Update app.py

app.py

@@ -50,6 +50,8 @@ BOTTOM_MARGIN_MM = 5
 PRINTABLE_HEIGHT_MM = PAGE_HEIGHT_MM - TOP_MARGIN_MM - BOTTOM_MARGIN_MM # ≈ 290 mm
 # Approximate height of a single text line in the calendar layout (in mm)
 LINE_HEIGHT_MM = 0.90 # 6.8 pt ≈ 0.9 mm
+# Estimated height of the day headers block (in mm) — measured empirically
+DAY_HEADERS_HEIGHT_MM = 5.0
 # ========================================== Shared CSS (Used in both HTML preview and PDF generation) ==========================================
 # CSS styles shared between the HTML preview and the PDF output.
 # Uses placeholders {{locations}}, {{start}}, {{end}}, {{time}} that are replaced at generation time.
@@ -348,7 +350,7 @@ def estimate_week_height(week_content, display_locs):
     max_lines = 0
     for day_html in week_content:
         content = day_html.split(' ', 1)[1] if ' ' in day_html else day_html
-        lines = content.count(' ') + 1
+        lines = content.count('<br>') + content.count('</div>') + 1  # Better estimate using tags
         if lines > max_lines:
             max_lines = lines
     base_height = 4.0
@@ -923,9 +925,8 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
     page_content = []
     week_content = []
     current_page_height = 0
-    is_first_page = True
   
-    # Static day headers (Mon-Sat)
+    # Static day headers (Mon-Sat) — now included on EVERY page
     day_headers = """
     <div class="week day-headers">
         <div class="week-number"></div>
@@ -949,14 +950,17 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
         if current.weekday() == 0 and len(week_content) > 0:
             finished_week = f'<div class="week"><div class="week-number">{week_num}</div>{"".join(week_content)}</div>'
             week_height = estimate_week_height(week_content, display_locs)
-            needed = week_height + (5 if len(page_content) == 0 else 1)
-          
-            # Page break logic: start new page if adding this week would overflow
+            
+            # Add day headers height only if this is the first week on the current page
+            extra_header = DAY_HEADERS_HEIGHT_MM if len(page_content) == 0 else 0
+            needed = week_height + extra_header + 1  # +1 mm buffer between weeks
+            
+            # Page break logic
             if current_page_height + needed > PRINTABLE_HEIGHT_MM:
-                html += f'<div class="page-group"><div class="week-group">{day_headers if is_first_page else ""}{"".join(page_content)}</div></div>'
+                # Close current page (always include headers)
+                html += f'<div class="page-group"><div class="week-group">{day_headers}{"".join(page_content)}</div></div>'
                 page_content = []
                 current_page_height = 0
-                is_first_page = False
           
             page_content.append(finished_week)
             current_page_height += needed
@@ -964,22 +968,21 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
             week_num += 1
       
         # Build the HTML for the current day
-        day_num = current.day
         day_html = f'<div class="day">{current.strftime("%m/%d")}'
       
-        # Global conflict warnings (same provider at multiple locations)
+        # Global conflict warnings
         if perform_conflict:
             for p, locs, cls, msg in check_provider_location_conflicts(providers_df, current, display_locs):
                 day_html += f'<div class="{cls}"><span class="warning-details">{msg}</span></div>'
       
-        # === NEW: Check if any clinic has "CLINIC_CLOSE!" on this day ===
+        # Check for CLINIC_CLOSE note anywhere
         has_clinic_close_anywhere = False
         if 'Note' in providers_df.columns:
             day_notes = providers_df[providers_df['Date'] == current]['Note'].dropna().astype(str).str.upper()
             if day_notes.str.contains('CLINIC_CLOSE').any():
                 has_clinic_close_anywhere = True
       
-        # Overall age coverage warning across main sites (only if no clinic is closed with "CLINIC_CLOSE!")
+        # Overall age coverage
         if perform_overall and not has_clinic_close_anywhere:
             missing, _ = check_overall_age_coverage(providers_df, info_df, current, display_locs)
             if missing:
@@ -990,7 +993,7 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
             loc_df = providers_df[(providers_df['Date'] == current) & (providers_df['Display_Location'] == loc)]
             loc_info = info_df[info_df['Location'] == loc]
           
