WkHTMLtoPDF is a tool to convert html reports to PDF. Odoo/OpenERP uses this tool to generate PDF reports. If you are using an unicode character in PDF report, that character has to be installed in font package in the system. If you are using Windows, Bangla characters are available by default. But that is not same for Linux.
Bangla & Assamese fonts are available for Linux, in a package called ‘lohit’.
To install Bangla fonts in CentOS/Redhat based system, use the following:
yum install lohit-bengali-fonts
To install in Debian/Ubuntu based system, use the following:
apt install fonts-lohit-beng-bengali
Once this is installed, PDF reports should start showing the Bangla characters properly.
In Odoo Qweb report, you probably want to change the time to something else, like add few hours, or change the timezone, or so on. To do that, you would need to use ‘t-esc’ template attribute of Odoo. Here is an example of how to add 6 hours to the original timestamp and then convert the time to string to show the value according to the user preference:
You are seeing a time in the Odoo form view, which is using the correct timezone, but when you try to download the report of the same form view data, you see, the timezone is changed to UTC or something else. How to fix this?
The issue appears when the user has configured a different timezone or has not, while the Odoo system uses a central timezone. Make sure to set the timezone for the user to same as the one Odoo uses. You may do so, from
Odoo >> Settings >> Users >> Select User >> Edit >> Preference >> Set Timezone
If this does not solve the problem, then the problem probably appears because Odoobot is set to use a different timezone, and your report is generated using ‘sudo()’ function. To set the timezone for Odoobot
Odoo >> Settings >> Users >> From the Filters, select ‘Inactive Users’ >> Click on Odoobot >> Edit >> Preference >> Set the Timezone
Installing Odoo 15 along with the CentOS 7 and the latest PGSQL repo has changed pretty a lot. I will try to cover solutions to a few errors along with the straightforward steps on installing Odoo 15 in CentOS 7.
First Step First
Update your CentOS 7 installation and install Epel-release
yum update -y
yum install epel-release
Install Python 3.8
We will use Python 3.8 for Odoo 15. We will use Software Collection Repository or SCL to install our Python binary. You may find details of SCL here:
Once done, you can now install Python 3.8 using the following:
yum install rh-python38 -y
Also, install python38-devel as Python.h is used to compile psycopg2 and python-ldap package. From Odoo 15, you need this to get going:
yum install rh-python38-python-devel -y
Note: The above is used to resolve an error like the following
fatal error: Python.h: No such file or directory
Now, we will install a few prerequisites to install Odoo 15. One difference between the old version installed and the new is that you need to load GCC-c+ now along with the GCC compiler. Otherwise, you will see an error like the following:
gcc: error trying to exec ‘cc1plus’: execvp: No such file or directory
So, to install the pre-requisites, run the following:
Once done, now can you initiate the PostgreSQL and start the database server
systemctl start postgresql-13.service
systemctl enable postgresql-13.service
# create the postgres user odoo
su - postgres -c "createuser -s odoo"
Brilliant, now, one more additional thing we need to resolve. With the latest Postgresql 13, you might still not be able to use the libpq. You need to install it manually. Otherwise, you will see an error like the following:
fatal error: libpq-fe.h: No such file or directory
To resolve this error, you need to install these libraries manually with the following command:
yum install libpq5 libpq5-devel -y
Remember to install libpq5-devel as the source of the libpq would be used to compile psycopg2.
Now, let’s move to the next step of installing wkhtmltox. The version for wkhtmltox has remained the same for pretty long. The following shall work till now:
yum localinstall wkhtmltox-0.12.6-1.centos7.x86_64.rpm
This specific tool is used to generate reports in Odoo, without this, you might not be able to use pdf/html reports using Qweb in Odoo.
Final Step: Install and Configure Odoo 15
We will here download the source from Github and install all the dependent packages. First, we switch to the user odoo
su - odoo
Now, clone the git repo for Odoo 15 to a folder odoo15 using the following:
Once done, now, we can enable python3.8 and create a virtual environment for our Odoo installation. First, enable the Python3.8 using scl:
scl enable rh-python38 bash
Now create a virtual environment for our Odoo15 installation:
python3 -m venv odoo15-venv
Activate the virtual environment we just created
Now, we upgrade the pip and install wheel package:
pip install --upgrade pip
pip3 install wheel
Now, before we can install the requirements file using pip3 package installer, here is an error you will face when compiling psycopg2
Error: pg_config executable not found.
