[Keymap] Drashna update for post Q2 merge (#17241)

11 files changed, 384 insertions, 469 deletions

diff --git a/users/drashna/keyrecords/autocorrection/autocorrection.c b/users/drashna/keyrecords/autocorrection/autocorrection.c
index c7e938a341..682d6fd49c 100644
--- a/users/drashna/keyrecords/autocorrection/autocorrection.c
+++ b/users/drashna/keyrecords/autocorrection/autocorrection.c
@@ -10,6 +10,13 @@
 #    pragma GCC push_options
 #    pragma GCC optimize("O0")
 #    include "autocorrection_data.h"
+#    ifndef AUTOCORRECTION_MIN_LENGTH
+#        define AUTOCORRECTION_MIN_LENGTH AUTOCORRECT_MIN_LENGTH
+#    endif
+#    ifndef AUTOCORRECTION_MAX_LENGTH
+#        define AUTOCORRECTION_MAX_LENGTH AUTOCORRECT_MAX_LENGTH
+#    endif
+#    define autocorrection_data autocorrect_data
 #    if AUTOCORRECTION_MIN_LENGTH < 4
 #        error Minimum Length is too short and may cause overflows
 #    endif
@@ -17,6 +24,138 @@
 #        error Dictionary size excees maximum size permitted
 #    endif
 
+static uint8_t typo_buffer[AUTOCORRECT_MAX_LENGTH] = {KC_SPC};
+static uint8_t typo_buffer_size                    = 1;
+
+/**
+ * @brief function for querying the enabled state of autocorrect
+ *
+ * @return true if enabled
+ * @return false if disabled
+ */
+bool autocorrect_is_enabled(void) {
+    return userspace_config.autocorrection;
+}
+
+/**
+ * @brief Enables autocorrect and saves state to eeprom
+ *
+ */
+void autocorrect_enable(void) {
+    userspace_config.autocorrection = true;
+    eeconfig_update_user(userspace_config.raw);
+}
+
+/**
+ * @brief Disables autocorrect and saves state to eeprom
+ *
+ */
+void autocorrect_disable(void) {
+    userspace_config.autocorrection = false;
+    typo_buffer_size                 = 0;
+    eeconfig_update_user(userspace_config.raw);
+}
+
+/**
+ * @brief Toggles autocorrect's status and save state to eeprom
+ *
+ */
+void autocorrect_toggle(void) {
+    userspace_config.autocorrection = !userspace_config.autocorrection;
+    typo_buffer_size                 = 0;
+    eeconfig_update_user(userspace_config.raw);
+}
+
+/**
+ * @brief handler for determining if autocorrect should process keypress
+ *
+ * @param keycode Keycode registered by matrix press, per keymap
+ * @param record keyrecord_t structure
+ * @param typo_buffer_size passed along to allow resetting of autocorrect buffer
+ * @param mods allow processing of mod status
+ * @return true Allow autocorection
+ * @return false Stop processing and escape from autocorrect.
+ */
+__attribute__((weak)) bool process_autocorrect_user(uint16_t *keycode, keyrecord_t *record, uint8_t *typo_buffer_size, uint8_t *mods) {
+    // See quantum_keycodes.h for reference on these matched ranges.
+    switch (*keycode) {
+        // Exclude these keycodes from processing.
+        case KC_LSFT:
+        case KC_RSFT:
+        case KC_CAPS:
+        case QK_TO ... QK_ONE_SHOT_LAYER_MAX:
+        case QK_LAYER_TAP_TOGGLE ... QK_LAYER_MOD_MAX:
+        case QK_ONE_SHOT_MOD ... QK_ONE_SHOT_MOD_MAX:
+            return false;
+
+        // Mask for base keycode from shifted keys.
+        case QK_LSFT ... QK_LSFT + 255:
+        case QK_RSFT ... QK_RSFT + 255:
+            if (*keycode >= QK_LSFT && *keycode <= (QK_LSFT + 255)) {
+                *mods |= MOD_LSFT;
+            } else {
+                *mods |= MOD_RSFT;
+            }
+            *keycode &= 0xFF; // Get the basic keycode.
+            return true;
+#ifndef NO_ACTION_TAPPING
+        // Exclude tap-hold keys when they are held down
+        // and mask for base keycode when they are tapped.
+        case QK_LAYER_TAP ... QK_LAYER_TAP_MAX:
+#    ifdef NO_ACTION_LAYER
+            // Exclude Layer Tap, if layers are disabled
+            // but action tapping is still enabled.
+            return false;
+#    endif
+        case QK_MOD_TAP ... QK_MOD_TAP_MAX:
+            // Exclude hold keycode
+            if (!record->tap.count) {
+                return false;
+            }
+            *keycode &= 0xFF;
+            break;
+#else
+        case QK_MOD_TAP ... QK_MOD_TAP_MAX:
+        case QK_LAYER_TAP ... QK_LAYER_TAP_MAX:
+            // Exclude if disabled
+            return false;
+#endif
+        // Exclude swap hands keys when they are held down
+        // and mask for base keycode when they are tapped.
+        case QK_SWAP_HANDS ... QK_SWAP_HANDS_MAX:
+#ifdef SWAP_HANDS_ENABLE
+            if (*keycode >= 0x56F0 || !record->tap.count) {
+                return false;
+            }
+            *keycode &= 0xFF;
+            break;
+#else
+            // Exclude if disabled
+            return false;
+#endif
+    }
+
+    // Disable autocorrect while a mod other than shift is active.
+    if ((*mods & ~MOD_MASK_SHIFT) != 0) {
+        *typo_buffer_size = 0;
+        return false;
+    }
+
+    return true;
+}
+
+/**
+ * @brief handling for when autocorrection has been triggered
+ *
+ * @param backspaces number of characters to remove
+ * @param str pointer to PROGMEM string to replace mistyped seletion with
+ * @return true apply correction
+ * @return false user handled replacement
+ */
+__attribute__((weak)) bool apply_autocorrect(uint8_t backspaces, const char *str) {
+    return true;
+}
+
 /**
  * @brief Process handler for autocorrect feature
  *
@@ -25,131 +164,124 @@
  * @return true Continue processing keycodes, and send to host
  * @return false Stop processing keycodes, and don't send to host
  */
-bool process_autocorrection(uint16_t keycode, keyrecord_t* record) {
-    static uint8_t typo_buffer[AUTOCORRECTION_MAX_LENGTH] = {KC_SPC};
-    static uint8_t typo_buffer_size                       = 1;
+bool process_autocorrection(uint16_t keycode, keyrecord_t *record) {
+    uint8_t mods = get_mods();
+#ifndef NO_ACTION_ONESHOT
+    mods |= get_oneshot_mods();
+#endif
 
