Implement comprehensive external calendar event deduplication and fixes
All checks were successful
Build and Push Docker Image / docker (push) Successful in 1m11s

- Add UID-based deduplication to prefer recurring events over single events with same UID
- Implement RRULE-generated instance detection to filter duplicate occurrences
- Add title normalization for case-insensitive matching and consolidation
- Fix external calendar refresh button with proper error handling and loading states
- Update context menu for external events to show only "View Event Details" option
- Add comprehensive multi-pass deduplication: UID → title consolidation → RRULE filtering

This resolves issues where Outlook calendars showed duplicate events with same UID
but different RRULE states (e.g., "Dragster Stand Up" appearing both as recurring
and single events).

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
Connor Johnstone
2025-09-04 15:35:42 -04:00
parent c6eea88002
commit b16603b50b
4 changed files with 582 additions and 42 deletions

View File

@@ -2,7 +2,7 @@ use axum::{
extract::{Path, State},
response::Json,
};
use chrono::{DateTime, Utc};
use chrono::{DateTime, Utc, Datelike};
use ical::parser::ical::component::IcalEvent;
use reqwest::Client;
use serde::Serialize;
@@ -138,6 +138,9 @@ fn parse_ics_content(ics_content: &str) -> Result<Vec<VEvent>, Box<dyn std::erro
}
}
// Deduplicate events based on UID, start time, and summary
// Outlook sometimes includes duplicate events (recurring exceptions may appear multiple times)
events = deduplicate_events(events);
Ok(events)
}
@@ -407,4 +410,439 @@ fn parse_datetime_with_tz(datetime_str: &str, tzid: Option<&str>) -> Option<Date
}
None
}
/// Deduplicate events based on UID, start time, and summary
/// Some calendar systems (like Outlook) may include duplicate events in ICS feeds
/// This includes both exact duplicates and recurring event instances that would be
/// generated by existing RRULE patterns, and events with same title but different
/// RRULE patterns that should be consolidated
fn deduplicate_events(mut events: Vec<VEvent>) -> Vec<VEvent> {
use std::collections::HashMap;
let original_count = events.len();
// First pass: Group by UID and prefer recurring events over single events with same UID
let mut uid_groups: HashMap<String, Vec<VEvent>> = HashMap::new();
for event in events.drain(..) {
// Debug logging to understand what's happening
println!("🔍 Event: '{}' at {} (RRULE: {}) - UID: {}",
event.summary.as_ref().unwrap_or(&"No Title".to_string()),
event.dtstart.format("%Y-%m-%d %H:%M"),
if event.rrule.is_some() { "Yes" } else { "No" },
event.uid
);
uid_groups.entry(event.uid.clone()).or_insert_with(Vec::new).push(event);
}
let mut uid_deduplicated_events = Vec::new();
for (uid, mut events_with_uid) in uid_groups.drain() {
if events_with_uid.len() == 1 {
// Only one event with this UID, keep it
uid_deduplicated_events.push(events_with_uid.into_iter().next().unwrap());
} else {
// Multiple events with same UID - prefer recurring over non-recurring
println!("🔍 Found {} events with UID '{}'", events_with_uid.len(), uid);
// Sort by preference: recurring events first, then by completeness
events_with_uid.sort_by(|a, b| {
let a_has_rrule = a.rrule.is_some();
let b_has_rrule = b.rrule.is_some();
match (a_has_rrule, b_has_rrule) {
(true, false) => std::cmp::Ordering::Less, // a (recurring) comes first
(false, true) => std::cmp::Ordering::Greater, // b (recurring) comes first
_ => {
// Both same type (both recurring or both single) - compare by completeness
event_completeness_score(b).cmp(&event_completeness_score(a))
}
}
});
// Keep the first (preferred) event
let preferred_event = events_with_uid.into_iter().next().unwrap();
println!("🔄 UID dedup: Keeping '{}' (RRULE: {})",
preferred_event.summary.as_ref().unwrap_or(&"No Title".to_string()),
if preferred_event.rrule.is_some() { "Yes" } else { "No" }
);
uid_deduplicated_events.push(preferred_event);
}
}
// Second pass: separate recurring and single events from UID-deduplicated set
let mut recurring_events = Vec::new();
