imessage_database/tables/
handle.rs

1/*!
2 This module represents common (but not all) columns in the `handle` table.
3*/
4
5use rusqlite::{Connection, Error, Result, Row, Statement};
6use std::collections::{BTreeSet, HashMap};
7
8use crate::{
9    error::table::TableError,
10    tables::table::{Cacheable, Deduplicate, Diagnostic, HANDLE, ME, Table},
11    util::output::{done_processing, processing},
12};
13
14/// Represents a single row in the `handle` table.
15#[derive(Debug)]
16pub struct Handle {
17    /// The unique identifier for the handle in the database
18    pub rowid: i32,
19    /// Identifier for a contact, i.e. a phone number or email address
20    pub id: String,
21    /// Field used to disambiguate divergent handles that represent the same contact
22    pub person_centric_id: Option<String>,
23}
24
25impl Table for Handle {
26    fn from_row(row: &Row) -> Result<Handle> {
27        Ok(Handle {
28            rowid: row.get("rowid")?,
29            id: row.get("id")?,
30            person_centric_id: row.get("person_centric_id").unwrap_or(None),
31        })
32    }
33
34    fn get(db: &Connection) -> Result<Statement, TableError> {
35        Ok(db.prepare(&format!("SELECT * from {HANDLE}"))?)
36    }
37
38    fn extract(handle: Result<Result<Self, Error>, Error>) -> Result<Self, TableError> {
39        match handle {
40            Ok(Ok(handle)) => Ok(handle),
41            Err(why) | Ok(Err(why)) => Err(TableError::QueryError(why)),
42        }
43    }
44}
45
46impl Cacheable for Handle {
47    type K = i32;
48    type V = String;
49    /// Generate a `HashMap` for looking up contacts by their IDs, collapsing
50    /// duplicate contacts to the same ID String regardless of service
51    ///
52    /// # Example:
53    ///
54    /// ```
55    /// use imessage_database::util::dirs::default_db_path;
56    /// use imessage_database::tables::table::{Cacheable, get_connection};
57    /// use imessage_database::tables::handle::Handle;
58    ///
59    /// let db_path = default_db_path();
60    /// let conn = get_connection(&db_path).unwrap();
61    /// let chatrooms = Handle::cache(&conn);
62    /// ```
63    fn cache(db: &Connection) -> Result<HashMap<Self::K, Self::V>, TableError> {
64        // Create cache for user IDs
65        let mut map = HashMap::new();
66        // Handle ID 0 is self in group chats
67        map.insert(0, ME.to_string());
68
69        // Create query
70        let mut statement = Handle::get(db)?;
71
72        // Execute query to build the Handles
73        let handles = statement.query_map([], |row| Ok(Handle::from_row(row)))?;
74
75        // Iterate over the handles and update the map
76        for handle in handles {
77            let contact = Handle::extract(handle)?;
78            map.insert(contact.rowid, contact.id);
79        }
80
81        // Condense contacts that share person_centric_id so their IDs map to the same strings
82        let dupe_contacts = Handle::get_person_id_map(db)?;
83        for contact in dupe_contacts {
84            let (id, new) = contact;
85            map.insert(id, new);
86        }
87
88        // Done!
89        Ok(map)
90    }
91}
92
93impl Deduplicate for Handle {
94    type T = String;
95
96    /// Given the initial set of duplicated handles, deduplicate them
97    ///
98    /// This returns a new hashmap that maps the real handle ID to a new deduplicated unique handle ID
99    /// that represents a single handle for all of the deduplicate handles.
100    ///
101    /// Assuming no new handles have been written to the database, deduplicated data is deterministic across runs.