-    if (keycode == AUTO_CTN) {
-        if (record->event.pressed) {
-            typo_buffer_size = 0;
-            userspace_config.autocorrection ^= 1;
-            eeconfig_update_user(userspace_config.raw);
+    if ((keycode >= AUTOCORRECT_ON && keycode <= AUTOCORRECT_TOGGLE) && record->event.pressed) {
+        if (keycode == AUTOCORRECT_ON) {
+            autocorrect_enable();
+        } else if (keycode == AUTOCORRECT_OFF) {
+            autocorrect_disable();
+        } else if (keycode == AUTOCORRECT_TOGGLE) {
+            autocorrect_toggle();
+        } else {
+            return true;
         }
+
         return false;
     }
 
-    if (!userspace_config.autocorrection) {
+    if (!autocorrect_is_enabled()) {
         typo_buffer_size = 0;
         return true;
     }
 
+    if (!record->event.pressed) {
+        return true;
+    }
+
+    // autocorrect keycode verification and extraction
+    if (!process_autocorrect_user(&keycode, record, &typo_buffer_size, &mods)) {
+        return true;
+    }
+
+    // keycode buffer check
     switch (keycode) {
-        case KC_LSFT:
-        case KC_RSFT:
-            return true;
-#    ifndef NO_ACTION_TAPPING
-        case QK_MOD_TAP ... QK_MOD_TAP_MAX:
-            if (((keycode >> 8) & 0xF) == MOD_LSFT) {
-                return true;
-            }
-#        ifndef NO_ACTION_LAYER
-        case QK_LAYER_TAP ... QK_LAYER_TAP_MAX:
-#        endif
-            if (record->event.pressed || !record->tap.count) {
-                return true;
-            }
-            keycode &= 0xFF;
+        case KC_A ... KC_Z:
+            // process normally
             break;
-#    endif
-#    ifdef SWAP_HANDS_ENABLE
-        case QK_SWAP_HANDS ... QK_SWAP_HANDS_MAX:
-            if (keycode >= 0x56F0 || record->event.pressed || !record->tap.count) {
-                return true;
-            }
-            keycode &= 0xFF;
+        case KC_1 ... KC_0:
+        case KC_TAB ... KC_SEMICOLON:
+        case KC_GRAVE ... KC_SLASH:
+            // Set a word boundary if space, period, digit, etc. is pressed.
+            keycode = KC_SPC;
             break;
-#    endif
-#    ifndef NO_ACTION_ONESHOT
-        case QK_ONE_SHOT_MOD ... QK_ONE_SHOT_MOD_MAX:
-            if ((keycode & 0xF) == MOD_LSFT) {
-                return true;
-            }
-#    endif
-        default:
-            // Disable autocorrection while a mod other than shift is active.
-            if (((get_mods() | get_oneshot_mods()) & ~MOD_MASK_SHIFT) != 0) {
-                typo_buffer_size = 0;
-                return true;
-            }
-            if (!record->event.pressed) {
-                return true;
-            }
-    }
-
-    // Subtract buffer for Backspace key, reset for other non-alpha.
-    if (!(KC_A <= keycode && keycode <= KC_Z)) {
-        if (keycode == KC_BSPC) {
+        case KC_ENTER:
+            // Behave more conservatively for the enter key. Reset, so that enter
+            // can't be used on a word ending.
+            typo_buffer_size = 0;
+            keycode          = KC_SPC;
+            break;
+        case KC_BSPC:
             // Remove last character from the buffer.
             if (typo_buffer_size > 0) {
                 --typo_buffer_size;
             }
             return true;
-        } else if (KC_1 <= keycode && keycode <= KC_SLSH && keycode != KC_ESC) {
-            // Set a word boundary if space, period, digit, etc. is pressed.
-            // Behave more conservatively for the enter key. Reset, so that enter
-            // can't be used on a word ending.
-            if (keycode == KC_ENT || (keycode == KC_MINUS && (get_mods() | get_oneshot_mods()) & MOD_MASK_SHIFT)) {
-                typo_buffer_size = 0;
+        case KC_QUOTE:
+            // Treat " (shifted ') as a word boundary.
+            if ((mods & MOD_MASK_SHIFT) != 0) {
+                keycode = KC_SPC;
             }
-            keycode = KC_SPC;
-        } else {
+            break;
+        default:
             // Clear state if some other non-alpha key is pressed.
             typo_buffer_size = 0;
             return true;
-        }
     }
 
