Implements option to generate prefix-free hints

- Declares --prefix-free CLI flag in main.py.
- Implements dynamic index offset shifting in marks.go to prevent
  generating hints that prefix other hints.
- Integrates prefix-free encoding in main.go.
- Includes tests.

Implements #10195
This commit is contained in:
Hector Tellez 2026-07-01 00:38:32 -07:00
parent f6d1b11b29
commit 5d1215b11d
5 changed files with 226 additions and 5 deletions

View file

@ -192,7 +192,12 @@ func main(_ *cli.Command, o *Options, args []string) (rc int, err error) {
text_style := fctx.SprintFunc(fmt.Sprintf("fg=%s bg=%s bold", o.HintsTextColor, o.HintsTextBackgroundColor))
highlight_mark := func(m *Mark, mark_text string) string {
hint := encode_hint(m.Index, alphabet)
var hint string
if o.PrefixFree {
hint = generate_prefix_free_hint(m.Index, alphabet)
} else {
hint = encode_hint(m.Index, alphabet)
}
if current_input != "" && !strings.HasPrefix(hint, current_input) {
return faint(mark_text)
}
@ -311,7 +316,13 @@ func main(_ *cli.Command, o *Options, args []string) (rc int, err error) {
if changed {
matches := []*Mark{}
for idx, m := range index_map {
if eh := encode_hint(idx, alphabet); strings.HasPrefix(eh, current_input) {
var eh string
if o.PrefixFree {
eh = generate_prefix_free_hint(idx, alphabet)
} else {
eh = encode_hint(idx, alphabet)
}
if strings.HasPrefix(eh, current_input) {
matches = append(matches, m)
}
}

View file

@ -201,6 +201,16 @@ The offset (from zero) at which to start hint numbering. Note that only numbers
greater than or equal to zero are respected.
--prefix-free
type=bool-set
Generate hints such that no hint is a prefix of another. For example, with
alphabet "abc" and 4 matches, hints are "b", "c", "aa", "ab", instead of "a",
"b", "c", "aa" (where "a" prefixes "aa"). This works by applying a dynamic
offset large enough so that the hints are prefix-free. When combined with a
:option:`--hints-offset` greater than 0, the effective offset is whichever is
larger.
--alphabet
The list of characters to use for hints. The default is to use numbers and
lowercase English alphabets. Specify your preference as a string of characters.

View file

@ -706,13 +706,28 @@ process_answer:
}
largest_index := ans[len(ans)-1].Index
offset := max(0, opts.HintsOffset)
if opts.PrefixFree {
alphabetLength := len(opts.Alphabet)
if alphabetLength == 0 {
alphabetLength = len(DEFAULT_HINT_ALPHABET)
}
offset = max(offset, hints_to_skip(len(ans), alphabetLength))
}
index_map = make(map[int]*Mark, len(ans))
for i := range ans {
m := &ans[i]
if opts.Ascending {
m.Index += offset
if opts.PrefixFree {
if opts.Ascending {
m.Index = i + offset + 1
} else {
m.Index = (largest_index - m.Index) + offset + 1
}
} else {
m.Index = largest_index - m.Index + offset
if opts.Ascending {
m.Index += offset
} else {
m.Index = largest_index - m.Index + offset
}
}
index_map[m.Index] = m
}

View file

@ -0,0 +1,62 @@
// License: GPLv3 Copyright: 2026, Kovid Goyal, <kovid at kovidgoyal.net>
package hints
// hints_to_skip calculates how many hints to skip based on the given n and
// alphabet size so that the next n hints generated by expanding the prefix tree
// of an alphabet of size alphabetSize will be prefix-free.
//
// The result is 1-based, meaning that the first hint to be skipped is the first
// non-empty hint.
//
// For example, if alphabet was "abc" and n was 5, the first non-empty hints
// from that alphabet would be: "a", "b", "c", "aa", "ab", "ac", ...
// the result of hints_to_skip would be 1, and this would mean that if you skip
// "a" from the list above, the next 5 hints are prefix-free:
// "b", "c", "aa", "ab", "ac".
//
// See issue #10195 for an informal proof on why this function works.
func hints_to_skip(n int, alphabetSize int) int {
if n < 2 {
n = 2
}
return (n - 2) / (alphabetSize - 1)
}
// generate_prefix_free_hint generates the hint corresponding to the given index
// in the breadth-first traversal of the prefix tree generated by the given
// alphabet.
//
// For example, given the alphabet "abc", the hints would be generated in
// the following order:
// 1. "a"
// 2. "b"
// 3. "c"
// 4. "aa"
// 5. "ab"
// 6. "ac"
// 7. "ba"
// 8. "bb"
// 9. "bc"
// 10. "ca"
// 11. "cb"
// 12. "cc"
// 13. "aaa"
// ...
//
// Indices are 1-based on non-empty hints; index=0 returns an empty (invalid)
// hint.
func generate_prefix_free_hint(index int, alphabet string) string {
l := len(alphabet)
hint := ""
index -= 1
for index >= 0 {
mod := index % l
char := string(alphabet[mod])
index /= l
hint = char + hint
index -= 1
}
return hint
}

