%global _empty_manifest_terminate_build 0 Name: python-dbt-extractor Version: 0.4.1 Release: 1 Summary: A tool to analyze and extract information from Jinja used in dbt projects. License: Apache-2.0 URL: https://github.com/dbt-labs/dbt-parser-generator/ Source0: https://mirrors.nju.edu.cn/pypi/web/packages/af/2e/a110b40212480fd02bff567ff84effea8b9937ccd6ebfad0f10a382183d2/dbt_extractor-0.4.1.tar.gz %description # dbt extractor ![demo app](demo/demo.gif) This repository contains a tool that processes the most common jinja value templates in dbt model files. The tool depends on tree-sitter and the tree-sitter-jinja2 library. # Strategy The current strategy is for this processor to be 100% certain when it can accurately extract values from a given model file. Anything less than 100% certainty returns an exception so that the model can be rendered with python Jinja instead. There are two cases we want to avoid because they would risk correctness to user's projects: 1. Confidently extracting values that would not be extracted by python jinja (false positives) 2. Confidently extracting a set of values that are missing values that python jinja would have extracted. (misses) If we instead error when we could have confidently extracted values, there is no correctness risk to the user. Only an opportunity to expand the rules to encompass this class of cases as well. Even though jinja in dbt is not a typed language, the type checker statically determines whether or not the current implementation can confidently extract values without relying on python jinja rendering, which is when these errors would otherwise surface. This type checker will become more permissive over time as this tool expands to include more dbt and jinja features. # Architecture This architecture is optimized for value extraction and for future flexibility. This architecture is expected to change, and is coded in fp-style stages to make those changes easier for the future. This processor is composed of several stages: 1. parser 2. type checker 3. extractor Additionally, the following tools utilize the above processor: 1. browser-based demo of dbt extraction as you type The tree-sitter parser is located in the tree-sitter-jinja2 library. The rust bindings are used to traverse the concrete syntax tree that tree-sitter creates in order to create a typed abstract syntax tree in the type checking stage. The errors in the type checking stage are not raised to the user, and are instead used by developers to debug tests. The parser is solely responsible for turning text into recognized values, while the type checker does arity checking, and enforces argument list types (e.g. nested function calls like `{{ config(my_ref=ref('table')) }}` will parse but not type check even though it is valid dbt syntax. The tool at this time doesn't have an agreed serialization to communicate refs as config values, but could in the future.) The extractor uses the typed abstract syntax tree to easily identify all the refs, sources, and configs present and extract them. ## Running The Demo App To see the full implementation extract dbt values live as you type in a browser, run: ``` make demo ``` It may take a moment for the demo to compile an optimized version of itself. Kill the server with ctrl+c to end the demo. ## Testing The Project ``` make test ``` ## Future Work - Refactor the tree-sitter jinja parser into its own repository to potentially open source and engage with the community on implementing improvements. - Remove ref, source, and config type checking as hard coded rules and instead read these function types from external function definition statements. - Create input path for a manifest file so it can be run on any project without additional pre-processing %package -n python3-dbt-extractor Summary: A tool to analyze and extract information from Jinja used in dbt projects. Provides: python-dbt-extractor BuildRequires: python3-devel BuildRequires: python3-setuptools BuildRequires: python3-pip BuildRequires: python3-cffi BuildRequires: gcc BuildRequires: gdb %description -n python3-dbt-extractor # dbt extractor ![demo app](demo/demo.gif) This repository contains a tool that processes the most common jinja value templates in dbt model files. The tool depends on tree-sitter and the tree-sitter-jinja2 library. # Strategy The current strategy is for this processor to be 100% certain when it can accurately extract values from a given model file. Anything less than 100% certainty returns an exception so that the model can be rendered with python Jinja instead. There are two cases we want to avoid because they would risk correctness to user's projects: 1. Confidently extracting values that would not be extracted by python jinja (false positives) 2. Confidently extracting a set of values that are missing values that python jinja would have extracted. (misses) If we instead error when we could have confidently extracted values, there is no correctness risk to the user. Only an opportunity to expand the rules to encompass this class of cases as well. Even though jinja in dbt is not a typed language, the type checker statically determines whether or not the current implementation can confidently extract values without relying on python jinja rendering, which is when these errors would otherwise surface. This type checker will become more permissive over time as this tool expands to include more dbt and jinja features. # Architecture This architecture is optimized for value extraction and for future flexibility. This architecture is expected to change, and is coded in fp-style stages to make those changes easier for the future. This processor is composed of several stages: 1. parser 2. type checker 3. extractor Additionally, the following tools utilize the above processor: 1. browser-based demo of dbt extraction as you type The tree-sitter parser is located in the tree-sitter-jinja2 library. The rust bindings are used to traverse the concrete syntax tree that tree-sitter creates in order to create