feat: develop a crate to parse cwl documents

.github/workflows/cd.yml

@@ -22,26 +22,25 @@ jobs:
         uses: orhun/git-cliff-action@v4
         id: git-cliff
       - name: Run git-cliff to generate changelog
-        run: git-cliff -vv --latest --no-exec --github-repo ${{ github.repository }} --config cliff.toml
         id: changelog
+        run: git-cliff -vv --latest --no-exec --github-repo ${{ github.repository }} --config cliff.toml
 
-   publish-crates-io:
-   name: Publish on crates.io
-   runs-on: ubuntu-22.04
-   needs: generate-changelog
-   steps:
-    - name: Checkout
-      uses: actions/checkout@v4
-    - name: Verify package
-      run: cargo verify-project
-    - name: Run tests
-      run: cargo test --all-features
-    - name: Set the release version
-      run: echo "RELEASE_VERSION=${GITHUB_REF:11}" >> "$GITHUB_ENV"
-    - name: Install Rust toolchain
-      uses: actions-rs/toolchain@v1
-      uses: dtolnay/rust-toolchain@stable
-    # - name: Publish the zefiro-core library
-    #   run: |
-    #     cargo publish --manifest-path zefiro-core/Cargo.toml \
-    #       --locked --token ${{ secrets.CARGO_REGISTRY_TOKEN }}
+  publish-crates-io:
+    name: Publish on crates.io
+    runs-on: ubuntu-22.04
+    needs: generate-changelog
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Verify package
+        run: cargo verify-project
+      - name: Run tests
+        run: cargo test --all-features
+      - name: Set the release version
+        run: echo "RELEASE_VERSION=${GITHUB_REF:11}" >> "$GITHUB_ENV"
+      - name: Install Rust toolchain
+        uses: dtolnay/rust-toolchain@stable
+      # - name: Publish the zefiro-core library
+      #   run: |
+      #     cargo publish --manifest-path zefiro-core/Cargo.toml \
+      #       --locked --token ${{ secrets.CARGO_REGISTRY_TOKEN }}

.github/workflows/ci.yml

@@ -21,9 +21,9 @@ jobs:
       - name: Checkout
         uses: actions/checkout@v4
       - name: Check
-        run: cargo check --offline --verbose
+        run: cargo check --locked --verbose
       - name: Check without default features
-        run: cargo check --offline --no-default-features --verbose
+        run: cargo check --locked --no-default-features --verbose
 
   typos:
     name: Typos

zefiro-core/Cargo.toml

@@ -4,4 +4,7 @@ version = "0.1.0"
 edition = "2021"
 rust-version = "1.83.0"
 
-[dependencies]
+[workspace]
+members = [
+    "zefiro-cwl-parser"
+]

zefiro-core/zefiro-cwl-parser/Cargo.toml

@@ -0,0 +1,26 @@
+[package]
+name = "zefiro-cwl-parser"
+version = "0.1.0"
+description = """
+"The Common Workflow Language (CWL) object model that used in `zefiro`.
+"""
+authors = [
+  "Nikita Syzrantsev <[email protected]>"
+]
+categories = ["Common Workflow Language", "CWL"]
+keywords = ["CWL", "CommanLineTool", "Workflow"]
+edition = "2021"
+homepage = "https://github.com/zefiroproj/zefiro"
+license = "Apache-2.0"
+readme = "README.md"
+repository = "https://github.com/zefiroproj/zefiro"
+rust-version = "1.83.0"
+
+[dependencies]
+anyhow = "1.0.95"
+serde = { version = "1.0.216", features = ["derive"] }
+serde_with = "3.12.0"
+serde_yaml = "0.9.34"
+
+[dev-dependencies]
+rstest = "0.24.0"

zefiro-core/zefiro-cwl-parser/README.md

@@ -0,0 +1,77 @@
+# zefiro-cwl-parser
+
+A Rust library for parsing and working with Common Workflow Language (CWL) documents.
+
+## Overview
+
+* Supports only some fields of CWL v1.2 specification (see description of structures in the code)
+* Can serialize and deserialize [CommandLineTool](https://www.commonwl.org/v1.2/CommandLineTool.html) and [Workflow](https://www.commonwl.org/v1.2/Workflow.html) documents
+
+## Usage
+
+Add this to your `Cargo.toml`:
+
+```toml
+[dependencies]
+zefiro-cwl-parser = <zefiro-cwl-parser>
+```
+
+
+### Parsing CWL Schema Documents
+
+```rust
+use zefiro_cwl_parser::CwlSchema;
+
+// Parse from file
+let schema = CwlSchema::from_path("workflow.yml")?;
+
+// Parse from string
+let yaml_str = r#"
+cwlVersion: v1.2
+class: CommandLineTool
+id: step
+inputs:
+    - id: in_file
+      type: File
+      inputBinding:
+        prefix: --in-file
+    - id: out_file
+      type: string
+      default: "output.txt"
+      inputBinding:
+        prefix: --out-file
+    - id: output_location_subdir
+      type: string
+      default: output/
+outputs:
+    - id: out_file
+      type: File
+      outputBinding:
+        glob: $(inputs.out_file)
+        outputEval: ${self[0].location += inputs.output_location_subdir; return self[0]}
+requirements:
+    - class: DockerRequirement
+      dockerPull: step-image-uri:1.0
+    - class: InlineJavascriptRequirement
+"#;
+let schema = CwlSchema::from_string(yaml_str)?;
+```
+
+
+### Parsing CWL Values Documents
+
+```rust
+use zefiro_cwl_parser::CwlValues;
+
+// Parse input values from file
+let values = CwlValues::from_path("values.yml")?;
+
+// Create values from string
+let yaml_input = r#"
+input_file:
+    class: File
+    location: 's3://bucket/input.txt'
+output_file: 'output.txt'
+"#;
+let values = CwlValues::from_string(yaml_input)?;
+```