View file

@ -0,0 +1,123 @@
// License: GPLv3 Copyright: 2026, Kovid Goyal, <kovid at kovidgoyal.net>
package hints
import (
"math"
"strings"
"testing"
"github.com/kovidgoyal/kitty"
)
func TestPrefixFreeHints(t *testing.T) {
// Test hints_to_skip
testsToSkip := []struct {
n, alphabetSize, expected int
}{
{5, 3, 1},
{6, 3, 2},
{4, 3, 1},
{3, 3, 0},
{2, 3, 0},
{1, 3, 0},
{0, 3, 0},
}
for _, tc := range testsToSkip {
actual := hints_to_skip(tc.n, tc.alphabetSize)
if actual != tc.expected {
t.Errorf("hints_to_skip(%d, %d) = %d, expected %d", tc.n, tc.alphabetSize, actual, tc.expected)
}
}
// Test prefix-free property verification
alphabets := []string{"abc", "0123456789"}
for _, alph := range alphabets {
l := len(alph)
for n := 1; n <= 10; n++ {
total_hints := int(math.Pow(float64(2), float64(n)))
skip := hints_to_skip(total_hints, l)
hints := make([]string, total_hints)
for i := 0; i < total_hints; i++ {
hints[i] = generate_prefix_free_hint(skip+1+i, alph)
}
// Verify that no hint is a prefix of another hint
for i := 0; i < total_hints; i++ {
for j := 0; j < total_hints; j++ {
if i == j {
continue
}
if strings.HasPrefix(hints[j], hints[i]) {
t.Errorf("For alphabet %q and n=%d: %q is a prefix of %q (skip=%d)", alph, total_hints, hints[i], hints[j], skip)
}
}
}
}
}
// HintsOffset + PrefixFree Test case 1: HintsOffset is smaller than dynamic
// offset (dynamic offset wins)
//
// With 4 matches, and alphabet size 3:
// hints_to_skip(4, 3) = 1.
// HintsOffset = 0.
// Effective offset = max(0, 1) = 1.
// Ascending = true.
// Match 0: 0 + 1 + 1 = 2 -> "b"
// Match 1: 1 + 1 + 1 = 3 -> "c"
// Match 2: 2 + 1 + 1 = 4 -> "aa"
// Match 3: 3 + 1 + 1 = 5 -> "ab"
opts1 := &Options{
Type: "url",
UrlPrefixes: "default",
Regex: kitty.HintsDefaultRegex,
PrefixFree: true,
Alphabet: "abc",
HintsOffset: 0,
Ascending: false, // default
}
_, marks1, _, err := find_marks("http://a.com http://b.com http://c.com http://d.com", opts1)
if err != nil {
t.Fatalf("find_marks failed: %v", err)
}
expectedCodes1 := []string{"ab", "aa", "c", "b"}
for i, m := range marks1 {
hint := generate_prefix_free_hint(m.Index, opts1.Alphabet)
if hint != expectedCodes1[i] {
t.Errorf("Case 1 - Match %d: got hint %q, expected %q", i, hint, expectedCodes1[i])
}
}
// HintsOffset + PrefixFree Test case 2: HintsOffset is larger than dynamic
// offset (HintsOffset wins)
//
// With 4 matches, and alphabet size 3:
// hints_to_skip(4, 3) = 1.
// HintsOffset = 3.
// Effective offset = max(3, 1) = 3.
// Ascending = true.
// Match 0: 0 + 3 + 1 = 4 -> "aa"
// Match 1: 1 + 3 + 1 = 5 -> "ab"
// Match 2: 2 + 3 + 1 = 6 -> "ac"
// Match 3: 3 + 3 + 1 = 7 -> "ba"
opts2 := &Options{
Type: "url",
UrlPrefixes: "default",
Regex: kitty.HintsDefaultRegex,
PrefixFree: true,
Alphabet: "abc",
HintsOffset: 3,
Ascending: true,
}
_, marks2, _, err := find_marks("http://a.com http://b.com http://c.com http://d.com", opts2)
if err != nil {
t.Fatalf("find_marks failed: %v", err)
}
expectedCodes2 := []string{"aa", "ab", "ac", "ba"}
for i, m := range marks2 {
hint := generate_prefix_free_hint(m.Index, opts2.Alphabet)
if hint != expectedCodes2[i] {
t.Errorf("Case 2 - Match %d: got hint %q, expected %q", i, hint, expectedCodes2[i])
}
}
}