattern to select values for. This accepts \* as a wildcard character. Default='\*' - **Raises** *AppError* -- if the server returns an exception - **Returns** A dict mapping **attribute** to **value**. - **gravatar\_url(\*args, \*\*kwargs)** *Deprecated* - Use avatar\_url. Added in version 0.3.26. - **group\_by\_id(group\_id)** Returns a group object based on its id - **group\_by\_name(groupname)** Returns a group object based on its name - **group\_data(force\_refresh=None)** Return administrators/sponsors/users and group type for all groups - **Parameters** **force\_refresh** -- If true, the returned data will be queried from the database, as opposed to memcached. - **Raises** *AppError* -- if the query failed on the server - **Returns** A dict mapping group names to the group type and the user IDs of the administrator, sponsors, and users of the group. Added in version 0.3.8. - **group\_members(groupname)** Return a list of people approved for a group. This method returns a list of people who are in the requested group. The people are all approved in the group. Unapproved people are not shown. The format of data is: ``` \[{'username': 'person1', 'role_type': 'user'}, \{'username': 'person2', 'role_type': 'sponsor'}] ``` role\_type can be one of 'user', 'sponsor', or 'administrator'. Added in version 0.3.2. Changed in version 0.3.21: Return a Bunch instead of a DictContainer - **property insecure** If this attribute is set to True, do not check server certificates against their CA's. This means that man-in-the-middle attacks are possible. You might turn this option on for testing against a local version of a server with a self-signed certificate but it should be off in production. - **people\_by\_groupname(groupname)** Return a list of persons for the given groupname. - **Parameters** **groupname** -- Name of the group to look up - **Returns** A list of person objects from the group. If the group contains no entries, then an empty list is returned. - **people\_by\_id()** *Deprecated* Use people\_by\_key() instead. Returns a dict relating user IDs to human\_name, email, username, and bugzilla email Changed in version 0.3.21: Return a Bunch instead of a DictContainer - **people\_by\_key(key='username', search='\*', fields=None)** Return a dict of people For example: ``` >>> ret_val = FASCLIENT.people_by_key( ... key='email', search='toshio*', fields=['id']) >>> ret_val.keys() a.badger@[...].com a.badger+test1@[...].com a.badger+test2@[...].com >>> ret_val.values() 100068 102023 102434 ``` - **Parameters** - • **key** -- Key used to organize the returned dictionary. Valid values are 'id', 'username', or 'email'. Default is 'username'. - • **search** -- Pattern to match usernames against. Defaults to the '\*' wildcard which matches everyone. - • **fields** -- Limit the data returned to a specific list of fields. The default is to retrieve all fields. Valid fields are: - • affiliation - • alias\_enabled - • bugzilla\_email - • certificate\_serial - • comments - • country\_code - • creation - • email - • emailtoken - • facsimile - • gpg\_keyid - • group\_roles - • human\_name - • id - • internal\_comments - • ircnick - • last\_seen - • latitude - • locale - • longitude - • memberships - • old\_password - • password - • password\_changed - • passwordtoken - • postal\_address - • privacy - • roles - • ssh\_key - • status - • status\_change - • telephone - • timezone - • unverified\_email - • username Note that for most users who access this data, many of these fields will be set to None due to security or privacy settings. - **Returns** a dict relating the key value to the fields. Changed in version 0.3.21: Return a Bunch instead of a DictContainer Changed in version 0.3.26: Fixed to return a list with both people who have signed the CLA and have not - **people\_query(constraints=None, columns=None)** Returns a list of dicts representing database rows - **Parameters** - • **constraints** -- A dictionary specifying WHERE constraints on columns - • **columns** -- A list of columns to be selected in the query - **Raises** *AppError* -- if the query failed on the server (most likely because the server was given a bad query) - **Returns** A list of dicts representing database rows (the keys of the dict are the columns requested) Added in version 0.3.12.1. - **person\_by\_id(person\_id)** Returns a person object based on its id - **person\_by\_username(username)** Returns a person object based on its username - **proxy = None** - **set\_config(username, application, attribute, value)** Set a config entry in FAS for the user. Note: authentication on the server will prevent anyone but the user or a fas admin from viewing or changing their configs. - **Parameters** - • **username** -- Username of the person - • **application** -- Application for which the config is set - • **attribute** -- The name of the config key that we're setting - • **value** -- The value to set this to - **Raises** *AppError* -- if the server returns an exception - **user\_data()** Return user data for all users in FAS Note: If the user is not authorized to see password hashes, '\*' is returned for the hash. - **Raises** *AppError* -- if the query failed on the server - **Returns** A dict mapping user IDs to a username, password hash, SSH public key, email address, and status. Added in version 0.3.8. - **user\_gencert()** Generate a cert for a user - **user\_id()** Returns a dict relating user IDs to usernames - **verify\_password(username, password)** Return whether the username and password pair are valid. - **Parameters** - • **username** -- username to try authenticating - • **password** -- password for the user - **Returns** True if the username/password are valid. False otherwise. ### Threadsafe Account System Access It is not safe to use a single instance of the *AccountSystem* object in multiple threads. This is because instance variables are used to hold some connection-specific information (for instance, the user who is logging in). For this reason, we also provide the *fedora.client.FasProxyClient* object. This is especially handy when writing authn and authz adaptors that talk to fas from a multithreaded webserver. - **class fedora.client.FasProxyClient(base\_url='https://admin.fedoraproject.org/accounts/', \*args, \*\*kwargs)** A threadsafe client to the Fedora Account System. - **get\_user\_info(auth\_params)** Retrieve information about a logged in user. - **Parameters** **auth\_params** -- Auth information for a particular user. For instance, this can be a username/password pair or a session\_id. Refer to **fedora.client.proxyclient.ProxyClient.send\_request()** for all the legal values for this. - **Returns** a tuple of session\_id and information about the user. - **Raises** *AuthError* -- if the auth\_params do not give access - **group\_list(auth\_params)** Retrieve a list of groups - **Parameters** **auth\_params** -- Auth information for a particular user. For instance, this can be a username/password pair or a session\_id. Refer to **fedora.client.proxyclient.ProxyClient.send\_request()** for all the legal values for this. - **Returns** a tuple of session\_id and information about groups. The groups information is in two fields: - **groups** contains information about each group - **memberships** contains information about which users are members of which groups - **Raises** *AuthError* -- if the auth\_params do not give access - **login(username, password)** Login to the Account System - **Parameters** - • **username** -- username to send to FAS - • **password** -- Password to verify the username with - **Returns** a tuple of the session id FAS has associated with the user and the user's account information. This is similar to what is returned by **fedora.client.proxyclient.ProxyClient.get\_user\_info()** - **Raises** *AuthError* -- if the username and password do not work - **logout(session\_id)** Logout of the Account System - **Parameters** **session\_id** -- a FAS session\_id to remove from FAS - **person\_by\_id(person\_id, auth\_params)** Retrieve information about a particular person - **Parameters** **auth\_params** -- Auth information for a particular user. For instance, this can be a username/password pair or a session\_id. Refer to **fedora.client.proxyclient.ProxyClient.send\_request()** for all the legal values for this. - **Returns** a tuple of session\_id and information about the user. - **Raises** - • *AppError* -- if the server returns an exception - • *AuthError* -- if the auth\_params do not give access - **refresh\_session(session\_id)** Try to refresh a session\_id to prevent it from timing out - **Parameters** **session\_id** -- FAS session\_id to refresh - **Returns** session\_id that FAS has set now - **verify\_password(username, password)** Return whether the username and password pair are valid. - **Parameters** - • **username** -- username to try authenticating - • **password** -- password for the user - **Returns** True if the username/password are valid. False otherwise. - **verify\_session(session\_id)** Verify that a session is active. - **Parameters** **session\_id** -- session\_id to verify is currently associated with a logged in user - **Returns** True if the session\_id is valid. False otherwise. ### Bodhi, the Update Server Bodhi is used to push updates from the build system to the download repositories. It lets packagers send packages to the testing repository or to the update repository. pythyon-fedora currently supports both the old Bodhi1 interface and the new Bodhi2 interface. By using **fedora.client.BodhiCLient**, the correct one should be returned to you depending on what is running live on Fedora Infrastructure servers. ## FEDORA SERVICES - **Authors** Toshio Kuratomi Luke Macken - **Date** 02 February 2009 - **For Version** 0\.3.x In the loosest sense, a Fedora Service is a web application that sends data that *BaseClient* is able to understand. This document defines things that a web application must currently do for *BaseClient* to understand it. ### TurboGears and fedora.tg All current Fedora Services are written in *TurboGears*. Examples in this document will be for that framework. However, other frameworks can be used to write a Service if they can correctly create the data that *BaseClient* needs. A Fedora Service differs from other web applications in that certain URLs for the web application are an API layer that the Service can send and receive *JSON* data from for *BaseClient* to interpret. This imposes certain constraints on what data can be sent and what data can be received from those URLs. The **fedora.tg** module contains functions that help you write code that communicated well with *BaseClient*. The *TurboGears* framework separates an application into *model*, *view*, and *controller* layers. The model is typically a database and holds the raw data that the application needs. The view formats the data before output. The controller makes decisions about what data to retrieve from the model and which view to pass it onto. The code that you'll most often need to use from **fedora.tg** operates on the *controller* layer but there's also code that works on the model and view behind the scenes. ### Controllers The *controller* is the code that processes an http request. It validates and processes requests and parameters sent by the user, gets data from the model to satisfy the request, and then passes it onto a view layer to be returned to the user. **fedora.tg.utils** contains several helpful functions for working with controllers. ### URLs as API In *TurboGears* and most web frameworks, a URL is a kind of API for accessing the data your web app provides. This data can be made available in multiple formats. *TurboGears* allows you to use one URL to serve multiple formats by specifying a query parameter or a special header. ### Selecting JSON Output A URL in *TurboGears* can serve double duty by returning multiple formats depending on how it is called. In most cases, the URL will return HTML or XHTML by default. By adding the query parameter, **tg\_format=json** you can switch from returning the default format to returning *JSON* data. You need to add an **@expose(allow\_json=True)** decorator \[1\] to your controller method to tell *TurboGears* that this controller should return *JSON* data: ``` @expose(allow_json=True) @expose(template='my.templates.amplifypage') def amplify(self, data): ``` **allow\_json=True** is a shortcut for this: ``` @expose("json", content_type="text/javascript", as_format="json", accept_format="text/javascript") ``` That means that the controller method will use the **json** template (uses TurboJson to marshal the returned data to *JSON*) to return data of type **text/javascript** when either of these conditions is met: a query param of **tg\_format=json** or an **Accept: text/javascript** header is sent. *BaseClient* in python-fedora 0.1.x and 0.2.x use the query parameter method of selecting *JSON* output. *BaseClient* 0.2.99.6 and 0.3.x use both the header method and the query parameter since the argument was made that the header method is friendlier to other web frameworks. 0.4 intends to use the header alone. If you use **allow\_json=True** this change shouldn't matter. If you specify the **@expose("json")** decorator only **accept\_format** is set. So it is advisable to include both **as\_format** and **accept\_format** in your decorator. Note that this is a limitation of *TurboGears* \<= 1.0.4.4. If your application is only going to run on *TurboGears* \> 1.0.4.4 you should be able to use a plain **@expose("json")** \[2\]. - \[1\] *http://docs.turbogears.org/1.0/ExposeDecorator#same-method-different-template* - \[2\] *http://trac.turbogears.org/ticket/1459#comment:4* ### Why Two Formats from a Single URL? When designing your URLs you might wonder why you'd want to return *JSON* and HTML from a single controller method instead of having two separate controller methods. For instance, separating the URLs into their own namespaces might seem logical: **/app/json/get\_user/USERNAME** as opposed to **/app/user/USERNAME**. Doing things with two URLs as opposed to one has both benefits and drawbacks. ### Benefits of One Method Handling Multiple Formats - • Usually less code as there's only one controller method - • When a user sees a page that they want to get data from, they can get it as *JSON* instead of screen scraping. - • Forces the application designer to think more about the API that is being provided to the users instead of just the needs of the web page they are creating. - • Makes it easier to see what data an application will need to implement an alternate interface since you can simply look at the template code to see what variables are being used on a particular page. ### Benefits of Multiple Methods for Each Format - • Avoids special casing for error handlers (See below) - • Separates URLs that you intend users to grab *JSON* data from URLs where you only want to display HTML. - • Allows the URLs that support *JSON* to concentrate on trimming the size of the data sent while