Skip to main content

SoundexWord

Function SoundexWord( const AText : string) : Word

Example​

procedure ScriptEvent(var Value: variant);
begin
  Value := SoundexWord('Robert');
end;

Usage​

SoundexWord encodes a four-character Soundex code into a Word.

Parameters

NameTypeDescription
ATextstring, constNon-empty text to encode.

Returns

The packed four-character code. For a normal initial A through Z, valid output fits within Word and can be reconstructed with DecodeSoundexWord.

Behaviour

All words with the same four-character Soundex code produce the same packed value. The result cannot recover the source spelling.

Errors

Normal non-empty input beginning with an ASCII letter has no expected exception. Empty or invalid-initial input can cause memory/range/conversion behaviour dependent on the deployed RTL checks and must be rejected by the caller.

Additional Technical Info​

SoundexWord generates a four-character Soundex code and packs the initial and three digits into an unsigned 16-bit Word. It is a compact Delphi representation, not a standard phonetic interchange number.

The example returns 21752, the packed form of R163. It is source-reviewed and is not executed by the documentation workflow.

Implementation

Velox binds directly to System.StrUtils.SoundexWord. The routine calls Soundex(AText, 4), then packs:

value = ordinal(initial) - ordinal('A')
value = value * 26 + first digit
value = value * 7  + second digit
value = value * 7  + third digit

For R163, this is 17 -> 443 -> 3107 -> 21752.

Edge cases and quirks

  • Empty input is unsafe. Soundex('', 4) returns an empty string, after which SoundexWord indexes characters 0 through 3 without an empty check. This is out-of-range access: it can raise when range checking applies or otherwise produce invalid, memory-dependent behaviour. It never defines an empty-input sentinel. Validate non-empty input first.
  • The first character is assumed to be A through Z. Punctuation, digits and non-Latin initials can create negative/out-of-range intermediate values and wrapped or range-dependent Word results.
  • The remaining Soundex limitations are inherited: ASCII-focused scoring, no Unicode normalisation and expected phonetic collisions.
  • The packed layout is Delphi-specific. Do not assume another Soundex implementation uses the same bases or first-character mapping.
  • Unlike SoundexInt, no length is stored because the code is always four characters.

Side effects

None. A temporary four-character code is allocated.

Performance and concurrency

The scan stops once four code characters are available or input ends, then performs constant-time packing. There is no shared mutable state.

Remarks

Require non-empty, suitably normalised name text before calling. Store the readable code when interoperability and diagnosis matter more than two-byte compactness.

Related entries

External references

Created 2026-07-15