fix: more validator fixes

Cargo.toml

@@ -124,6 +124,7 @@ lettre = { version = "0.11.19", default-features = false, features = [
   "tokio1",
   "tokio1-rustls",
 ] }
+lingua = "=1.8.0"
 linkify = "0.11.0"
 lz4_flex = { version = "0.11.5", default-features = false, features = [
   "checked-decode",
@@ -245,7 +246,6 @@ webauthn-rs = "0.5.5"
 webauthn-rs-proto = "0.5.5"
 webp = { version = "0.3.1", default-features = false }
 webview2-com = "0.38.0"  # Should be updated in lockstep with wry
-whatlang = "0.18.0"
 whoami = "1.6.1"
 windows = "=0.61.3"  # Locked on 0.61 until we can update windows-core to 0.62
 windows-core = "=0.61.2"  # Locked on 0.61 until webview2-com updates to 0.62

apps/frontend/src/components/ui/moderation/ModerationProjectNags.vue

@@ -142,6 +142,7 @@ import {
 	type MessageDescriptor,
 	useVIntl,
 } from '@modrinth/ui'
+import { isStaff } from '@modrinth/utils'
 import type { Component } from 'vue'
 import { computed, nextTick, onBeforeUnmount, onMounted, ref, watch } from 'vue'
 
@@ -223,6 +224,7 @@ const messages = defineMessages({
 
 const { formatMessage } = useVIntl()
 const { addNotification } = injectNotificationManager()
+const flags = useFeatureFlags()
 
 const props = withDefaults(defineProps<Props>(), {
 	versions: () => [],
@@ -420,6 +422,15 @@ function isNagComplete(nag: Nag): boolean {
 const visibleNags = computed<Nag[]>(() => {
 	const finalNags = applicableNags.value.filter((nag) => !isNagComplete(nag))
 
+	if (
+		isProcessing.value &&
+		isStaff(props.currentMember?.user) &&
+		!flags.value.alwaysShowPublishingChecklistForStaff &&
+		!finalNags.some((nag) => nag.status === 'required')
+	) {
+		return []
+	}
+
 	if (props.project.status === 'draft') {
 		finalNags.push({
 			id: 'submit-for-review',

apps/frontend/src/composables/featureFlags.ts

@@ -53,6 +53,7 @@ export const DEFAULT_FEATURE_FLAGS = validateValues({
 	alwaysIgnoreErrorBanner: false,
 	showViewProdRouteBanner: false,
 	showModeratorProjectMemberUi: false,
+	alwaysShowPublishingChecklistForStaff: false,
 	archonApiStaging: false,
 	showHostingAccessInstanceAuditLog: false,
 	versionDevInfoCollapsed: true,

apps/labrinth/Cargo.toml

@@ -72,6 +72,7 @@ image = { workspace = true, features = [
 itertools = { workspace = true }
 json-patch = { workspace = true }
 lettre = { workspace = true }
+lingua = { workspace = true }
 linkify = { workspace = true }
 modrinth-content-management = { workspace = true }
 modrinth-util = { workspace = true, features = ["decimal", "sentry", "utoipa"] }
@@ -134,7 +135,6 @@ validator = { workspace = true, features = ["derive"] }
 webauthn-rs = { workspace = true, features = ["conditional-ui", "danger-allow-state-serialisation"] }
 webauthn-rs-proto = { workspace = true }
 webp = { workspace = true }
-whatlang = { workspace = true }
 woothee = { workspace = true }
 xredis = { workspace = true }
 yaserde = { workspace = true, features = ["derive"] }

apps/labrinth/src/routes/v3/projects/validate.rs

@@ -67,12 +67,17 @@ pub(crate) async fn ensure_project_is_valid_for_review(
     .collect::<Vec<_>>();
     let project = Project::from(reloaded_project.clone());
 
