Upload 8 files

constants.py

@@ -0,0 +1,10 @@
+"""
+Constants for timezone mappings and configuration
+"""
+
+POD_TIMEZONE_MAPPINGS = {
+    "JFK": "US/Eastern",
+    "LAX": "US/Pacific",
+    "ORD": "US/Central",
+    "MIA": "US/Eastern"
+}

entry_milestone_exporter.py

@@ -8,15 +8,61 @@ import pandas as pd
 import sys
 from datetime import datetime
 from typing import List, Dict, Optional
+import pytz
 
 # Add src directory to path
 sys.path.append('src')
 
 from notion_integration import NotionIntegration
+from constants import POD_TIMEZONE_MAPPINGS
 
 class EntryMilestoneExporter:
     """Class to export entry milestone data to Excel format."""
 
+    @staticmethod
+    def convert_utc_to_pod_timezone(utc_dt: datetime, pod: Optional[str]) -> datetime:
+        """
+        Convert UTC datetime to POD local timezone without timezone info.
+
+        Args:
+            utc_dt: UTC datetime object
+            pod: Port of discharge (JFK, LAX, ORD, MIA)
+
+        Returns:
+            Timezone-naive datetime in POD local time (or UTC if POD not found)
+        """
+        if not utc_dt or pd.isna(utc_dt):
+            return utc_dt
+
+        if not pod or pod not in POD_TIMEZONE_MAPPINGS:
+            # Keep as UTC if POD is missing or not in mappings
+            if hasattr(utc_dt, 'tzinfo') and utc_dt.tzinfo is not None:
+                return utc_dt.replace(tzinfo=None)
+            return utc_dt
+
+        try:
+            # Ensure UTC timezone
+            if hasattr(utc_dt, 'tzinfo'):
+                if utc_dt.tzinfo is None:
+                    utc_dt = pytz.utc.localize(utc_dt)
+                else:
+                    utc_dt = utc_dt.astimezone(pytz.utc)
+            else:
+                utc_dt = pytz.utc.localize(utc_dt)
+
+            # Convert to POD timezone
+            timezone_name = POD_TIMEZONE_MAPPINGS[pod]
+            pod_tz = pytz.timezone(timezone_name)
+            local_dt = utc_dt.astimezone(pod_tz)
+
+            # Remove timezone info
+            return local_dt.replace(tzinfo=None)
+        except:
+            # Fallback to UTC if conversion fails
+            if hasattr(utc_dt, 'tzinfo') and utc_dt.tzinfo is not None:
+                return utc_dt.replace(tzinfo=None)
+            return utc_dt
+
     # Define all Excel columns based on the template
     EXCEL_COLUMNS = [
         'MAWB', 'Entry', 'POD', 'handover_doc', 'receipt_scan', 'CBP Line',
@@ -62,8 +108,8 @@ class EntryMilestoneExporter:
 
         return unique_mawbs
 
-    def extract_entry_data(self, entry_record: Dict) -> Dict:
-        """Extract relevant data from a Notion entry record."""
+    def extract_entry_data(self, entry_record: Dict, pod: Optional[str] = None) -> Dict:
+        """Extract relevant data from a Notion entry record with timezone conversion."""
         properties = entry_record.get("properties", {})
         data = {}
 
@@ -75,7 +121,7 @@ class EntryMilestoneExporter:
         if "entry_number" in properties and properties["entry_number"].get("rich_text"):
             data["entry_number"] = properties["entry_number"]["rich_text"][0].get("text", {}).get("content", "")
 
-        # Extract dates
+        # Extract dates with timezone conversion
         date_fields = [
             "ams_submit_date", "entry_submit_date", "cbp_hold_date", "cbp_release_date",
             "cpsc_review_date", "cpsc_release_date"
@@ -87,7 +133,9 @@ class EntryMilestoneExporter:
                 if date_str:
                     try:
                         dt = datetime.fromisoformat(date_str.replace('Z', '+00:00'))
-                        data[field] = dt.replace(tzinfo=None)
+                        dt_naive = dt.replace(tzinfo=None)
+                        # Convert to POD timezone
+                        data[field] = self.convert_utc_to_pod_timezone(dt_naive, pod)
                     except:
                         data[field] = date_str
                 else:
@@ -245,7 +293,16 @@ class EntryMilestoneExporter:
         # Process data
         excel_rows = []
         for entry in entries:
-            entry_data = self.extract_entry_data(entry)
+            # Get MAWB to look up POD
+            mawb_for_lookup = ""
+            if "mawb" in entry.get("properties", {}) and entry["properties"]["mawb"].get("title"):
+                mawb_for_lookup = entry["properties"]["mawb"]["title"][0].get("text", {}).get("content", "")
+
+            # Get POD from mawb_info
+            pod = mawb_info.get(mawb_for_lookup, {}).get("pod", None) if mawb_for_lookup else None
+
+            # Extract entry data with POD timezone conversion
+            entry_data = self.extract_entry_data(entry, pod)
             mawb = entry_data.get("mawb", "")
 
             if mawb in mawb_info:

mawb_milestone_exporter.py

@@ -8,11 +8,13 @@ import sys
 import pandas as pd
 from datetime import datetime
 from typing import List, Dict, Optional
+import pytz
 
 # Add src directory to path
 sys.path.append('src')
 
 from notion_integration import NotionIntegration
+from constants import POD_TIMEZONE_MAPPINGS
 
 
 class MAWBMilestoneExporter:
@@ -22,6 +24,50 @@ class MAWBMilestoneExporter:
         """Initialize with Notion integration instance."""
         self.notion = notion_integration
 
+    @staticmethod
+    def convert_utc_to_pod_timezone(utc_dt: datetime, pod: Optional[str]) -> datetime:
+        """
+        Convert UTC datetime to POD local timezone without timezone info.
+
+        Args:
+            utc_dt: UTC datetime object
+            pod: Port of discharge (JFK, LAX, ORD, MIA)
+
+        Returns:
+            Timezone-naive datetime in POD local time (or UTC if POD not found)
+        """
+        if not utc_dt or pd.isna(utc_dt):
+            return utc_dt
+
+        if not pod or pod not in POD_TIMEZONE_MAPPINGS:
+            # Keep as UTC if POD is missing or not in mappings
+            if hasattr(utc_dt, 'tzinfo') and utc_dt.tzinfo is not None:
+                return utc_dt.replace(tzinfo=None)
+            return utc_dt
+
+        try:
+            # Ensure UTC timezone
+            if hasattr(utc_dt, 'tzinfo'):
+                if utc_dt.tzinfo is None:
+                    utc_dt = pytz.utc.localize(utc_dt)
+                else:
+                    utc_dt = utc_dt.astimezone(pytz.utc)
+            else:
+                utc_dt = pytz.utc.localize(utc_dt)
+
+            # Convert to POD timezone
+            timezone_name = POD_TIMEZONE_MAPPINGS[pod]
+            pod_tz = pytz.timezone(timezone_name)
+            local_dt = utc_dt.astimezone(pod_tz)
+
+            # Remove timezone info
+            return local_dt.replace(tzinfo=None)
+        except:
+            # Fallback to UTC if conversion fails
+            if hasattr(utc_dt, 'tzinfo') and utc_dt.tzinfo is not None:
+                return utc_dt.replace(tzinfo=None)
+            return utc_dt
+
     def parse_mawb_input(self, mawb_input: str) -> List[str]:
         """Parse MAWB input string into a list of MAWBs."""
         if not mawb_input or not mawb_input.strip():
@@ -260,28 +306,32 @@ class MAWBMilestoneExporter:
                 if cpsc_release_dt:
                     cpsc_release_dates_for_applicable.append(cpsc_release_dt)
 
-        # Apply aggregation rules
-        result["T01 Entry Submission"] = min(entry_submit_dates) if entry_submit_dates else None
-        result["T01 CBP Exam"] = min(cbp_hold_dates) if cbp_hold_dates else None
-        result["T01 CPSC/PGA Review"] = min(cpsc_review_dates) if cpsc_review_dates else None
+        # Apply aggregation rules and convert to POD timezone
+        result["T01 Entry Submission"] = self.convert_utc_to_pod_timezone(
+            min(entry_submit_dates), pod) if entry_submit_dates else None
+        result["T01 CBP Exam"] = self.convert_utc_to_pod_timezone(
+            min(cbp_hold_dates), pod) if cbp_hold_dates else None
+        result["T01 CPSC/PGA Review"] = self.convert_utc_to_pod_timezone(
+            min(cpsc_review_dates), pod) if cpsc_review_dates else None
 
-        # Conditional aggregation
+        # Conditional aggregation with timezone conversion
         if total_entries > 0 and len(cbp_release_dates) == total_entries:
-            result["T01 CBP Release"] = max(cbp_release_dates)
+            result["T01 CBP Release"] = self.convert_utc_to_pod_timezone(max(cbp_release_dates), pod)
         else:
             result["T01 CBP Release"] = None
 
         if len(cpsc_applicable_entries) > 0 and len(cpsc_release_dates_for_applicable) == len(cpsc_applicable_entries):
-            result["T01 CPSC/PGA release"] = max(cpsc_release_dates_for_applicable)
+            result["T01 CPSC/PGA release"] = self.convert_utc_to_pod_timezone(
+                max(cpsc_release_dates_for_applicable), pod)
         else:
             result["T01 CPSC/PGA release"] = None
 
         if total_entries > 0 and len(final_release_dates) == total_entries:
-            result["Customs Release"] = max(final_release_dates)
+            result["Customs Release"] = self.convert_utc_to_pod_timezone(max(final_release_dates), pod)
         else:
             result["Customs Release"] = None
 
-        # Aggregate milestone record data
+        # Aggregate milestone record data with timezone conversion
         warehouse_in_dates = []
         cargo_ready_dates = []
         warehouse_out_dates = []
@@ -294,9 +344,12 @@ class MAWBMilestoneExporter:
             if dt := self.extract_datetime_from_notion_field(milestone_record, "warehouse_out"):
                 warehouse_out_dates.append(dt)
 
-        result["PDA scan"] = min(warehouse_in_dates) if warehouse_in_dates else None
-        result["Cargo ready for pickup"] = min(cargo_ready_dates) if cargo_ready_dates else None
-        result["Ship out completed"] = min(warehouse_out_dates) if warehouse_out_dates else None
+        result["PDA scan"] = self.convert_utc_to_pod_timezone(
+            min(warehouse_in_dates), pod) if warehouse_in_dates else None
+        result["Cargo ready for pickup"] = self.convert_utc_to_pod_timezone(
+            min(cargo_ready_dates), pod) if cargo_ready_dates else None
+        result["Ship out completed"] = self.convert_utc_to_pod_timezone(
+            min(warehouse_out_dates), pod) if warehouse_out_dates else None
 
         return result
 

notion_exporter.py

@@ -13,11 +13,13 @@ from typing import Dict, List, Optional, Any
 import numpy as np
 import time
 import random
+import pytz
 
 # Add src directory to Python path
 sys.path.append('src')
 
 from notion_integration import NotionIntegration
+from constants import POD_TIMEZONE_MAPPINGS
 
 
 def load_notion_config_from_env() -> Dict:
@@ -51,6 +53,50 @@ def load_notion_config_from_env() -> Dict:
     return config
 
