Importing stock locations correct CSV format.

[StevenIsaacs]

I have a CSV to define stock locations but am encountering an error when attempting to import. I need help understanding what the correct format for the CSV should be in this case.

parent: Incorrect type. Expected pk value, received str

This confuses me because a CSV only contains strings. Here are the first three lines of the CSV. The error occurs on the third line and all lines following it.

Name,Parent,Description,Structural,Location
02-01-cc-iiii,,"DIV=01 HQ:",1,02-01-cc-iiii
02-01-02-iiii,02-01-cc-iiii,"CST=02 DIVISION Assets:",1,02-01-02-iiii

Version info:

Instance Name	InvenTree
Server Version	0.17.6
API Version	294
Database	postgresql
Docker Mode	Server is deployed using docker

Thanks

[StevenIsaacs]

From examining the code I believe a pk must be an integer. I changed the CSV accordingly. The first three lines are now:

Name,Parent,Description,Structural,Location
101000000,,"02-01-cc-iiii:DIV=01 HQ:",1,1101000000
101020000,101000000,"02-01-02-iiii:CST=02 DIVISION Assets:",1,1101020000

This still produces an error at the third line but now it is:
Invalid pk "101000000" - object does not exist.

Shouldn't 101000000 have been created by the second line? Is this a bug?

[SchrodingersGat]

"Parent" should be the ID of the parent location - so an integer number which is unique to the database.

[matmair]

you can only reference things that exist in the database. You are referencing something that is not already in the db; you would have to split this into several imports right now

[SchrodingersGat]

If I understand correctly you want to be able to reference one importing line from within another importing line?

This is currently not possible - although we have an outline for an implementation here #8203

Currently, the imported dataset must point to records already in the database.

[StevenIsaacs]

Then what is the purpose of the 'stocklocation' import model?

I was really hoping to automate the process of creating stock locations and part categories. This would save a tremendous amount of work.

The reason for this is to use identifiers in the accounting software (in this case GnuCash) for accounts representing ASSETS to also reference part categories and the corresponding stock locations -- labeled accordingly. I already have a Python script to generate the COA for import into GnuCash as a CSV and also generate the asset lists as a separate CSV to be imported into Inventree (the first lines of which I've shown in my previous posts.

I took a look at #8203 and it aligns with what I want to achieve. However, I have no clue how to get started being completely new to the code base.

[StevenIsaacs]

I took a look at the example on the import doc page (https://docs.inventree.org/en/latest/settings/import/). There is a CSV shown in one of the screen shots but nothing that shows the contents of the CSV. Can someone provide a CSV that works for import of either part categories (like the example) or stock locations (my current problem)? I might be able to work out what I need to do to solve my problem.

I did see in #8166 (comment) the line:

The only thing which is currently possible is to import the categories "level by level" or by just creating them manually.

This would be a bit tedious but I think doable in my case. I can change my Python script to output a series of CSVs -- one for each "level" of the hierarchy. The problem I will have with this approach is pre-determining the PK numbers for the parents.

The import apparently demands a valid PK for a parent reference. Is there a way to determine the PK corresponding to the name of the parent category or location? I can't find a UI screen which shows the PK number for an item. I suppose I could dig in the database itself but that would be a major PITA.

An Alternative Approach

I found this regarding the Python Interface and the examples there: https://docs.inventree.org/en/latest/api/python/python/
This could be the solution I need. Would it make sense to install a plugin based on the examples to do this? After looking at the examples more carefully I realized there's no need for a plugin -- I think. Instead, I maybe could use a Python venv, pip install inventree and then hack to my heart's content without worry of breaking the backend -- all debuggable in vscode (rubbing hands together and grinning).

btw: I tried several ERP solutions none of which checked boxes for my situation like Inventree does. I'm determined to solve this problem either by improving my understanding of how Inventree works or writing or changing some code somewhere. If some code is required I'll need some pointers to save time and I will contribute my solution. I'm a solo trying to start a business with virtually zero budget. I need to turn my attention to product development, instead of being sys admin or dev, so need a quick solution.

Thanks

[StevenIsaacs]

I did a small experiment in Python which went without a hitch and stepping with the debugger in vscode. In my experiment I was able to retrieve and examine existing stock locations and retrieve their PK numbers as well as their names. Having this information will allow me to both detect name collisions and determine parent PK numbers when creating new stock locations. Looking at the objects (using dir()) I see categories are pretty much the same. I'm loving the fact that Inventree is Python based.

[StevenIsaacs]

@SchrodingersGat @matmair

Here's a gist for the script I wrote to MASS import stock locations and part categories into Inventree. This is first pass so comments appreciated. I used a virtual environment created using vscode to develop and test. The Inventree Python API made it possible to condense the problem into not very many lines.

https://gist.github.com/StevenIsaacs/77ed1ca5542cae919d94361f83988ad5

I tested this on my own instance of Inventree. The invoke backup and invoke restore commands were very useful for resetting test data.

Thanks for everyone's hard work which made it relatively easy for me to implement this.