-    if has_required_nags_with_context(
-        &project,
-        &versions,
-        &available_categories,
-        &disclosures,
-    ) {
+    let has_required_nags = web::block(move || {
+        has_required_nags_with_context(
+            &project,
+            &versions,
+            &available_categories,
+            &disclosures,
+        )
+    })
+    .await
+    .wrap_internal_err("validating project for review")?;
+    if has_required_nags {
         return Err(ApiError::Request(eyre!(
             "project must have no required validation nags before or while under review"
         )));
@@ -157,12 +162,15 @@ pub async fn validate(
     .collect::<Vec<_>>();
     let project = Project::from(project);
 
-    Ok(web::Json(ProjectValidationResponse {
-        nags: validate_project(
+    let nags = web::block(move || {
+        validate_project(
             &project,
             &versions,
             &available_categories,
             &disclosures,
-        ),
-    }))
+        )
+    })
+    .await
+    .wrap_internal_err("validating project")?;
+    Ok(web::Json(ProjectValidationResponse { nags }))
 }

apps/labrinth/src/validate/project/description.rs

@@ -7,14 +7,17 @@ use super::text::{
     ProfanityKind, contains_description_spam, extract_description_blocks,
     extract_description_text, find_banned_description_link,
     has_image_without_alt_text, has_sufficient_english_blocks,
-    js_string_length, non_standard_text_ratio, normalize_project_field_text,
-    profanity_matches, project_requires_english,
+    is_confidently_non_english_short_text, js_string_length,
+    non_standard_text_ratio, normalize_project_field_text, profanity_matches,
+    project_requires_english, project_text_similarity,
 };
 use super::{ProjectNag, ProjectNagKind, ProjectNagSeverity};
 
 use crate::models::projects::Project;
 
 const MIN_DESCRIPTION_CHARS: usize = 125;
+const MIN_DESCRIPTION_LANGUAGE_CHARS: usize = 35;
+const MAX_DESCRIPTION_SUMMARY_SIMILARITY: f64 = 0.8;
 const MAX_PROFANITY_COUNT: usize = 2;
 const NON_STANDARD_TEXT_FAILURE_THRESHOLD: f64 = 0.05;
 
@@ -91,6 +94,14 @@ pub(super) fn validate(project: &Project) -> Vec<ProjectNag> {
             );
         }
     }
+    if project_text_similarity(&normalized_text, &project.summary)
+        >= MAX_DESCRIPTION_SUMMARY_SIMILARITY
+    {
+        nags.push(ProjectNag::new(
+            ProjectNagKind::ProjectDescriptionMatchesSummary,
+            ProjectNagSeverity::Required,
+        ));
+    }
     if has_spam {
         nags.push(ProjectNag::new(
             ProjectNagKind::ProjectDescriptionSpam,
@@ -151,8 +162,11 @@ fn is_non_english_text(
     text: &str,
     blocks: &[String],
 ) -> bool {
+    let length = js_string_length(text);
     project_requires_english(project)
-        && js_string_length(text) >= MIN_DESCRIPTION_CHARS
+        && length >= MIN_DESCRIPTION_LANGUAGE_CHARS
+        && (length >= MIN_DESCRIPTION_CHARS
+            || is_confidently_non_english_short_text(&blocks.join(" ")))
         && !has_sufficient_english_blocks(blocks)
 }
 

apps/labrinth/src/validate/project/description/markdown.rs

@@ -6,10 +6,10 @@ use regex::Regex;
 use unicode_segmentation::UnicodeSegmentation;
 
