feat: adds validation and linting

Author: claymcleod

Cargo.lock

@@ -1,47 +1,101 @@
 version = "v1"
 
+[[repositories]]
+organization = "stjudecloud"
+name = "workflows"
+
 [[repositories]]
 organization = "PacificBiosciences"
 name = "HiFi-human-WGS-WDL"
 
 [[repositories]]
-organization = "biowdl"
-name = "tasks"
+organization = "chanzuckerberg"
+name = "czid-workflows"
 
 [[repositories]]
-organization = "stjudecloud"
-name = "workflows"
+organization = "biowdl"
+name = "tasks"
 
-[[repositories]]
-organization = "chanzuckerberg"
-name = "czid-workflows"
+[[ignored_errors]]
+document = "biowdl/tasks:bcftools.wdl"
+error = '''validation error: [v1::001] invalid escape character '\_' in string at line 114:75'''
 
 [[ignored_errors]]
 document = "biowdl/tasks:bedtools.wdl"
-error = """
-  --> 29:67
-   |
-29 |         String memory = \"~{512 + ceil(size([inputBed, faidx], \"MiB\"))}MiB\"
-   |                                                                   ^---
-   |
-   = expected WHITESPACE or OPTION"""
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 27:48'''
 
 [[ignored_errors]]
 document = "biowdl/tasks:bowtie.wdl"
-error = """
-  --> 40:58
-   |
-40 |         String memory = \"~{5 + ceil(size(indexFiles, \"GiB\"))}GiB\"
-   |                                                          ^---
-   |
-   = expected WHITESPACE or OPTION"""
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 63:32'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:centrifuge.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 122:57'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:common.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 275:45'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:fastqc.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 59:42'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:gatk.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 127:57'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:hisat2.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 60:34'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:multiqc.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 133:45'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:picard.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 643:51'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:sambamba.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 91:44'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:samtools.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 80:42'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:umi-tools.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 95:49'''
+
+[[ignored_errors]]
+document = "biowdl/tasks:umi.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 39:60'''
 
 [[ignored_errors]]
 document = "stjudecloud/workflows:template/task-templates.wdl"
 error = """
+parse error:
+
   --> 17:25
    |
 17 |         Int memory_gb = <>
    |                         ^---
    |
    = expected WHITESPACE, COMMENT, or expression"""
+
+[[ignored_errors]]
+document = "stjudecloud/workflows:tools/bwa.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 34:17'''
+
+[[ignored_errors]]
+document = "stjudecloud/workflows:tools/fq.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 123:17'''
+
+[[ignored_errors]]
+document = "stjudecloud/workflows:tools/kraken2.wdl"
+error = '''validation error: [v1::001] invalid escape character '\.' in string at line 335:17'''
+
+[[ignored_errors]]
+document = "stjudecloud/workflows:workflows/rnaseq/rnaseq-standard-fastq.wdl"
+error = '''validation error: [v1::001] invalid escape character '\*' in string at line 55:241'''

Gauntlet.toml

@@ -0,0 +1 @@
+format_code_in_doc_comments = true

rustfmt.toml

@@ -19,7 +19,8 @@ pest_derive = { workspace = true }
 reqwest = { version = "0.11.22", optional = true }
 serde = { workspace = true }
 serde_with = { workspace = true, optional = true }
-tokio = { version = "1.33.0", features = ["full"], optional = true}
+to_snake_case = "0.1.1"
+tokio = { version = "1.33.0", features = ["full"], optional = true }
 toml = { workspace = true, optional = true }
 
 [features]
@@ -36,7 +37,7 @@ binaries = [
     "reqwest",
     "serde_with",
     "tokio",
-    "toml"
+    "toml",
 ]
 
 [[bin]]

wdl-grammar/Cargo.toml

@@ -1,8 +1,8 @@
 //! `wdl-grammar create-test`
 
 use clap::Parser;
+use log::warn;
 use pest::iterators::Pair;
-use pest::Parser as _;
 use pest::RuleType;
 
 use wdl_grammar as grammar;
@@ -18,8 +18,8 @@ pub enum Error {
     /// Multiple root nodes parsed.
     MultipleRootNodes,
 