 
+def convert_utc_to_pod_timezone(utc_dt: datetime, pod: Optional[str]) -> datetime:
+    """
+    Convert UTC datetime to POD local timezone without timezone info.
+
+    Args:
+        utc_dt: UTC datetime object
+        pod: Port of discharge (JFK, LAX, ORD, MIA)
+
+    Returns:
+        Timezone-naive datetime in POD local time (or UTC if POD not found)
+    """
+    if not utc_dt or pd.isna(utc_dt):
+        return utc_dt
+
+    if not pod or pod not in POD_TIMEZONE_MAPPINGS:
+        # Keep as UTC if POD is missing or not in mappings
+        if hasattr(utc_dt, 'tz') and utc_dt.tz is not None:
+            return utc_dt.tz_localize(None)
+        return utc_dt
+
+    try:
+        # Ensure UTC timezone
+        if hasattr(utc_dt, 'tz'):
+            if utc_dt.tz is None:
+                utc_dt = pytz.utc.localize(utc_dt)
+            else:
+                utc_dt = utc_dt.tz_convert(pytz.utc)
+        else:
+            utc_dt = pytz.utc.localize(utc_dt)
+
+        # Convert to POD timezone
+        timezone_name = POD_TIMEZONE_MAPPINGS[pod]
+        pod_tz = pytz.timezone(timezone_name)
+        local_dt = utc_dt.astimezone(pod_tz)
+
+        # Remove timezone info
+        return local_dt.replace(tzinfo=None)
+    except:
+        # Fallback to UTC if conversion fails
+        if hasattr(utc_dt, 'tz') and utc_dt.tz is not None:
+            return utc_dt.tz_localize(None)
+        return utc_dt
+
+
 class NotionExcelExporter:
     def __init__(self, slow_mode: bool = False, page_size: int = 50):
         """Initialize the exporter with Notion configuration from environment."""
@@ -337,6 +383,10 @@ class NotionExcelExporter:
 
         for record in records:
             props = record.get('properties', {})
+
+            # Extract POD for timezone conversion
+            pod = self._extract_rich_text(props.get('pod', {}))
+
             row = {
                 'ID': record.get('id', ''),
             }
@@ -357,13 +407,13 @@ class NotionExcelExporter:
                 elif 'select' in prop_data:
                     row[prop_name] = self._extract_select(prop_data)
                 elif 'date' in prop_data:
-                    row[prop_name] = self._format_datetime(prop_data)
+                    row[prop_name] = self._format_datetime(prop_data, pod)
                 elif 'checkbox' in prop_data:
                     row[prop_name] = self._extract_checkbox(prop_data)
                 elif 'rollup' in prop_data:
                     row[prop_name] = self._extract_rollup(prop_data)
                 elif 'formula' in prop_data:
-                    row[prop_name] = self._extract_formula(prop_data)
+                    row[prop_name] = self._extract_formula(prop_data, pod)
                 elif 'relation' in prop_data:
                     # Extract relation as count
                     relation_list = prop_data.get('relation', [])
@@ -392,16 +442,16 @@ class NotionExcelExporter:
                 elif 'last_edited_by' in prop_data:
                     row[prop_name] = self._extract_last_edited_by(prop_data)
                 elif 'created_time' in prop_data:
-                    row[prop_name] = self._format_datetime_iso(prop_data.get('created_time', ''))
+                    row[prop_name] = self._format_datetime_iso(prop_data.get('created_time', ''), pod)
                 elif 'last_edited_time' in prop_data:
-                    row[prop_name] = self._format_datetime_iso(prop_data.get('last_edited_time', ''))
+                    row[prop_name] = self._format_datetime_iso(prop_data.get('last_edited_time', ''), pod)
                 else:
                     # Unknown property type, store as string
                     row[prop_name] = str(prop_data)
 
-            # Add system metadata fields
-            row['Created'] = self._format_datetime_iso(record.get('created_time', ''))
-            row['Last Modified'] = self._format_datetime_iso(record.get('last_edited_time', ''))
+            # Add system metadata fields with timezone conversion
+            row['Created'] = self._format_datetime_iso(record.get('created_time', ''), pod)
+            row['Last Modified'] = self._format_datetime_iso(record.get('last_edited_time', ''), pod)
 
             data.append(row)
 
