3mcloud/requirement-walker

Walks python requirement.txt files.

Requirement Walker

A simple python package which makes it easy to crawl/parse/walk over the requirements within a requirements.txt file. It can handle nested requirement files, i.e. -r ./nested_path/other_reqs.txt and handle paths to local pip packages (but cannot currently parse their requirements): ./pip_package/my_pip_package # requirement-walk: local-package-name=my-package. Comments within the requirement files can also be preserved.

Installation

pip install requirement-walker

Arguments

Arguments for requirement-walker are parsed from the comments within the requirements.txt files.

Arguments should follow the pattern of:

flat-earth==1.1.1 # requirement-walker: {arg1_name}={arg1_val}
bigfoot==0.0.1 # He is real requirement-walker: {arg1_name}={arg1_val}|{arg2_name}={arg2_val1},{arg2_val2}

Available arguments:

Name	Expect # of Values	Discription
local-package-name	0 or 1	If a requirement is a path to a local pip package, then provide this argument to tell the walker that its local. You can optionally tell provide the name of the pip package which can be used when filtering requirements. (See Example Workflow)
root-relative	1	Can be provided along with local-package-name or can be stand alone with any -r requirements. When the walker sees a relative path for a requirement, it will use this provided value instead of the value actually in that line of the requirements.txt file when saving to a file.

Example Workflow

Lets walk through a complex example. Note, I am only doing the requirement.txt files like this to give a detailed example. I do NOT recommend you do requirements like this.

Folder Structure

walk_requirements.py
example_application
│   README.md
│   project_requirements.txt
│
└───lambdas
│   │   generic_reqs.txt
│   │
│   └───s3_event_lambda
│   │   │   s3_lambda_reqs.txt
│   │   │   ...
│   │   │
│   │   └───src
│   │       │   ...
│   │
│   └───api_lambda
│       │   api_lambda_reqs.txt
│       │   ...
│       │
│       └───src
│           │   ...
│
└───pip_packages
    └───orm_models
        │   setup.py
        │
        └───orm_models
        │   |   ...
        │
        └───tests
            |   ...

NOTE: This package CANNOT currently parse a setup.py file to walk its requirements but we can keep track of the path to the local requirement.

walk_requirements.py

Assuming requirement-walker is already installed in a virtual environment or locally such that it can be imported.

These files can also be found in ./test/examples/example_application.

""" Example Script """
# Assuming I am running this script in the directory it is within above.

# Built In
import logging

# 3rd Party
from requirement_walker import RequirementFile

# Owned


if __name__ == '__main__':
    FORMAT = '[%(asctime)s] {%(pathname)s:%(lineno)d} %(levelname)s - %(message)s'
    logging.basicConfig(format=FORMAT, level=logging.DEBUG)
    req_file = RequirementFile('./example_application/project_requirements.txt')
    # RequirementFile has a magic method __iter__ written for it so it can be iterated over.
    # Outputs found down below
    print("Output 1:", *req_file, sep='\n') # This will print the file basically as is
    print("---------------------------------------------")
    print("Output 2:", *req_file.iter_recursive(), sep='\n') # This will print all reqs in without -r
    # You can also send the reqs to a single file via:
    # req_file.to_single_file(path_to_output_to)
    # That method accepts, no_empty_lines and no_comment_only_lines as arguments.

project_requirements.txt

# One-lining just to show multiple -r works on one line, -r is the only thing that works on one line.
-r ./lambdas/s3_event_lambda/s3_lambda_reqs.txt --requirement=./lambdas/api_lambda/api_lambda_reqs.txt # comment

./pip_packages/orm_models # requirement-walker: local-package-name=orm-models
orm @ git+ssh://git@github.com/ORG/orm.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link
orm2 @ git+https://github.com/ORG/orm2.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link
orm3 @ git+http://github.com/ORG/orm3.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link

generic_reqs.txt

moto==1.3.16.dev67
pytest==6.1.2
pytest-cov==2.10.1
pylint==2.6.0
docker==4.4.0
coverage==4.5.4
# Some other stuff

# Add empty line

s3_lambda_reqs.txt

-r ./../generic_reqs.txt
./../../pip_packages/orm_models # requirement-walker: local-package-name|root-relative=./pip_packages/orm_models

api_lambda_reqs.txt

-r ./../generic_reqs.txt
./../../pip_packages/orm_models # requirement-walker: local-package-name|root-relative=./pip_packages/orm_models

Output

... Logs omitted ...
Output 1:
# One-lining just to show multiple -r works on one line, -r is the only thing that works on one line.
-r C:\Users\{UserName}\Repos\3mcloud\requirement-walker\tests\examples\example_application\lambdas\s3_event_lambda\s3_lambda_reqs.txt # comment
-r C:\Users\{UserName}\Repos\3mcloud\requirement-walker\tests\examples\example_application\lambdas\api_lambda\api_lambda_reqs.txt # comment

