Update src/streamlit_app.py

src/streamlit_app.py

@@ -2,13 +2,15 @@
 Health Matrix – AI Command Center (Light UI)
 Single-file Streamlit app with embedded CSS + UKG helpers.
 - Keeps all original functionality
-- New Light UI cards/timeline inspired by your style (21).css
+- Business Case section + Agent Runline (vertical execution log)
+- Extended shift criteria (role/specialty/mandatory training/ward/language)
 Usage: streamlit run merged_app.py
 """
 
 import os
 import json
 import pathlib
+import time
 from typing import List, Optional, Iterable, Dict, Any
 from io import StringIO
 
@@ -44,7 +46,7 @@ p,li{color:var(--t2);font-size:1.05em}
 .stDataFrame th{background:#003b70 !important;color:#fff !important}
 
 /* ================= Hero ================= */
-.hero{position:relative;text-align:center;padding:72px 0 32px;overflow:hidden}
+.hero{position:relative;text-align:center;padding:72px 0 18px;overflow:hidden}
 .hero img{width:200px;height:auto;margin-bottom:14px}
 .hero h1{font-size:34px;line-height:42px;font-weight:900;margin:0 0 10px}
 .ai-headline{
@@ -52,7 +54,7 @@ p,li{color:var(--t2);font-size:1.05em}
   -webkit-background-clip:text;background-clip:text;-webkit-text-fill-color:transparent;
   background-size:200% 100%;animation:aiTitle 8s ease-in-out infinite;
 }
-.hero p{font-size:16px;line-height:26px;color:var(--t2);max-width:760px;margin:0 auto}
+.hero p{font-size:16px;line-height:26px;color:var(--t2);max-width:760px;margin:0 auto 8px}
 .hero::before,.hero::after{content:"";position:absolute;inset:-20%;pointer-events:none;filter:blur(42px);opacity:.18}
 .hero::before{
   background:
@@ -82,7 +84,7 @@ p,li{color:var(--t2);font-size:1.05em}
   box-shadow:0 0 0 0 rgba(54,186,1,.6);animation:pulse 1.4s infinite}
 .status-done{background:#E8F7ED;border-color:#B6E2C4}.status-done .pulse-dot{background:var(--ok);box-shadow:none}
 
-/* ================= Mini Timeline ================= */
+/* ================= Mini Timeline (kept for reuse) ================= */
 .ai-timeline-light{display:flex;justify-content:space-between;gap:10px;margin:18px 0 8px}
 .ai-timeline-light .step{flex:1;position:relative;text-align:center;color:var(--t1);font-size:12px;padding-top:18px}
 .ai-timeline-light .step::before{content:"";position:absolute;top:0;left:50%;transform:translateX(-50%);
@@ -102,6 +104,20 @@ p,li{color:var(--t2);font-size:1.05em}
 .stat-icon{font-size:24px;margin-bottom:.2rem;color:var(--blue)}
 .stat-label{color:#334155;font-size:.95rem}.stat-value{font-size:1.6rem;font-weight:800;color:var(--t1)}
 
+/* === Agent Runline (vertical execution log) === */
+.run-wrap{margin:10px 0 20px}
+.run-legend{display:flex;gap:8px;flex-wrap:wrap;margin:6px 0 10px}
+.run-legend .chip{font-size:12px;padding:6px 10px;border:1px solid var(--line);border-radius:999px;background:#fff}
+.run{position:relative;margin:0;padding:0 0 0 18px;list-style:none}
+.run::before{content:"";position:absolute;left:6px;top:0;bottom:0;width:2px;background:#E2E8F0}
+.run li{position:relative;margin:0 0 12px;padding:8px 12px 8px 12px;background:#fff;border:1px solid var(--line);border-radius:12px;box-shadow:0 6px 14px rgba(0,0,0,.05)}
+.run li::before{content:"";position:absolute;left:-13px;top:12px;width:12px;height:12px;border-radius:50%;background:#94A3B8;border:2px solid #fff;box-shadow:0 0 0 2px #E2E8F0}
+.run li.done::before{background:var(--green)}
+.run li.current::before{background:var(--blue)}
+.run h5{margin:0 0 4px;font-size:14px}
+.run .meta{display:flex;gap:10px;font-size:12px;color:#64748B}
+.run .desc{margin-top:6px;color:#334155;font-size:13px}
+
 /* ================= Animations ================= */
 @keyframes aiGradient{0%{background-position:0% 50%}50%{background-position:100% 50%}100%{background-position:0% 50%}}
 @keyframes aiTitle{0%{background-position:0 0}50%{background-position:100% 0}100%{background-position:0 0}}
@@ -253,36 +269,75 @@ def fetch_employees(employee_ids: Iterable[int]) -> pd.DataFrame:
     df = pd.DataFrame(records)
     for col in ["personNumber", "organizationPath", "phoneNumber", "fullName", "Certifications"]:
         if col not in df.columns: df[col] = []
+    # Optionals for matching
+    if "Languages" not in df.columns:
+        df["Languages"] = [[] for _ in range(len(df))]
     df["JobRole"] = df["organizationPath"].apply(lambda p: p.split("/")[-1] if isinstance(p, str) and p else "")
     return df
 