@@ -413,22 +463,26 @@ class NotionExcelExporter:
 
         for record in records:
             props = record.get('properties', {})
+
+            # Try to extract POD if available in entry record, otherwise None (keep as UTC)
+            pod = self._extract_rich_text(props.get('pod', {})) if 'pod' in props else None
+
             row = {
                 'ID': record.get('id', ''),
                 'Entry Number': self._extract_rich_text(props.get('entry_number', {})),
                 'MAWB': self._extract_title(props.get('mawb', {})),
-                'Entry Submit Date': self._format_datetime(props.get('entry_submit_date', {})),
-                'CBP Release Date': self._format_datetime(props.get('cbp_release_date', {})),
-                'CBP Hold Date': self._format_datetime(props.get('cbp_hold_date', {})),
+                'Entry Submit Date': self._format_datetime(props.get('entry_submit_date', {}), pod),
+                'CBP Release Date': self._format_datetime(props.get('cbp_release_date', {}), pod),
+                'CBP Hold Date': self._format_datetime(props.get('cbp_hold_date', {}), pod),
                 'CBP Hold Line': self._extract_rich_text(props.get('cbp_hold_line', {})),
-                'CPSC Review Date': self._format_datetime(props.get('cpsc_review_date', {})),
+                'CPSC Review Date': self._format_datetime(props.get('cpsc_review_date', {}), pod),
                 'CPSC Review Line': self._extract_rich_text(props.get('cpsc_review_line', {})),
-                'CPSC Release Date': self._format_datetime(props.get('cpsc_release_date', {})),
-                'Estimated Review Date': self._format_datetime(props.get('estimated_review_date', {})),
+                'CPSC Release Date': self._format_datetime(props.get('cpsc_release_date', {}), pod),
+                'Estimated Review Date': self._format_datetime(props.get('estimated_review_date', {}), pod),
                 'Milestone Task Status': self._extract_rollup(props.get('milestone_task_status', {})),
-                'Final Release': self._extract_formula(props.get('final_release', {})),
-                'Created': self._format_datetime_iso(record.get('created_time', '')),
-                'Last Modified': self._format_datetime_iso(record.get('last_edited_time', ''))
+                'Final Release': self._extract_formula(props.get('final_release', {}), pod),
+                'Created': self._format_datetime_iso(record.get('created_time', ''), pod),
+                'Last Modified': self._format_datetime_iso(record.get('last_edited_time', ''), pod)
             }
             data.append(row)
 
@@ -440,6 +494,10 @@ class NotionExcelExporter:
 
         for record in records:
             props = record.get('properties', {})
+
+            # Try to extract POD if available, otherwise None (keep as UTC)
+            pod = self._extract_rich_text(props.get('pod', {})) if 'pod' in props else None
+
             row = {
                 'ID': record.get('id', ''),
                 'MAWB': self._extract_title(props.get('mawb', {})),
@@ -447,8 +505,8 @@ class NotionExcelExporter:
                 'Customer': self._extract_rich_text(props.get('customer', {})),
                 'Entry Type': self._extract_rich_text(props.get('entry_type', {})),
                 'Done': self._extract_checkbox(props.get('done', {})),
-                'Created': self._format_datetime_iso(record.get('created_time', '')),
-                'Last Modified': self._format_datetime_iso(record.get('last_edited_time', '')),
+                'Created': self._format_datetime_iso(record.get('created_time', ''), pod),
+                'Last Modified': self._format_datetime_iso(record.get('last_edited_time', ''), pod),
                 'Last Edited By': self._extract_last_edited_by(props.get('Last edited by', {}))
             }
             data.append(row)
@@ -461,6 +519,10 @@ class NotionExcelExporter:
 