URLs that only return HTML can return whole objects. - • Organization can be better if you don't have to include all of the pages that may only be useful for user interface elements. Personal use has found that allowing *JSON* requests on one controller method works well for cases where you want the user to get data and for traditional form based user interaction. AJAX requests have been better served via dedicated methods. ### Return Values The toplevel of the return values should be a dict. This is the natural return value for *TurboGears* applications. ### Marshaling All data should be encoded in *JSON* before being returned. This is normally taken care of automatically by *TurboGears* and simplejson. If you are returning non-builtin objects you may have to define an *\_\_json\_\_()* method. ### Unicode simplejson (and probably other *JSON* libraries) will take care of encoding Unicode strings to *JSON* so be sure that you are passing Unicode strings around rather than encoded byte strings. ### Error Handling In python, error conditions are handled by raising an exception. However, an exception object will not propagate automatically through a return from the server. Instead we set several special variables in the returned data to inform *BaseClient* of any errors. At present, when *BaseClient* receives an error it raises an exception of its own with the exception information from the server inside. Raising the same exception as the server is being investigated but may pose security risks so hasn't yet been implemented. ### exc All URLs which return *JSON* data should set the **exc** variable when the method fails unexpectedly (a database call failed, a place where you would normally raise an exception, or where you'd redirect to an error page if a user was viewing the HTML version of the web app). **exc** should be set to the name of an exception and *tg\_flash* set to the message that would normally be given to the exception's constructor. If the return is a success (expected values are being returned from the method or a value was updated successfully) **exc** may either be unset or set to **None**. ### tg\_flash When viewing the HTML web app, **tg\_flash** can be set with a message to display to the user either on the next page load or via an AJAX handler. When used in conjunction with *JSON*, **exc=EXCEPTIONNAME**, and *BaseClient*, **tg\_flash** should be set to an error message that the client can use to identify what went wrong or display to the user. It's equivalent to the message you would normally give when raising an exception. ### Authentication Errors Errors in authentication are a special case. Instead of returning an error with **exc='AuthError'** set, the server should return with **response.status =** **403**. *BaseClient* will see the 403 and raise an *AuthError*. This is the signal for the client to ask the user for new credentials (usually a new username and password). **NOTE:** Upstream *TurboGears* has switched to sending a 401 for authentication problems. However, their use of 401 is against the http specification (It doesn't set the 'WWW-Authentication' header) and it causes problems for konqueror and webkit based browsers so we probably will not be switching. ### Performing Different Actions when Returning JSON So far we've run across three features of *TurboGears* that provide value to a web application but don't work when returning *JSON* data. We provide a function that can code around this. **fedora.tg.utils.request\_format()** will return the format that the page is being returned as. Code can use this to check whether *JSON* output is expected and do something different based on it: ``` output = {'tg_flash': 'An Error Occurred'} if fedora.tg.utils.request_format() == 'json': output['exc'] = 'ServerError' else: output['tg_template'] = 'my.templates.error' return output ``` In this example, we return an error through our "exception" mechanism if we are returning *JSON* and return an error page by resetting the template if not. ### Redirects Redirects do not play well with *JSON* \[3\] because *TurboGears* is unable to turn the function returned from the redirect into a dictionary that can be turned into *JSON*. Redirects are commonly used to express errors. This is actually better expressed using *tg\_template* because that method leaves the URL intact. That allows the end user to look for spelling mistakes in their URL. If you need to use a redirect, the same recipe as above will allow you to split your code paths. - \[3\] Last checked in TurboGears 1.0.4 ### tg\_template Setting what template is returned to a user by setting tg\_template in the return dict (for instance, to display an error page without changing the URL) is a perfectly valid way to use *TurboGears*. Unfortunately, since *JSON* is simply another template in *TurboGears* you have to be sure not to interfere with the generation of *JSON* data. You need to check whether *JSON* was requested using **fedora.tg.utils.request\_format()** and only return a different template if that's not the case. The recipe above shows how to do this. ### Validators Validators are slightly different than the issues we've encountered so far. Validators are used to check and convert parameters sent to a controller method so that only good data is dealt with in the controller method itself. The problem is that when a validator detects a parameter that is invalid, it performs a special internal redirect to a method that is its **error\_handler**. We can't intercept this redirect because it happens in the decorators before our method is invoked. So we have to deal with the aftermath of the redirect in the **error\_handler** method: ``` class NotNumberValidator(turbogears.validators.FancyValidator): messages = {'Number': 'Numbers are not allowed'} def to_python(self, value, state=None): try: number = turbogears.validators.Number(value.strip()) except: return value raise validators.Invalid(self.message('Number', state), value, state) class AmplifyForm(turbogears.widgets.Form): template = my.templates.amplifyform submit_text = 'Enter word to amplify' fields = [ turbogears.widgets.TextField(name='data', validator=NotNumberValidator()) ] amplify_form = AmplifyForm() class mycontroller(RootController): @expose(template='my.templates.errorpage', allow_json=True) def no_numbers(self, data): errors = fedora.tg.utils.jsonify_validation_errors() if errors: return errors # Construct a dict to return the data error message as HTML via # the errorpage template pass @validate(form=amplify_form) @error_handler('no_numbers') @expose(template='my.templates.amplifypage', allow_json=True) def amplify(self, data): return dict(data=data.upper()) ``` When a user hits **amplify()**'s URL, the validator checks whether **data** is a number. If it is, it redirects to the error\_handler, **no\_numbers()**. **no\_numbers()** will normally return HTML which is fine if we're simply hitting **amplify()** from a web browser. If we're hitting it from a *BaseClient* app, however, we need it to return *JSON* data instead. To do that we use **jsonify\_validation\_errors()** which checks whether there was a validation error and whether we need to return *JSON* data. If both of those are true, it returns a dictionary with the validation errors. This dictionary is appropriate for returning from the controller method in response to a *JSON* request. **WARNING:** When defining @error\_handler() order of decorators can be important. The short story is to always make @validate() and @error\_handler() the