let mut single_events = Vec::new();
for event in uid_deduplicated_events.drain(..) {
if event.rrule.is_some() {
recurring_events.push(event);
} else {
single_events.push(event);
}
}
// Third pass: Group recurring events by normalized title and consolidate different RRULE patterns
let mut title_groups: HashMap<String, Vec<VEvent>> = HashMap::new();
for event in recurring_events.drain(..) {
let title = normalize_title(event.summary.as_ref().unwrap_or(&String::new()));
title_groups.entry(title).or_insert_with(Vec::new).push(event);
}
let mut deduplicated_recurring = Vec::new();
for (title, events_with_title) in title_groups.drain() {
if events_with_title.len() == 1 {
// Single event with this title, keep as-is
deduplicated_recurring.push(events_with_title.into_iter().next().unwrap());
} else {
// Multiple events with same title - consolidate or deduplicate
println!("🔍 Found {} events with title '{}'", events_with_title.len(), title);
// Check if these are actually different recurring patterns for the same logical event
let consolidated = consolidate_same_title_events(events_with_title);
deduplicated_recurring.extend(consolidated);
}
}
// Fourth pass: filter single events, removing those that would be generated by recurring events
let mut deduplicated_single = Vec::new();
let mut seen_single: HashMap<String, usize> = HashMap::new();
for event in single_events.drain(..) {
let normalized_title = normalize_title(event.summary.as_ref().unwrap_or(&String::new()));
let dedup_key = format!(
"{}|{}",
event.dtstart.format("%Y%m%dT%H%M%S"),
normalized_title
);
// First check for exact duplicates among single events
if let Some(&existing_index) = seen_single.get(&dedup_key) {
let existing_event: &VEvent = &deduplicated_single[existing_index];
let current_completeness = event_completeness_score(&event);
let existing_completeness = event_completeness_score(existing_event);
if current_completeness > existing_completeness {
println!("🔄 Replacing single event: Keeping '{}' over '{}'",
event.summary.as_ref().unwrap_or(&"No Title".to_string()),
existing_event.summary.as_ref().unwrap_or(&"No Title".to_string())
);
deduplicated_single[existing_index] = event;
} else {
println!("🚫 Discarding duplicate single event: Keeping existing '{}'",
existing_event.summary.as_ref().unwrap_or(&"No Title".to_string())
);
}
continue;
}
// Check if this single event would be generated by any recurring event
let is_rrule_generated = deduplicated_recurring.iter().any(|recurring_event| {
// Check if this single event matches the recurring event's pattern (use normalized titles)
let single_title = normalize_title(event.summary.as_ref().unwrap_or(&String::new()));
let recurring_title = normalize_title(recurring_event.summary.as_ref().unwrap_or(&String::new()));
if single_title != recurring_title {
return false; // Different events
}
// Check if this single event would be generated by the recurring event
would_event_be_generated_by_rrule(recurring_event, &event)
});
if is_rrule_generated {
println!("🚫 Discarding RRULE-generated instance: '{}' at {} would be generated by recurring event",
event.summary.as_ref().unwrap_or(&"No Title".to_string()),
event.dtstart.format("%Y-%m-%d %H:%M")
);
} else {
// This is a unique single event
seen_single.insert(dedup_key, deduplicated_single.len());
deduplicated_single.push(event);
}
}
// Combine recurring and single events
let mut result = deduplicated_recurring;
result.extend(deduplicated_single);
println!("📊 Deduplication complete: {} -> {} events ({} recurring, {} single)",
original_count, result.len(),
result.iter().filter(|e| e.rrule.is_some()).count(),
result.iter().filter(|e| e.rrule.is_none()).count()
);
result
}
/// Normalize title for grouping similar events
fn normalize_title(title: &str) -> String {
title.trim()
.to_lowercase()
.chars()
.filter(|c| c.is_alphanumeric() || c.is_whitespace())
.collect::<String>()
.split_whitespace()
.collect::<Vec<&str>>()
.join(" ")
}
/// Consolidate events with the same title but potentially different RRULE patterns