         for record in records:
             props = record.get('properties', {})
+
+            # Try to extract POD if available, otherwise None (keep as UTC)
+            pod = self._extract_rich_text(props.get('pod', {})) if 'pod' in props else None
+
             row = {
                 'ID': record.get('id', ''),
                 'MAWB': self._extract_title(props.get('mawb', {})),
@@ -469,15 +531,15 @@ class NotionExcelExporter:
                 'Pallets': self._extract_number(props.get('pallets', {})),
                 'GWT': self._extract_number(props.get('gwt', {})),
                 'CWT': self._extract_number(props.get('cwt', {})),
-                'Warehouse In': self._format_datetime(props.get('warehouse_in', {})),
-                'Cargo Ready': self._format_datetime(props.get('cargo_ready', {})),
-                'Warehouse Out': self._format_datetime(props.get('warehouse_out', {})),
-                'ETD': self._format_datetime(props.get('etd', {})),
-                'ETA': self._format_datetime(props.get('eta', {})),
+                'Warehouse In': self._format_datetime(props.get('warehouse_in', {}), pod),
+                'Cargo Ready': self._format_datetime(props.get('cargo_ready', {}), pod),
+                'Warehouse Out': self._format_datetime(props.get('warehouse_out', {}), pod),
+                'ETD': self._format_datetime(props.get('etd', {}), pod),
+                'ETA': self._format_datetime(props.get('eta', {}), pod),
                 'Cartons Left at Airline': self._extract_number(props.get('cartons_left_at_airline', {})),
                 'Cartons Left at Warehouse': self._extract_number(props.get('cartons_left_at_warehouse', {})),
-                'Created': self._format_datetime_iso(record.get('created_time', '')),
-                'Last Modified': self._format_datetime_iso(record.get('last_edited_time', ''))
+                'Created': self._format_datetime_iso(record.get('created_time', ''), pod),
+                'Last Modified': self._format_datetime_iso(record.get('last_edited_time', ''), pod)
             }
             data.append(row)
 
@@ -550,8 +612,8 @@ class NotionExcelExporter:
 
         return ''
 
-    def _extract_formula(self, prop: Dict) -> str:
-        """Extract formula property value - return computed result."""
+    def _extract_formula(self, prop: Dict, pod: Optional[str] = None):
+        """Extract formula property value - return computed result with timezone conversion for dates."""
         if not prop or 'formula' not in prop:
             return ''
         formula_obj = prop['formula']
@@ -568,11 +630,19 @@ class NotionExcelExporter:
         elif formula_obj.get('type') == 'date':
             date_obj = formula_obj.get('date')
             if date_obj and date_obj.get('start'):
-                return date_obj['start']
+                try:
+                    dt = pd.to_datetime(date_obj['start'])
+                    # Remove timezone info for Excel compatibility
+                    if dt.tz is not None:
+                        dt = dt.tz_convert('UTC').tz_localize(None)
+                    # Convert to POD timezone if provided
+                    return convert_utc_to_pod_timezone(dt, pod)
+                except (ValueError, TypeError):
+                    return pd.NaT
         return ''
 
-    def _format_datetime(self, prop: Dict) -> pd.Timestamp:
-        """Extract date property value and convert to Excel-friendly datetime."""
+    def _format_datetime(self, prop: Dict, pod: Optional[str] = None) -> pd.Timestamp:
+        """Extract date property value and convert to Excel-friendly datetime with POD timezone."""
         if not prop or 'date' not in prop:
             return pd.NaT
         date_obj = prop['date']
@@ -583,12 +653,13 @@ class NotionExcelExporter:
             # Remove timezone info for Excel compatibility
             if dt.tz is not None:
                 dt = dt.tz_convert('UTC').tz_localize(None)
-            return dt
+            # Convert to POD timezone if provided
+            return convert_utc_to_pod_timezone(dt, pod)
         except (ValueError, TypeError):
             return pd.NaT
 
-    def _format_datetime_iso(self, iso_string: str) -> pd.Timestamp:
-        """Convert ISO datetime string to Excel-friendly datetime."""
+    def _format_datetime_iso(self, iso_string: str, pod: Optional[str] = None) -> pd.Timestamp:
+        """Convert ISO datetime string to Excel-friendly datetime with POD timezone."""
         if not iso_string:
             return pd.NaT
         try:
@@ -596,7 +667,8 @@ class NotionExcelExporter:
             # Remove timezone info for Excel compatibility
             if dt.tz is not None:
                 dt = dt.tz_convert('UTC').tz_localize(None)
-            return dt
+            # Convert to POD timezone if provided
+            return convert_utc_to_pod_timezone(dt, pod)
         except (ValueError, TypeError):
             return pd.NaT