a typed abstract syntax tree in the type checking stage. The errors in the type checking stage are not raised to the user, and are instead used by developers to debug tests. The parser is solely responsible for turning text into recognized values, while the type checker does arity checking, and enforces argument list types (e.g. nested function calls like `{{ config(my_ref=ref('table')) }}` will parse but not type check even though it is valid dbt syntax. The tool at this time doesn't have an agreed serialization to communicate refs as config values, but could in the future.) The extractor uses the typed abstract syntax tree to easily identify all the refs, sources, and configs present and extract them. ## Running The Demo App To see the full implementation extract dbt values live as you type in a browser, run: ``` make demo ``` It may take a moment for the demo to compile an optimized version of itself. Kill the server with ctrl+c to end the demo. ## Testing The Project ``` make test ``` ## Future Work - Refactor the tree-sitter jinja parser into its own repository to potentially open source and engage with the community on implementing improvements. - Remove ref, source, and config type checking as hard coded rules and instead read these function types from external function definition statements. - Create input path for a manifest file so it can be run on any project without additional pre-processing %package help Summary: Development documents and examples for dbt-extractor Provides: python3-dbt-extractor-doc %description help # dbt extractor ![demo app](demo/demo.gif) This repository contains a tool that processes the most common jinja value templates in dbt model files. The tool depends on tree-sitter and the tree-sitter-jinja2 library. # Strategy The current strategy is for this processor to be 100% certain when it can accurately extract values from a given model file. Anything less than 100% certainty returns an exception so that the model can be rendered with python Jinja instead. There are two cases we want to avoid because they would risk correctness to user's projects: 1. Confidently extracting values that would not be extracted by python jinja (false positives) 2. Confidently extracting a set of values that are missing values that python jinja would have extracted. (misses) If we instead error when we could have confidently extracted values, there is no correctness risk to the user. Only an opportunity to expand the rules to encompass this class of cases as well. Even though jinja in dbt is not a typed language, the type checker statically determines whether or not the current implementation can confidently extract values without relying on python jinja rendering, which is when these errors would otherwise surface. This type checker will become more permissive over time as this tool expands to include more dbt and jinja features. # Architecture This architecture is optimized for value extraction and for future flexibility. This architecture is expected to change, and is coded in fp-style stages to make those changes easier for the future. This processor is composed of several stages: 1. parser 2. type checker 3. extractor Additionally, the following tools utilize the above processor: 1. browser-based demo of dbt extraction as you type The tree-sitter parser is located in the tree-sitter-jinja2 library. The rust bindings are used to traverse the concrete syntax tree that tree-sitter creates in order to create a typed abstract syntax tree in the type checking stage. The errors in the type checking stage are not raised to the user, and are instead used by developers to debug tests. The parser is solely responsible for turning text into recognized values, while the type checker does arity checking, and enforces argument list types (e.g. nested function calls like `{{ config(my_ref=ref('table')) }}` will parse but not type check even though it is valid dbt syntax. The tool at this time doesn't have an agreed serialization to communicate refs as config values, but could in the future.) The extractor uses the typed abstract syntax tree to easily identify all the refs, sources, and configs present and extract them. ## Running The Demo App To see the full implementation extract dbt values live as you type in a browser, run: ``` make demo ``` It may take a moment for the demo to compile an optimized version of itself. Kill the server with ctrl+c to end the demo. ## Testing The Project ``` make test ``` ## Future Work - Refactor the tree-sitter jinja parser into its own repository to potentially open source and engage with the community on implementing improvements. - Remove ref, source, and config type checking as hard coded rules and instead read these function types from external function definition statements. - Create input path for a manifest file so it can be run on any project without additional pre-processing %prep %autosetup -n dbt-extractor-0.4.1 %build %py3_build %install %py3_install install -d -m755 %{buildroot}/%{_pkgdocdir} if [ -d doc ]; then cp -arf doc %{buildroot}/%{_pkgdocdir}; fi if [ -d docs ]; then cp -arf docs %{buildroot}/%{_pkgdocdir}; fi if [ -d example ]; then cp -arf example %{buildroot}/%{_pkgdocdir}; fi if [ -d examples ]; then cp -arf examples %{buildroot}/%{_pkgdocdir}; fi pushd %{buildroot} if [ -d usr/lib ]; then find usr/lib -type f -printf "/%h/%f\n" >> filelist.lst fi if [ -d usr/lib64 ]; then find usr/lib64 -type f -printf "/%h/%f\n" >> filelist.lst fi if [ -d usr/bin ]; then find usr/bin -type f -printf "/%h/%f\n" >> filelist.lst fi if [ -d usr/sbin ]; then find usr/sbin -type f -printf "/%h/%f\n" >> filelist.lst fi touch doclist.lst if [ -d usr/share/man ]; then find usr/share/man -type f -printf "/%h/%f.gz\n" >> doclist.lst fi popd mv %{buildroot}/filelist.lst . mv %{buildroot}/doclist.lst . %files -n python3-dbt-extractor -f filelist.lst %dir %{python3_sitearch}/* %files help -f doclist.lst %{_docdir}/* %changelog * Mon Apr 10 2023 Python_Bot - 0.4.1-1 - Package Spec generated