/// This handles cases where calendar systems provide multiple recurring definitions
/// for the same logical meeting (e.g., one RRULE for Tuesdays, another for Thursdays)
fn consolidate_same_title_events(events: Vec<VEvent>) -> Vec<VEvent> {
if events.is_empty() {
return events;
}
// Log the RRULEs we're working with
for event in &events {
if let Some(rrule) = &event.rrule {
println!("🔍 RRULE for '{}': {}",
event.summary.as_ref().unwrap_or(&"No Title".to_string()),
rrule
);
}
}
// Check if all events have similar time patterns and could be consolidated
let first_event = &events[0];
let base_time = first_event.dtstart.time();
let base_duration = if let Some(end) = first_event.dtend {
Some(end.signed_duration_since(first_event.dtstart))
} else {
None
};
// Check if all events have the same time and duration
let can_consolidate = events.iter().all(|event| {
let same_time = event.dtstart.time() == base_time;
let same_duration = match (event.dtend, base_duration) {
(Some(end), Some(base_dur)) => end.signed_duration_since(event.dtstart) == base_dur,
(None, None) => true,
_ => false,
};
same_time && same_duration
});
if !can_consolidate {
println!("🚫 Cannot consolidate events - different times or durations");
// Just deduplicate exact duplicates
return deduplicate_exact_recurring_events(events);
}
// Try to detect if these are complementary weekly patterns
let weekly_events: Vec<_> = events.iter()
.filter(|e| e.rrule.as_ref().map_or(false, |r| r.contains("FREQ=WEEKLY")))
.collect();
if weekly_events.len() >= 2 && weekly_events.len() == events.len() {
// All events are weekly - try to consolidate into a single multi-day weekly pattern
if let Some(consolidated) = consolidate_weekly_patterns(&events) {
println!("✅ Successfully consolidated {} weekly patterns into one", events.len());
return vec![consolidated];
}
}
// If we can't consolidate, just deduplicate exact matches and keep the most complete one
println!("🚫 Cannot consolidate - keeping most complete event");
let deduplicated = deduplicate_exact_recurring_events(events);
// If we still have multiple events, keep only the most complete one
if deduplicated.len() > 1 {
let best_event = deduplicated.into_iter()
.max_by_key(|e| event_completeness_score(e))
.unwrap();
println!("🎯 Kept most complete event: '{}'",
best_event.summary.as_ref().unwrap_or(&"No Title".to_string())
);
vec![best_event]
} else {
deduplicated
}
}
/// Deduplicate exact recurring event matches
fn deduplicate_exact_recurring_events(events: Vec<VEvent>) -> Vec<VEvent> {
use std::collections::HashMap;
let mut seen: HashMap<String, usize> = HashMap::new();
let mut deduplicated = Vec::new();
for event in events {
let dedup_key = format!(
"{}|{}|{}",
event.dtstart.format("%Y%m%dT%H%M%S"),
event.summary.as_ref().unwrap_or(&String::new()),
event.rrule.as_ref().unwrap_or(&String::new())
);
if let Some(&existing_index) = seen.get(&dedup_key) {
let existing_event: &VEvent = &deduplicated[existing_index];
let current_completeness = event_completeness_score(&event);
let existing_completeness = event_completeness_score(existing_event);
if current_completeness > existing_completeness {
println!("🔄 Replacing exact duplicate: Keeping more complete event");
deduplicated[existing_index] = event;
}
} else {
seen.insert(dedup_key, deduplicated.len());
deduplicated.push(event);
}
}
deduplicated
}
/// Attempt to consolidate multiple weekly RRULE patterns into a single pattern
fn consolidate_weekly_patterns(events: &[VEvent]) -> Option<VEvent> {
use std::collections::HashSet;
let mut all_days = HashSet::new();
let mut base_event = None;
for event in events {
let Some(rrule) = &event.rrule else { continue; };
if !rrule.contains("FREQ=WEEKLY") {
continue;
}
// Extract BYDAY if present
if let Some(byday_part) = rrule.split(';').find(|part| part.starts_with("BYDAY=")) {
let days_str = byday_part.strip_prefix("BYDAY=").unwrap_or("");
for day in days_str.split(',') {
all_days.insert(day.trim().to_string());
}
} else {
// If no BYDAY specified, use the weekday from the start date
let weekday = match event.dtstart.weekday() {
chrono::Weekday::Mon => "MO",
chrono::Weekday::Tue => "TU",
chrono::Weekday::Wed => "WE",
chrono::Weekday::Thu => "TH",
chrono::Weekday::Fri => "FR",
chrono::Weekday::Sat => "SA",
chrono::Weekday::Sun => "SU",
};
all_days.insert(weekday.to_string());
}
// Use the first event as the base (we already know they have same time/duration)
if base_event.is_none() {
base_event = Some(event.clone());
}
}
if all_days.is_empty() || base_event.is_none() {
return None;
}
// Create consolidated RRULE
let mut base = base_event.unwrap();
let days_list: Vec<_> = all_days.into_iter().collect();
let byday_str = days_list.join(",");
// Build new RRULE with consolidated BYDAY
let new_rrule = if let Some(existing_rrule) = &base.rrule {
// Remove existing BYDAY and add our consolidated one
let parts: Vec<_> = existing_rrule.split(';')
.filter(|part| !part.starts_with("BYDAY="))
.collect();
format!("{};BYDAY={}", parts.join(";"), byday_str)
} else {
format!("FREQ=WEEKLY;BYDAY={}", byday_str)
};
base.rrule = Some(new_rrule);
println!("🔗 Consolidated weekly pattern: BYDAY={}", byday_str);
Some(base)
}
/// Check if a single event would be generated by a recurring event's RRULE
fn would_event_be_generated_by_rrule(recurring_event: &VEvent, single_event: &VEvent) -> bool {
let Some(rrule) = &recurring_event.rrule else {
return false; // No RRULE to check against
};
// Parse basic RRULE patterns
if rrule.contains("FREQ=DAILY") {
// Daily recurrence
let interval = extract_interval_from_rrule(rrule).unwrap_or(1);
let days_diff = (single_event.dtstart.date_naive() - recurring_event.dtstart.date_naive()).num_days();
if days_diff >= 0 && days_diff % interval as i64 == 0 {
// Check if times match (allowing for timezone differences within same day)
let recurring_time = recurring_event.dtstart.time();
let single_time = single_event.dtstart.time();
return recurring_time == single_time;
}
} else if rrule.contains("FREQ=WEEKLY") {
// Weekly recurrence
let interval = extract_interval_from_rrule(rrule).unwrap_or(1);
let days_diff = (single_event.dtstart.date_naive() - recurring_event.dtstart.date_naive()).num_days();
// First check if it's the same day of week and time
let recurring_weekday = recurring_event.dtstart.weekday();
let single_weekday = single_event.dtstart.weekday();
let recurring_time = recurring_event.dtstart.time();
let single_time = single_event.dtstart.time();
if recurring_weekday == single_weekday && recurring_time == single_time && days_diff >= 0 {
// Calculate how many weeks apart they are
let weeks_diff = days_diff / 7;
// Check if this falls on an interval boundary
return weeks_diff % interval as i64 == 0;
}
} else if rrule.contains("FREQ=MONTHLY") {
// Monthly recurrence - simplified check
let months_diff = (single_event.dtstart.year() - recurring_event.dtstart.year()) * 12
+ (single_event.dtstart.month() as i32 - recurring_event.dtstart.month() as i32);
if months_diff >= 0 {
let interval = extract_interval_from_rrule(rrule).unwrap_or(1) as i32;
if months_diff % interval == 0 {
// Same day of month and time
return recurring_event.dtstart.day() == single_event.dtstart.day()
&& recurring_event.dtstart.time() == single_event.dtstart.time();
}
}
}
false
}
/// Extract INTERVAL value from RRULE string, defaulting to 1 if not found
fn extract_interval_from_rrule(rrule: &str) -> Option<u32> {
for part in rrule.split(';') {
if part.starts_with("INTERVAL=") {
return part.strip_prefix("INTERVAL=")
.and_then(|s| s.parse().ok());
}
}
Some(1) // Default interval is 1 if not specified
}
/// Calculate a completeness score for an event based on how many optional fields are filled
fn event_completeness_score(event: &VEvent) -> u32 {
let mut score = 0;
if event.summary.is_some() { score += 1; }
if event.description.is_some() { score += 1; }
if event.location.is_some() { score += 1; }
if event.dtend.is_some() { score += 1; }
if event.rrule.is_some() { score += 1; }
if !event.categories.is_empty() { score += 1; }
if !event.alarms.is_empty() { score += 1; }
if event.organizer.is_some() { score += 1; }
if !event.attendees.is_empty() { score += 1; }
score
}