Performance Optimization of Template Variable Replacement in Alipay Card Package

The Alipay card package stores user membership cards and coupons, rendering them through static templates combined with dynamic data. The original implementation replaces variables delimited by '$' using repeated indexOf and substring calls, which becomes a performance bottleneck as the template is processed for each request.

Version V1 removes the indexOf and substring calls by first extracting all variables with a double‑pointer scan and then replacing them in a second pass, improving performance over the baseline.

Version V2 adds a cache (using Google Guava) that maps a template ID to its pre‑extracted variable set, eliminating the need to re‑parse the template on every request and further reducing latency.

Version V3 replaces the costly String.replace calls with a StringBuilder ‑based assembly, but discovers that the unordered Set used for variable storage breaks replacement order; switching to an ordered List restores correctness.

Version V4 combines the insights from V3 and removes the remaining String.replace usage entirely, using only StringBuilder for substitution, which yields a performance gain of more than ten times compared with the original code.

Version V5 further simplifies the approach by discarding the cache, keeping the ordered variable list, and using StringBuilder to replace String.replace , achieving the best balance between readability and speed, as demonstrated by a 1‑million‑iteration benchmark.

Performance comparisons (Figures 7, 11, 12) show the hierarchy V4 > V3 > V5 > V2 > V1 > original, confirming that String.replace is the dominant cost and that StringBuilder dramatically reduces both CPU time and garbage‑collection pressure.

The article concludes that a simple refactor—replacing String.replace with StringBuilder and optionally adding a lightweight cache—can yield >10× speed improvements, leading to substantial resource savings in large‑scale deployments.