102    ///
103    /// # Example:
104    ///
105    /// ```
106    /// use imessage_database::util::dirs::default_db_path;
107    /// use imessage_database::tables::table::{Cacheable, Deduplicate, get_connection};
108    /// use imessage_database::tables::handle::Handle;
109    ///
110    /// let db_path = default_db_path();
111    /// let conn = get_connection(&db_path).unwrap();
112    /// let chatrooms = Handle::cache(&conn).unwrap();
113    /// let deduped_chatrooms = Handle::dedupe(&chatrooms);
114    /// ```
115    fn dedupe(duplicated_data: &HashMap<i32, Self::T>) -> HashMap<i32, i32> {
116        let mut deduplicated_participants: HashMap<i32, i32> = HashMap::new();
117        let mut participant_to_unique_participant_id: HashMap<Self::T, i32> = HashMap::new();
118
119        // Build cache of each unique set of participants to a new identifier:
120        let mut unique_participant_identifier = 0;
121
122        // Iterate over the values in a deterministic order
123        let mut sorted_dupes: Vec<(&i32, &Self::T)> = duplicated_data.iter().collect();
124        sorted_dupes.sort_by(|(a, _), (b, _)| a.cmp(b));
125
126        for (participant_id, participant) in sorted_dupes {
127            if let Some(id) = participant_to_unique_participant_id.get(participant) {
128                deduplicated_participants.insert(participant_id.to_owned(), id.to_owned());
129            } else {
130                participant_to_unique_participant_id
131                    .insert(participant.to_owned(), unique_participant_identifier);
132                deduplicated_participants
133                    .insert(participant_id.to_owned(), unique_participant_identifier);
134                unique_participant_identifier += 1;
135            }
136        }
137        deduplicated_participants
138    }
139}
140
141impl Diagnostic for Handle {
142    /// Emit diagnostic data for the Handles table
143    ///
144    /// Get the number of handles that are duplicated
145    ///
146    /// The `person_centric_id` is used to map handles that represent the
147    /// same contact across ids (numbers, emails, etc) and across
148    /// services (iMessage, Jabber, iChat, SMS, etc)
149    ///
150    /// In some databases, `person_centric_id` may not be available.
151    ///
152    /// # Example:
153    ///
154    /// ```
155    /// use imessage_database::util::dirs::default_db_path;
156    /// use imessage_database::tables::table::{Diagnostic, get_connection};
157    /// use imessage_database::tables::handle::Handle;
158    ///
159    /// let db_path = default_db_path();
160    /// let conn = get_connection(&db_path).unwrap();
161    /// Handle::run_diagnostic(&conn);
162    /// ```
163    fn run_diagnostic(db: &Connection) -> Result<(), TableError> {
164        let query = concat!(
165            "SELECT COUNT(DISTINCT person_centric_id) ",
166            "FROM handle ",
167            "WHERE person_centric_id NOT NULL"
168        );
169
170        if let Ok(mut rows) = db.prepare(query) {
171            processing();
172
173            let count_dupes: Option<i32> = rows.query_row([], |r| r.get(0))?;
174
175            done_processing();
176
177            if let Some(dupes) = count_dupes {
178                if dupes > 0 {
179                    println!("Handle diagnostic data:");
180                    println!("    Contacts with more than one ID: {dupes}");
181                }
182            }
183        }
184
185        Ok(())
186    }
187}
188
189impl Handle {
190    /// The handles table does not have a lot of information and can have many duplicate values.
191    ///
192    /// This method generates a hashmap of each separate item in this table to a combined string
193    /// that represents all of the copies, so any handle ID will always map to the same string
194    /// for a given chat participant
195    fn get_person_id_map(db: &Connection) -> Result<HashMap<i32, String>, TableError> {
196        let mut person_to_id: HashMap<String, BTreeSet<String>> = HashMap::new();
197        let mut row_to_id: HashMap<i32, String> = HashMap::new();
198        let mut row_data: Vec<(String, i32, String)> = vec![];
199
200        // Build query
201        let query = concat!(
202            "SELECT DISTINCT A.person_centric_id, A.rowid, A.id ",
203            "FROM handle A ",
204            "INNER JOIN handle B ON B.id = A.id ",
205            "WHERE A.person_centric_id NOT NULL ",
206            "ORDER BY A.person_centric_id",
207        );
208        let statement = db.prepare(query);
209
210        if let Ok(mut statement) = statement {
211            // Cache the results of the query in memory
212            let contacts = statement.query_map([], |row| {
213                let person_centric_id: String = row.get(0)?;
214                let rowid: i32 = row.get(1)?;
215                let id: String = row.get(2)?;
216                Ok((person_centric_id, rowid, id))
217            })?;
218
219            for contact in contacts {
220                row_data.push(contact?);
221            }
222
223            // First pass: generate a map of each person_centric_id to its matching ids
224            for contact in &row_data {
225                let (person_centric_id, _, id) = contact;
226                if let Some(set) = person_to_id.get_mut(person_centric_id) {
227                    set.insert(id.to_owned());
228                } else {
229                    let mut set = BTreeSet::new();
230                    set.insert(id.to_owned());
231                    person_to_id.insert(person_centric_id.to_owned(), set);
232                }
233            }
234
235            // Second pass: point each ROWID to the matching ids
236            for contact in &row_data {
237                let (person_centric_id, rowid, _) = contact;
238                let data_to_insert = match person_to_id.get_mut(person_centric_id) {
239                    Some(person) => person.iter().cloned().collect::<Vec<String>>().join(" "),
240                    None => panic!("Attempted to resolve contact with no person_centric_id!"),
241                };
242                row_to_id.insert(rowid.to_owned(), data_to_insert);
243            }
244        }
245
246        Ok(row_to_id)
247    }
248}
249
250#[cfg(test)]
251mod tests {
252    use crate::tables::{handle::Handle, table::Deduplicate};
253    use std::collections::{HashMap, HashSet};
254
255    #[test]
256    fn test_can_dedupe() {
257        let mut input: HashMap<i32, String> = HashMap::new();
258        input.insert(1, String::from("A")); // 0
259        input.insert(2, String::from("A")); // 0
260        input.insert(3, String::from("A")); // 0
261        input.insert(4, String::from("B")); // 1
262        input.insert(5, String::from("B")); // 1
263        input.insert(6, String::from("C")); // 2
264
265        let output = Handle::dedupe(&input);
266        let expected_deduped_ids: HashSet<i32> = output.values().copied().collect();
267        assert_eq!(expected_deduped_ids.len(), 3);
268    }
269
270    #[test]
271    // Simulate 3 runs of the program and ensure that the order of the deduplicated contacts is stable
272    fn test_same_values() {
273        let mut input_1: HashMap<i32, String> = HashMap::new();
274        input_1.insert(1, String::from("A"));
275        input_1.insert(2, String::from("A"));
276        input_1.insert(3, String::from("A"));
277        input_1.insert(4, String::from("B"));
278        input_1.insert(5, String::from("B"));
279        input_1.insert(6, String::from("C"));
280
281        let mut input_2: HashMap<i32, String> = HashMap::new();
282        input_2.insert(1, String::from("A"));
283        input_2.insert(2, String::from("A"));
284        input_2.insert(3, String::from("A"));
285        input_2.insert(4, String::from("B"));
286        input_2.insert(5, String::from("B"));
287        input_2.insert(6, String::from("C"));
288
289        let mut input_3: HashMap<i32, String> = HashMap::new();
290        input_3.insert(1, String::from("A"));
291        input_3.insert(2, String::from("A"));
292        input_3.insert(3, String::from("A"));
293        input_3.insert(4, String::from("B"));
294        input_3.insert(5, String::from("B"));
295        input_3.insert(6, String::from("C"));
296
297        let mut output_1 = Handle::dedupe(&input_1)
298            .into_iter()
299            .collect::<Vec<(i32, i32)>>();
300        let mut output_2 = Handle::dedupe(&input_2)
301            .into_iter()
302            .collect::<Vec<(i32, i32)>>();
303        let mut output_3 = Handle::dedupe(&input_3)
304            .into_iter()
305            .collect::<Vec<(i32, i32)>>();
306
307        output_1.sort_unstable();
308        output_2.sort_unstable();
309        output_3.sort_unstable();
310
311        assert_eq!(output_1, output_2);
312        assert_eq!(output_1, output_3);
313        assert_eq!(output_2, output_3);
314    }
315}
imessage_database/tables/handle.rs

imessage_database/tables/
handle.rs