 # =============================================================================
 # Decision logic + timelines
 # =============================================================================
+def _has_any(tokens: List[str], haystack: str) -> bool:
+    hay = (haystack or "").lower()
+    return any(t.strip().lower() in hay for t in tokens if t and t.strip())
+
+def _has_all_in_list(need: List[str], have: List[str]) -> bool:
+    have_l = [h.lower() for h in (have or [])]
+    return all(n.strip().lower() in have_l for n in need if n and n.strip())
+
 def is_eligible(row: pd.Series, shift: pd.Series) -> bool:
-    role_match = row.get("JobRole", "").strip().lower() == shift["RequiredSkill"].strip().lower()
-    cert_ok = shift["RequiredCert"].strip().lower() in [c.lower() for c in row.get("Certifications", [])]
-    return role_match or cert_ok  # keep original relaxed rule
+    """Eligibility using expanded criteria."""
+    role_req = str(shift.get("RoleRequired","")).strip().lower()
+    specialty = str(shift.get("Specialty","")).replace("-", " ").strip().lower()
+    mandatory = [t.strip() for t in str(shift.get("MandatoryTraining","")).replace(",", "/").split("/") if t.strip()]
+    ward = str(shift.get("WardDepartment","")).strip().lower()
+    lang_req = str(shift.get("LanguageRequirement","")).strip().lower()
+
+    role_ok = (row.get("JobRole","").strip().lower() == role_req) or _has_any([role_req], row.get("organizationPath",""))
+    spec_ok = (not specialty) or _has_any([specialty], row.get("JobRole","")) \
+              or _has_any([specialty], row.get("organizationPath","")) \
+              or _has_any([specialty], " ".join(row.get("Certifications", [])))
+    training_ok = (not mandatory) or _has_all_in_list(mandatory, row.get("Certifications", []))
+    ward_ok = (not ward) or _has_any([ward], row.get("organizationPath",""))
+    lang_ok = (not lang_req) or (lang_req in [l.lower() for l in (row.get("Languages") or [])])
+
+    return role_ok and spec_ok and training_ok and ward_ok and lang_ok
 
