implement fuzzing

Notes:

- The fuzz seed is now derived from the card id and # of reps, so
if a card is undone and done again, the same fuzz will be used.
- The intervals shown on the answer buttons now include the fuzz, instead
of hiding it from the user. This will prevent questions about due dates
being different to what was shown on the buttons, but will create
questions about due dates being different for cards with the same
interval, and some people may find it distracting for learning cards.
The new approach is easier to reason about, but time will tell
whether it's a net gain or not.
- The env var we were using to shift the clock away from rollover for
unit tests has been repurposed to also disable fuzzing, which simplifies
the tests.
- Cards in filtered decks without scheduling now have the preview delay
fuzzed.
- Sub-day learning cards are mostly fuzzed like before, but will apply
the up-to-5-minutes of fuzz regardless of the time of day.
- The answer buttons now round minute values, as the fuzz on short
intervals is distracting.

pylib/tests/run_pytest.py

@@ -3,7 +3,7 @@ import sys
 
 import pytest
 
-os.environ["SHIFT_CLOCK_HACK"] = "1"
+os.environ["ANKI_TEST_MODE"] = "1"
 
 if __name__ == "__main__":
     folder = os.path.join(os.path.dirname(__file__), "..", "tests")

rslib/src/sched/answering.rs

@@ -67,6 +67,11 @@ impl AnswerContext {
             lapse_multiplier: self.config.inner.lapse_multiplier,
             minimum_lapse_interval: self.config.inner.minimum_lapse_interval,
             in_filtered_deck: self.deck.is_filtered(),
+            preview_step: if let DeckKind::Filtered(deck) = &self.deck.kind {
+                deck.preview_delay
+            } else {
+                0
+            },
         }
     }
 
@@ -540,7 +545,7 @@ impl Collection {
             config: self.deck_config_for_card(card)?,
             timing,
             now: TimestampSecs::now(),
-            fuzz_seed: None,
+            fuzz_seed: get_fuzz_seed(card),
         })
     }
 
@@ -552,3 +557,13 @@ impl Collection {
         Ok(current.next_states(&state_ctx))
     }
 }
+
+/// Return a consistent seed for a given card at a given number of reps.
+/// If in test environment, disable fuzzing.
+fn get_fuzz_seed(card: &Card) -> Option<u64> {
+    if *crate::timestamp::TESTING {
+        None
+    } else {
+        Some((card.id.0 as u64).wrapping_add(card.reps as u64))
+    }
+}

rslib/src/sched/states/filtered.rs

@@ -28,7 +28,7 @@ impl FilteredState {
 
     pub(crate) fn next_states(self, ctx: &StateContext) -> NextCardStates {
         match self {
-            FilteredState::Preview(state) => state.next_states(),
+            FilteredState::Preview(state) => state.next_states(ctx),
             FilteredState::Rescheduling(state) => state.next_states(ctx),
         }
     }

rslib/src/sched/states/learning.rs

@@ -33,20 +33,20 @@ impl LearnState {
     fn answer_again(self, ctx: &StateContext) -> LearnState {
         LearnState {
             remaining_steps: ctx.steps.remaining_for_failed(),
-            scheduled_secs: ctx.steps.again_delay_secs_learn(),
+            scheduled_secs: ctx.with_learning_fuzz(ctx.steps.again_delay_secs_learn()),
         }
     }
 
     fn answer_hard(self, ctx: &StateContext) -> CardState {
         if let Some(hard_delay) = ctx.steps.hard_delay_secs(self.remaining_steps) {
             LearnState {
-                scheduled_secs: hard_delay,
+                scheduled_secs: ctx.with_learning_fuzz(hard_delay),
                 ..self
             }
             .into()
         } else {
             ReviewState {
-                scheduled_days: ctx.graduating_interval_good,
+                scheduled_days: ctx.fuzzed_graduating_interval_good(),
                 ..Default::default()
             }
             .into()
@@ -57,12 +57,12 @@ impl LearnState {
         if let Some(good_delay) = ctx.steps.good_delay_secs(self.remaining_steps) {
             LearnState {
                 remaining_steps: ctx.steps.remaining_for_good(self.remaining_steps),
-                scheduled_secs: good_delay,
+                scheduled_secs: ctx.with_learning_fuzz(good_delay),
             }
             .into()
         } else {
             ReviewState {
-                scheduled_days: ctx.graduating_interval_good,
+                scheduled_days: ctx.fuzzed_graduating_interval_good(),
                 ..Default::default()
             }
             .into()
@@ -71,7 +71,7 @@ impl LearnState {
 
     fn answer_easy(self, ctx: &StateContext) -> ReviewState {
         ReviewState {
-            scheduled_days: ctx.graduating_interval_easy,
+            scheduled_days: ctx.fuzzed_graduating_interval_easy(),
             ..Default::default()
         }
     }

rslib/src/sched/states/mod.rs

@@ -12,6 +12,9 @@ pub(crate) mod rescheduling_filter;
 pub(crate) mod review;
 pub(crate) mod steps;
 