Now the problem is understandable, pg_config file is usually placed under the binary folder of pgsql which is:
For some reason, our installer fails to identify this. To solve the issue, we first, load this in the $PATH variable before running pip3 for requirements.
Now, you can run the pip3 installer:
pip3 install -r odoo15/requirements.txt
This shall be complete without any error if you have solved the solutions I had given above. If any of them are missed, you should double-check all the mentioned errors above.
Now exit the venv:
deactivate && exit ; exit
Now, the first step for configuration, edit the /etc/odoo.conf file
Paste the following:
; This is the password that allows database operations:
admin_passwd = set_the_password_to_create_odoo_database
db_host = False
db_port = False
db_user = odoo
db_password = False
addons_path = /opt/odoo/odoo15/addons
; You can enable log file with uncommenting the next line
; logfile = /var/log/odoo15/odoo.log
Replace ‘set_the_password_to_create_odoo_database’ with the one you want to use to allow odoo installer to create the database for you.
Odoo15 Service File
Now, we will create a service file to start/stop/restart our Odoo 15 installation.
We have a product_id column that contains which product we are referring to, a qty_done column, which contains the number of products sold, and sale_id refers to the sale order where the product_id was dispatched. Now the example table contains how many of each product dispatched in different sale orders. We can write a query, to group the product_id and the total number of products sold using aggregation
select product_id, sum(qty_done) as total_sold from sale_line group by product_id;
We can see, how group by can give you the benefit of aggregation. But have you noticed, we have a column called ‘sale_id’ that doesn’t go with any aggregation function like ‘count’, ‘sum’ or ‘avg’. Can we use them for any purpose?
Sometimes, you may be interested to track down the orders responsible for a set of products using group by when each products will contain the list of orders responsible for causing the total sell. Can we accomplish that in one query?
We actually can. We can concat distinct ids and put them in a resulting array instead of our known aggregation function like ‘count’ or ‘sum’. Postgres provides a array aggregation function called ‘array_agg’, it can be used to produce such result. Have a look at the following query:
select product_id, sum(qty_done) as total_sold, array_agg(sale_id) as sale_ids from sale_line group by product_id;
As you can see, we have made postgres return us an array that contains the sale_ids for consisting total_sold even though, sale_id column wasn’t conventionally aggregable or let’s say different than the usual numeric pattern.
This technique can have many use cases. For example if you have a different model/table for managing sale orders, purchase orders, pickings and invoices while, your sale orders contains the price information, and the pickings doesn’t, but picking contains the actual number of products got dispatched, then you may first aggregate the total_sold from the picking table, then produce an average price using the sale_ids you have produced from the picking table using group by and array aggregation, then merge them. This seems quite complicated, I understand, but again, it’s all about putting a break to your reporting time, and manage scalability of your application, putting a constant cost based reporting algorithm using single sql query, even if you process 10 times more orders in future.
stock_picking is one of the core model for Odoo if you using Odoo for ‘Inventory’. Now, if you use ‘Sales’ module, then stock_picking would be extended with a new field for model and column for database table, naming ‘sale_id’. This can be used to detect if the picking is originated from a sale order or not. But if you install ‘Purchase’ module, then stock_picking model is extended with ‘purchase_id’ like the ‘sale_id’ for purchases, but the database isn’t expanded with a column like ‘Sale’ module.
What does this mean?
This means, if you use Odoo ORM, only then, you may use purchase_id of a stock_picking. An example could be like the following. Let’s say, we would like to pick the pickings that originated from purchase orders, aka, GRN, we could use something like this:
This works, only if you are not trying to make a report from a huge lot of pickings, purchase orders and sale orders, when you want to use SQL statement to produce efficient joins and generate the report quickly.
Let me demonstrate what I meant
We know, stock_picking has a field called sale_id and also this also belongs to the database column as well. Hence, to get all the pickings belongs to sale order, we may first use the ORM:
query = """select * from stock_picking where sale_id is not null"""
result = self.env.cr.fetchall()
Now, the second example is not only faster, but also, it allows you to extend the facility further to use joins or select specific field of a table result, which is only possible using ‘read’ Odoo ORM method, again, domain specification is not permissible like it is available in ‘search’.