-            # Detect holiday or school-closed status from notes
+            # Detect holiday or school-closed
             is_holiday = is_school = False
             if not loc_df.empty:
                 notes = loc_df['Note'].dropna().str.strip().str.upper().tolist()
@@ -999,7 +1002,6 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
                 elif notes and all(n == 'SCHOOL CLOSED' for n in notes) and loc in NO_AGE_CHECK_LOCATIONS:
                     is_school = True
           
-            # Remove completely empty rows
             loc_df = loc_df[~((loc_df['Start_Time'].isna()) & (loc_df['End_Time'].isna()) & (loc_df['Note'].isna() | (loc_df['Note'] == '')))]
           
             if not loc_df.empty or is_holiday or is_school:
@@ -1009,7 +1011,6 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
                 elif is_school:
                     day_html += '<div class="holiday-message">School Closed!</div>'
                 else:
-                    # === NEW: Check for clinic closure ===
                     if is_clinic_closed(providers_df, current, loc):
                         day_html += '<div class="clinic-closed-warning">Clinic Closed!</div>'
                     else:
@@ -1020,8 +1021,6 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
                             info_row = loc_info[loc_info['Provider'] == r['Name']]
                             name = info_row['Last_Name'].iloc[0] if not info_row.empty else r['Name']
                             tstr = get_time_string(r)
-                          
-                            # Color coding based on age coverage
                             if r['Name'] in full:
                                 color = "#ff6347"
                             elif r['Name'] in under:
@@ -1038,17 +1037,14 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
                             day_html += f'<span style="{style}">{name}: {tstr}</span><br>'
                         day_html += '</div>'
                       
-                        # Operational hour coverage warning
                         if check_operation_coverage_flag and loc not in NO_OPERATION_CHECK_LOCATIONS:
                             gaps = check_operation_time_coverage(providers_df, current, loc)
                             if gaps:
                                 day_html += f'<div class="operation-warning"><span class="warning-details">Missing: {", ".join(gaps)}</span></div>'
                       
-                        # Per-location age coverage warning
                         if check_age_coverage_flag and missing and loc not in NO_AGE_CHECK_LOCATIONS:
                             day_html += f'<div class="warning"><span class="warning-details">Missing: {", ".join(missing)}</span></div>'
                       
-                        # MA scheduling and ratio check
                         if check_ma_mismatch_flag:
                             ma_loc_df = ma_df[(ma_df['Date'] == current) & (ma_df['Display_Location'] == loc)]
                             ma_loc_df = ma_loc_df[~((ma_loc_df['Start_Time'].isna()) & (ma_loc_df['End_Time'].isna()) & (ma_loc_df['Note'].isna() | (ma_loc_df['Note'] == '')))]
@@ -1073,23 +1069,25 @@ def combine_schedules(provider_info_file, provider_files, ma_files, start_date,
         week_content.append(day_html)
         current += pd.Timedelta(days=1)
   
-    # Finalize any remaining week
+    # Finalize remaining week
     if week_content:
         finished_week = f'<div class="week"><div class="week-number">{week_num}</div>{"".join(week_content)}</div>'
         week_height = estimate_week_height(week_content, display_locs)
-        needed = week_height + (5 if len(page_content) == 0 else 1)
+        extra_header = DAY_HEADERS_HEIGHT_MM if len(page_content) == 0 else 0
+        needed = week_height + extra_header + 1
+        
         if current_page_height + needed > PRINTABLE_HEIGHT_MM:
-            html += f'<div class="page-group"><div class="week-group">{day_headers if is_first_page else ""}{"".join(page_content)}</div></div>'
+            html += f'<div class="page-group"><div class="week-group">{day_headers}{"".join(page_content)}</div></div>'
             page_content = []
             current_page_height = 0
-            is_first_page = False
+        
         page_content.append(finished_week)
   
-    # Close current page
+    # Close final page
     if page_content:
-        html += f'<div class="page-group"><div class="week-group">{day_headers if is_first_page else ""}{"".join(page_content)}</div></div>'
+        html += f'<div class="page-group"><div class="week-group">{day_headers}{"".join(page_content)}</div></div>'
   
-    # Add weekly clinical hours summary table if requested
+    # Add weekly hours table
     if show_weekly_hours:
         wh, wt = calculate_weekly_hours(providers_df, info_df, start_obj, end_obj, display_locs)
         html += '<div class="hours-table-section" style="break-before: page;">'