Commit b623e51e authored by Wohlgemuth, Jason's avatar Wohlgemuth, Jason
Browse files

feat: Initial check capability for DOCX files

parent 61a07828

acorn-cli/src/cli/arguments.rs

+3 −0
Original line number Diff line number Diff line
@@ -113,6 +113,9 @@ pub enum Standard { 
    /// Dublin Core Metadata Initiative (DCMI)

    #[display("dcmi")]

    Dcmi,

    /// DOCX-derived text

    #[display("docx")]

    Docx,

    /// InvenioRDM

    #[display("invenio")]

    Invenio,

acorn-cli/src/commands/check/mod.rs

+5 −1
Original line number Diff line number Diff line
@@ -8,7 +8,7 @@ use acorn::io::unique_file_extensions; 
use acorn::prelude::{exit, PathBuf};

use acorn::schema::research_activity::ResearchActivity;

use acorn::schema::standard::cff::Cff;

use acorn::schema::standard::text::Text;

use acorn::schema::standard::text::{Docx, Text};

use acorn::util::constants::ENV_READABILITY_METRIC;

use acorn::util::{print_values_as_table, regex_join, Label};

use acorn::{fail, skip};
@@ -92,6 +92,7 @@ fn apply_early_exit_policy(results: Vec<Check>, category: &CheckCategory, option 
async fn collect(paths: &[PathBuf], check_options: &CheckOptions) -> Vec<Check> {

    match infer_standard(paths) {

        | Standard::Cff => collect_checks_for::<Cff>(paths, check_options).await,

        | Standard::Docx => collect_checks_for::<Docx>(paths, check_options).await,

        | Standard::Rads => collect_checks_for::<ResearchActivity>(paths, check_options).await,

        | Standard::Text => collect_checks_for::<Text>(paths, check_options).await,

        | _ => unimplemented!(),
@@ -158,6 +159,9 @@ fn infer_standard(paths: &[PathBuf]) -> Standard { 
    if extensions.len() == 1 && extensions.contains(&"cff".to_string()) {

        warn!("=> {} Inferred standard (CFF)", Label::using());

        Standard::Cff

    } else if extensions.len() == 1 && extensions.contains(&"docx".to_string()) {

        warn!("=> {} Inferred standard (DOCX)", Label::using());

        Standard::Docx

    } else if extensions.len() == 1 && extensions.contains(&"txt".to_string()) {

        warn!("=> {} Inferred standard (Text)", Label::using());

        Standard::Text

acorn-lib/src/analyzer/check.rs

+3 −0
Original line number Diff line number Diff line
@@ -94,6 +94,9 @@ pub enum Standard { 
    /// InvenioRDM

    #[display("invenio")]

    Invenio,

    /// DOCX-derived text

    #[display("docx")]

    Docx,

    /// Plain text

    #[display("text")]

    Text,

acorn-lib/src/analyzer/mod.rs

+36 −1
Original line number Diff line number Diff line
@@ -13,7 +13,7 @@ use crate::prelude::{self, create_dir_all, remove_file, write, Command, CommandO 
use crate::schema::pid::{PersistentIdentifier, PersistentIdentifierParse, DOI};

use crate::schema::research_activity::ResearchActivity;

use crate::schema::standard::cff::{Cff, Identifier, IdentifierType, Reference};

use crate::schema::standard::text::Text;

use crate::schema::standard::text::{Docx, Text};

use crate::schema::{Organization, ProgrammingLanguage, Website};

use crate::util::constants::{APPLICATION, CUSTOM_VALE_PACKAGE_NAME, DEFAULT_VALE_PACKAGE_URL, DEFAULT_VALE_ROOT, VALE_RELEASES_URL, VALE_VERSION};

use crate::util::{is_uri_or_path, Constant, Label, SemanticVersion, StringConversion};
@@ -233,6 +233,41 @@ impl Analysis for Cff { 
    }