     // Rotate oldest character if buffer is full.
-    if (typo_buffer_size >= AUTOCORRECTION_MAX_LENGTH) {
-        memmove(typo_buffer, typo_buffer + 1, AUTOCORRECTION_MAX_LENGTH - 1);
-        typo_buffer_size = AUTOCORRECTION_MAX_LENGTH - 1;
+    if (typo_buffer_size >= AUTOCORRECT_MAX_LENGTH) {
+        memmove(typo_buffer, typo_buffer + 1, AUTOCORRECT_MAX_LENGTH - 1);
+        typo_buffer_size = AUTOCORRECT_MAX_LENGTH - 1;
     }
 
     // Append `keycode` to buffer.
     typo_buffer[typo_buffer_size++] = keycode;
     // Return if buffer is smaller than the shortest word.
-    if (typo_buffer_size < AUTOCORRECTION_MIN_LENGTH) {
+    if (typo_buffer_size < AUTOCORRECT_MIN_LENGTH) {
         return true;
     }
 
-    // Check for typo in buffer using a trie stored in `autocorrection_data`.
+    // Check for typo in buffer using a trie stored in `autocorrect_data`.
     uint16_t state = 0;
-    uint8_t  code  = pgm_read_byte(autocorrection_data + state);
-    for (uint8_t i = typo_buffer_size - 1; i >= 0; --i) {
+    uint8_t  code  = pgm_read_byte(autocorrect_data + state);
+    for (int8_t i = typo_buffer_size - 1; i >= 0; --i) {
         uint8_t const key_i = typo_buffer[i];
 
-        if (code & 64) {  // Check for match in node with multiple children.
+        if (code & 64) { // Check for match in node with multiple children.
             code &= 63;
-            for (; code != key_i; code = pgm_read_byte(autocorrection_data + (state += 3))) {
+            for (; code != key_i; code = pgm_read_byte(autocorrect_data + (state += 3))) {
                 if (!code) return true;
             }
             // Follow link to child node.
-            state = (pgm_read_byte(autocorrection_data + state + 1) | pgm_read_byte(autocorrection_data + state + 2) << 8);
+            state = (pgm_read_byte(autocorrect_data + state + 1) | pgm_read_byte(autocorrect_data + state + 2) << 8);
             // Check for match in node with single child.
         } else if (code != key_i) {
             return true;
-        } else if (!(code = pgm_read_byte(autocorrection_data + (++state)))) {
+        } else if (!(code = pgm_read_byte(autocorrect_data + (++state)))) {
             ++state;
         }
 
-        code = pgm_read_byte(autocorrection_data + state);
+        // Stop if `state` becomes an invalid index. This should not normally
+        // happen, it is a safeguard in case of a bug, data corruption, etc.
+        if (state >= DICTIONARY_SIZE) {
+            return true;
+        }
+
+        code = pgm_read_byte(autocorrect_data + state);
 
-        if (code & 128) {  // A typo was found! Apply autocorrection.
-            const uint8_t backspaces = code & 63;
-            for (uint8_t i = 0; i < backspaces; ++i) {
-                tap_code(KC_BSPC);
+        if (code & 128) { // A typo was found! Apply autocorrect.
+            const uint8_t backspaces = (code & 63) + !record->event.pressed;
+            if (apply_autocorrect(backspaces, (char const *)(autocorrect_data + state + 1))) {
+                for (uint8_t i = 0; i < backspaces; ++i) {
+                    tap_code(KC_BSPC);
+                }
+                send_string_P((char const *)(autocorrect_data + state + 1));
             }
-            send_string_P((char const*)(autocorrection_data + state + 1));
 
             if (keycode == KC_SPC) {
                 typo_buffer[0]   = KC_SPC;
@@ -166,5 +298,7 @@ bool process_autocorrection(uint16_t keycode, keyrecord_t* record) {
 #    pragma GCC pop_options
 #else
 #    pragma message "Warning!!! Autocorrect is not corretly setup!"
-bool process_autocorrection(uint16_t keycode, keyrecord_t* record) { return true; }
+bool process_autocorrection(uint16_t keycode, keyrecord_t* record) {
+    return true;
+}
 #endif
diff --git a/users/drashna/keyrecords/autocorrection/autocorrection.h b/users/drashna/keyrecords/autocorrection/autocorrection.h
index cea93159ae..8946b91f1f 100644
--- a/users/drashna/keyrecords/autocorrection/autocorrection.h
+++ b/users/drashna/keyrecords/autocorrection/autocorrection.h
@@ -7,4 +7,11 @@
 
 #include "drashna.h"
 
-bool process_autocorrection(uint16_t keycode, keyrecord_t* record);
+bool process_autocorrection(uint16_t keycode, keyrecord_t *record);
+bool process_autocorrect_user(uint16_t *keycode, keyrecord_t *record, uint8_t *typo_buffer_size, uint8_t *mods);
+bool apply_autocorrect(uint8_t backspaces, const char *str);
+
+bool autocorrect_is_enabled(void);
+void autocorrect_enable(void);
+void autocorrect_disable(void);
+void autocorrect_toggle(void);
diff --git a/users/drashna/keyrecords/autocorrection/make_autocorrection_data.py b/users/drashna/keyrecords/autocorrection/make_autocorrection_data.py
index 54fd9ba594..0dd9b78b9c 100755
--- a/users/drashna/keyrecords/autocorrection/make_autocorrection_data.py
+++ b/users/drashna/keyrecords/autocorrection/make_autocorrection_data.py
@@ -1,4 +1,4 @@
-# Copyright 2021 Google LLC
+# Copyright 2021-2022 Google LLC
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -29,6 +29,7 @@ Each line of the dict file defines one typo and its correction with the syntax
 Example:
 
   :thier        -> their
+  dosen't       -> doesn't
   fitler        -> filter
   lenght        -> length
   ouput         -> output
@@ -42,7 +43,7 @@ https://getreuer.info/posts/keyboards/autocorrection
 
 import sys
 import textwrap
-from typing import Any, Dict, List, Tuple
+from typing import Any, Dict, Iterator, List, Tuple
 
 try:
   from english_words import english_words_lower_alpha_set as CORRECT_WORDS
@@ -51,85 +52,67 @@ except ImportError:
         'correctly spelled word. To check for this, install the english_words '
         'package and rerun this script:\n\n  pip install english_words\n')
   # Use a minimal word list as a fallback.
-  CORRECT_WORDS = ('information', 'available', 'international', 'language',
-                   'loosest', 'reference', 'wealthier', 'entertainment',
-                   'association', 'provides', 'technology', 'statehood')
+  CORRECT_WORDS = ('apparent', 'association', 'available', 'classification',
+                   'effect', 'entertainment', 'fantastic', 'information',
+                   'integrate', 'international', 'language', 'loosest',
+                   'manual', 'nothing', 'provides', 'reference', 'statehood',
+                   'technology', 'virtually', 'wealthier', 'wonderful')
 
 KC_A = 4
 KC_SPC = 0x2c
+KC_QUOT = 0x34
+
+TYPO_CHARS = dict(
+  [
+    ("'", KC_QUOT),
+    (':', KC_SPC),  # "Word break" character.
+  ] +
+  # Characters a-z.
+  [(chr(c), c + KC_A - ord('a')) for c in range(ord('a'), ord('z') + 1)]
+)
+
 
 def parse_file(file_name: str) -> List[Tuple[str, str]]:
   """Parses autocorrections dictionary file.
 
   Each line of the file defines one typo and its correction with the syntax
   "typo -> correction". Blank lines or lines starting with '#' are ignored. The
-  function validates that typos only have characters a-z and that typos are not
-  substrings of other typos, otherwise the longer typo would never trigger.
+  function validates that typos only have characters in TYPO_CHARS, that
+  typos are not substrings of other typos, and checking that typos don't trigger
+  on CORRECT_WORDS.
 
