relay_base_schema/
data_category.rs

1//! Defines the [`DataCategory`] type that classifies data Relay can handle.
2
3use std::fmt;
4use std::str::FromStr;
5
6use serde::{Deserialize, Serialize};
7
8use crate::events::EventType;
9
10/// An error that occurs if a number cannot be converted into a [`DataCategory`].
11#[derive(Debug, PartialEq, thiserror::Error)]
12#[error("Unknown numeric data category {0} can not be converted into a DataCategory.")]
13pub struct UnknownDataCategory(pub u8);
14
15/// Classifies the type of data that is being ingested.
16#[derive(Clone, Copy, Debug, Eq, Hash, Ord, PartialEq, PartialOrd, Deserialize, Serialize)]
17#[serde(rename_all = "snake_case")]
18#[repr(i8)]
19pub enum DataCategory {
20    /// Reserved and unused.
21    Default = 0,
22    /// Error events and Events with an `event_type` not explicitly listed below.
23    Error = 1,
24    /// Transaction events.
25    Transaction = 2,
26    /// Events with an event type of `csp`, `hpkp`, `expectct` and `expectstaple`.
27    Security = 3,
28    /// An attachment. Quantity is the size of the attachment in bytes.
29    Attachment = 4,
30    /// Session updates. Quantity is the number of updates in the batch.
31    Session = 5,
32    /// Profile
33    ///
34    /// This is the category for processed profiles (all profiles, whether or not we store them).
35    Profile = 6,
36    /// Session Replays
37    Replay = 7,
38    /// DEPRECATED: A transaction for which metrics were extracted.
39    ///
40    /// This category is now obsolete because the `Transaction` variant will represent
41    /// processed transactions from now on.
42    TransactionProcessed = 8,
43    /// Indexed transaction events.
44    ///
45    /// This is the category for transaction payloads that were accepted and stored in full. In
46    /// contrast, `transaction` only guarantees that metrics have been accepted for the transaction.
47    TransactionIndexed = 9,
48    /// Monitor check-ins.
49    Monitor = 10,
50    /// Indexed Profile
51    ///
52    /// This is the category for indexed profiles that will be stored later.
53    ProfileIndexed = 11,
54    /// Span
55    ///
56    /// This is the category for spans from which we extracted metrics from.
57    Span = 12,
58    /// Monitor Seat
59    ///
60    /// Represents a monitor job that has scheduled monitor checkins. The seats are not ingested
61    /// but we define it here to prevent clashing values since this data category enumeration
62    /// is also used outside of Relay via the Python package.
63    MonitorSeat = 13,
64    /// User Feedback
65    ///
66    /// Represents a User Feedback processed.
67    /// Currently standardized on name UserReportV2 to avoid clashing with the old UserReport.
68    /// TODO(jferg): Rename this to UserFeedback once old UserReport is deprecated.
69    UserReportV2 = 14,
70    /// Metric buckets.
71    MetricBucket = 15,
72    /// SpanIndexed
73    ///
74    /// This is the category for spans we store in full.
75    SpanIndexed = 16,
76    /// ProfileDuration
77    ///
78    /// This data category is used to count the number of milliseconds per indexed profile chunk,
79    /// excluding UI profile chunks.
80    ProfileDuration = 17,
81    /// ProfileChunk
82    ///
83    /// This is a count of profile chunks received. It will not be used for billing but will be
84    /// useful for customers to track what's being dropped.
85    ProfileChunk = 18,
86    /// MetricSecond
87    ///
88    /// Reserved by billing to summarize the bucketed product of metric volume
89    /// and metric cardinality. Defined here so as not to clash with future
90    /// categories.
91    MetricSecond = 19,
92    /// Replay Video
93    ///
94    /// This is the data category for Session Replays produced via a video recording.
95    DoNotUseReplayVideo = 20,
96    /// This is the data category for Uptime monitors.
97    Uptime = 21,
98    /// Counts the number of individual attachments, as opposed to the number of bytes in an attachment.
99    AttachmentItem = 22,
100    /// LogItem
101    ///
102    /// This is the category for logs for which we store the count log events for users for measuring
103    /// missing breadcrumbs, and count of logs for rate limiting purposes.
104    LogItem = 23,
105    /// LogByte
106    ///
107    /// This is the category for logs for which we store log event total bytes for users.