 def gpt_decide(shift: pd.Series, eligible_df: pd.DataFrame) -> Dict[str, str]:
     if not HAS_OPENAI or openai is None or not os.getenv("OPENAI_API_KEY"):
         return {"action": "skip"}
     emp_list = []
     for _, r in eligible_df.iterrows():
-        role_match = shift['RequiredSkill'] in (r.get("organizationPath") or "")
-        cert_ok = shift['RequiredCert'] in (r.get("Certifications") or [])
+        role_match = str(shift.get('RoleRequired','')) in (r.get("organizationPath") or "") or \
+                     (r.get("JobRole","") == str(shift.get('RoleRequired','')))
+        mand_tokens = [t.strip() for t in str(shift.get("MandatoryTraining","")).replace(",", "/").split("/") if t.strip()]
+        cert_ok = _has_all_in_list(mand_tokens, r.get("Certifications") or [])
+        score = int(bool(role_match)) + int(bool(cert_ok))
         emp_list.append({
-            "fullName": r["fullName"], "phoneNumber": r["phoneNumber"],
-            "organizationPath": r["organizationPath"], "Certifications": r["Certifications"],
-            "matchScore": int(role_match) + int(cert_ok)
+            "fullName": r.get("fullName",""),
+            "phoneNumber": r.get("phoneNumber",""),
+            "organizationPath": r.get("organizationPath",""),
+            "Certifications": r.get("Certifications",[]),
+            "matchScore": score
         })
     prompt = f"""
 أنت مساعد ذكي للمناوبات:
-تفاصيل الشفت: القسم: {shift['Department']} | المهارة: {shift['RequiredSkill']} | الشهادة: {shift['RequiredCert']} | الوقت: {shift['ShiftTime']}
-الموظفون المؤهلون جزئياً/كلياً: {emp_list if emp_list else 'لا يوجد'}
-اختر بتنسيق JSON فقط:
+تفاصيل الشفت:
+- القسم: {shift.get('Department','')}
+- الدور المطلوب: {shift.get('RoleRequired','')}
+- التخصص/الخبرة: {shift.get('Specialty','')}
+- التدريب الإلزامي: {shift.get('MandatoryTraining','')}
+- نوع الشفت/المدة: {shift.get('ShiftType','')} / {shift.get('ShiftDuration','')}
+- القسم/الورد: {shift.get('WardDepartment','')}
+- اللغة المطلوبة: {shift.get('LanguageRequirement','')}
+- الوقت: {shift.get('ShiftTime','')}
+
+الموظفون المؤهلون (matchScore=0..2): {emp_list if emp_list else 'لا يوجد'}
+
+اختر JSON فقط:
 {{"action":"assign","employee":"الاسم"}} أو {{"action":"notify","employee":"الاسم"}} أو {{"action":"skip"}}
-    """.strip()
+""".strip()
     try:
         resp = openai.ChatCompletion.create(
             model="gpt-4",
@@ -297,12 +352,11 @@ def gpt_decide(shift: pd.Series, eligible_df: pd.DataFrame) -> Dict[str, str]:
         return {"action": "skip"}
 
 def render_mini_timeline(events: List[Dict[str, str]]) -> str:
-    # Convert events to a compact “done/next” mini timeline
+    # kept in case you want the compact view
     steps = []
     for i, ev in enumerate(events):
         title = ev.get("title","")
         done = "done" if i < len(events)-1 else ""  # last is current
-        # short label
         label = (title.replace("🔍","").replace("✅","").replace("📋","")
                       .replace("🤖","").replace("📬","").strip())
         steps.append(f'<div class="step {done}">{label}</div>')
@@ -312,57 +366,94 @@ def render_mini_timeline(events: List[Dict[str, str]]) -> str:
     '''
     return html
 
+def render_runline(events: List[Dict[str, Any]]) -> str:
+    def li(ev):
+        cls = ev.get("status","done")
+        dur = ev.get("dur_ms")
+        dur_txt = f"{int(dur)} ms" if isinstance(dur,(int,float)) else "—"
+        return f"""
+        <li class="{cls}">
+          <h5>{ev.get('title','')}</h5>
+          <div class="meta"><span>Duration: {dur_txt}</span></div>
+          <div class="desc">{ev.get('desc','')}</div>
+        </li>"""
+    html = '<div class="run-wrap"><div class="run-legend">' \
+           '<span class="chip">● Current</span><span class="chip">● Done</span></div>' \
+           '<ul class="run">{items}</ul></div>'
+    return html.replace("{items}", "".join(li(e) for e in events))
+
 def run_agent(employee_ids: Iterable[int], df_shifts: pd.DataFrame):
-    events: List[Dict[str, str]] = []
+    events: List[Dict[str, Any]] = []
     shift_assignment_results: List[tuple] = []
     reasoning_rows: List[Dict[str, Any]] = []
 
-    # 1) Employees
-    events.append({"icon":"🔍","title":"Fetching employee data","desc":"Loading employee information from UKG API..."})
+    # 1) Fetch employees
+    t0 = time.perf_counter()
     df_employees = fetch_employees(employee_ids)
-    if df_employees.empty:
-        events.append({"icon":"⚠️","title":"No Employee Data","desc":"No employees returned or credentials invalid."})
-    else:
-        events.append({"icon":"✅","title":"Employee Data Loaded","desc":f"{len(df_employees)} employee(s) loaded successfully."})
+    dur = (time.perf_counter() - t0) * 1000
+    loaded_txt = f"{len(df_employees)} employee(s) loaded successfully." if not df_employees.empty else "No employees returned or credentials invalid."
+    events.append({"title":"Fetch employees", "desc":loaded_txt, "status":"done", "dur_ms":max(int(dur),1)})
 
     # 2) Evaluate shifts
-    events.append({"icon":"📋","title":"Evaluating Shifts","desc":"Matching employees to shift requirements..."})
+    t_eval = time.perf_counter()
+    events.append({"title":"Evaluate open shifts", "desc":"Matching employees vs. role/specialty/training/ward…", "status":"done"})
+
     for _, shift in df_shifts.iterrows():
         eligible = df_employees[df_employees.apply(lambda r: is_eligible(r, shift), axis=1)] if not df_employees.empty else pd.DataFrame()
-        # 3) Decision
-        events.append({"icon":"🤖","title":"Running AI Decision","desc":"Determining best action for the shift..."})
+
+        # 3) AI decision
+        t_ai = time.perf_counter()
+        events.append({"title":"AI decision", "desc":"Select assign / notify / skip", "status":"done"})  # duration after call
         decision = gpt_decide(shift, eligible)
+        ai_dur = (time.perf_counter() - t_ai) * 1000
+        events[-1]["dur_ms"] = max(int(ai_dur),1)
+
         if decision.get("action") == "assign":
             emp = decision.get("employee")
-            events.append({"icon":"✅","title":f"Assigned {emp}","desc":f"{emp} -> {shift['Department']} at {shift['ShiftTime']}"})
+            events.append({"title":f"Assigned {emp}", "desc":f"{emp} → {shift.get('Department','')} ({shift.get('ShiftType','')}, {shift.get('ShiftDuration','')})", "status":"done"})
             shift_assignment_results.append((emp, shift["ShiftID"], "✅ Auto-Filled"))
         elif decision.get("action") == "notify":
             emp = decision.get("employee")
-            events.append({"icon":"📬","title":f"Notify {emp}","desc":f"Send notification to {emp} for the shift"})
+            events.append({"title":f"Notify {emp}", "desc":f"Send notification for {shift['ShiftID']}", "status":"done"})
             shift_assignment_results.append((emp, shift["ShiftID"], "📨 Notify"))
         else:
-            events.append({"icon":"⚠️","title":"Skipped","desc":"No eligible employees available or decision skipped."})
+            events.append({"title":"Skipped", "desc":"No eligible employees or decision skipped", "status":"done"})
             shift_assignment_results.append(("❌ No eligible", shift["ShiftID"], "⚠️ Skipped"))
 
-        # Reasoning rows
+        # Reasoning rows (explanatory)
         for _, emp_row in df_employees.iterrows():
-            role_match = shift["RequiredSkill"].strip().lower() == emp_row.get("JobRole", "").strip().lower()
-            cert_ok = shift["RequiredCert"].strip().lower() in [c.lower() for c in emp_row.get("Certifications", [])]
-            avail_ok = True; ot_ok = True
-            status = "✅ Eligible" if all([role_match, cert_ok, avail_ok, ot_ok]) else "❌ Not Eligible"
+            role_match = str(shift.get("RoleRequired","")).strip().lower() == emp_row.get("JobRole","").strip().lower()
+            spec = str(shift.get("Specialty","")).replace("-", " ").strip().lower()
+            spec_ok = any(s for s in [spec] if s and (
+                s in (emp_row.get("JobRole","")+emp_row.get("organizationPath","")+" ".join(emp_row.get("Certifications",[]))).lower()
+            ))
+            training_need = [t.strip().lower() for t in str(shift.get("MandatoryTraining","")).replace(",", "/").split("/") if t.strip()]
+            training_ok = all(t in [c.lower() for c in emp_row.get("Certifications",[])] for t in training_need)
+            ward_ok = not str(shift.get("WardDepartment","")).strip() or str(shift.get("WardDepartment","")).lower() in (emp_row.get("organizationPath","").lower())
+            status = "✅ Eligible" if all([role_match or spec_ok, training_ok, ward_ok]) else "❌ Not Eligible"
             reasoning_rows.append({
                 "Employee": emp_row.get("fullName",""),
                 "ShiftID": shift["ShiftID"],
                 "Eligible": status,
                 "Reasoning": " | ".join([
-                    "✅ Role Match" if role_match else "❌ Role",
-                    "✅ Cert" if cert_ok else "❌ Cert",
-                    "✅ Avail" if avail_ok else "❌ Avail",
-                    "✅ OT" if ot_ok else "❌ OT",
+                    "✅ Role" if role_match else "❌ Role",
+                    "✅ Specialty" if spec_ok else "❌ Specialty",
+                    "✅ Training" if training_ok else "❌ Training",
+                    "✅ Ward" if ward_ok else "❌ Ward",
+                    f"Lang: {shift.get('LanguageRequirement','') or '—'}"
                 ]),
                 "Certifications": ", ".join(emp_row.get("Certifications", []))
             })
-    events.append({"icon":"📊","title":"Summary Ready","desc":"AI finished processing shifts."})
+
+    # set evaluation duration
+    eval_dur = (time.perf_counter() - t_eval) * 1000
+    for e in events:
+        if e["title"] == "Evaluate open shifts":
+            e["dur_ms"] = max(int(eval_dur),1)
+            break
+
+    # 4) Summary
+    events.append({"title":"Summary ready", "desc":"AI finished processing shifts", "status":"current"})
     return events, shift_assignment_results, reasoning_rows
 
 # =============================================================================
@@ -384,6 +475,23 @@ def main() -> None:
 """, unsafe_allow_html=True
     )
 
+    # --- Business Case (top) ---
+    st.markdown("""
+<div class="card" style="max-width:980px;margin:0 auto 10px;">
+  <h4 style="margin:0 0 6px;">Business Case — Open Shifts Auto‑Fulfillment</h4>
+  <p style="margin:0;">
+    Health Matrix AI Agent يراقب الشفتات المفتوحة ويطابقها تلقائيًا مع الموارد المؤهّلة وفق المعايير:
+    <b>Role</b>، <b>Specialty</b>، <b>Mandatory Training</b>، <b>Shift Type/Duration</b>، <b>Ward/Department</b>، <b>Language</b>.
+    ثم يقرّر <b>Assign</b> أو <b>Notify</b> تلقائيًا ويعرض سجل التنفيذ خطوة‑بخطوة.
+  </p>
+  <ul style="margin:8px 0 0 18px;color:#334155;font-size:14px;line-height:22px">
+    <li>تقليل وقت شَغل الشفتات الحرجة (ICU/ED)</li>
+    <li>تحسين الالتزام باللوائح والتدريب الإلزامي</li>
+    <li>شفافية كاملة عبر Agent Runline (الخطوات + المدد)</li>
+  </ul>
+</div>
+""", unsafe_allow_html=True)
+
     # --- Button + status bar (initial) ---
     status_placeholder = st.empty()
     status_placeholder.markdown(
@@ -397,9 +505,10 @@ def main() -> None:
 """, unsafe_allow_html=True
     )
 
-    # Demo data (unchanged)
+    # Demo data (updated with new fields)
     employee_ids_default = [850, 825]
-    shift_data = "ShiftID,Department,RequiredSkill,RequiredCert,ShiftTime\nS101,ICU,MedSurg RN, Registered Nurse,2025-06-04 07:00"
+    shift_data = """ShiftID,Department,RoleRequired,Specialty,MandatoryTraining,ShiftType,ShiftDuration,WardDepartment,LanguageRequirement,ShiftTime
+S101,ICU,Registered Nurse,ICU-trained,BLS/ALS certified,Night,12h,A&E,Arabic,2025-06-04 07:00"""
     df_shifts_default = pd.read_csv(StringIO(shift_data))
 
     # --- CTA ---
@@ -433,10 +542,9 @@ def main() -> None:
 """, unsafe_allow_html=True
             )
 
-            # --- Mini Timeline (horizontal) ---
-            st.markdown("### 🕒 Timeline")
-            tl_html = render_mini_timeline(events)
-            components.html(tl_html, height=96, scrolling=False)
+            # --- Agent Runline (vertical) ---
+            st.markdown("### 🕒 Agent Runline")
+            components.html(render_runline(events), height=360, scrolling=True)
 
             # --- Stat row (quick KPIs) ---
             assigned = sum(1 for e in events if e["title"].startswith("Assigned"))
@@ -470,9 +578,9 @@ def main() -> None:
   <h4>How it works</h4>
   <ol style="margin:6px 0 0 18px;color:#334155">
     <li>Fetch employees & certifications from UKG demo API</li>
-    <li>Match against shift skill & required cert</li>
+    <li>Match against role/specialty/mandatory training/ward/language</li>
     <li>Use GPT (if key set) to assign/notify or skip</li>
-    <li>Show a clean timeline + KPIs + detailed reasoning</li>
+    <li>Show an execution Runline + KPIs + detailed reasoning</li>
   </ol>
 </div>
 """, unsafe_allow_html=True