   Args:
     file_name: String, path of the autocorrections dictionary.
   Returns:
     List of (typo, correction) tuples.
   """
-
+  correct_words = ('information', 'available', 'international', 'language', 'loosest', 'reference', 'wealthier', 'entertainment', 'association', 'provides', 'technology', 'statehood')
   autocorrections = []
   typos = set()
-  line_number = 0
-  for line in open(file_name, 'rt'):
-    line_number += 1
-    line = line.strip()
-    if line and line[0] != '#':
-      # Parse syntax "typo -> correction", using strip to ignore indenting.
-      tokens = [token.strip() for token in line.split('->', 1)]
-      if len(tokens) != 2 or not tokens[0]:
-        print(f'Error:{line_number}: Invalid syntax: "{line}"')
+  for line_number, typo, correction in parse_file_lines(file_name):
+    if typo in typos:
+      print(f'Warning:{line_number}: Ignoring duplicate typo: "{typo}"')
+      continue
+
+    # Check that `typo` is valid.
+    if not(all([c in TYPO_CHARS for c in typo])):
+      print(f'Error:{line_number}: Typo "{typo}" has '
+            'characters other than ' + ''.join(TYPO_CHARS.keys()))
+      sys.exit(1)
+    for other_typo in typos:
+      if typo in other_typo or other_typo in typo:
+        print(f'Error:{line_number}: Typos may not be substrings of one '
+              f'another, otherwise the longer typo would never trigger: '
+              f'"{typo}" vs. "{other_typo}".')
         sys.exit(1)
+    if len(typo) < 5:
+      print(f'Warning:{line_number}: It is suggested that typos are at '
+            f'least 5 characters long to avoid false triggers: "{typo}"')
 
-      typo, correction = tokens
-      typo = typo.lower()  # Force typos to lowercase.
-      typo = typo.replace(' ', ':')
-
-      if typo in typos:
-        print(f'Warning:{line_number}: Ignoring duplicate typo: "{typo}"')
-        continue
+    check_typo_against_dictionary(typo, line_number, correct_words)
 
-      # Check that `typo` is valid.
-      if not(all([ord('a') <= ord(c) <= ord('z') or c == ':' for c in typo])):
-        print(f'Error:{line_number}: Typo "{typo}" has '
-              'characters other than a-z and :.')
-        sys.exit(1)
-      for other_typo in typos:
-        if typo in other_typo or other_typo in typo:
-          print(f'Error:{line_number}: Typos may not be substrings of one '
-                f'another, otherwise the longer typo would never trigger: '
-                f'"{typo}" vs. "{other_typo}".')
-          sys.exit(1)
-      if len(typo) < 5:
-        print(f'Warning:{line_number}: It is suggested that typos are at '
-              f'least 5 characters long to avoid false triggers: "{typo}"')
-
-      if typo.startswith(':') and typo.endswith(':'):
-        if typo[1:-1] in CORRECT_WORDS:
-          print(f'Warning:{line_number}: Typo "{typo}" is a correctly spelled '
-                'dictionary word.')
-      elif typo.startswith(':') and not typo.endswith(':'):
-        for word in CORRECT_WORDS:
-          if word.startswith(typo[1:]):
-            print(f'Warning:{line_number}: Typo "{typo}" would falsely trigger '
-                  f'on correctly spelled word "{word}".')
-      elif not typo.startswith(':') and typo.endswith(':'):
-        for word in CORRECT_WORDS:
-          if word.endswith(typo[:-1]):
-            print(f'Warning:{line_number}: Typo "{typo}" would falsely trigger '
-                  f'on correctly spelled word "{word}".')
-      elif not typo.startswith(':') and not typo.endswith(':'):
-        for word in CORRECT_WORDS:
-          if typo in word:
-            print(f'Warning:{line_number}: Typo "{typo}" would falsely trigger '
-                  f'on correctly spelled word "{word}".')
-
-      autocorrections.append((typo, correction))
-      typos.add(typo)
+    autocorrections.append((typo, correction))
+    typos.add(typo)
 
   return autocorrections
 
@@ -152,6 +135,47 @@ def make_trie(autocorrections: List[Tuple[str, str]]) -> Dict[str, Any]:
   return trie
 