}

#[async_trait]

impl Analysis for Docx {

    fn standard() -> Standard {

        Standard::Docx

    }

    async fn check_prose(paths: &[PathBuf], options: Option<&CheckOptions>) -> Vec<Check> {

        check_prose_for::<Self>(paths, options).await

    }

    async fn check_quality(paths: &[PathBuf], _options: Option<&CheckOptions>) -> Vec<Check> {

        paths

            .par_iter()

            .map(|path| match Self::read(path) {

                | Ok(_) => check_ok!(CheckCategory::Quality),

                | Err(why) => check_err!(

                    CheckCategory::Quality,

                    message: "Cannot read DOCX data",

                    context: why.to_string()

                ),

            })

            .collect()

    }

    async fn check_readability(paths: &[PathBuf], options: Option<&CheckOptions>) -> Vec<Check> {

        check_readability_for::<Self>(paths, options)

    }

    async fn check_schema(_paths: &[PathBuf], _options: Option<&CheckOptions>) -> Vec<Check> {

        vec![]

    }

    #[cfg(feature = "std")]

    async fn check_websites(_paths: &[PathBuf], _options: Option<&CheckOptions>) -> Vec<Check> {

        vec![]

    }

    fn output_path(path: &Path, _data: &Self) -> PathBuf {

        standard_project_folder("check", None).join(path.to_path_buf().file_name_with_parent())

    }

}

#[async_trait]

impl Analysis for ResearchActivity {

    fn standard() -> Standard {

        Standard::ResearchActivityData

acorn-lib/src/schema/standard/tests/text.rs

+42 −3
Original line number Diff line number Diff line 
#[cfg(feature = "std")]

use crate::io::InputOutput;

use crate::schema::standard::text::Text;

use crate::schema::standard::text::{Docx, Text};

#[cfg(feature = "std")]

use crate::test_utils::unique_path;

use crate::util::{ToMarkdown, ToProse};

use crate::test_utils::{fixture_path, unique_path};

use crate::util::{Unstructured, ToMarkdown, ToProse};

use validator::Validate;



#[test]
@@ -13,6 +13,7 @@ fn test_text_to_prose_and_markdown_passthrough() { 
    };

    assert_eq!(data.to_prose(), "Line one\n\nLine two");

    assert_eq!(data.to_markdown(), "Line one\n\nLine two");

    assert_eq!(data.content(), "Line one\n\nLine two");

}

#[test]

fn test_text_validate_is_noop() {
@@ -43,3 +44,41 @@ fn test_text_input_output_rejects_unsupported_extension() { 
    let result = Text::read(output);

    assert!(result.is_err());

}



#[test]

fn test_docx_to_prose_and_markdown_passthrough() {

    let data = Docx {

        content: "Line one\n\nLine two".to_string(),

    };

    assert_eq!(data.to_prose(), "Line one\n\nLine two");

    assert_eq!(data.to_markdown(), "Line one\n\nLine two");

    assert_eq!(data.content(), "Line one\n\nLine two");

}

#[test]

fn test_docx_validate_is_noop() {

    let data = Docx {

        content: "any content should validate".to_string(),

    };

    assert!(data.validate().is_ok());

}

#[cfg(feature = "std")]

#[test]

fn test_docx_input_output_read_docx_fixture() {

    let source = fixture_path("acorn.docx");

    let result = Docx::read(source).expect("failed to read docx file");

    assert!(result.content.contains("ACORN"));

}

#[cfg(feature = "std")]

#[test]

fn test_docx_input_output_write_rejects_docx_extension() {

    let output = unique_path("docx-io", "docx");

    if let Some(parent) = output.parent() {

        std::fs::create_dir_all(parent).expect("failed to create test_artifacts directory");

    }

    let source = Docx {

        content: "plain text content".to_string(),

    };

    let result = source.write(output.clone());

    assert!(result.is_err());

    let _cleanup = std::fs::remove_file(output);

}