108    LogByte = 24,
109    /// Profile duration of a UI profile.
110    ///
111    /// This data category is used to count the number of milliseconds per indexed UI profile
112    /// chunk.
113    ///
114    /// See also: [`Self::ProfileDuration`]
115    ProfileDurationUi = 25,
116    /// UI Profile Chunk.
117    ///
118    /// This data category is used to count the number of milliseconds per indexed UI profile
119    /// chunk.
120    ///
121    /// See also: [`Self::ProfileChunk`]
122    ProfileChunkUi = 26,
123    /// This is the data category to count Seer Autofix run events.
124    SeerAutofix = 27,
125    /// This is the data category to count Seer Scanner run events.
126    SeerScanner = 28,
127    //
128    // IMPORTANT: After adding a new entry to DataCategory, go to the `relay-cabi` subfolder and run
129    // `make header` to regenerate the C-binding. This allows using the data category from Python.
130    // Rerun this step every time the **code name** of the variant is updated.
131    //
132    /// Any other data category not known by this Relay.
133    #[serde(other)]
134    Unknown = -1,
135}
136
137impl DataCategory {
138    /// Returns the data category corresponding to the given name.
139    pub fn from_name(string: &str) -> Self {
140        // TODO: This should probably use serde.
141        match string {
142            "default" => Self::Default,
143            "error" => Self::Error,
144            "transaction" => Self::Transaction,
145            "security" => Self::Security,
146            "attachment" => Self::Attachment,
147            "session" => Self::Session,
148            "profile" => Self::Profile,
149            "profile_indexed" => Self::ProfileIndexed,
150            "replay" => Self::Replay,
151            "transaction_processed" => Self::TransactionProcessed,
152            "transaction_indexed" => Self::TransactionIndexed,
153            "monitor" => Self::Monitor,
154            "span" => Self::Span,
155            "log_item" => Self::LogItem,
156            "log_byte" => Self::LogByte,
157            "monitor_seat" => Self::MonitorSeat,
158            "feedback" => Self::UserReportV2,
159            "user_report_v2" => Self::UserReportV2,
160            "metric_bucket" => Self::MetricBucket,
161            "span_indexed" => Self::SpanIndexed,
162            "profile_duration" => Self::ProfileDuration,
163            "profile_duration_ui" => Self::ProfileDurationUi,
164            "profile_chunk" => Self::ProfileChunk,
165            "profile_chunk_ui" => Self::ProfileChunkUi,
166            "metric_second" => Self::MetricSecond,
167            "replay_video" => Self::DoNotUseReplayVideo,
168            "uptime" => Self::Uptime,
169            "attachment_item" => Self::AttachmentItem,
170            "seer_autofix" => Self::SeerAutofix,
171            "seer_scanner" => Self::SeerScanner,
172            _ => Self::Unknown,
173        }
174    }
175
176    /// Returns the canonical name of this data category.
177    pub fn name(self) -> &'static str {
178        // TODO: This should probably use serde.
179        match self {
180            Self::Default => "default",
181            Self::Error => "error",
182            Self::Transaction => "transaction",
183            Self::Security => "security",
184            Self::Attachment => "attachment",
185            Self::Session => "session",
186            Self::Profile => "profile",
187            Self::ProfileIndexed => "profile_indexed",
188            Self::Replay => "replay",
189            Self::DoNotUseReplayVideo => "replay_video",
190            Self::TransactionProcessed => "transaction_processed",
191            Self::TransactionIndexed => "transaction_indexed",
192            Self::Monitor => "monitor",
193            Self::Span => "span",
194            Self::LogItem => "log_item",
195            Self::LogByte => "log_byte",
196            Self::MonitorSeat => "monitor_seat",
197            Self::UserReportV2 => "feedback",
198            Self::MetricBucket => "metric_bucket",
199            Self::SpanIndexed => "span_indexed",
200            Self::ProfileDuration => "profile_duration",
201            Self::ProfileDurationUi => "profile_duration_ui",
202            Self::ProfileChunk => "profile_chunk",
203            Self::ProfileChunkUi => "profile_chunk_ui",
204            Self::MetricSecond => "metric_second",
205            Self::Uptime => "uptime",
206            Self::AttachmentItem => "attachment_item",
207            Self::SeerAutofix => "seer_autofix",
208            Self::SeerScanner => "seer_scanner",
209            Self::Unknown => "unknown",
210        }
211    }
212
213    /// Returns true if the DataCategory refers to an error (i.e an error event).
214    pub fn is_error(self) -> bool {
215        matches!(self, Self::Error | Self::Default | Self::Security)
216    }
217
218    /// Returns the numeric value for this outcome.
219    pub fn value(self) -> Option<u8> {
220        // negative values (Internal and Unknown) cannot be sent as
221        // outcomes (internally so!)
222        (self as i8).try_into().ok()
223    }
224
225    /// Returns a dedicated category for indexing if this data can be converted to metrics.
226    ///
227    /// This returns `None` for most data categories.
228    pub fn index_category(self) -> Option<Self> {
229        match self {
230            Self::Transaction => Some(Self::TransactionIndexed),
231            Self::Span => Some(Self::SpanIndexed),
232            Self::Profile => Some(Self::ProfileIndexed),
233            _ => None,
234        }
235    }
236
237    /// Returns `true` if this data category is an indexed data category.
238    pub fn is_indexed(self) -> bool {
239        matches!(
240            self,
241            Self::TransactionIndexed | Self::SpanIndexed | Self::ProfileIndexed
242        )
243    }
244}
245
246impl fmt::Display for DataCategory {
247    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
248        write!(f, "{}", self.name())
249    }
250}
251
252impl FromStr for DataCategory {
253    type Err = ();
254
255    fn from_str(string: &str) -> Result<Self, Self::Err> {
256        Ok(Self::from_name(string))
257    }
258}
259
260impl From<EventType> for DataCategory {
261    fn from(ty: EventType) -> Self {
262        match ty {
263            EventType::Default | EventType::Error | EventType::Nel => Self::Error,
264            EventType::Transaction => Self::Transaction,
265            EventType::Csp | EventType::Hpkp | EventType::ExpectCt | EventType::ExpectStaple => {
266                Self::Security
267            }
268            EventType::UserReportV2 => Self::UserReportV2,
269        }
270    }
271}
272
273impl TryFrom<u8> for DataCategory {
274    type Error = UnknownDataCategory;
275
276    fn try_from(value: u8) -> Result<Self, UnknownDataCategory> {
277        match value {
278            0 => Ok(Self::Default),
279            1 => Ok(Self::Error),
280            2 => Ok(Self::Transaction),
281            3 => Ok(Self::Security),
282            4 => Ok(Self::Attachment),
283            5 => Ok(Self::Session),
284            6 => Ok(Self::Profile),
285            7 => Ok(Self::Replay),
286            8 => Ok(Self::TransactionProcessed),
287            9 => Ok(Self::TransactionIndexed),
288            10 => Ok(Self::Monitor),
289            11 => Ok(Self::ProfileIndexed),
290            12 => Ok(Self::Span),
291            13 => Ok(Self::MonitorSeat),
292            14 => Ok(Self::UserReportV2),
293            15 => Ok(Self::MetricBucket),
294            16 => Ok(Self::SpanIndexed),
295            17 => Ok(Self::ProfileDuration),
296            18 => Ok(Self::ProfileChunk),
297            19 => Ok(Self::MetricSecond),
298            20 => Ok(Self::DoNotUseReplayVideo),
299            21 => Ok(Self::Uptime),
300            22 => Ok(Self::AttachmentItem),
301            23 => Ok(Self::LogItem),
302            24 => Ok(Self::LogByte),
303            25 => Ok(Self::ProfileDurationUi),
304            26 => Ok(Self::ProfileChunkUi),
305            27 => Ok(Self::SeerAutofix),
306            28 => Ok(Self::SeerScanner),
307            other => Err(UnknownDataCategory(other)),
308        }
309    }
310}
311
312#[cfg(test)]
313mod tests {
314    use super::*;
315
316    #[test]
317    pub fn test_last_variant_conversion() {
318        // If this test fails, update the numeric bounds so that the first assertion
319        // maps to the last variant in the enum and the second assertion produces an error
320        // that the DataCategory does not exist.
321        assert_eq!(DataCategory::try_from(28), Ok(DataCategory::SeerScanner));
322        assert_eq!(DataCategory::try_from(29), Err(UnknownDataCategory(29)));
323    }
324}