 
+def parse_file_lines(file_name: str) -> Iterator[Tuple[int, str, str]]:
+  """Parses lines read from `file_name` into typo-correction pairs."""
+
+  line_number = 0
+  for line in open(file_name, 'rt'):
+    line_number += 1
+    line = line.strip()
+    if line and line[0] != '#':
+      # Parse syntax "typo -> correction", using strip to ignore indenting.
+      tokens = [token.strip() for token in line.split('->', 1)]
+      if len(tokens) != 2 or not tokens[0]:
+        print(f'Error:{line_number}: Invalid syntax: "{line}"')
+        sys.exit(1)
+
+      typo, correction = tokens
+      typo = typo.lower()  # Force typos to lowercase.
+      typo = typo.replace(' ', ':')
+
+      yield line_number, typo, correction
+
+
+def check_typo_against_dictionary(typo: str, line_number: int, correct_words) -> None:
+  """Checks `typo` against English dictionary words."""
+
+  if typo.startswith(':') and typo.endswith(':'):
+    if typo[1:-1] in correct_words:
+      print(f'Warning:{line_number}: Typo "{typo}" is a correctly spelled dictionary word.')
+  elif typo.startswith(':') and not typo.endswith(':'):
+    for word in correct_words:
+      if word.startswith(typo[1:]):
+            print(f'Warning:{line_number}: Typo "{typo}" would falsely trigger on correctly spelled word "{word}".')
+  elif not typo.startswith(':') and typo.endswith(':'):
+    for word in correct_words:
+      if word.endswith(typo[:-1]):
+            print(f'Warning:{line_number}: Typo "{typo}" would falsely trigger on correctly spelled word "{word}".')
+  elif not typo.startswith(':') and not typo.endswith(':'):
+    for word in correct_words:
+      if typo in word:
+        print(f'Warning:{line_number}: Typo "{typo}" would falsely trigger on correctly spelled word "{word}".')
+
+
 def serialize_trie(autocorrections: List[Tuple[str, str]],
                    trie: Dict[str, Any]) -> List[int]:
   """Serializes trie and correction data in a form readable by the C code.
@@ -165,7 +189,7 @@ def serialize_trie(autocorrections: List[Tuple[str, str]],
   table = []
 
   # Traverse trie in depth first order.
-  def traverse(trie_node):
+  def traverse(trie_node: Dict[str, Any]) -> Dict[str, Any]:
     if 'LEAF' in trie_node:  # Handle a leaf trie node.
       typo, correction = trie_node['LEAF']
       word_boundary_ending = typo[-1] == ':'
@@ -200,37 +224,35 @@ def serialize_trie(autocorrections: List[Tuple[str, str]],
 
   traverse(trie)
 
-  def serialize(e):
-    def kc_code(c):
-      if ord('a') <= ord(c) <= ord('z'):
-        return ord(c) - ord('a') + KC_A
-      elif c == ':':
-        return KC_SPC
-      else:
-        raise ValueError(f'Invalid character: {c}')
-
-    encode_link = lambda link: [link['byte_offset'] & 255,
-                                link['byte_offset'] >> 8]
-
+  def serialize(e: Dict[str, Any]) -> List[int]:
     if not e['links']:  # Handle a leaf table entry.
       return e['data']
     elif len(e['links']) == 1:  # Handle a chain table entry.
-      return list(map(kc_code, e['chars'])) + [0] #+ encode_link(e['links'][0]))
+      return [TYPO_CHARS[c] for c in e['chars']] + [0]
     else:  # Handle a branch table entry.
       data = []
       for c, link in zip(e['chars'], e['links']):
-        data += [kc_code(c) | (0 if data else 64)] + encode_link(link)
+        data += [TYPO_CHARS[c] | (0 if data else 64)] + encode_link(link)
       return data + [0]
 
   byte_offset = 0
   for e in table:  # To encode links, first compute byte offset of each entry.
     e['byte_offset'] = byte_offset
     byte_offset += len(serialize(e))
-    assert 0 <= byte_offset <= 0xffff
 
   return [b for e in table for b in serialize(e)]  # Serialize final table.
 
