add LLInterpreter.has_pending_stop()

parser/src/constraint.rs

@@ -118,6 +118,10 @@ impl Constraint {
         Ok(())
     }
 
+    pub fn has_pending_stop(&self) -> bool {
+        self.delayed_stop
+    }
+
     /// This computes token sampling mask.
     /// It typically takes up to a millisecond for a 100k tokenizer.
     /// It will return an error when the order of calls is violated.

python/llguidance/_lib.pyi

@@ -135,3 +135,7 @@ class LLInterpreter:
         This is experimental and breaks tests when used instead of post_process().
         """
 
+    def has_pending_stop(self) -> bool:
+        """
+        If true, next mid_process() call will return stop
+        """

rust/src/py.rs

@@ -120,6 +120,10 @@ impl LLInterpreter {
     fn post_process(&mut self, sampled_token: Option<TokenId>) -> PyResult<(u32, Vec<TokenId>)> {
         self.advance_parser(sampled_token)
     }
+
+    fn has_pending_stop(&self) -> bool {
+        self.inner.has_pending_stop()
+    }
 }
 
 #[derive(Serialize, Deserialize)]

scripts/install-deps.sh

@@ -2,5 +2,5 @@
 
 # installing guidance for deps
 pip install pytest guidance huggingface_hub tokenizers jsonschema maturin[zig] \
-    torch transformers bitsandbytes
+    torch transformers bitsandbytes ipython
 pip uninstall -y guidance