+use rand::prelude::*;
+use rand::rngs::StdRng;
+
 pub use {
     filtered::FilteredState, learning::LearnState, new::NewState, normal::NormalState,
     preview_filter::PreviewState, relearning::RelearnState,
@@ -75,6 +78,64 @@ pub(crate) struct StateContext<'a> {
 
     // filtered
     pub in_filtered_deck: bool,
+    pub preview_step: u32,
+}
+
+impl<'a> StateContext<'a> {
+    pub(crate) fn with_review_fuzz(&self, interval: f32) -> u32 {
+        // fixme: floor() is to match python
+        let interval = interval.floor();
+        if let Some(seed) = self.fuzz_seed {
+            let mut rng = StdRng::seed_from_u64(seed);
+            let (lower, upper) = if interval < 2.0 {
+                (1.0, 1.0)
+            } else if interval < 3.0 {
+                (2.0, 3.0)
+            } else if interval < 7.0 {
+                fuzz_range(interval, 0.25, 0.0)
+            } else if interval < 30.0 {
+                fuzz_range(interval, 0.15, 2.0)
+            } else {
+                fuzz_range(interval, 0.05, 4.0)
+            };
+            if lower >= upper {
+                lower
+            } else {
+                rng.gen_range(lower, upper)
+            }
+        } else {
+            interval
+        }
+        .round() as u32
+    }
+
+    /// Add up to 25% increase to seconds, but no more than 5 minutes.
+    pub(crate) fn with_learning_fuzz(&self, secs: u32) -> u32 {
+        if let Some(seed) = self.fuzz_seed {
+            let mut rng = StdRng::seed_from_u64(seed);
+            let upper_exclusive = secs + ((secs as f32) * 0.25).min(300.0).floor() as u32;
+            if secs >= upper_exclusive {
+                secs
+            } else {
+                rng.gen_range(secs, upper_exclusive)
+            }
+        } else {
+            secs
+        }
+    }
+
+    pub(crate) fn fuzzed_graduating_interval_good(&self) -> u32 {
+        self.with_review_fuzz(self.graduating_interval_good as f32)
+    }
+
+    pub(crate) fn fuzzed_graduating_interval_easy(&self) -> u32 {
+        self.with_review_fuzz(self.graduating_interval_easy as f32)
+    }
+}
+
+fn fuzz_range(interval: f32, factor: f32, minimum: f32) -> (f32, f32) {
+    let delta = (interval * factor).max(minimum).max(1.0);
+    (interval - delta, interval + delta + 1.0)
 }
 
 #[derive(Debug)]

rslib/src/sched/states/preview_filter.rs

@@ -1,7 +1,7 @@
 // Copyright: Ankitects Pty Ltd and contributors
 // License: GNU AGPL, version 3 or later; http://www.gnu.org/licenses/agpl.html
 
-use super::{IntervalKind, NextCardStates, NormalState};
+use super::{IntervalKind, NextCardStates, NormalState, StateContext};
 
 #[derive(Debug, Clone, Copy, PartialEq)]
 pub struct PreviewState {
@@ -14,10 +14,14 @@ impl PreviewState {
         IntervalKind::InSecs(self.scheduled_secs)
     }
 
-    pub(crate) fn next_states(self) -> NextCardStates {
+    pub(crate) fn next_states(self, ctx: &StateContext) -> NextCardStates {
         NextCardStates {
             current: self.into(),
-            again: self.into(),
+            again: PreviewState {
+                scheduled_secs: ctx.with_learning_fuzz(ctx.preview_step * 60),
+                ..self
+            }
+            .into(),
             hard: self.original_state.into(),
             good: self.original_state.into(),
             easy: self.original_state.into(),