 
+def encode_link(link: Dict[str, Any]) -> List[int]:
+  """Encodes a node link as two bytes."""
+  byte_offset = link['byte_offset']
+  if not (0 <= byte_offset <= 0xffff):
+    print('Error: The autocorrection table is too large, a node link exceeds '
+          '64KB limit. Try reducing the autocorrection dict to fewer entries.')
+    sys.exit(1)
+  return [byte_offset & 255, byte_offset >> 8]
+
+
 def write_generated_code(autocorrections: List[Tuple[str, str]],
                          data: List[int],
                          file_name: str) -> None:
@@ -242,7 +264,10 @@ def write_generated_code(autocorrections: List[Tuple[str, str]],
     file_name: String, path of the output C file.
   """
   assert all(0 <= b <= 255 for b in data)
-  typo_len = lambda e: len(e[0])
+
+  def typo_len(e: Tuple[str, str]) -> int:
+    return len(e[0])
+
   min_typo = min(autocorrections, key=typo_len)[0]
   max_typo = max(autocorrections, key=typo_len)[0]
   generated_code = ''.join([
@@ -252,9 +277,8 @@ def write_generated_code(autocorrections: List[Tuple[str, str]],
                    for typo, correction in autocorrections)),
     f'\n#define AUTOCORRECTION_MIN_LENGTH {len(min_typo)}  // "{min_typo}"\n',
     f'#define AUTOCORRECTION_MAX_LENGTH {len(max_typo)}  // "{max_typo}"\n\n',
-    f'#define DICTIONARY_SIZE {len(data)}\n\n',
-    textwrap.fill('static const uint8_t autocorrection_data[DICTIONARY_SIZE] PROGMEM = {%s};' % (
-      ', '.join(map(str, data))), width=120, subsequent_indent='    '),
+    textwrap.fill('static const uint8_t autocorrection_data[%d] PROGMEM = {%s};' % (
+      len(data), ', '.join(map(str, data))), width=80, subsequent_indent='  '),
     '\n\n'])
 