-    /// A parsing error from Pest.
-    Parse(Box<dyn std::error::Error>),
+    /// An error parsing the grammar.
+    GrammarV1(grammar::Error<grammar::v1::Rule>),
 
     /// Unknown rule name.
     UnknownRule {
@@ -35,11 +35,11 @@ impl std::fmt::Display for Error {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         match self {
             Error::Common(err) => write!(f, "{err}"),
+            Error::GrammarV1(err) => write!(f, "grammar parse error: {err}"),
             Error::MultipleRootNodes => write!(f, "multiple root nodes found"),
             Error::UnknownRule { name, grammar } => {
                 write!(f, "unknown rule '{name}' for grammar {grammar}")
             }
-            Error::Parse(err) => write!(f, "parse error: {err}"),
         }
     }
 }
@@ -84,11 +84,15 @@ pub fn create_test(args: Args) -> Result<()> {
         .unwrap_or_else(|| get_contents_stdin().map_err(Error::Common))?;
 
     let mut parse_tree = match args.specification_version {
-        grammar::Version::V1 => {
-            grammar::v1::Parser::parse(rule, &input).map_err(|err| Error::Parse(Box::new(err)))?
-        }
+        grammar::Version::V1 => grammar::v1::parse(rule, &input).map_err(Error::GrammarV1)?,
     };
 
+    if let Some(warnings) = parse_tree.warnings() {
+        for warning in warnings {
+            warn!("{}", warning);
+        }
+    }
+
     let root = match parse_tree.len() {
         // SAFETY: this should not be possible, as parsing just successfully
         // completed. As such, we should always have at least one parsed

wdl-grammar/src/commands/create_test.rs

@@ -8,7 +8,6 @@ use clap::Parser;
 use colored::Colorize as _;
 use log::debug;
 use log::trace;
-use pest::Parser as _;
 
 pub mod config;
 pub mod document;
@@ -81,12 +80,16 @@ pub struct Args {
     #[arg(short, long)]
     config_file: Option<PathBuf>,
 
+    /// Don't load any configuration from the cache.
+    #[arg(short, long, global = true)]
+    no_cache: bool,
+
     /// Only errors are printed to the stderr stream.
     #[arg(short, long, global = true)]
     quiet: bool,
 
     /// Overwrites the configuration file.
-    #[arg(short, long, global = true)]
+    #[arg(long, global = true)]
     save_config: bool,
 
     /// Silences printing detailed error information.
@@ -97,6 +100,10 @@ pub struct Args {
     #[arg(long, global = true)]
     skip_remote: bool,
 
+    /// Displays warnings as part of the report output.
+    #[arg(long, global = true)]
+    show_warnings: bool,
+
     /// The Workflow Description Language (WDL) specification version to use.
     #[arg(value_name = "VERSION", short = 's', long, default_value_t, value_enum)]
     specification_version: grammar::Version,
@@ -108,9 +115,16 @@ pub struct Args {
 
 /// Main function for this subcommand.
 pub async fn gauntlet(args: Args) -> Result<()> {
-    let path = args.config_file.unwrap_or(Config::default_path());
-    let mut config =
-        Config::load_or_new(path, args.specification_version).map_err(Error::Config)?;
+    let mut config = match args.no_cache {
+        true => {
+            debug!("Skipping loading from cache.");
+            Config::default()
+        }
+        false => {
+            let path = args.config_file.unwrap_or(Config::default_path());
+            Config::load_or_new(path, args.specification_version).map_err(Error::Config)?
+        }
+    };
 
     if let Some(repositories) = args.repositories {
         config.repositories_mut().extend(
@@ -157,13 +171,33 @@ pub async fn gauntlet(args: Args) -> Result<()> {
 