first decorators of your method. The longer version is that this is known to cause errors with the json request not being honored or skipping identity checks when the method is its own error handler. ### Expected Methods Certain controller methods are necessary in order for *BaseClient* to properly talk to your service. *TurboGears* can quickstart an application template for you that sets most of these variables correctly: ``` $ tg-admin quickstart -i -s -p my my # edit my/my/controllers.py ``` ### login() You need to have a **login()** method in your application's root. This method allows *BaseClient* to authenticate against your Service: ``` @expose(template="my.templates.login") + @expose(allow_json=True) def login(self, forward_url=None, previous_url=None, \*args, \**kw): if not identity.current.anonymous \ and identity.was_login_attempted() \ and not identity.get_identity_errors(): + # User is logged in + if 'json' == fedora.tg.utils.request_format(): + return dict(user=identity.current.user) + if not forward_url: + forward_url = turbogears.url('/') raise redirect(forward_url) ``` ### For non-TurboGears Implementors If you are implementing a server in a non-TurboGears framework, note that one of the ways to reach the **login()** method is through special parameters parsed by the *TurboGears* framework. *BaseClient* uses these parameters instead of invoking the **login()** method directly as it saves a round trip when authenticating to the server. It will be necessary for you to implement handling of these parameters (passed via **POST**) on your application as well. The parameters are: **user\_name**, **password**, and **login**. When these three parameters are sent to the server, the server authenticates the user and records their information before deciding what information to return to them from the URL. ### logout() The **logout()** method is similar to **login()**. It also needs to be modified to allow people to connect to it via *JSON*: ``` - @expose() + @expose(allow_json=True) def logout(self): identity.current.logout() + if 'json' in fedora.tg.utils.request_format(): + return dict() raise redirect("/") ``` ### CSRF Protection For an overview of CSRF and how to protect *TurboGears* 1 based services, look at this document. ### Using SABase **fedora.tg.json** contains several functions that help to convert *SQLAlchemy* objects into *JSON*. For the most part, these do their work behind the scenes. The **SABase** object, however, is one that you might need to take an active role in using. When you return an *SQLAlchemy* object in a controller to a template, the template is able to access any of the relations mapped to it. So, instead of having to construct a list of people records from a table and the the list of groups that each of them are in you can pass in the list of people and let your template reference the relation properties to get the groups. This is extremely convenient for templates but has a negative effect when returning *JSON*. Namely, the default methods for marshaling *SQLAlchemy* objects to *JSON* only return the attributes of the object, not the relations that are linked to it. So you can easily run into a situation where someone querying the *JSON* data for a page will not have all the information that a template has access to. SABase fixes this by allowing you to specify relations that your *SQLAlchemy* backed objects should marshal as *JSON* data. Further information on SABase can be found in the API documentation: ``` pydoc fedora.tg.json ``` ### Example SABase is a base class that you can use when defining objects in your project's model. So the first step is defining the classes in your model to inherit from SABase: ``` from fedora.tg.json import SABase from sqlalchemy import Table, Column, Integer, String, MetaData, ForeignKey from turbogears.database import metadata, mapper class Person(SABase): pass PersonTable = Table('person', metadata Column('name', String, primary_key=True), ) class Address(SABase): pass AddressTable = Table ( Column('id', Integer, primary_key=True), Column('street', string), Column('person_id', Integer, ForeignKey('person.name') ) mapper(PersonTable, Person) mapper(AddressTable, Address, properties = { person: relation(Person, backref = 'addresses'), }) ``` The next step is to tell SABase which properties should be copied (this allows you to omit large trees of objects when you only need the data from a few of them): ``` @expose('my.templates.about_me') @expose(allow_json=True) def my_info(self): person = Person.query.filter_by(name='Myself').one() person.jsonProps = {'Person': ['addresses']} return dict(myself=person} ``` Now, when someone requests *JSON* data from my\_info, they should get back a record for person that includes a property addresses. Addresses will be a list of address records associated with the person. ### How it Works SABase adds a special *\_\_json\_\_()* method to the class. By default, this method returns a dict with all of the attributes that are backed by fields in the database. Adding entries to jsonProps adds the values for those properties to the returned dict as well. If you need to override the *\_\_json\_\_()* method in your class you probably want to call SABase's *\_\_json\_\_()* unless you know that neither you nor any future subclasses will need it. ### Using \_\_json\_\_() Sometimes you need to return an object that isn't a basic python type (list, tuple, dict, number. string, etc). When that occurs, *simplejson* won't know how to marshal the data into *JSON* until you write own method to transform the values. If this method is named \_\_json\_\_(), *TurboGears* will automatically perform the conversion when you return the object. Example: ``` class MyObject(object): def _init__(self, number): self.someNumber = number self.cached = None def _calc_data(self): if not self.cached: self.cached = self.someNumber * 2 return self.cached twiceData = property(_calc_data) def __json__(self): return {'someNumber': self.someNumber, 'twiceData': self.twiceData} ``` In this class, you have a variable and a property. If you were to return it from a controller method without defining the \_\_json\_\_() method, *TurboGears* would give you an error that it was unable to adapt the object to *JSON*. The *JSON* method transforms the object into a dict with sensibly named values for the variable and property so that simplejson is able to marshal the data to *JSON*. Note that you will often have to choose between space (more data takes more bandwidth to deliver to the end user) and completeness (you need to return enough data so the client isn't looking for another method that can complete its needs) when returning data. ## AUTHENTICATION TO FAS The *Fedora Account System* has a *JSON* interface that we make use of to authenticate users in our web apps. Currently, there are two modes of operation. Some web apps have *single sign-on* capability with *Fedora Account System*. These are the *TurboGears* applications that use the **jsonfasprovider**. Other apps do not have *single sign-on* but they do connect to *Fedora Account System* to verify the username and password so changing the password in *Fedora Account System* changes it everywhere. ### TurboGears Identity Provider 2 An identity provider with *CSRF* protection. This will install as a TurboGears identity plugin. To use it, set the following in your **APPNAME/config/app.cfg** file: ``` identity.provider='jsonfas2' visit.manager='jsonfas2' ``` **SEE ALSO:** CSRF-Protection ### Turbogears Identity