Add message auto-censorship

src/groups/calls/events.rs

@@ -1,15 +1,18 @@
-use std::{collections::HashMap, sync::Arc};
+use std::sync::Arc;
 
 use async_trait::async_trait;
-use linkify::LinkFinder;
 use twilight_model::{gateway::payload::incoming::MessageCreate, id::Id};
 use twilight_validate::request::webhook_username;
 
 use crate::{
-    database::models::{Call, Channel}, discord::{images, sanitizing::SanitizeDiscordMarkdown}, framework::{
+    database::models::{Call, Channel},
+    discord::images,
+    framework::{
         events::traits::{EventResult, OnMessageHandler},
         runtime::Runtime,
-    }, groups::calls::moderation::escape_message, state::State
+    },
+    groups::calls::moderation::escape_message,
+    state::State,
 };
 
 #[derive(Debug)]

src/groups/calls/lists/suggestive-language.txt

@@ -0,0 +1 @@
+slap that ass

src/groups/calls/moderation.rs

@@ -62,11 +62,11 @@ pub fn escape_message(event: &MessageCreate) -> String {
         message = message.replace(url, &"\\#".repeat(url.len()));
     }
 
-    let normalized = normalize(message.clone());
+    let (censored, rules) = moderate_message(&message);
 
-    log::debug!("Escaped and normalized message: {normalized:?}");
+    log::debug!("Escaped and normalized message: {censored:?}");
 
-    message.sanitize(false)
+    censored
 }
 
 #[derive(Debug)]
@@ -113,13 +113,16 @@ pub fn normalize(message: String) -> NormalizedMessage {
         for char in unidecoded.chars() {
             if char.is_alphanumeric() {
                 normalized.push(char);
-                mappings.push(i);
+            } else {
+                normalized.push(' ');
             }
+
+            mappings.push(i);
         }
     }
 
     NormalizedMessage {
-        message: normalized,
+        message: normalized.to_ascii_lowercase(),
         mappings,
     }
 }
@@ -127,8 +130,12 @@ pub fn normalize(message: String) -> NormalizedMessage {
 fn regex_from_phrase(phrase: &str) -> String {
     let mut pattern = String::new();
 
-    for ch in phrase.replace(" ", "").chars() {
+    for (i, ch) in phrase.replace(" ", "").chars().enumerate() {
-        pattern.push_str(&format!("{ch}+\\s*"));
+        if i == phrase.len() - 1 {
+            pattern.push_str(&format!("{ch}+"));
+        } else {
+            pattern.push_str(&format!("{ch}+\\s*"));
+        }
     }
 
     let pattern = format!("\\b{pattern}\\b");
@@ -170,19 +177,44 @@ pub fn censor_message(
 ) -> String {
     let mut censored = original.to_string();
 
-    for (findings, rule) in analysis {
+    // Collect all ranges to censor
-        for finding in findings {
+    let mut ranges = Vec::new();
-            while let Some(start) = normalized.message.find(finding) {
-                let end = start + finding.len();
 
-                todo!();
+    for (findings, _) in analysis {
+        for finding in findings {
+            for (index, _) in normalized.message.match_indices(finding) {
+                let start_index = normalized.mappings.get(index).cloned().unwrap_or(0);
+                let end_index = normalized.mappings.get(index + finding.len()).cloned().unwrap_or(original.len());
+
+                ranges.push((start_index, end_index));
             }
         }
     }
 
+    // Sort ranges in reverse order to avoid shifting indices
+    ranges.sort_by(|a, b| b.0.cmp(&a.0));
+
+    for (start_index, end_index) in ranges {
+        log::trace!("Censoring from {start_index} to {end_index}");
+        if start_index < end_index && end_index <= censored.len() {
+            censored.replace_range(start_index..end_index, "[censored]");
+        }
+    }
+
     censored
 }
 
+pub fn moderate_message(original: &str) -> (String, Vec<Arc<dyn Rule>>) {
+    let normalized = normalize(original.to_string());
+
+    log::trace!("Normalized message: {:?}", normalized.message);
+
+    let analysis = analyze_message(&normalized.message);
+    let censored = censor_message(original, normalized, &analysis);
+
+    (censored, analysis.iter().map(|(_, r)| r.clone()).collect())
+}
+
 #[derive(Debug, PartialEq, Eq, Clone, Copy)]
 pub enum Action {
     BlockAndPenalty,
@@ -218,6 +250,8 @@ impl HateSpeechAndSlursRule {
 
         let regex = regex_from_phrases_list(&phrases);
 
+        log::trace!("Hate Speech and Slurs Rule regex: {regex}");
+
         Self(regex)
     }
 }
@@ -248,6 +282,8 @@ impl ThreatsAndSelfHarmRule {
 
         let regex = regex_from_phrases_list(&phrases);
 
+        log::trace!("Threats and Self Harm Rule regex: {regex}");
+
         Self(regex)
     }
 }
@@ -278,6 +314,8 @@ impl ExplicitSexualLanguageRule {
 
         let regex = regex_from_phrases_list(&phrases);
 
+        log::trace!("Explicit Sexual Language Rule regex: {regex}");
+
         Self(regex)
     }
 }
@@ -307,6 +345,8 @@ impl SuggestiveLanguageRule {
 
         let regex = regex_from_phrases_list(&phrases);
 
+        log::trace!("Suggestive Language Rule regex: {regex}");
+
         Self(regex)
     }
 }