We are able to do things like the following with the sql:
query = """select sale_order.name, stock_picking.name from stock_picking left join on stock_picking.sale_id = sale_order.id where stock_picking.sale_id is not null"""
result = self.env.cr.fetchall()
This would give you a result of each sale order with it’s picking name. To produce a result like the above using ORM is costly as it follows ‘N+1’ algorithm, hence inefficient in making reports or scaling the software.
Now, we understand, we are able to use such field and make the reports efficient using SQL as sale_id is distinctively available in the database. But what if you want to check how the product has been purchased, and then sold? Then, we also need purchase_order model to connect to our above query, right? But unfortunately, as ‘Purchase’ module doesn’t add a column purchase_id, we are unable to use this directly.
So, how can we still use purchase_id in the SQL Query to generate report in Odoo?
First, we need to see, how purchase_id is added in Odoo.
purchase_id is added in stock_picking model in the ‘purchase_stock’ module. If you open the following file:
you may see, how purchase_id is defined as related Many2one field:
A related field in Odoo, is like a pointer, a syntactic sugar of foreign key for less used fields. If the field is highly used, this might cause performance issue, as Odoo has to do multiple lookups unlike direct lookup for a related field. Now, get to the point, purchase_id is related to ‘move_lines.purchase_line_id.order_id’. This is a long relation. Let me go one by one:
move_lines : stock_picking has an One2many relation with stock.move model, that derives the available moves for the picking.
purchase_line_id: Each move line derived from a purchase order line, and while doing so, it keeps the ID of the purchase order line in a foreign key of stock.move model, namely purchase_line_id.
order_id: Each purchase_order_line has a foreign key with the purchase.order model kept in order_id field.
Now, we know, how the purchase_id derives the purchase_order id using the following relation:
Picking > Moves > Purchase Order Line > Purchase Order
Now we can use the following kind of relation for detecting purchase order from stock picking:
select purchase_order.name, stock_picking.name from stock_picking left join stock_move on stock_move.picking_id = stock_picking.id left join purchase_order_line on purchase_order_line.id = stock_move.purchase_line_id left join purchase_order on purchase_order.id = purchase_order_line.order_id where stock_move.purchase_line_id is not null group by stock_picking.name, purchase_order.name
Here, we are able to get the picking and purchase in relation with one query. This concept can be used to derive many data, like, let’s say, you would like to see, how many of your products are purchased, then, sold and returned, all can be done in few queries, without having N+1 problem.
When looking at the html report in Odoo, locale fonts look ok, but if you download the Qweb report to print in pdf format, it prints gibberish. How to fix that?
Odoo uses a templating engine for reporting called ‘Qweb’. Qweb can be used to generate two types of reports. One is HTML and the other is PDF. Odoo primarily uses Qweb engine to generate the HTML code. After that, it uses a tool called ‘wkhtmltopdf’ to convert the report to pdf and make it printable. Now when, we look at the HTML version of the report, fonts are shown based on Unicode supported browsers or the fonts you have installed on your computer. But when you try to convert this to PDF using wkhtmltopdf, that tool has to have exclusive access to those fonts to be able to convert them from HTML to pdf for you. As wkhtmlpdf command runs in the server you have installed Odoo, hence, you would need to install the font package in the server.
In my case, I required to install Bengali fonts. For CentOS, it is available under the lohit package, that contains several indian fonts including bengali. To install bengali font package in CentOS 7, use the following command:
yum install lohit-bengali* -y
Once done, your wkhtmltopdf should be able to read the bengali fonts from your html/qweb templates and able to convert them to PDF for you.
res.users and res.partners tables are two base tables of odoo. If you would like to inherit and extend them, remember, you can’t do it from the Odoo user view. The reason is, when you do an upgrade from the user view, it has to be something that works over base module, not base module itself. Hence, you will get a 500 error or internal server error when trying to upgrade the module.
We will make a simple module for res.users to extend the model to add a field called ‘access_token’ for each user, and generate a key automatically when a user is added.
I will only post the model file and the view file here. I expect you already know how to write an Odoo module.
This is my res_users.py file
from odoo import api, fields, models
import string, random
_inherit = 'res.users'
access_token = fields.Text(string='API Access Token', default=False)
def create(self, vals):
key = ''.join(random.choices(string.ascii_lowercase + string.digits, k = 48))
user_id = super(res_users_ex, self).create(vals)
user_id.access_token = key
The next one is the binary of odoo with it’s location, which should be odoo-bin. With the parameter -d, you give your database name, and with the parameter -u, you need to give your module name. After you run the command, you should see no ‘Error’ or Red marked line in your console. If not, it shall be upgraded. Now do control + c, and start your odoo again to see the new fields being visible in your Users tab.
In one line, context managers are an efficient way of handling resources in Python. So, what kind of resources are they? It could be any logical resource that you are using for your software, a common one, is database connections, or the files or in few cases, locks for concurrency control.
How exactly Context Manager is efficient?
If we talk about efficient programs, there could be several meanings of it in computer science. For our case, we mean efficient by writing less code, or more specifically, not writing repetitive codes for managing resources. There are some pinpoint benefits of using less repetitive codes other than writing more codes, which is purely technical. You do not forget to perform a step, that is essential if you do it from one source. Let’s find an example. You have a code, that connects to a remote FTP and uploads some content, once done, it closes the connection. If you do the process in several places, you might miss out to close the FTP connection in a place, that is accessed several times by the users. If that is so you might run out of the FTP connection pool on a random day. It is essential to close the connection after you are done with the connection to free the resources. Context Managers can help you write a code, that does the job for you, without the need to remember closing the connection each time.
The ‘with’ statement in Python
Before we can go deeper with Context Managers, we need to learn something about ‘with’ statement in python. ‘with’ is a special statement in python, that does jobs automatically for you. One, it calls a method of setting class, that is called ‘__enter__’ when it calls it, and the other, it automatically calls ‘__exit__’ method when it completes running its code. Let’s do some coding now
with open('test.txt') as f:
In the above code, we are opening a file using the ‘with’ statement of python, then writing some texts in it. But wait, we haven’t closed the file, did you notice? Isn’t it necessary to close a file after opening it to free up the system file descriptors? Absolutely, but using the ‘with’ statement, do it for you even if you don’t do it in your code. In Python, the open() method for opening a file, can be used as Context Manager. For Context Manager, two methods are essential, one is when you set up the call, which is ‘__enter__’ and the other is, when you end the code, that is ‘__exit__’. ‘with’ statement in python is created to be used for Context Managers. As I said, the open() method can be used as Context Manager, which means, Python has both of these methods defined by default with the open() method, and can be used using the ‘with’ statement.
Why and when do we need Context Managers?
Before jumping into, how can we do context managers, let’s understand, if we understand the need of context managers properly with an example! The primary purpose of a context manager is to write cleaner nonrepetitive codes. Do we need this often? Yes, we do. A common example would be in setting up your database connections. If you are setting up a database connection and clean up the things once done, you may create a context manager to do that. There are more complex database use cases of context managers. Let’s focus on one of them. Let’s say, you would like to utilize the database ‘SAVEPOINT’ in a cleaner way, manage the release, and rollback for concurrent transactions based on the savepoint you create dynamically, a clean technique would be to use context manager. Pseudocode could be like the following for this kind of context manager
NAME = UUID()
SQL.EXECUTE('SAVEPOINT ' + UUID)
EXECUTE('ROLLBACK TO SAVEPOINT ' + UUID)
EXECUTE('RELEASE SAVEPOINT ' + UUID)
What this context manager is trying to do, is generating a savepoint with a name for you. Once done, it yields the code you instruct it to run after the ‘with’ statement. If you create an exception from those codes, it rollback and sends the exception to the main program else if not it releases the savepoint and gives control to the code after the ‘with’ statement. This is technically the most efficient way of using Savepoint for SQL in your code. Similarly so, we can acquire and release locks for concurrent control using Context managers, or processing an API that had a setup call and an end/cleanup call.
How can we write Context Manager in Python?
There are two ways you can do it. One is using Python class, and the other use, using contextlib and the contextmanager decorator. Let’s first check out, how to do it using Python class to understand the concept better.