rslib/src/sched/states/relearning.rs

@@ -33,11 +33,11 @@ impl RelearnState {
     }
 
     fn answer_again(self, ctx: &StateContext) -> CardState {
-        if let Some(learn_interval) = ctx.relearn_steps.again_delay_secs_relearn() {
+        if let Some(again_delay) = ctx.relearn_steps.again_delay_secs_relearn() {
             RelearnState {
                 learning: LearnState {
                     remaining_steps: ctx.relearn_steps.remaining_for_failed(),
-                    scheduled_secs: learn_interval,
+                    scheduled_secs: ctx.with_learning_fuzz(again_delay),
                 },
                 review: ReviewState {
                     scheduled_days: self.review.failing_review_interval(ctx),
@@ -52,13 +52,13 @@ impl RelearnState {
     }
 
     fn answer_hard(self, ctx: &StateContext) -> CardState {
-        if let Some(learn_interval) = ctx
+        if let Some(hard_delay) = ctx
             .relearn_steps
             .hard_delay_secs(self.learning.remaining_steps)
         {
             RelearnState {
                 learning: LearnState {
-                    scheduled_secs: learn_interval,
+                    scheduled_secs: ctx.with_learning_fuzz(hard_delay),
                     ..self.learning
                 },
                 review: ReviewState {
@@ -73,13 +73,13 @@ impl RelearnState {
     }
 
     fn answer_good(self, ctx: &StateContext) -> CardState {
-        if let Some(learn_interval) = ctx
+        if let Some(good_delay) = ctx
             .relearn_steps
             .good_delay_secs(self.learning.remaining_steps)
         {
             RelearnState {
                 learning: LearnState {
-                    scheduled_secs: learn_interval,
+                    scheduled_secs: ctx.with_learning_fuzz(good_delay),
                     remaining_steps: ctx
                         .relearn_steps
                         .remaining_for_good(self.learning.remaining_steps),

rslib/src/sched/states/review.rs

@@ -1,9 +1,6 @@
 // Copyright: Ankitects Pty Ltd and contributors
 // License: GNU AGPL, version 3 or later; http://www.gnu.org/licenses/agpl.html
 
-use rand::prelude::*;
-use rand::rngs::StdRng;
-
 use crate::revlog::RevlogReviewKind;
 
 use super::{
@@ -81,11 +78,11 @@ impl ReviewState {
             lapses: self.lapses + 1,
         };
 
-        if let Some(learn_interval) = ctx.relearn_steps.again_delay_secs_relearn() {
+        if let Some(again_delay) = ctx.relearn_steps.again_delay_secs_relearn() {
             RelearnState {
                 learning: LearnState {
                     remaining_steps: ctx.relearn_steps.remaining_for_failed(),
-                    scheduled_secs: learn_interval,
+                    scheduled_secs: ctx.with_learning_fuzz(again_delay),
                 },
                 review: again_review,
             }
@@ -143,16 +140,18 @@ impl ReviewState {
         // fixme: floor() is to match python
 
         let hard_interval =
-            constrain_passing_interval(ctx, current_interval * hard_factor, hard_minimum);
+            constrain_passing_interval(ctx, current_interval * hard_factor, hard_minimum, true);
         let good_interval = constrain_passing_interval(
             ctx,
             (current_interval + (days_late / 2.0).floor()) * self.ease_factor,
             hard_interval + 1,
+            true,
         );
         let easy_interval = constrain_passing_interval(
             ctx,
             (current_interval + days_late) * self.ease_factor * ctx.easy_multiplier,
             good_interval + 1,
+            true,
         );
 
         (hard_interval, good_interval, easy_interval)
@@ -169,11 +168,16 @@ impl ReviewState {
         let hard_interval = {
             let factor = ctx.hard_multiplier;
             let half_usual = factor / 2.0;
-            constrain_passing_interval(ctx, (elapsed * factor).max(scheduled * half_usual), 0)
+            constrain_passing_interval(
+                ctx,
+                (elapsed * factor).max(scheduled * half_usual),
+                0,
+                false,
+            )
         };
 
         let good_interval =
-            constrain_passing_interval(ctx, (elapsed * self.ease_factor).max(scheduled), 0);
+            constrain_passing_interval(ctx, (elapsed * self.ease_factor).max(scheduled), 0, false);
 
         let easy_interval = {
             // currently flooring() f64s to match python output
@@ -184,6 +188,7 @@ impl ReviewState {
                 ((elapsed as f64 * self.ease_factor as f64).max(scheduled as f64) * reduced_bonus)
                     .floor() as f32,
                 0,
+                false,
             )
         };
 
