revert: undo 2 main-only commits to unblock staging-promote merge (#2297)

Reverts:
- 6f7575de Fix Telegram UTF-16 message splitting (#1961)
- 7be3b910 [codex] Label migration PRs with DB MIGRATION (#1967)

Keeps f0db0a3d (registry version bumps) intact.

These changes were made directly on main and conflict with staging-promote.
Both already exist in staging and will flow back to main via the promote merge.

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.github/labeler.yml

@@ -1,5 +1,5 @@
-# Labels for actions/labeler@v6
-# Maps file path globs to labels. Multiple labels can apply per PR.
+# Scope labels for actions/labeler@v6
+# Maps file path globs to scope labels. Multiple labels can apply per PR.
 
 "scope: agent":
   - changed-files:
@@ -164,9 +164,3 @@
       - any-glob-to-any-file:
           - Cargo.toml
           - Cargo.lock
-
-"DB MIGRATION":
-  - changed-files:
-      - any-glob-to-any-file:
-          - migrations/**
-          - src/db/libsql_migrations.rs

.github/scripts/create-labels.sh

@@ -62,9 +62,6 @@ create "scope: ci"            "546E7A" "CI/CD workflows"
 create "scope: docs"          "78909C" "Documentation"
 create "scope: dependencies"  "90A4AE" "Dependency updates"
 
-echo "==> Creating coordination labels..."
-create "DB MIGRATION"         "C62828" "PR adds or modifies PostgreSQL or libSQL migration definitions"
-
 echo "==> Creating workflow labels..."
 create "skip-regression-check" "9E9E9E" "Acknowledged: fix without regression test"
 

.github/workflows/pr-label-scope.yml

@@ -6,20 +6,13 @@ on:
 
 permissions:
   contents: read
-  issues: write
   pull-requests: write
 
 jobs:
   scope:
     runs-on: ubuntu-latest
     steps:
-      - name: Ensure DB MIGRATION label exists
-        env:
-          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          REPO: ${{ github.repository }}
-        run: gh label create "DB MIGRATION" --repo "$REPO" --color C62828 --description "PR adds or modifies PostgreSQL or libSQL migration definitions" --force
-
-      - uses: actions/labeler@8558fd74291d67161a8a78ce36a881fa63b766a9 # v5
+      - uses: actions/labeler@v5
         with:
           configuration-path: .github/labeler.yml
           sync-labels: false          # additive only — never remove scope labels

channels-src/telegram/src/lib.rs

@@ -363,26 +363,6 @@ const TELEGRAM_STATUS_MAX_CHARS: usize = 600;
 /// Telegram's hard limit for message text length.
 const TELEGRAM_MAX_MESSAGE_LEN: usize = 4096;
 
-fn utf16_code_unit_len(text: &str) -> usize {
-    text.encode_utf16().count()
-}
-
-fn prefix_within_utf16_limit(text: &str, max_units: usize) -> usize {
-    let mut units = 0;
-    let mut end = 0;
-
-    for (byte_idx, ch) in text.char_indices() {
-        let ch_units = ch.len_utf16();
-        if units + ch_units > max_units {
-            break;
-        }
-        units += ch_units;
-        end = byte_idx + ch.len_utf8();
-    }
-
-    end
-}
-
 fn truncate_status_message(input: &str, max_chars: usize) -> String {
     let mut iter = input.chars();
     let truncated: String = iter.by_ref().take(max_chars).collect();
@@ -393,7 +373,7 @@ fn truncate_status_message(input: &str, max_chars: usize) -> String {
     }
 }
 
-/// Split a long message into chunks that fit within Telegram's 4096 UTF-16-unit limit.
+/// Split a long message into chunks that fit within Telegram's 4096-char limit.
 ///
 /// Tries to split at the most natural boundary available (in priority order):
 /// 1. Double newline (paragraph break)
@@ -402,7 +382,7 @@ fn truncate_status_message(input: &str, max_chars: usize) -> String {
 /// 4. Word boundary (space)
 /// 5. Hard cut at the limit (last resort for pathological input)
 fn split_message(text: &str) -> Vec<String> {
-    if utf16_code_unit_len(text) <= TELEGRAM_MAX_MESSAGE_LEN {
+    if text.chars().count() <= TELEGRAM_MAX_MESSAGE_LEN {
         return vec![text.to_string()];
     }
 
@@ -410,8 +390,13 @@ fn split_message(text: &str) -> Vec<String> {
     let mut remaining = text;
 
     while !remaining.is_empty() {
-        // Find the longest UTF-8 prefix that fits within Telegram's UTF-16 limit.
-        let window_bytes = prefix_within_utf16_limit(remaining, TELEGRAM_MAX_MESSAGE_LEN);
+        // Count chars to find the byte offset for our window.
+        let window_bytes = remaining
+            .char_indices()
+            .take(TELEGRAM_MAX_MESSAGE_LEN)
+            .last()
+            .map(|(byte_idx, ch)| byte_idx + ch.len_utf8())
+            .unwrap_or(remaining.len());
 
         if window_bytes >= remaining.len() {
             // Remainder fits entirely.
@@ -419,24 +404,10 @@ fn split_message(text: &str) -> Vec<String> {
             break;
         }
 
-        if window_bytes == 0 {
-            // Defensive fallback: make progress even if a future caller uses a
-            // smaller limit than a single scalar value can fit within.
-            let first_char_len = remaining
-                .chars()
-                .next()
-                .map(|ch| ch.len_utf8())
-                .unwrap_or(remaining.len());
-            chunks.push(remaining[..first_char_len].to_string());
-            remaining = &remaining[first_char_len..];
-            continue;
-        }
-
         let window = &remaining[..window_bytes];
 
         // 1. Double newline — best paragraph boundary
-        let split_at = window
-            .rfind("\n\n")
+        let split_at = window.rfind("\n\n")
             // 2. Single newline
             .or_else(|| window.rfind('\n'))
             // 3. Sentence-ending punctuation followed by space.
@@ -446,9 +417,9 @@ fn split_message(text: &str) -> Vec<String> {
             .or_else(|| {
                 let bytes = window.as_bytes();
                 // Search backwards for '. ', '! ', '? '
-                (1..bytes.len())
-                    .rev()
-                    .find(|&i| matches!(bytes[i - 1], b'.' | b'!' | b'?') && bytes[i] == b' ')
+                (1..bytes.len()).rev().find(|&i| {
+                    matches!(bytes[i - 1], b'.' | b'!' | b'?') && bytes[i] == b' '
+                })
             })
             // 4. Word boundary (last space)
             .or_else(|| window.rfind(' '))
@@ -456,11 +427,7 @@ fn split_message(text: &str) -> Vec<String> {
             .unwrap_or(window_bytes);
 
         // Avoid empty chunks (e.g. text starting with \n\n).
-        let split_at = if split_at == 0 {
-            window_bytes
-        } else {
-            split_at
-        };
+        let split_at = if split_at == 0 { window_bytes } else { split_at };
 
         // Trim whitespace at chunk boundaries for clean Telegram display.
         // Note: this drops leading/trailing spaces at split points, which is
@@ -1391,13 +1358,7 @@ fn send_response(
 
     for (i, chunk) in chunks.into_iter().enumerate() {
         // Try Markdown, fall back to plain text on parse errors
-        let result = send_message(
-            chat_id,
-            &chunk,
-            reply_to,
-            Some("Markdown"),
-            message_thread_id,
-        );
+        let result = send_message(chat_id, &chunk, reply_to, Some("Markdown"), message_thread_id);
 
         let msg_id = match result {
             Ok(id) => {
@@ -2260,10 +2221,6 @@ export!(TelegramChannel);
 mod tests {
     use super::*;
 
-    fn utf16_len(text: &str) -> usize {
-        text.encode_utf16().count()
-    }
-
     #[test]
     fn test_split_message_short() {
         let text = "Hello, world!";
@@ -2291,7 +2248,7 @@ mod tests {
         let chunks = split_message(&text);
         assert!(chunks.len() > 1, "expected multiple chunks");
         for chunk in &chunks {
-            assert!(utf16_len(chunk) <= TELEGRAM_MAX_MESSAGE_LEN);
+            assert!(chunk.chars().count() <= TELEGRAM_MAX_MESSAGE_LEN);
         }
         // Rejoined chunks must equal the original text exactly.
         let rejoined = chunks.join(" ");
@@ -2307,7 +2264,7 @@ mod tests {
         assert!(text.len() > TELEGRAM_MAX_MESSAGE_LEN);
         let chunks = split_message(&text);
         for chunk in &chunks {
-            assert!(utf16_len(chunk) <= TELEGRAM_MAX_MESSAGE_LEN);
+            assert!(chunk.chars().count() <= TELEGRAM_MAX_MESSAGE_LEN);
         }
     }
 
@@ -2337,7 +2294,7 @@ mod tests {
         let chunks = split_message(&text);
         assert!(chunks.len() >= 2);
         for chunk in &chunks {
-            assert!(utf16_len(chunk) <= TELEGRAM_MAX_MESSAGE_LEN);
+            assert!(chunk.chars().count() <= TELEGRAM_MAX_MESSAGE_LEN);
         }
         // Rejoined must preserve all characters
         let rejoined: String = chunks.concat();
@@ -2354,25 +2311,12 @@ mod tests {
         let chunks = split_message(&text);
         assert!(chunks.len() >= 2);
         for chunk in &chunks {
-            assert!(utf16_len(chunk) <= TELEGRAM_MAX_MESSAGE_LEN);
+            assert!(chunk.chars().count() <= TELEGRAM_MAX_MESSAGE_LEN);
             // Every char should be a complete emoji
             assert!(chunk.chars().all(|c| c == '\u{1F600}'));
         }
     }
 
-    #[test]
-    fn test_split_message_exact_utf16_limit_for_surrogate_pairs() {
-        let emoji = "\u{1F600}"; // 😀
-        let text = emoji.repeat(TELEGRAM_MAX_MESSAGE_LEN);
-
-        let chunks = split_message(&text);
-
-        assert_eq!(chunks.len(), 2);
-        assert!(chunks
-            .iter()
-            .all(|chunk| utf16_len(chunk) <= TELEGRAM_MAX_MESSAGE_LEN));
-    }
-
     #[test]
     fn test_clean_message_text() {
         // Without bot_username: strips any leading @mention