Provider 1 These methods are **deprecated** because they do not provide the *CSRF* protection of *TurboGears Identity Provider 2*. Please use that identity provider instead. ### Django Authentication Backend ### Flask Auth Plugin ### Flask FAS OpenId Auth Plugin The flask\_openid provider is an alternative to the flask\_fas auth plugin. It leverages our FAS-OpenID server to do authn and authz (group memberships). Note that not every feature is available with a generic OpenID provider -- the plugin depends on the OpenID provider having certain extensions in order to provide more than basic OpenID auth. - • Any compliant OpenID server should allow you to use the basic authn features of OpenID OpenID authentication core: *http://openid.net/specs/openid-authentication-2\_0.html* - • Retrieving simple information about the user such as username, human name, email is done with sreg: *http://openid.net/specs/openid-simple-registration-extension-1\_0.html* which is an extension supported by many providers. - • Advanced security features such as requiring a user to re-login to the OpenID provider or specifying that the user login with a hardware token requires the PAPE extension: *http://openid.net/specs/openid-provider-authentication-policy-extension-1\_0.html* - • To get groups information, the provider must implement the *https://dev.launchpad.net/OpenIDTeams* extension. - • We have extended the teams extension so you can request a team name of **\_FAS\_ALL\_GROUPS\_** to retrieve all the groups that a user belongs to. Without this addition to the teams extension you will need to manually configure which groups you are interested in knowing about. See the documentation for how to do so. - • Retrieving information about whether a user has signed a CLA (For Fedora, this is the Fedora Project Contributor Agreement). *http://fedoraproject.org/specs/open\_id/cla* If the provider you use does not support one of these extensions, the plugin should still work but naturally, it will return empty values for the information that the extension would have provided. ### FAS Flask OpenID Auth Plugin - **Authors** Patrick Uiterwjk - **Date** 18 February 2013 - **For Version** 0\.3.x The *Fedora Account System* has a *OpenID* provider that applications can use to authenticate users in web apps. For our *Flask* applications we have an identity provider that uses this OpenID service to authenticate users. It is almost completely compatible with *Flask Auth Plugin* except that it does not use the username/password provided by the client application (it is silently ignored). It can be configured to use any OpenID authentication service that implements the OpenID Teams Extension, Simple Registration Extension and CLA Extension. ### Configuration The FAS OpenID auth plugin has several config values that can be used to control how the auth plugin functions. You can set these in your application's config file. - **FAS\_OPENID\_ENDPOINT** Set this to the OpenID endpoint url you are authenticating against. Default is "*http://id.fedoraproject.org/*" - **FAS\_CHECK\_CERT** When set, this will check the SSL Certificate for the FAS server to make sure that it is who it claims to be. This is useful to set to False when testing against a local FAS server but should always be set to True in production. Default: True ### Sample Application The following is a sample, minimal flask application that uses fas\_flask for authentication: ``` #!/usr/bin/python -tt # Flask-FAS-OpenID - A Flask extension for authorizing users with OpenID # Primary maintainer: Patrick Uiterwijk # # Copyright (c) 2012-2013, Red Hat, Inc., Patrick Uiterwijk # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions are met: # # * Redistributions of source code must retain the above copyright notice, this # list of conditions and the following disclaimer. # * Redistributions in binary form must reproduce the above copyright notice, # this list of conditions and the following disclaimer in the documentation # and/or other materials provided with the distribution. # * Neither the name of the Red Hat, Inc. nor the names of its contributors may # be used to endorse or promote products derived from this software without # specific prior written permission. # # THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ''AS IS'' AND ANY # EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED # WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE # DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE FOR ANY # DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES # (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; # LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON # ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS # SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. # This is a sample application. import flask from flask_fas_openid import fas_login_required, cla_plus_one_required, FAS # Set up Flask application app = flask.Flask(__name__) # Set up FAS extension fas = FAS(app) # Application configuration # SECRET_KEY is necessary for the Flask session system. It needs to be secret to # make the sessions secret but if you have multiple servers behind # a load balancer, the key needs to be the same on each. app.config['SECRET_KEY'] = 'change me!' # Other configuration options for Flask-FAS-OpenID: # FAS_OPENID_ENDPOINT: the OpenID endpoint URL # (default http://id.fedoraproject.org/) # FAS_CHECK_CERT: check the SSL certificate of FAS (default True) # You should use these options' defaults for production applications! app.config['FAS_OPENID_ENDPOINT'] = 'http://id.fedoraproject.org/' app.config['FAS_CHECK_CERT'] = True # Inline templates keep this test application all in one file. Don't do this in # a real application. Please. TEMPLATE_START = """

Flask-FAS-OpenID test app

{% if g.fas_user %}

Hello, {{ g.fas_user.username }} — Log out {% else %}

You are not logged in — Log in {% endif %} — Main page

""" @app.route('/') def index(): data = TEMPLATE_START data += '

Check if you are cla+1

' % \ flask.url_for('claplusone') data += '

See a secret message (requires login)

' % \ flask.url_for('secret') return flask.render_template_string(data) @app.route('/login', methods=['GET', 'POST']) def auth_login(): # Your application should probably do some checking to make sure the URL # given in the next request argument is sane. (For example, having next set # to the login page will cause a redirect loop.) Some more information: # http://flask.pocoo.org/snippets/62/ if 'next' in flask.request.args: next_url = flask.request.args['next'] else: next_url = flask.url_for('index') # If user is already logged in, return them to where they were last if flask.g.fas_user: return flask.redirect(next_url) return fas.login(return_url=next_url) @app.route('/logout') def logout(): if flask.g.fas_user: fas.logout() return flask.redirect(flask.url_for('index')) # This demonstrates the use of the fas_login_required decorator. The # secret message can only be viewed by those who are logged in. @app.route('/secret') @fas_login_required def secret(): data = TEMPLATE_START + '

Be sure to drink your Ovaltine

' return flask.render_template_string(data) # This demonstrates checking for group membership inside of a function. # The flask_fas adapter also provides a cla_plus_one_required decorator that # can restrict a url so that you can only access it from an account that has # cla +1. @app.route('/claplusone') @cla_plus_one_required def claplusone(): data = TEMPLATE_START data += '

Your account is cla+1.

