Run3 extract request details

Extract Request Details

Inputs

• input_json_path: string, path to /root/test_input.json
• today_date: string, today's date in YYYY-MM-DD format

Outputs

• requests: list of objects, each containing:
  • message_id: string (unique identifier)
  • recipient_email: string
  • duration_minutes: integer
  • date_constraint: object with:
    • type: string ("specific_date", "date_range", "day_of_week", "relative", "none")
    • start_date: YYYY-MM-DD (or null if no constraint)
    • end_date: YYYY-MM-DD (or null if single date or none)
    • day_names: list of strings (e.g., ["Monday", "Tuesday"]) if recurring day constraint
    • relative_offset_days: integer (e.g., 1 for "tomorrow", 0 for "today")
  • time_constraint: object with:
    • type: string ("specific_time", "time_window", "time_of_day", "none")
    • start_time: HH:MM (24-hour) or null
    • end_time: HH:MM (24-hour) or null
    • period: string ("morning", "afternoon", "evening") or null
  • raw_email_body: string (original message text)

Algorithm

1. Load and Parse JSON

• Read /root/test_input.json
• Validate JSON structure; expect array of email objects
• For each email, extract: message_id, from (or sender_email), subject, body (or message_body)

2. Parse Duration

• Search email body for patterns:
  • "X hour" → duration_minutes = X * 60
  • "X minute" → duration_minutes = X
  • "X min" → duration_minutes = X
  • "30 mins", "1.5 hours" → parse decimal
  • "half hour" → duration_minutes = 30
  • If no duration found, flag as error and set duration_minutes = null

3. Parse Date Constraint

• Search for date/day patterns in body:
  • Specific date: "January 8", "Jan 8, 2026", "01/08/2026"
    • Parse and normalize to YYYY-MM-DD
    • If year omitted, apply year inference logic (see below)
    • Type: "specific_date", set start_date = end_date
  • Date range: "Jan 5-7", "January 5 to 7", "between Jan 5 and Jan 8"
    • Type: "date_range", extract start_date and end_date
    • Apply year inference if needed
  • Day of week: "next Monday", "any Tuesday or Wednesday", "Mondays"
    • Type: "day_of_week", populate day_names list
    • Set relative_offset_days if "next" or "this" qualifier present
  • Relative: "tomorrow", "next week", "in 3 days"
    • Type: "relative", calculate offset from today_date
  • None found: type = "none", all date fields null

4. Parse Time Constraint

• Search for time patterns in body:
  • Specific time: "9:00 AM", "14:30", "2:30 PM"
    • Type: "specific_time", set start_time, set end_time = null (will be calculated from duration)
  • Time window: "9:00 AM - 10:30 AM", "between 2 PM and 4 PM"
    • Type: "time_window", extract start_time and end_time
  • Time of day: "morning", "afternoon", "evening", "before noon", "after 5 PM"
    • Type: "time_of_day", extract period or convert to approximate start_time/end_time:
      • Morning: 06:00 - 12:00
      • Afternoon: 12:00 - 17:00
      • Evening: 17:00 - 21:00
  • None found: type = "none", all time fields null

5. Year Inference Logic

• If a date is parsed without a year (e.g., "January 8"):
  • Extract month and day from parsed date
  • If (month, day) >= (today_date.month, today_date.day), assume current year
  • Else, assume next year
  • Example: if today is 2026-01-15, "January 8" → 2027-01-08; "January 20" → 2026-01-20

6. Extract Recipient Email

• Use from, sender_email, or sender field from JSON object
• Validate format (should contain "@")

7. Return Results

• Compile all requests into list and return