./pip_packages/orm_models # requirement-walker: local-package-name=orm-models
orm@ git+ssh://git@github.com/ORG/orm.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link
orm2@ git+https://github.com/ORG/orm2.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link
orm3@ git+http://github.com/ORG/orm3.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link
---------------------------------------------
Output 2:
# One-lining just to show multiple -r works on one line, -r is the only thing that works on one line.
moto==1.3.16.dev67
pytest==6.1.2
pytest-cov==2.10.1
pylint==2.6.0
docker==4.4.0
coverage==4.5.4
# Some other stuff

# Add empty line
./pip_packages/orm_models # requirement-walker: local-package-name|root-relative=./pip_packages/orm_models
moto==1.3.16.dev67
pytest==6.1.2
pytest-cov==2.10.1
pylint==2.6.0
docker==4.4.0
coverage==4.5.4
# Some other stuff

# Add empty line
./pip_packages/orm_models # requirement-walker: local-package-name|root-relative=./pip_packages/orm_models

./pip_packages/orm_models # requirement-walker: local-package-name=orm-models
orm@ git+ssh://git@github.com/ORG/orm.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link
orm2@ git+https://github.com/ORG/orm2.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link
orm3@ git+http://github.com/ORG/orm3.git@5e2b6d14f00ffbd473dfe8b8602b79e37266568c # git link

NOTE: Duplicates are NOT filtered out. You can do this on your own if you want using entry.requirement.name to filter them out as you iterate.

Failed Parsing

Sometimes the requirement parser fails. For example, maybe it tries parsing a -e or maybe you do a local pip package but don't provide local-package-name. If this happens, please open an issue; however, you should still be able to code yourself around the issue or use the walker till a fix is implemented. The walker aims to store as much information as it can, even in cases of failure. See the following example.

requirements.txt

astroid==2.4.2
attrs==20.3.0
aws-xray-sdk==2.6.0
boto==2.49.0
./local_pips/my_package # This will cause a failed requirement step
boto3==1.16.2
botocore==1.19.28
certifi==2020.11.8
cffi==1.14.4
./pip_packages/orm_models # requirement-walker: local-package-name

Code

""" Example Script """

# Built In
import logging

# 3rd Party
from requirement_walker import RequirementFile

# Owned

if __name__ == '__main__':
    FORMAT = '[%(asctime)s] {%(pathname)s:%(lineno)d} %(levelname)s - %(message)s'
    logging.basicConfig(format=FORMAT, level=logging.DEBUG)
    entries = RequirementFile('./requirements.txt')
    print(*entries, sep='\n')

Code Output

... logs omitted ...
astroid==2.4.2
attrs==20.3.0
aws-xray-sdk==2.6.0
boto==2.49.0
./local_pips/my_package # This will cause a failed requirement step
boto3==1.16.2
botocore==1.19.28
certifi==2020.11.8
cffi==1.14.4
./pip_packages/orm_models # requirement-walker: local-package-name

Note that it still printed correctly, but if you look at the logs you will see what happened:

WARNING  requirement_walker.walker:walker.py:148 Unable to parse requirement. Doing simple FailedRequirement where name=failed_req and url=./local_pips/my_package. Will still output.

If you want, you can refine requirements by looking at class instances:

""" Example Script """

# Built In
import logging

# 3rd Party
from requirement_walker import RequirementFile, LocalPackageRequirement, FailedRequirement

# Owned

if __name__ == '__main__':
    FORMAT = '[%(asctime)s] {%(pathname)s:%(lineno)d} %(levelname)s - %(message)s'
    logging.basicConfig(format=FORMAT, level=logging.ERROR)
    for entry in RequirementFile('./requirements.txt'):
        # `requirement` can be one of: `None, FailedRequirement, LocalPackageRequirement`
        if isinstance(entry.requirement, FailedRequirement):
            print("This requirement was a failed req.", entry)
        elif isinstance(entry.requirement, LocalPackageRequirement):
            print("This requirement was a local req.", entry)
        # If a entry is a requirement file, `requirement` will be None
        # and `requirement_file` will have a value other then None.
        elif isinstance(entry.requirement_file, RequirementFile):
            print("This entry is another requirement file.", entry)
# Ouput:
# This requirement was a failed req. ./local_pips/my_package # This will cause a failed requirement step
# This requirement was a local req. ./pip_packages/orm_models # requirement-walker: local-package-name

What is an Entry?

We define an entry as a single line within a requirements.txt file which could be empty, only have a comment, only have a requirement, be a reference to another requirement file, or have a mixture of a requirement/requirement file and a comment.

An Entry object has four main attributes but will not have them all at the same time:

• comment: Union[Comment, None]
• requirement: Union[pkg_resources.Requirement, FailedRequirement, LocalPackageRequirement, None]
• proxy_requirement: Union[_ProxyRequirement, None]
• requirement_file: [RequirementFile, None].

When attributes have values:

• If all of these attributes are set to None then the line the entry represents was an empty line.
• If requirement has a value then proxy_requirement will as well but requirement_file will NOT.
• If requirement_file has a value then requirement and proxy_requirement will NOT.
• A comment can exist on its own (a line with only a comment) or a comment can exist with either requirement or requirement_file.

Note, you will mainly work with requirement NOT proxy_requirement, but there may be cases where the package does not behave properly, in which cases proxy_requirement will hold all the other information pulled by the walker than you can use to code your way out of the mess.