First, we want to emulate the way Python uses ‘open’ method as context manager using our own context manager class. A context manager class that can be used using ‘with’ statement could be like the following:
def __init__(self, name):
self.name = name
self.f = open(self.name)
def __exit__(self, exc_type, exc_val, traceback):
with Open_A_File('test.txt') as f:
In our class ‘Open_A_File()’, we have 3 methods. Our constructor __init__ method and other two methods are __enter__ and __exit__. When we used Open_A_File() using ‘with’ statement with a parameter, it setup our filename variable for the class using constructor, and then, calls the __enter__ method. It then opens the file and returns the file object. When it returns the file object, we catch it as ‘f’ to use in our code under the ‘with’ statement. We then write the code and the codes within ‘with’ statement ends, hence the __exit__ is automatically called, that closes the file object by calling ‘close()’. We can technically convert any class into a Context Manager and use them using ‘with’ statement if can define the methods to do while entering and exiting the class when called directly with the ‘with’ statement.
Other than class, we can use Context Managers using a function, through the use of contextlib library. This is the most used method of using Context Managers. We used the idea of this, in our pseudocode while demonstrating earlier. Let’s rewrite the above code using contextlib below
from contextlib import contextmanager
f = open(file)
with Open_A_File('test.txt') as f:
We first import contextmanager decorator from contextlib and then, we define a normal function. Although, there exists a ‘yield’ statement. For context managers, the statements that exist before yield would execute on __enter__ method, and the statements after yield would execute on __exit__ method. If you want to return anything to the ‘with’ call, then you need to specify that after yield, as we did in yield f, that means, we returned the file object to the ‘wite’ statement. the yield would replace the code, we run after the ‘with’ statement, like the f.write() in our case.
Hope this make sense. For confusion, or in case you would like to add some, do comment below. Thanks for reading.
There are times, when you might suddenly see your Odoo is shutdown automatically, without warning. Once you enable to logging, you could see an error like the following:
virtual real time limit (151/120s) reached.
or in full details like the following
2021-04-22 06:46:44,054 32685 WARNING ? odoo.service.server: Thread <Thread(odoo.service.http.request.140015617943296, started 140015617943296)> virtual real time limit (151/120s) reached.
2021-04-22 06:46:44,054 32685 INFO ? odoo.service.server: Dumping stacktrace of limit exceeding threads before reloading
2021-04-22 06:46:44,060 32685 INFO ? odoo.tools.misc:
# Thread: <Thread(odoo.service.http.request.140015617943296, started 140015617943296)> (db:n/a) (uid:n/a) (url:n/a)
File: "/opt/rh/rh-python36/root/usr/lib64/python3.6/threading.py", line 884, in _bootstrap
File: "/opt/rh/rh-python36/root/usr/lib64/python3.6/threading.py", line 916, in _bootstrap_inner
File: "/opt/rh/rh-python36/root/usr/lib64/python3.6/threading.py", line 864, in run
File: "/opt/rh/rh-python36/root/usr/lib64/python3.6/socketserver.py", line 654, in process_request_thread
File: "/opt/rh/rh-python36/root/usr/lib64/python3.6/socketserver.py", line 364, in finish_request
self.RequestHandlerClass(request, client_address, self)
File: "/opt/rh/rh-python36/root/usr/lib64/python3.6/socketserver.py", line 724, in __init__
File: "/opt/odoo/odoo14-venv/lib64/python3.6/site-packages/werkzeug/serving.py", line 329, in handle
rv = BaseHTTPRequestHandler.handle(self)
File: "/opt/rh/rh-python36/root/usr/lib64/python3.6/http/server.py", line 418, in handle
File: "/opt/odoo/odoo14-venv/lib64/python3.6/site-packages/werkzeug/serving.py", line 360, in handle_one_request
self.raw_requestline = self.rfile.readline()
File: "/opt/rh/rh-python36/root/usr/lib64/python3.6/socket.py", line 586, in readinto
2021-04-22 06:46:44,060 32685 INFO ? odoo.service.server: Initiating server reload
This is because Odoo is killing zombie processes and probably mistakenly crashing your Odoo completely while doing so. The parameter that is used for this purpose, can be found in Odoo documentation:
Prevents the worker from taking longer than <limit> seconds to process a request. If the limit is exceeded, the worker is killed.
Differs from --limit-time-cpu in that this is a “wall time” limit including e.g. SQL queries.
Defaults to 120.
You may start your service command with something like –limit-time-real 100000 to avoid Odoo from auto killing processes. A command could look like the following if you edit your service file located at: