IT4Change/gradido
3
0
Fork 0
mirror of https://github.com/IT4Change/gradido.git synced 2025-12-13 07:45:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #2252 from gradido/1574-Concept_to_introduce_Gradido-ID

Browse Source 
1574 concept to introduce gradidoID and change password encryption
This commit is contained in:
clauspeterhuebner 2022-11-02 22:52:24 +01:00 committed by GitHub
commit 1bdb41f3a2
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 127 additions and 68 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

195
docu/Concepts/TechnicalRequirements/UC_Introduction_of_Gradido-ID.md
View File

@ -10,30 +10,27 @@ Additionally the Gradido-ID allows to administrade any user account data like ch
The formalized definition of the Gradido-ID can be found in the document [BenutzerVerwaltung#Gradido-ID](../BusinessRequirements/BenutzerVerwaltung#Gradido-ID).
## Steps of Introduction
## 1st Stage
To Introduce the Gradido-ID there are several steps necessary. The first step is to define a proper database schema with additional columns and tables followed by data migration steps to add or initialize the new columns and tables by keeping valid data at all.
The 1st stage of introducing the Gradido-ID contains several steps. The first step is to define a proper database schema with additional columns and tables followed by data migration steps to add or initialize the new columns and tables by keeping valid data at all.
The second step is to decribe all concerning business logic processes, which have to be adapted by introducing the Gradido-ID.
The second step is to decribe all concerning business logic processes, which have to be adapted by introducing the Gradido-ID and handling the attributes of the new user_contacts table.
### Database-Schema
#### Users-Table
The entity users has to be changed by adding the following columns.
The entity users has to be changed by adding the following columns. The column State gives a hint about the working state including the ticket number.
| Column | Type | Description |
| ------------------------ | ------ | ----------------------------------------------------------------------------------------------------------------- |
| gradidoID | String | technical unique key of the user as UUID (version 4) |
| alias | String | a business unique key of the user |
| passphraseEncryptionType | int | defines the type of encrypting the passphrase: 1 = email (default), 2 = gradidoID, ... |
| emailID | int | technical foreign key to the entry with type Email and contactChannel=maincontact of the new entity UserContacts |
| State | Column | Type | Description |
| -------------- | --------- | ------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| *done #2125* | gradidoID | String | technical unique key of the user as UUID (version 4) |
| *done #2125* | alias | String | a business unique key of the user |
| *done #2165* | emailID | int | technical foreign key to the UserContacts-Table with the entry of type Email, which will be interpreted as the maincontact from the Users table point of view |
##### Email vs emailID
The existing column `email`, will now be changed to the primary email contact, which will be stored as a contact entry in the new `UserContacts` table. It is necessary to decide if the content of the `email `will be changed to the foreign key `emailID `to the contact entry with the email address or if the email itself will be kept as a denormalized and duplicate value in the `users `table.
The preferred and proper solution will be to add a new column `Users.emailId `as foreign key to the `UsersContact `entry and delete the `Users.email` column after the migration of the email address in the `UsersContact `table.
The existing column `email`, will now be changed to the primary email contact, which will be stored as a contact entry in the new `UserContacts` table.
#### new UserContacts-Table
@ -55,17 +52,21 @@ A new entity `UserContacts `is introduced to store several contacts of different
| phone | String | defines the address of a contact entry of type Phone |
| contactChannels | String | define the contact channel as comma separated list for which this entry is confirmed by the user e.g. main contact (default), infomail, contracting, advertisings, ... |
##### ToDo:
The UserContacts, expecially the email contacts, will for future be categorized to communication channels for example to allow the user to define which information he will get on which email-contact (aspects of administration, contract, advertising, etc.)
### Database-Migration
After the adaption of the database schema and to keep valid consistent data, there must be several steps of data migration to initialize the new and changed columns and tables.
#### Initialize GradidoID
#### Initialize GradidoID (done #2125)
In a one-time migration create for each entry of the `Users `tabel an unique UUID (version4).
#### Primary Email Contact
#### Primary Email Contact (done #1798)
In a one-time migration read for each entry of the `Users `table the `Users.id` and `Users.email`, select from the table `login_email_opt_in` the entry with the `login_email_opt_in.user_id` = `Users.id` and create a new entry in the `UsersContact `table, by initializing the contact-values with:
In a one-time migration read for each entry of the `Users `table the `Users.id` and `Users.email` and create for it a new entry in the `UserContacts `table, by initializing the contact-values with:
* id = new technical key
* type = Enum-Email
@ -88,72 +89,130 @@ After this one-time migration and a verification, which ensures that all data ar
The following logic or business processes has to be adapted for introducing the Gradido-ID
#### Read-Write Access of Users-Table especially Email
#### Capturing of alias
To avoid using the email as primary identifier it is necessary to introduce a capturing of the alias. It is not a good solution to create for existing users an individual alias by a migration. So each user should capture his own alias during registration- and/or login-process.
These requirements are described in the concept document [../BusinessRequirements/UC_Set_UserAlias.md]() **(done #2144)** and the implementation of these requirements will be the prerequisite for changing the login-process from single email-identifier to the future identifiers alias / gradidoID / email.
#### Read-Write Access of Users-Table especially Email (done #1798)
The ORM mapping has to be adapted to the changed and new database schema.
#### Registration Process
#### Create and Update User Processes
The logic of the registration process has to be adapted by
The logic of the create and update user process has to be adapted by
* initializing the `Users.userID` with a unique UUID
* creating a new `UsersContact `entry with the given email address and *maincontact* as `usedChannel `
* set `emailID `in the `Users `table as foreign key to the new `UsersContact `entry
* set `Users.passphraseEncrpytionType = 2` and encrypt the passphrase with the `Users.userID` instead of the `UsersContact.email`
* creating a new User including with a unique UUID-V4 **(done #2125)**
* creating a new `UserContacts `entry with the given email address **(#2165)**
* set `emailID `in the `Users `table as foreign key to the new `UserContacts `entry **(#2165)**
* handling the new emailXXX attributes in the `user_contacts `table previously in the `email_opt_in `table **(#2165)**
#### Login Process
#### Search User Processes (#2165)
The logic of the login process has to be adapted by
The logic of all processes where the user is searched has to be adapted by
* search the users data by reading the `Users `and the `UsersContact` table with the email (or alias as soon as the user can maintain his profil with an alias) as input
* depending on the `Users.passphraseEncryptionType` decrypt the stored password
* = 1 : with the email
* = 2 : with the userID
* always search a *user* with its relation "emailContact" to load the associated userContact with his email
* a search user by *email* has to be implemented by searching a `userContact `for the given *email* and its relation "user" to load the associated user to this email
#### Password Processes (#2165)
The logic of all password processes has to be adapted by
* read the *emailXXX* attributes out of the `user_contacts `table instead of previoulsy from the `email_opt_in `table
* writing or updating the *emailXXX* attributes now in the `user_contact `table instead of previously in the `email_opt_in `table
* the logic how to de/encrypt the password will not part of this 1st stage of introduction of the gradidoID. This will be part of the 2nd stage
## 2nd Stage
In the 2nd stage of this topic the password handling during registration and login process will be changed. These change must keep the current active password handling where the email is part of the encryption as long as all users are shifted to the new logic of password handling where the gradidoID will part of the encryption. This means there must be a kind of versioning which type of password encryption is used. Because some users will not login for a long time, which causes to use the old password encryption at their login process or in the future there could be the requirement to change the password handling to newer and safer algorithms.
### Database-Schema
#### Users-Table
The entity *users* has to be changed by
| Action | Column | Type | Description |
| :----: | ---------------------- | ---------- | ----------------------------------------------------------------------------------- |
| add | passwordEncryptionType | int | defines the type of encrypting the password: default 1 = email, 2 = gradidoID, ... |
| delete | public_key | binary(32) | before deletion verify and ensure that realy not in use even for encryption type 1 |
| delete | privkey | binary(80) | before deletion verify and ensure that realy not in use even for encryption type 1 |
| delete | email_hash | binary(32) | before deletion verify and ensure that realy not in use even for encryption type 1 |
| delete | passphrase | text | before deletion verify and ensure that realy not in use even for encryption type 1 |
### Adaption of BusinessLogic
#### Password En/Decryption
The logic of the password en/decryption has to be adapted by encapsulate the logic to be controlled with an input parameter. The input parameter can be the email or the userID.
The logic of the existing password en/decryption has to be shifted out of the ***UserResolver.js*** file in separated file(s). This separated file will be placed in the package-directory `backend/src/password` and named ***emailEncryptor.js***. As the name express the password encryption uses the `email `attribute.
For the new password encryption logic a new file named ***gradidoIDEncryptor.js*** has to be created in the package-directory `backend/src/password`, which uses the *gradidoID* instead of the *email* for the password encryption. As soon as a user is changed to this encryption type with the *gradidoID*, it will be possible for him to change his *email* in his gradido-profile without any effect on his password encryption.
For possible future requirements of newer and safer encryption logic additional files can be placed in the same directory with an expressiv file name for the new encryption type.
All these `xxxEncryptor `files has to implement the following API, but with possibly different parameter types, depending on the encryption requirements:
| API | emailEncryptor | gradidoIDEncryptor | return | description |
| ------------------------- | ---------------- | ------------------ | ------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------- |
| **encryptPassword** | dbUser, password | dbUser, password | encrypted password | process the encryption with<br />the encryptor specific attributs <br />out of the dbUser and the original <br />password entered by the user |
| **verifyPassword** | dbUser, password | dbUser, password | boolean | process the decryption with<br />the encryptor specific attributs <br />out of the dbUser and the original <br />password entrered by the user |
| **isPassword** | password | password | boolean | verifiy the formal rules of the original<br />password entered by the user |
Which of the *xxxEncryptor* implementations will be used, depends on the value of the attribute `user.passwordEncryptionType`, which has to be interpreted before. To encapsulate this logic from the general business logic the ***Encryptor.js*** will be created with the same API as the specific *encryptor* classes, but it will interpret the attribute `dbUser.passwordEncryptionType` to select and invoke the correct *encryptor* implementation and to decide if an upgrade to a newer *encryptor* class should be done.
The new Enum `PasswordEncryptionType `with the increasing values:
* 1 = emailEncryptor
* 2 = gradidoIDEncryptor
* ... = ?
will be used to define the order, which encryptor implementation is the oldest and the newest. That means if a user is still not using the newest *encryptor* for his password encryption the logic will implicit start a change to the newest *encryptor*. In all business processes, where the user enters his password the invokation of the ***Encryptor.js*** has to be introduced, because without the original entered password from the user no *encryptor* upgrade can be done.
#### Registration Process
The backend logic of the registration process has to be adapted
* the ***UserResolver.createUser*** logic has to be changed by setting for a new user the attribut `Users.passwordEncrpytionType = 2`
* As soon as the user activates the email-confirmation link `https://gradido.net/checkEmail/` the application frontend invokes
* at first the ***UserResolver.queryOptIn*** method, which will not be necessary, because the same checks about the given *emailOptIn*-code will be done a 2nd time in the invocation of *UserResolver.setPassword*
* at second the ***UserResolver.setPassword*** method, which has to be changed
* to use the new ***Encryptor.isPassword*** to validate the formal rules of the given password
* to remove all cryptographic logic like passphrase and key pair generation and password hashing to the new ***emailEncryptor.js***
* to introduce the invocation of the new ***Encryptor.encryptPassword*** in the existing logic flow
#### Login Process
The logic of the login process has to be adapted in frontend and backend
* Frontend
* The login dialog has to be changed at the email input component
* the new label contains "Email / Alias / GradidoID"
* the validation of the input field has to be changed to accept the input of one of these three possible values
* in case of failed validation an expressiv error message for the specific given input has to be shown (for more details about the rules for alias and gradidoID see the concepts [UC_SetUserAlias.md](../BusinessRequirements/UC_SetUserAlias.md) and [BenutzerVerwaltung#Gradido-ID](../BusinessRequirements/BenutzerVerwaltung#Gradido-ID)).
* The signature of the backend invocation ***UserResolver.login*** has to be changed to accept all three variants of identifiers
* depending on the implemented backend solution the frontend has to detect and initialize the correct parameter settings
* Backend
* The signature of the backend invocation ***UserResolver.login*** has to be changed to accept all three variants of identifiers
* solution-A: the first parameter *email* is renamed to *identifier* and the backend has to detect which type of identifier is given
* solution-B: two additional parameters *alias* and *gradidoID* are inserted in the type ***UnsecureLoginArgs*** and the frontend has to decide, which type of identifier is given and initialize the correct parameter
* **TODO**: solution-A is preferred?
* The logic of ***UserResolver.login*** has to be changed by
* in case of solution-A for the signature, the given identifier has to be detected for the correct user searching
* the user to be searched by the given identifier (email / alias / gradidoID)
* if a user could be found all the existing checks will be done as is, except the public and private key check, which will be removed
* for the password check the new ***Encryptor.isPassword*** and ***Encryptor.verifyPassword*** has to be invoked; all existing cryptographic logic has to be deleted
#### Change Password Process
The logic of change password has to be adapted by
There are two ways to change a user password.
* if the `Users.passphraseEncryptionType` = 1, then
The first one is the *Forget-Password process*, which will use the same backend invocation with activating the email link like the *Registration Process* to set the password; for details see description above.
* read the users email address from the `UsersContact `table
* give the email address as input for the password decryption of the existing password
* use the `Users.userID` as input for the password encryption for the new password
* change the `Users.passphraseEnrycptionType` to the new value =2
* if the `Users.passphraseEncryptionType` = 2, then
The second one is the *Update-Userinfo process*, which invokes the ***UserResolver.updateUserInfos***. This method has to be changed in the *password check block* by
* give the `Users.userID` as input for the password decryption of the existing password
* use the `Users.userID` as input for the password encryption fo the new password
#### Search- and Access Logic
A new logic has to be introduced to search the user identity per different input values. That means searching the user data must be possible by
* searching per email (only with maincontact as contactchannel)
* searching per userID
* searching per alias
#### Identity-Mapping
A new mapping logic will be necessary to allow using unmigrated APIs like GDT-servers api. So it must be possible to give this identity-mapping logic the following input to get the respective output:
* email -> userID
* email -> gradidoID
* email -> alias
* userID -> gradidoID
* userID -> email
* userID -> alias
* alias -> gradidoID
* alias -> email
* alias -> userID
* gradidoID -> email
* gradidoID -> userID
* gradidoID -> alias
#### GDT-Access
To use the GDT-servers api the used identifier for GDT has to be switch from email to userID.
* removing all the cryptographic logic and
* invoke the new ***Encryptor.isPassword*** for the given *newPassword* and if valid then
* invoke the new ***Encryptor.verifyPassword*** for the given *oldPassword* and if valid then
* invoke the new ***Encryptor.encryptPassword*** for the given *newPassword*