zefiro-core/zefiro-cwl-parser/examples/data/clt-step-schema.yml

@@ -0,0 +1,32 @@
+cwlVersion: v1.2
+class: CommandLineTool
+id: step
+inputs:
+  - id: in_file
+    type: File
+    inputBinding:
+      prefix: --in-file
+  - id: out_file
+    type: string
+    default: "output.txt"
+    inputBinding:
+      prefix: --out-file
+  - id: output_location_subdir
+    type: string
+    default: output/
+outputs:
+  - id: out_file
+    type: File
+    outputBinding:
+      glob: $(inputs.out_file)
+      outputEval: ${self[0].location += inputs.output_location_subdir; return self[0]}
+requirements:
+  - class: DockerRequirement
+    dockerPull: step-image-uri:1.0
+  - class: ResourceRequirement
+    coresMin: 2
+    outdirMin: 1000
+    ramMin: 1024
+  - class: InlineJavascriptRequirement
+  - class: ToolTimeLimit
+    timelimit: $(60*60*1)

zefiro-core/zefiro-cwl-parser/examples/data/clt-step-values.yml

@@ -0,0 +1,4 @@
+in_file:
+  class: File
+  location: '{{ inputLocation }}/output/input-{{ readgroup }}.txt'
+out_file: 'output.txt'

zefiro-core/zefiro-cwl-parser/examples/data/wf-step-schema.yml

@@ -0,0 +1,64 @@
+class: Workflow
+cwlVersion: v1.2
+id: step
+inputs:
+  - id: step__in_file
+    type:
+      type: array
+      items: File
+  - id: step__out_file
+    type: string
+outputs:
+  - id: step__out_file
+    type: File
+    outputSource: step/out_file
+requirements:
+  - class: InlineJavascriptRequirement
+  - class: ScatterFeatureRequirement
+steps:
+  - id: step
+    in:
+      - id: in_file
+        source: step__in_file
+      - id: out_file
+        source: step__out_file
+    out:
+      - id: out_file
+    run:
+      cwlVersion: v1.2
+      class: CommandLineTool
+      id: step1
+      inputs:
+        - id: in_file
+          type: File
+          inputBinding:
+            prefix: --in-file
+        - id: out_file
+          type: string
+          default: output.txt
+          inputBinding:
+            prefix: --out-file
+        - id: output_location_subdir
+          type: string
+          default: output/
+      outputs:
+        - id: out_file
+          type: File
+          outputBinding:
+            glob: $(inputs.out_file)
+            outputEval: ${self[0].location += inputs.output_location_subdir; return self[0]}
+      requirements:
+        - class: DockerRequirement
+          dockerPull: step1-image:1.0
+        - class: ResourceRequirement
+          coresMin: 2
+          outdirMin: 1000
+          ramMin: 1024
+        - class: InlineJavascriptRequirement
+        - class: ToolTimeLimit
+          timelimit: $(60*60*1)
+        - class: WorkReuse
+          enableReuse: true
+    scatter:
+      - in_file
+    scatterMethod: dotproduct

zefiro-core/zefiro-cwl-parser/src/lib.rs

@@ -0,0 +1,5 @@
+pub mod schema;
+pub mod values;
+
+pub use crate::schema::document::CwlSchema;
+pub use crate::values::document::CwlValues;

zefiro-core/zefiro-cwl-parser/src/main.rs

@@ -0,0 +1,20 @@
+use zefiro_cwl_parser::values::{document::CwlValues, types::CwlValueType};
+use zefiro_cwl_parser::schema::document::CwlSchema;
+
+fn main() {
+    let file_path = "examples/data/clt-step-values.yml";
+    let values = CwlValues::from_path(file_path).expect("Failed to deserialize CWL values document");
+    for (key, value) in values.iter() {
+        match value {
+            CwlValueType::File(value) => println!("{} {:?} {:?}", key, value, value.get_location()),
+            _ => println!("{} {:?}", key, value)
+        }
+    }
+
+    let file_path = "examples/data/wf-step-schema.yml";
+    let schema = CwlSchema::from_path(file_path).expect("Failed to deserialize CWL values document");
+    println!("{:?}", schema);
+    // for (key, value) in schema.iter() {
+    //     println!("{:?}", value);
+    // }
+}