' return flask.render_template_string(data) if __name__ == '__main__': app.run(debug=True) ``` ### FAS Who Plugin for TurboGears2 ### FASWho Plugin - **Authors** Luke Macken Toshio Kuratomi - **Date** 3 September 2011 This plugin provides authentication to the Fedora Account System using the *repoze.who* WSGI middleware. It is designed for use with *TurboGears2* but it may be used with any *repoze.who* using application. Like *TurboGears Identity Provider 2*, faswho has builtin *CSRF* protection. This protection is implemented as a second piece of middleware and may be used with other *repoze.who* authentication schemes. ### Authenticating against FAS with TurboGears2 Setting up authentication against FAS in *TurboGears2* is very easy. It requires one change to be made to **app/config/app\_cfg.py**. This change will take care of registering faswho as the authentication provider, enabling *CSRF* protection, switching **tg.url()** to use **fedora.ta2g.utils.url()** instead, and allowing the *\_csrf\_token* parameter to be given to any URL. ### Using CSRF middleware with other Auth Methods This section needs to be made clearer so that apps like mirrormanager can be ported to use this. ### Templates The **fedora.tg2.utils** module contains some templates to help you write *CSRF* aware login forms and buttons. You can use the **fedora\_template()** function to integrate them into your templates: The templates themselves come in two flavors. One set for use with mako and one set for use with genshi. ### Mako ### Genshi ## JAVASCRIPT - **Authors** Toshio Kuratomi - **Date** 26 February 2009 - **For Version** 0\.3.x python-fedora currently provides some JavaScript files to make coding *Fedora Services* easier. In the future we may move these to their own package. ### *fedora.dojo* *Module author: Toshio Kuratomi \* *Added in version 0.3.10.* *Dojo is one of several JavaScript Toolkits. It aims to be a standard* *library for JavaScript. The fedora.dojo module has JavaScript code* *that make use of Dojo to do their work. It is most appropriate to use* *when the Dojo libraries are being used as the JavaScript library for* *the app. However, it is well namespaced and nothing should prevent it from* *being used in other apps as well.* ## API DOCUMENTATION *This API Documentation is currently a catch-all. We're going to merge the API* *docs into the hand created docs as we have time to integrate them.* ### Client *fedora.client is used to interact with Fedora Services.* *Changed in version 0.3.21: Deprecate DictContainer in favor of bunch.Bunch* *Changed in version 0.3.35: Add the openid clients* *Module author: Ricky Zhou \* *Module author: Luke Macken \* *Module author: Toshio Kuratomi \* - **exception fedora.client.AppError(name, message, extras=None)** *Error condition that the server is passing back to the client.* - **exception fedora.client.AuthError** *Error during authentication. For instance, invalid password.* - **exception fedora.client.CLAError** *CLA Error* - **class fedora.client.DictContainer(\*args, \*\*kwargs)** - **exception fedora.client.FASError** *FAS Error* - **exception fedora.client.FedoraClientError** *Base Exception for problems which originate within the Clients.* *This should be the base class for any exceptions that the Client* *generates. For instance, if the client performs validation before passing* *the data on to the Fedora Service.* *Problems returned while talking to the Services should be returned via a* *FedoraServiceError instead.* - **exception fedora.client.FedoraServiceError** *Base Exception for any problem talking with the Service.* *When the Client gets an error talking to the server, an exception of this* *type is raised. This can be anything in the networking layer up to an* *error returned from the server itself.* - **exception fedora.client.ServerError(url, status, msg)** *Unable to talk to the server properly.* *This includes network errors and 500 response codes. If the error was* *generated from an http response,* **code** *is the HTTP response code.* *Otherwise,* **code** *will be -1.* ### Generic Clients ### BaseClient - **class fedora.client.BaseClient(base\_url, useragent=None, debug=False, insecure=False, username=None, password=None, httpauth=None, session\_cookie=None, session\_id=None, session\_name='tg-visit', cache\_session=True, retries=None, timeout=None)** *A client for interacting with web services.* - **logout()** *Logout from the server.* - **send\_request(method, req\_params=None, file\_params=None, auth=False, retries=None, timeout=None, \*\*kwargs)** *Make an HTTP request to a server method.* *The given method is called with any parameters set in req\_params. If* *auth is True, then the request is made with an authenticated session* *cookie.* - **Parameters** - • **method** *-- Method to call on the server. It's a url fragment that* *comes after the base\_url set in \_\_init\_\_().* - • **req\_params** *-- Extra parameters to send to the server.* - • **file\_params** *-- dict of files where the key is the name of the* *file field used in the remote method and the value is the local* *path of the file to be uploaded. If you want to pass multiple* *files to a single file field, pass the paths as a list of paths.* - • **auth** *-- If True perform auth to the server, else do not.* - • **retries** *-- if we get an unknown or possibly transient error from* *the server, retry this many times. Setting this to a negative* *number makes it try forever. Default to use the* **retries** *value set on the instance or in* **\_\_init\_\_()** *(which defaults* *to zero, no retries).* - • **timeout** *-- A float describing the timeout of the connection. The* *timeout only affects the connection process itself, not the* *downloading of the response body. Default to use the* **timeout** *value set on the instance or in* **\_\_init\_\_()** *(which defaults to 120s).* - **Return type** *Bunch* - **Returns** *The data from the server* *Changed in version 0.3.21: \* Return data as a Bunch instead of a DictContainer* *\* Add file\_params to allow uploading files* *Changed in version 0.3.33: \* Added the timeout kwarg* - **property session\_cookie** *Deprecated, use session\_id instead.* *The session cookie is saved in a file in case it is needed in* *consecutive runs of BaseClient.* - **property session\_id** *The session\_id.* *The session id is saved in a file in case it is needed in consecutive* *runs of BaseClient.* ### ProxyClient - **class fedora.client.ProxyClient(base\_url, useragent=None, session\_name='tg-visit', session\_as\_cookie=True, debug=False, insecure=False, retries=None, timeout=None)** *A client to a Fedora Service. This class is optimized to proxy multiple* *users to a service. ProxyClient is designed to be threadsafe so that* *code can instantiate one instance of the class and use it for multiple* *requests for different users from different threads.* *If you want something that can manage one user's connection to a Fedora* *Service, then look into using BaseClient instead.* *This class has several attributes. These may be changed after* *instantiation however, please note that this class is intended to be* *threadsafe. Changing these values when another thread may affect more* *than just the thread that you are making the change in. (For instance,* *changing the debug option could cause other threads to start logging debug* *messages in the middle of a method.)* - **base\_url** *Initial portion of the url to contact the server. It is highly* *recommended not to change this value unless you know that no other* *threads are accessing this ProxyClient instance.* - **useragent** *Changes the useragent string that is reported to the web server.* - **session\_name** *Name of the cookie that holds the authentication value.* - **session\_as\_cookie** *If* **True***, then the session information is saved locally as* *a cookie. This is here for backwards compatibility. New code should* *set this to* **False** *when constructing the ProxyClient.* - **debug** *If* **True***, then more verbose logging is performed to aid in* *debugging issues.* - **insecure** *If* **True** *then the connection to the server is not checked to be* *sure that any SSL certificate information is valid. That means that* *a remote host can lie about who it is. Useful for development but* *should not be used in production code.