 static HTML_HEADER: LazyLock<Regex> = LazyLock::new(|| {
-    Regex::new(r"(?is)<h[1-3]\b[^>]*>(.*?)</h[1-3]>").unwrap()
+    Regex::new(r"(?is)<h[1-6]\b[^>]*>(.*?)</h[1-6]>").unwrap()
 });
 static ADJACENT_HTML_HEADERS: LazyLock<Regex> =
-    LazyLock::new(|| Regex::new(r"(?is)</h([1-3])>\s*<h([1-3])\b").unwrap());
+    LazyLock::new(|| Regex::new(r"(?is)</h([1-6])>\s*<h([1-6])\b").unwrap());
 static TRAILING_HTML_HEADER: LazyLock<Regex> = LazyLock::new(|| {
     Regex::new(r"(?is)</h[1-6]>\s*(?:</[a-z][^>]*>\s*)*$").unwrap()
 });
@@ -101,7 +101,6 @@ impl<'a> DescriptionMarkdown<'a> {
         let markdown_headers = self
             .headings
             .iter()
-            .filter(|heading| is_primary_heading(heading.level))
             .filter(|heading| header_is_long(&heading.text))
             .count();
         let without_code = self.replace_code(" ");
@@ -130,8 +129,7 @@ impl<'a> DescriptionMarkdown<'a> {
                 let [previous, current] = headings else {
                     return false;
                 };
-                is_primary_heading(previous.level)
-                    && previous.level == current.level
+                previous.level == current.level
                     && self.markdown[previous.range.end..current.range.start]
                         .trim()
                         .is_empty()
@@ -180,13 +178,6 @@ fn header_is_long(header: &str) -> bool {
     rendered.graphemes(true).count() > 80
 }
 
-fn is_primary_heading(level: HeadingLevel) -> bool {
-    matches!(
-        level,
-        HeadingLevel::H1 | HeadingLevel::H2 | HeadingLevel::H3
-    )
-}
-
 #[cfg(test)]
 mod tests {
     use super::DescriptionMarkdown;
@@ -250,4 +241,44 @@ image: "![](/missing-alt.png)"
                 .has_adjacent_same_level_headers()
         );
     }
+
+    #[test]
+    fn all_heading_levels_are_validated() {
+        for level in 1..=6 {
+            let prefix = "#".repeat(level);
+            let long_text = "heading ".repeat(12);
+            for (long_header, adjacent_headers, separated_headers) in [
+                (
+                    format!("{prefix} {long_text}"),
+                    format!("{prefix} First\n\n{prefix} Second"),
+                    format!(
+                        "{prefix} First\n\n```yaml\n# comment\n```\n\n{prefix} Second"
+                    ),
+                ),
+                (
+                    format!("<h{level}>{long_text}</h{level}>"),
+                    format!(
+                        "<h{level}>First</h{level}>\n\n<h{level}>Second</h{level}>"
+                    ),
+                    format!(
+                        "<h{level}>First</h{level}>\n\n```yaml\n# comment\n```\n\n<h{level}>Second</h{level}>"
+                    ),
+                ),
+            ] {
+                let markdown = DescriptionMarkdown::parse(&long_header);
+                assert_eq!(markdown.long_header_count(), 1, "{long_header}");
+                assert!(markdown.ends_with_header(), "{long_header}");
+                assert!(
+                    DescriptionMarkdown::parse(&adjacent_headers)
+                        .has_adjacent_same_level_headers(),
+                    "{adjacent_headers}"
+                );
+                assert!(
+                    !DescriptionMarkdown::parse(&separated_headers)
+                        .has_adjacent_same_level_headers(),
+                    "{separated_headers}"
+                );
+            }
+        }
+    }
 }

apps/labrinth/src/validate/project/disclosures.rs

@@ -7,15 +7,18 @@ pub(super) fn validate(
     project: &crate::models::projects::Project,
     disclosures: Option<&[ProjectDisclosure]>,
 ) -> Vec<ProjectNag> {
-    let (project_type, _) =
-        LegacyProject::get_project_type(&project.project_types);
-    let mut nags = vec![
-        ProjectNag::new(
-            ProjectNagKind::CheckDisclosures,
-            ProjectNagSeverity::Suggestion,
-        )
-        .with_details(serde_json::json!({ "project_type": project_type })),
-    ];
+    let mut nags = Vec::new();
+    if disclosures.is_some_and(|disclosures| disclosures.is_empty()) {
+        let (project_type, _) =
+            LegacyProject::get_project_type(&project.project_types);
+        nags.push(
+            ProjectNag::new(
+                ProjectNagKind::CheckDisclosures,
+                ProjectNagSeverity::Suggestion,
+            )
+            .with_details(serde_json::json!({ "project_type": project_type })),
+        );
+    }
 
     if disclosures.is_some_and(|disclosures| {
         disclosures.iter().any(disclosure_has_paired_html)