@@ -191,44 +196,21 @@ impl ReviewState {
     }
 }
 
-fn fuzz_range(interval: f32, factor: f32, minimum: f32) -> (f32, f32) {
-    let delta = (interval * factor).max(minimum).max(1.0);
-    (interval - delta, interval + delta)
-}
-
 /// Transform the provided hard/good/easy interval.
 /// - Apply configured interval multiplier.
 /// - Apply fuzz.
 /// - Ensure it is at least `minimum`, and at least 1.
 /// - Ensure it is at or below the configured maximum interval.
-fn constrain_passing_interval(ctx: &StateContext, interval: f32, minimum: u32) -> u32 {
+fn constrain_passing_interval(ctx: &StateContext, interval: f32, minimum: u32, fuzz: bool) -> u32 {
     // fixme: floor is to match python
-    let interval = interval.floor();
+    let interval = interval.floor() * ctx.interval_multiplier;
-    with_review_fuzz(ctx.fuzz_seed, interval * ctx.interval_multiplier)
+    let interval = if fuzz {
+        ctx.with_review_fuzz(interval)
+    } else {
+        interval.floor() as u32
+    };
+    interval
         .max(minimum)
         .min(ctx.maximum_review_interval)
         .max(1)
 }
-
-fn with_review_fuzz(seed: Option<u64>, interval: f32) -> u32 {
-    // fixme: floor() is to match python
-    let interval = interval.floor();
-    if let Some(seed) = seed {
-        let mut rng = StdRng::seed_from_u64(seed);
-        let (lower, upper) = if interval < 2.0 {
-            (1.0, 1.0)
-        } else if interval < 3.0 {
-            (2.0, 3.0)
-        } else if interval < 7.0 {
-            fuzz_range(interval, 0.25, 0.0)
-        } else if interval < 30.0 {
-            fuzz_range(interval, 0.15, 2.0)
-        } else {
-            fuzz_range(interval, 0.05, 4.0)
-        };
-        rng.gen_range(lower, upper + 1.0)
-    } else {
-        interval
-    }
-    .round() as u32
-}

rslib/src/sched/timespan.rs

@@ -6,7 +6,7 @@ use crate::i18n::{tr_args, I18n, TR};
 /// Short string like '4d' to place above answer buttons.
 pub fn answer_button_time(seconds: f32, i18n: &I18n) -> String {
     let span = Timespan::from_secs(seconds).natural_span();
-    let args = tr_args!["amount" => span.as_rounded_unit()];
+    let args = tr_args!["amount" => span.as_rounded_unit_for_answer_buttons()];
     let key = match span.unit() {
         TimespanUnit::Seconds => TR::SchedulingAnswerButtonTimeSeconds,
         TimespanUnit::Minutes => TR::SchedulingAnswerButtonTimeMinutes,
@@ -114,13 +114,26 @@ impl Timespan {
     /// truncates to one decimal place.
     pub fn as_rounded_unit(self) -> f32 {
         match self.unit {
-            // seconds/days as integer
+            // seconds/minutes/days as integer
             TimespanUnit::Seconds | TimespanUnit::Days => self.as_unit().round(),
             // other values shown to 1 decimal place
             _ => (self.as_unit() * 10.0).round() / 10.0,
         }
     }
 
+    /// Round seconds, minutes and days to integers, otherwise
+    /// truncates to one decimal place.
+    pub fn as_rounded_unit_for_answer_buttons(self) -> f32 {
+        match self.unit {
+            // seconds/minutes/days as integer
+            TimespanUnit::Seconds | TimespanUnit::Minutes | TimespanUnit::Days => {
+                self.as_unit().round()
+            }
+            // other values shown to 1 decimal place
+            _ => (self.as_unit() * 10.0).round() / 10.0,
+        }
+    }
+
     pub fn unit(self) -> TimespanUnit {
         self.unit
     }
@@ -161,7 +174,7 @@ mod test {
         let log = log::terminal();
         let i18n = I18n::new(&["zz"], "", log);
         assert_eq!(answer_button_time(30.0, &i18n), "30s");
-        assert_eq!(answer_button_time(70.0, &i18n), "1.2m");
+        assert_eq!(answer_button_time(70.0, &i18n), "1m");
         assert_eq!(answer_button_time(1.1 * MONTH, &i18n), "1.1mo");
     }
 

rslib/src/timestamp.rs

@@ -52,7 +52,7 @@ impl TimestampMillis {
 }
 
 lazy_static! {
-    static ref TESTING: bool = env::var("SHIFT_CLOCK_HACK").is_ok();
+    pub(crate) static ref TESTING: bool = env::var("ANKI_TEST_MODE").is_ok();
 }
 
 fn elapsed() -> time::Duration {