             match config.version() {
                 grammar::Version::V1 => {
-                    match grammar::v1::Parser::parse(grammar::v1::Rule::document, &content) {
-                        Ok(_) => {
-                            trace!("{}: successfully parsed.", document_identifier);
-                            report
-                                .register(document_identifier, Status::Success)
-                                .map_err(Error::InputOutput)?;
-                        }
+                    match grammar::v1::parse(grammar::v1::Rule::document, &content) {
+                        Ok(tree) => match tree.warnings() {
+                            Some(warnings) => {
+                                trace!(
+                                    "{}: successfully parsed with {} warnings.",
+                                    document_identifier,
+                                    warnings.len()
+                                );
+                                report
+                                    .register(document_identifier, Status::Warning)
+                                    .map_err(Error::InputOutput)?;
+
+                                if args.show_warnings {
+                                    for warning in warnings {
+                                        report
+                                            .report_warning(warning)
+                                            .map_err(Error::InputOutput)?;
+                                    }
+                                }
+                            }
+                            None => {
+                                trace!("{}: succesfully parsed.", document_identifier,);
+                                report
+                                    .register(document_identifier, Status::Success)
+                                    .map_err(Error::InputOutput)?;
+                            }
+                        },
                         Err(err) => {
                             let actual_error = err.to_string();
 
@@ -273,7 +307,7 @@ pub async fn gauntlet(args: Args) -> Result<()> {
         println!(
             "\n{}\n",
             "Undetected expected errors: you should remove these from your \
-            Config.toml or run this command with the `-s` option!"
+            Config.toml or run this command with the `--save-config` option!"
                 .red()
                 .bold()
         );

wdl-grammar/src/commands/gauntlet.rs

@@ -58,6 +58,7 @@ type Result<T> = std::result::Result<T, Error>;
 /// configuration itself. Notably, the path to the configuration file should
 /// _not_ be part of the serialized configuration value. Thus, I split the
 /// concept of the path and the actual configuration into two different structs.
+#[derive(Default)]
 pub struct Config {
     /// The path to the configuration file.
     path: PathBuf,

wdl-grammar/src/commands/gauntlet/config.rs

@@ -27,7 +27,7 @@ pub type Repositories = HashSet<repository::Identifier>;
 ///
 /// This object stores the actual configuration values for this subcommand.
 #[serde_as]
-#[derive(Debug, Deserialize, Serialize)]
+#[derive(Debug, Default, Deserialize, Serialize)]
 pub struct Inner {
     /// The WDL version.
     pub(super) version: grammar::Version,

wdl-grammar/src/commands/gauntlet/config/inner.rs

@@ -2,9 +2,12 @@
 
 use std::collections::HashMap;
 
+use colored::Colorize as _;
 use indexmap::IndexMap;
 
-use colored::Colorize as _;
+use wdl_grammar as grammar;
+
+use grammar::core::lint;
 
 use crate::commands::gauntlet::repository;
 use crate::gauntlet::document;
@@ -198,6 +201,21 @@ impl<T: std::io::Write> Report<T> {
         Ok(())
     }
 
+    /// Report a warning for a registered result.
+    pub fn report_warning(&mut self, warning: &lint::Warning) -> std::io::Result<()> {
+        if self.section != Section::Summary {
+            panic!(
+                "cannot report a warning when the report phase is {:?}",
+                self.section
+            );
+        }
+
+        writeln!(self.inner, "  ↳ {}", warning)?;
+        self.printed = true;
+
+        Ok(())
+    }
+
     /// Reports all unexpected errors for a repository report.
     pub fn report_unexpected_errors_for_repository(
         &mut self,
@@ -294,12 +312,12 @@ impl<T: std::io::Write> Report<T> {
         {
             0 => {}
             1 => with.push(String::from("1 mismatch error")),
-            v => with.push(format!("{} mismatched errors", v)),
+            v => with.push(format!("{} mismatch errors", v)),
         };
 
         match results.get(&Status::Warning).copied() {
-            Some(1) => with.push(String::from("1 error with warnings")),
-            Some(v) => with.push(format!("{} errors with warnings", v)),
+            Some(1) => with.push(String::from("1 test containing warnings")),
+            Some(v) => with.push(format!("{} tests containing warnings", v)),
             None => {}
         }