   with open(file_name, 'wt') as f:
diff --git a/users/drashna/keyrecords/caps_word.c b/users/drashna/keyrecords/caps_word.c
deleted file mode 100644
index a152b2387b..0000000000
--- a/users/drashna/keyrecords/caps_word.c
+++ /dev/null
@@ -1,139 +0,0 @@
-#include "caps_word.h"
-
-static bool caps_word_active = false;
-
-#if CAPS_WORD_IDLE_TIMEOUT > 0
-#    if CAPS_WORD_IDLE_TIMEOUT < 100 || CAPS_WORD_IDLE_TIMEOUT > 30000
-// Constrain timeout to a sensible range. With the 16-bit timer, the longest
-// representable timeout is 32768 ms, rounded here to 30000 ms = half a minute.
-#        error "caps_word: CAPS_WORD_IDLE_TIMEOUT must be between 100 and 30000 ms"
-#    endif
-
-static uint16_t idle_timer = 0;
-
-void caps_word_task(void) {
-    if (caps_word_active && timer_expired(timer_read(), idle_timer)) {
-        caps_word_set(false);
-    }
-}
-#endif // CAPS_WORD_IDLE_TIMEOUT > 0
-
-bool process_caps_word(uint16_t keycode, keyrecord_t* record) {
-#ifndef NO_ACTION_ONESHOT
-    const uint8_t mods = get_mods() | get_oneshot_mods();
-#else
-    const uint8_t mods = get_mods();
-#endif // NO_ACTION_ONESHOT
-
-    if (!caps_word_active) {
-        // Pressing both shift keys at the same time enables caps word.
-        if ((mods & MOD_MASK_SHIFT) == MOD_MASK_SHIFT) {
-            caps_word_set(true); // Activate Caps Word.
-            return false;
-        }
-        return true;
-    } else {
-#if CAPS_WORD_IDLE_TIMEOUT > 0
-        idle_timer = record->event.time + CAPS_WORD_IDLE_TIMEOUT;
-#endif // CAPS_WORD_IDLE_TIMEOUT > 0
-    }
-
-    if (!record->event.pressed) {
-        return true;
-    }
-
-    if (!(mods & ~MOD_MASK_SHIFT)) {
-        switch (keycode) {
-            // Ignore MO, TO, TG, TT, and OSL layer switch keys.
-            case QK_MOMENTARY ... QK_MOMENTARY_MAX:
-            case QK_TO ... QK_TO_MAX:
-            case QK_TOGGLE_LAYER ... QK_TOGGLE_LAYER_MAX:
-            case QK_LAYER_TAP_TOGGLE ... QK_LAYER_TAP_TOGGLE_MAX:
-            case QK_ONE_SHOT_LAYER ... QK_ONE_SHOT_LAYER_MAX:
-                return true;
-
-#ifndef NO_ACTION_TAPPING
-            case QK_MOD_TAP ... QK_MOD_TAP_MAX:
-                if (record->tap.count == 0) {
-                    // Deactivate if a mod becomes active through holding a mod-tap key.
-                    caps_word_set(false);
-                    return true;
-                }
-                keycode &= 0xff;
-                break;
-
-#    ifndef NO_ACTION_LAYER
-            case QK_LAYER_TAP ... QK_LAYER_TAP_MAX:
-#    endif // NO_ACTION_LAYER
-                if (record->tap.count == 0) {
-                    return true;
-                }
-                keycode &= 0xff;
-                break;
-#endif // NO_ACTION_TAPPING
-
-#ifdef SWAP_HANDS_ENABLE
-            case QK_SWAP_HANDS ... QK_SWAP_HANDS_MAX:
-                if (keycode > 0x56F0 || record->tap.count == 0) {
-                    return true;
-                }
-                keycode &= 0xff;
-                break;
-#endif // SWAP_HANDS_ENABLE
-        }
-
-        if (caps_word_press_user(keycode)) {
-            return true;
-        }
-    }
-
-    caps_word_set(false); // Deactivate Caps Word.
-    return true;
-}
-
-void caps_word_set(bool active) {
-    if (active != caps_word_active) {
-        if (active) {
-            clear_mods();
-#ifndef NO_ACTION_ONESHOT
-            clear_oneshot_mods();
-#endif // NO_ACTION_ONESHOT
-#if CAPS_WORD_IDLE_TIMEOUT > 0
-            idle_timer = timer_read() + CAPS_WORD_IDLE_TIMEOUT;
-#endif // CAPS_WORD_IDLE_TIMEOUT > 0
-        } else if ((get_weak_mods() & MOD_BIT(KC_LSFT)) != 0) {
-            // If the weak shift mod is still on, turn it off and send an update to
-            // the host computer.
-            del_weak_mods(MOD_BIT(KC_LSFT));
-            send_keyboard_report();
-        }
-
-        caps_word_active = active;
-        caps_word_set_user(active);
-    }
-}
-
-bool caps_word_get(void) {
-    return caps_word_active;
-}
-
-__attribute__((weak)) void caps_word_set_user(bool active) {}
-
-__attribute__((weak)) bool caps_word_press_user(uint16_t keycode) {
-    switch (keycode) {
-        // Keycodes that continue Caps Word, with shift applied.
-        case KC_A ... KC_Z:
-            add_weak_mods(MOD_BIT(KC_LSFT)); // Apply shift to the next key.
-            return true;
-
-        // Keycodes that continue Caps Word, without shifting.
-        case KC_1 ... KC_0:
-        case KC_BSPC:
-        case KC_MINS:
-        case KC_UNDS:
-            return true;
-
-        default:
-            return false; // Deactivate Caps Word.
-    }
-}
diff --git a/users/drashna/keyrecords/caps_word.h b/users/drashna/keyrecords/caps_word.h
deleted file mode 100644
index 4279d7e831..0000000000
--- a/users/drashna/keyrecords/caps_word.h
+++ /dev/null
@@ -1,85 +0,0 @@
-// Copyright 2021 Google LLC.
-// SPDX-License-Identifier: Apache-2.0
-
-#pragma once
-
-#include "drashna.h"
-
-// Call this function from `process_record_user()` to implement Caps Word.
-bool process_caps_word(uint16_t keycode, keyrecord_t* record);
-
-// If CAPS_WORD_IDLE_TIMEOUT is set, call `caps_word_task()` from
-// `matrix_scan_user()` as described above.
-//
-// If CAPS_WORD_IDLE_TIMEOUT isn't set, calling this function has no effect (but
-// will still compile).
-#if CAPS_WORD_IDLE_TIMEOUT > 0
-void caps_word_task(void);
-#else
-static inline void caps_word_task(void) {}
-#endif
-
-// Activates or deactivates Caps Word. For instance activate Caps Word with a
-// combo by defining a `COMBO_ACTION` that calls `caps_word_set(true)`:
-//
-// void process_combo_event(uint16_t combo_index, bool pressed) {
-//   switch(combo_index) {
-//     case CAPS_COMBO:
-//       if (pressed) {
-//         caps_word_set(true);  // Activate Caps Word.
-//       }
-//       break;
-//
-//     // Other combos...
-//   }
-// }
-void caps_word_set(bool active);
-
-// Returns whether Caps Word is currently active.
-bool caps_word_get(void);
-
-// An optional callback that gets called when Caps Word turns on or off. This is
-// useful to represent the current Caps Word state, e.g. by setting an LED or
-// playing a sound. In your keymap, define
-//
-//   void caps_word_set_user(bool active) {
-//     if (active) {
-//       // Do something when Caps Word activates.
-//     } else {
-//       // Do something when Caps Word deactivates.
-//     }
-//   }
-void caps_word_set_user(bool active);
-
-// An optional callback which is called on every key press while Caps Word is
-// active. When the key should be shifted (that is, a letter key), the callback
-// should call `add_weak_mods(MOD_BIT(KC_LSFT))` to shift the key. The callback
-// also determines whether the key should continue Caps Word. Returning true
-// continues the current "word", while returning false is "word breaking" and
-// deactivates Caps Word. The default callback is
-//
-//   bool caps_word_press_user(uint16_t keycode) {
-//     switch (keycode) {
-//       // Keycodes that continue Caps Word, with shift applied.
-//       case KC_A ... KC_Z:
-//         add_weak_mods(MOD_BIT(KC_LSFT));  // Apply shift to the next key.