qishibo/AnotherRedisDesktopManager · warning

Something wrong with your vector

Error message

Something wrong with your vector

What it means

Client-side validation in the similarity dialog's 'vector' mode. parseVectorText (KeyContentVector.vue:315) splits this.simVectorText on commas, trims each token, and maps it through Number(); it returns null when the text is blank or when any token is not a finite number (NaN, Infinity, empty token from a trailing/double comma). runSimilarity() aborts with this toast before any Redis call when the parse yields null.

Source

Thrown at src/components/contents/KeyContentVector.vue:484

    },
    runSimilarity() {
      const { client, redisKey } = this;
      const count = this.simCount;
      let args;

      // search by element name
      if (this.simMode === 'ele') {
        const element = (this.simElement || '').trim();
        if (!element) {
          return this.$message.error('Element is required');
        }
        args = ['VSIM', redisKey, 'ELE', element, 'WITHSCORES', 'WITHATTRIBS', 'COUNT', count];
      }
      // search by vector
      else {
        const vector = this.parseVectorText(this.simVectorText);
        if (!vector) {
          return this.$message.error('Something wrong with your vector');
        }
        if (this.dim && vector.length !== this.dim) {
          return this.$message.error(`Vector dimension must be ${this.dim}`);
        }
        args = ['VSIM', redisKey, 'VALUES', vector.length, ...vector, 'WITHSCORES', 'WITHATTRIBS', 'COUNT', count];
      }

      this.simLoading = true;

      client.call(...args).then((reply) => {
        this.simResults = this.parseSimReply(reply);
        this.simLoading = false;
      }).catch((e) => {
        this.simLoading = false;
        this.$message.error(e.message);
      });
    },
    parseSimReply(reply) {

View on GitHub (pinned to c149855106)

Solutions

  1. Reformat the vector as comma-separated bare numbers, e.g. 0.1,0.2,0.3 — no brackets, no quotes
  2. If pasting from Python/NumPy, strip brackets and convert: ','.join(map(str, vec)) or str(vec.tolist())
  3. Check for trailing commas, double commas, semicolons, or localized decimal separators (0,1 instead of 0.1) — all produce NaN tokens
  4. Verify no stray characters (quotes, 'tensor(...)', dtype suffix) remain after pasting

Example fix

// before: only comma-separated input parses
parseVectorText(text) {
  if (!text.trim()) return null;
  const vector = text.split(',').map(v => Number(v.trim()));
  if (!vector.length || vector.some(n => !Number.isFinite(n))) return null;
  return vector;
}

// after: also accept whitespace/bracket-separated pastes
parseVectorText(text) {
  if (!text) return null;
  const cleaned = text.replace(/[\[\]()]/g, '');
  const vector = cleaned.split(/[\s,;]+/).filter(t => t.length).map(Number);
  if (!vector.length || vector.some(n => !Number.isFinite(n))) return null;
  return vector;
}
Defensive patterns

Strategy: validation

Validate before calling

// validate before issuing VSIM VALUES
const vector = this.parseVectorText(this.simVectorText);
if (!vector) {
  // tell the user WHY: empty vs non-numeric tokens
  const tokens = this.simVectorText.split(',').map(s => s.trim()).filter(Boolean);
  const bad = tokens.filter(t => !Number.isFinite(Number(t)));
  this.vectorError = bad.length ? `Non-numeric component: ${bad[0]}` : 'Vector is empty';
  return;
}

Type guard

/** True when text parses to a finite-number vector of the expected length. */
function isParsableVector(text, dim) {
  if (!text || !text.trim()) return false;
  const parts = text.split(',').map(s => Number(s.trim()));
  return parts.length > 0 && parts.every(n => Number.isFinite(n)) && (!dim || parts.length === dim);
}

Try / catch

null

Prevention

When it happens

Trigger: Calling VSIM in VALUES mode with simVectorText that is (a) empty/whitespace-only, (b) space-separated instead of comma-separated numbers like '0.1 0.2 0.3', (c) contains a non-numeric token such as 'abc' or 'null', or (d) has a trailing/doubled comma producing an empty token ('0.1,0.2,').

Common situations: Copying an embedding from Python (repr uses spaces: [0.1, 0.2] sometimes survives but NumPy array str '[-0.01130762 0.04266454 ...]' fails because tokens are space-separated); pasting JSON array text with brackets '[' ']'; pasting scientific notation is fine but '1e' or '--1' is not; empty clipboard paste.

Related errors

AI-assisted analysis of qishibo/AnotherRedisDesktopManager@c149855106 (2026-08-22). Data as JSON: /api/errors/77df0d4b274ca4b1. Report an issue: GitHub.