* - **retries** *Setting this to a positive integer will retry failed requests to the* *web server this many times. Setting to a negative integer will retry* *forever.* - **timeout** *A float describing the timeout of the connection. The timeout only* *affects the connection process itself, not the downloading of the* *response body. Defaults to 120 seconds.* *Changed in version 0.3.33: Added the timeout attribute* - **property debug** *When True, we log extra debugging statements. When False, we only log* *errors.* - **log = \** - **send\_request(method, req\_params=None, auth\_params=None, file\_params=None, retries=None, timeout=None)** *Make an HTTP request to a server method.* *The given method is called with any parameters set in* **req\_params***.* *If auth is True, then the request is made with an authenticated session* *cookie. Note that path parameters should be set by adding onto the* *method, not via* **req\_params***.* - **Parameters** - • **method** *-- Method to call on the server. It's a url fragment that* *comes after the base\_url set in \_\_init\_\_(). Note that any* *parameters set as extra path information should be listed here,* *not in* **req\_params***.* - • **req\_params** *-- dict containing extra parameters to send to the* *server* - • **auth\_params** *--* *dict containing one or more means of authenticating* *to the server. Valid entries in this dict are:* - **cookie** **Deprecated** *Use* **session\_id** *instead. If both* **cookie** *and* **session\_id** *are set, only* **session\_id** *will* *be used. A* **Cookie.SimpleCookie** *to send as a session cookie* *to the server* - **session\_id** *Session id to put in a cookie to construct an identity* *for the server* - **username** *Username to send to the server* - **password** *Password to use with username to send to the server* - **httpauth** *If set to* **basic** *then use HTTP Basic Authentication* *to send the username and password to the server. This may be* *extended in the future to support other httpauth types than* **basic***.* *Note that cookie can be sent alone but if one of username or* *password is set the other must as well. Code can set all of these* *if it wants and all of them will be sent to the server. Be careful* *of sending cookies that do not match with the username in this* *case as the server can decide what to do in this case.* - • **file\_params** *-- dict of files where the key is the name of the* *file field used in the remote method and the value is the local* *path of the file to be uploaded. If you want to pass multiple* *files to a single file field, pass the paths as a list of paths.* - • **retries** *-- if we get an unknown or possibly transient error from* *the server, retry this many times. Setting this to a negative* *number makes it try forever. Default to use the retries* *value set on the instance or in* **\_\_init\_\_()***.* - • **timeout** *-- A float describing the timeout of the connection. The* *timeout only affects the connection process itself, not the* *downloading of the response body. Defaults to the timeout* *value set on the instance or in* **\_\_init\_\_()***.* - **Returns** *If ProxyClient is created with session\_as\_cookie=True (the* *default), a tuple of session cookie and data from the server.* *If ProxyClient was created with session\_as\_cookie=False, a tuple* *of session\_id and data instead.* - **Return type** *tuple of session information and data from server* *Changed in version 0.3.17: No longer send tg\_format=json parameter. We rely solely on the* *Accept: application/json header now.* *Changed in version 0.3.21: \* Return data as a Bunch instead of a DictContainer* *\* Add file\_params to allow uploading files* *Changed in version 0.3.33: Added the timeout kwarg* ### OpenIdBaseClient - **class fedora.client.OpenIdBaseClient(base\_url, login\_url=None, useragent=None, debug=False, insecure=False, openid\_insecure=False, username=None, cache\_session=True, retries=None, timeout=None, retry\_backoff\_factor=0)** *A client for interacting with web services relying on openid auth.* - **has\_cookies()** - **login(username, password, otp=None)** *Open a session for the user.* *Log in the user with the specified username and password* *against the FAS OpenID server.* - **Parameters** - • **username** *-- the FAS username of the user that wants to log in* - • **password** *-- the FAS password of the user that wants to log in* - • **otp** *-- currently unused. Eventually a way to send an otp to the* *API that the API can use.* - **send\_request(method, auth=False, verb='POST', \*\*kwargs)** *Make an HTTP request to a server method.* *The given method is called with any parameters set in req\_params. If* *auth is True, then the request is made with an authenticated session* *cookie.* - **Parameters** - • **method** *-- Method to call on the server. It's a url fragment that* *comes after the* **base\_url** *set in* **\_\_init\_\_()***.* - • **auth** *-- If True perform auth to the server, else do not.* - • **req\_params** *-- Extra parameters to send to the server.* - • **file\_params** *-- dict of files where the key is the name of the* *file field used in the remote method and the value is the local* *path of the file to be uploaded. If you want to pass multiple* *files to a single file field, pass the paths as a list of paths.* - • **verb** *-- HTTP verb to use. GET and POST are currently supported.* *POST is the default.* - **property session\_key** - **fedora.client.openidbaseclient.requires\_login(func)** *Decorator function for get or post requests requiring login.* *Decorate a controller method that requires the user to be authenticated.* *Example:* ``` from fedora.client.openidbaseclient import requires_login @requires_login def rename_user(new_name): user = new_name # [...] ``` ### OpenIdProxyClient - **class fedora.client.OpenIdProxyClient(base\_url, login\_url=None, useragent=None, session\_name='session', debug=False, insecure=False, openid\_insecure=False, retries=None, timeout=None)** *A client to a Fedora Service. This class is optimized to proxy multiple* *users to a service. OpenIdProxyClient is designed to be usable by code* *that creates a single instance of this class and uses it in multiple* *threads. However it is not completely threadsafe. See the information* *on setting attributes below.* *If you want something that can manage one user's connection to a Fedora* *Service, then look into using OpenIdBaseClient* *instead.* *This class has several attributes. These may be changed after* *instantiation. Please note, however, that changing these values when* *another thread is utilizing the same instance may affect more than just* *the thread that you are making the change in. (For instance, changing* *the debug option could cause other threads to start logging debug* *messages in the middle of a method.)* - **base\_url** *Initial portion of the url to contact the server. It is highly* *recommended not to change this value unless you know that no other* *threads are accessing this OpenIdProxyClient instance.* - **useragent** *Changes the useragent string that is reported to the web server.* - **session\_name** *Name of the cookie that holds the authentication value.* - **debug** *If* **True***, then more verbose logging is performed to aid in* *debugging issues.* - **insecure** *If* **True** *then the connection to the server is not checked to be* *sure that any SSL certificate information is valid. That means that* *a remote host can lie about who it is. Useful for development but* *should not be used in production code.* - **retries** *Setting this to a positive integer will retry failed requests to the* *web server this many times. Setting to a negative integer will retry* *forever.* - **timeout** *A float describing the timeout of the connection. The timeout only* *affects the connection process itself, not the downloading of the* *response body. Defaults to 120 seconds.* - **property debug** *When True, we log extra debugging statements. When False, we only log* *errors.* - **login(username, password, otp=None)** *Open a session for the user.* *Log in the user with the specified username and password* *against the FAS OpenID server.* - **Parameters** - • **username** *-- the FAS username of the user that wants to log in* - • **password** *-- the FAS password of the user that wants to log in* - • **otp** *-- currently unused. Eventually a way to send an otp to the* *API that the API can use.* - **Returns** *a tuple containing both the response from the OpenID* *provider and the session used to by this provider.* - **send\_request(method, verb='POST', req\_params=None, auth\_params=None, file\_params=None, retries=None, timeout=None, headers=None)** *Make an HTTP request to a server method.* *The given method is called with any parameters set in* **req\_params***.* *If auth is True, then the request is made with an authenticated* *session cookie. Note that path parameters should be set by adding* *onto the method, not via* **req\_params***.* - **Parameters** - • **method** *-- Method to call on the server. It's a url fragment that* *comes after the base\_url set in \_\_init\_\_(). Note that any* *parameters set as extra path information should be listed here,* *not in* **req\_params***.* - • **req\_params** *-- dict containing extra parameters to send to the* *server* - • **auth\_params** *--* *dict containing one or more means of* *authenticating to the server. Valid entries in this dict are:* - **cookie** **Deprecated** *Use* **session\_id** *instead. If both* **cookie** *and* **session\_id** *are set, only* **session\_id** *will be used. A* **Cookie.SimpleCookie** *to send as a* *session cookie to the server* - **session\_id** *Session id to put in a cookie to construct an* *identity for the server* - **username** *Username to send to the server* - **password** *Password to use with username to send to the server* - **httpauth** *If set to* **basic** *then use HTTP Basic Authentication* *to send the username and password to the server. This may* *be extended in the future to support other httpauth types* *than* **basic***.* *Note that cookie can be sent alone but if one of username or* *password is set the other must as well. Code can set all of* *these if it wants and all of them will be sent to the server.* *Be careful of sending cookies that do not match with the* *username in this case as the server can decide what to do in* *this case.* - • **file\_params** *-- dict of files where the key is the name of the* *file field used in the remote method and the value is the local* *path of the file to be uploaded. If you want to pass multiple* *files to a single file field, pass the paths as a list of paths.* - • **retries** *-- if we get an unknown or possibly transient error* *from the server, retry this many times. Setting this to a* *negative number makes it try forever. Default to use the* *retries value set on the instance or in* **\_\_init\_\_()***.* - • **timeout** *-- A float describing the timeout of the connection.* *The timeout only affects the connection process itself, not the* *downloading of the response body. Defaults to the timeout* *value set on the instance or in* **\_\_init\_\_()***.* - • **headers** *-- A dictionary containing specific headers to add to* *the request made.* - **Returns** *A tuple of session\_id and data.* - **Return type** *tuple of session information and data from server* ### Clients for Specific Services ### Wiki - **class fedora.client.Wiki(base\_url='https://fedoraproject.org/w/', \*args, \*\*kwargs)** - **api\_high\_limits = False** - **check\_api\_limits()** *Checks whether you have the 'apihighlimits' right or not.* - **fetch\_all\_revisions(start=1, flags=True, timestamp=True, user=True, size=False, comment=True, content=False, title=True, ignore\_imported\_revs=True, ignore\_wikibot=False, callback=None)** *Fetch data for all revisions. This could take a long time. You can* *start at a specific revision by modifying the 'start' keyword argument.* *To ignore revisions made by "ImportUser" and "Admin" set* *ignore\_imported\_revs to True (this is the default). To ignore edits* *made by Wikibot set ignore\_wikibot to True (False is the default).* *Modifying the remainder of the keyword arguments will return less/more* *data.* - **get\_recent\_changes(now, then, limit=500)** *Get recent wiki changes from now until then* - **login(username, password)** - **print\_recent\_changes(days=7, show=10)** ### Service ### Transforming SQLAlchemy Objects into JSON ## GLOSSARY - **controller** *In MVC design, the controller is in charge of things. It takes* *processes events and decides what data to ask the model for,* *manipulates the data according to the information in the event, and* *decides which view to send the results to to be rendered.* - **CSRF** *Cross-site request forgery is a* *technique where a malicious website can gain access to another web* *site by hijaacking a currently open session that the user has open to* *the site. This technique can also affect identification via SSL* *Certificates or anything else that the browser sends to the server* *automatically when a request is made.* **SEE ALSO:** *CSRF-Protection* - **Dojo** *Dojo is a JavaScript toolkit that aims to be a standard library for* *JavaScript. It provides a small core library with useful functions* *and an expanded set of scripts that can be added that provide widgets* *and other features.* **SEE ALSO:** *http://www.dojotoolkit.org* - **double submit** *A strategy to foil CSRF attacks. This strategy involves* *sending the value of the authentication cookie (or something derivable* *only from knowing the value of the authentication cookie) in the body* *of the request. Since the Same Origin Policy prevents a web* *site other than the one originating the cookie from reading what's in* *the cookie, the server can be reasonably assured that the request does* *not originate from an unknown request on another website. Note that* *this and other anti-CSRF measures do not protect against spoofing or* *getting a user to actively click on a link on an attacked website by* *mistake.* - **flask** *A simple Python web framework that we're using in parts of Fedora* *Infrastructure. It provides good documentation and simplicity in its* *design.* **SEE ALSO:** *http://flask.pocoo.org/docs/* - **JSON** *JavaScript Object Notation is a format for* *marshalling data. It is based on a subset of JavaScript that is used* *to declare objects. Compared to xml, JSON is a lightweight, easily* *parsed format.* **SEE ALSO:** *Wikipedia's JSON Entry* - **model** *In MVC design, the layer that deals directly with the data.* - **OpenID** *A specification for single sign on to web services where the* *authentication server and the application seeking to have the user* *authenticated do not need to have complete trust in each other.* **SEE ALSO:** *http://openid.net/get-an-openid/what-is-openid/* - **Same Origin Policy** *A web browser security policy that prevents one website from reading:* *1) the cookies from another website* *2) the response body from another website* **SEE ALSO:** *http://en.wikipedia.org/wiki/Same\_origin\_policy* - **single sign-on** *A feature that allows one login to authenticate a user for multiple* *applications. So logging into one application will authenticate you* *for all the applications that support the same single-sign-on* *infrastructure.* - **TurboGears** *A Python web framework that most of Fedora Infrastructure's apps are* *built on.* **SEE ALSO:** *http://www.turbogears.org/* - **TurboGears2** *The successor to TurboGears, TurboGears2 provides a very* *similar framework to coders but has some notable differences. It is* *based on pylons and paste so it is much more tightly integrated with* *WSGI. The differences with :ref\`TurboGears\`1 are largely with* *the organization of code and how to configure the application.* **SEE ALSO:** *http://www.turbogears.org/* - **view** *In MVC design, the layer that takes care of formatting and rendering* *data for the consumer. This could be displaying the data as an html* *page or marshalling it into JSON objects.* - **WSGI** *WSGI is an interface between web servers and web frameworks that* *originated in the Python community. WSGI lets different components* *embed each other even if they were originally written for different* *python web frameworks.* **SEE ALSO:** *http://en.wikipedia.org/wiki/Web\_Server\_Gateway\_Interface* - • *Glossary* - • *Index* - • *Module Index* - • *Search Page* ## AUTHOR *Author name not set* ## COPYRIGHT *2007-2020 Red Hat, Inc.*