feat: support custom dictionaries for IK tokenizer

[jbjvhvhh]

Task Description

Add user-defined dictionary support for the IK full-text tokenizer.

Solution Description

Support FULLTEXT_DICT tables, dictionary refresh, and IK parser properties for custom main, stopword, and quantifier dictionaries. Add dictionary validation, TOKENIZE() integration, dynamic cache refresh, and DDL dependency protection.

Passed Regressions

• observer build passed.
• Module-layer check passed with no upward edges.
• ik_custom_dict.test passed.

Upgrade Compatibility

Existing IK indexes continue to use built-in dictionaries by default. The legacy quanitfier_table spelling remains compatible.

Other Information

Dictionary refresh only affects newly indexed data. Existing data requires rebuilding the full-text index.

Release Note

IK full-text indexes and TOKENIZE() now support user-defined dictionaries.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

jbjvhvhh seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[LINxiansheng]

Document AI & IK Custom Dictionary Score

Document AI Functions Score
===========================
score: 0.00 / 100
load_file: 0 / 50
ai_split_document: 0 / 50

IK Custom Dictionary Score
==========================
score: 100.00 / 100
ik_custom_dict: 100 / 100

FTS Large Benchmark Score

FTS Large Benchmark Score
=========================
score: 0.00 / 100
mean_improvement: -0.02%
full_score_improvement: 50.00%

build_improvement: 3.22%
  build_ik_all_sec: baseline=35.2836, current=33.44, improvement=5.23%
  build_ik_content_sec: baseline=28.3764, current=27.445, improvement=3.28%
  build_beng_en_sec: baseline=14.7578, current=14.586, improvement=1.16%
tokenize_improvement: -2.86%
  tokenize_ik_avg_ms: baseline=0.76478, current=0.7822, improvement=-2.28%
  tokenize_beng_avg_ms: baseline=0.42262, current=0.4372, improvement=-3.45%
query_improvement: -0.43%
  query_cn_avg_ms: baseline=16.6628, current=16.826, improvement=-0.98%
  query_beng_avg_ms: baseline=24.3042, current=24.3752, improvement=-0.29%
  query_mixed_avg_ms: baseline=17.5593, current=17.5705, improvement=-0.06%
  query_limit_avg_ms: baseline=16.2334, current=16.2964, improvement=-0.39%

FTS Large Benchmark Report

========================================
FTS Large Benchmark Report
========================================
timestamp:              2026-07-14 02:02:41
label:                  vldb-ci-29299059459-1
git_head:               b216bb9
git_dirty:              0
rows:                   20000
batch:                  500
rounds:                 3000
query_rounds:           200
samples:                3
warmup:                 30
skip_load:              0
----------------------------------------
select1_avg_ms:         0.2161
select1_stdev_ms:       0.0144
raw_load_sec:           1.465
raw_load_rows_per_sec:  13651.9
build_ik_all_sec:       33.440
build_ik_content_sec:   27.445
build_beng_en_sec:      14.586
build_total_sec:        75.488
----------------------------------------
tokenize_ik_avg_ms:     0.7822
tokenize_ik_median_ms:  0.7838
tokenize_ik_stdev_ms:   0.0029
tokenize_beng_avg_ms:   0.4372
tokenize_beng_median_ms:0.4215
tokenize_beng_stdev_ms: 0.0242
----------------------------------------
query_cn_hits:          8001
query_cn_avg_ms:        16.8260
query_cn_stdev_ms:      0.0201
query_beng_hits:        11000
query_beng_avg_ms:      24.3752
query_beng_stdev_ms:    0.0090
query_mixed_hits:       7332
query_mixed_avg_ms:     17.5705
query_mixed_stdev_ms:   0.0127
query_limit_hits:       20
query_limit_avg_ms:     16.2964
query_limit_stdev_ms:   0.0061
========================================

Workflow run