Update main.py

main.py

@@ -353,21 +353,68 @@ def _filter_context_for_query(context: str, query: str) -> tuple[str, dict]:
     }
 
 # ---------- NEW: intent-specific line extractors (steps/navigation/errors) ----------
-STEP_LINE_REGEX = re.compile(r"^\s*(?:\d+\.\s+|[•\-]\s+)")
+
+STEP_LINE_REGEX = re.compile(r"^\s*(?:\d+[\.\)]\s+|[•\-]\s+)", re.IGNORECASE)
 NAV_LINE_REGEX  = re.compile(r"(navigate\s+to|>\s*)", re.IGNORECASE)
 
+# Common imperative verbs across SOPs (add more if you want, optional)
+PROCEDURE_VERBS = [
+    "log in", "select", "scan", "verify", "confirm", "print",
+    "move", "complete", "click", "open", "navigate", "choose",
+    "enter", "update", "save", "delete", "create", "attach", "assign"
+]
+VERB_START_REGEX = re.compile(r"^\s*(?:" + "|".join([re.escape(v) for v in PROCEDURE_VERBS]) + r")\b", re.IGNORECASE)
+
+# Lines that clearly are NOT steps when user intent is 'steps'
+NON_PROCEDURAL_STARTS = [
+    "to ensure", "as per", "purpose", "pre-requisites", "prerequisites", "overview", "introduction"
+]
+NON_PROC_REGEX = re.compile(r"^\s*(?:" + "|".join([re.escape(v) for v in NON_PROCEDURAL_STARTS]) + r")\b", re.IGNORECASE)
+
+def _is_procedural_line(ln: str) -> bool:
+    """
+    A line is procedural if:
+      - it starts with a number/bullet, OR
+      - it starts with an imperative verb (Log in, Select, Scan, etc.)
+    and it does not look like Purpose/Pre-Requisites/Overview.
+    Bullets are kept only if they contain an action verb (to avoid prereq bullets).
+    """
+    s = (ln or "").strip()
+    if not s:
+        return False
+    # Exclude clearly non-procedural lines
+    if NON_PROC_REGEX.match(s):
+        return False
+
+    # Numbered/bulleted lines
+    if STEP_LINE_REGEX.match(s):
+        # Keep bullet only if an action verb appears somewhere in the line
+        if s.lstrip().startswith(("•", "-")):
+            return bool(VERB_START_REGEX.search(s))
+        return True
+
+    # Imperative verb lines (covers Word lists where 1. isn't part of the text)
+    if VERB_START_REGEX.match(s):
+        return True
+
+    # Allow navigation lines (even if not numbered)
+    if NAV_LINE_REGEX.search(s):
+        return True
+
+    return False
+
 def _extract_steps_only(text: str, max_lines: int = 12) -> str:
     """
-    Keep only numbered/bulleted lines in original order.
-    Accepts formats like '1. ...', '2. ...', '• ...', '- ...'.
+    Keep only procedural lines (numbered/bulleted or imperative verb starts) in original order.
     """
     lines = [ln.strip() for ln in (text or "").splitlines() if ln.strip()]
     kept = []
     for ln in lines:
-        if STEP_LINE_REGEX.match(ln):
+        if _is_procedural_line(ln):
             kept.append(ln)
             if len(kept) >= max_lines:
                 break
+    # If nothing matched (rare), return the original “concise” filtered text
     return "\n".join(kept).strip() if kept else (text or "").strip()
 
 def _extract_navigation_only(text: str, max_lines: int = 6) -> str:
@@ -384,6 +431,7 @@ def _extract_errors_only(text: str, max_lines: int = 10) -> str:
     lines = [ln.strip() for ln in (text or "").splitlines() if ln.strip()]
     kept = []
     for ln in lines:
+        # Keep error/resolution bullets or imperative fixes (verify, check, etc.)
         if STEP_LINE_REGEX.match(ln) or ln.lower().startswith(("error", "resolution", "fix", "verify", "check")):
             kept.append(ln)
             if len(kept) >= max_lines: