Merge pull request TencentBlueKing#5225 from hingbong/issue-#5100

feat: 流水线页面 按名称A-Z 支持中文按拼音排序 TencentBlueKing#5100 替换汉字拼音转换工具 提高多音字正确率

src/backend/ci/build.gradle.kts

@@ -83,8 +83,8 @@ allprojects {
                 entry("poi")
                 entry("poi-ooxml")
             }
-            dependencySet("com.hankcs:${Versions.HanLP}") {
-                entry("hanlp")
+            dependencySet("com.github.taptap:${Versions.PinyinPlus}") {
+                entry("pinyin-plus")
             }
         }
     }

src/backend/ci/buildSrc/src/main/kotlin/constants/Versions.kt

@@ -33,5 +33,5 @@ object Versions {
     const val Elasticsearch = "7.4.0"
     const val Lucene = "8.2.0"
     const val Feign = "11.6" // TODO 修复这个BUG:https://github.com/OpenFeign/feign/pull/1347 , 后续SpringCloud升级后可以去掉
-    const val HanLP = "portable-1.8.2"
+    const val PinyinPlus = "1.0"
 }

src/backend/ci/core/common/common-util/build.gradle.kts

@@ -31,4 +31,5 @@ dependencies {
     implementation("commons-codec:commons-codec")
     implementation("com.google.guava:guava")
     implementation("org.jolokia:jolokia-core")
+    implementation("com.github.taptap:pinyin-plus")
 }

src/backend/ci/core/common/common-util/src/main/kotlin/com/tencent/devops/common/util/PinyinUtil.kt

@@ -0,0 +1,50 @@
+package com.tencent.devops.common.util
+
+import com.github.houbb.heaven.util.lang.CharUtil
+import com.github.houbb.heaven.util.lang.StringUtil
+import com.github.houbb.segment.support.segment.result.impl.SegmentResultHandlers
+import com.taptap.pinyin.ResourceLoad
+import com.taptap.pinyin.Word
+import com.taptap.pinyin.analyzer.WordAnalyzer
+import com.taptap.pinyin.utils.Utils
+import java.util.HashMap
+import java.util.StringJoiner
+
+object PinyinUtil {
+
+    private val words: HashMap<String, Word> = ResourceLoad.loadCedict()
+    private val wordAnalyzer = WordAnalyzer.newInstance()
+
+    /**
+     *  com.taptap.pinyin.PinyinPlus#to(java.lang.String, boolean) 修改此方法逻辑, 改为词组之间不带分隔符
+     *  返回拼音字符串, 不存在拼音的返回原字符
+     */
+    @Suppress("NestedBlockDepth")
+    fun toPinyin(text: String): String {
+        if (StringUtil.isBlank(text)) return text
+        var word = words[text]
+        return if (word != null) {
+            Utils.trim(word.pinyinNoTone)
+        } else {
+            val joiner = StringJoiner("")
+            val segmentResult = wordAnalyzer.segment(text, SegmentResultHandlers.word())
+            for (segmentStr in segmentResult) {
+                word = words[segmentStr]
+                if (word != null) {
+                    joiner.add(Utils.trim(word.pinyinNoTone))
+                } else {
+                    val characterList = StringUtil.toCharacterList(segmentStr)
+                    for (character in characterList) {
+                        if (CharUtil.isChinese(character)) {
+                            word = words[character.toString()]
+                            joiner.add(word!!.pinyinNoTone)
+                        } else {
+                            joiner.add(character.toString())
+                        }
+                    }
+                }
+            }
+            joiner.toString()
+        }
+    }
+}

src/backend/ci/core/common/common-util/src/test/kotlin/com/tencent/devops/common/util/PinyinUtilTest.kt

@@ -0,0 +1,17 @@
+package com.tencent.devops.common.util
+
+import org.junit.Test
+
+class PinyinUtilTest {
+
+    @Test
+    fun test() {
+        assert("chongzhiliushuixianzhuangtai TEST.テスト-jinめ_ceshi" ==
+            PinyinUtil.toPinyin("重置流水线状态 TEST.テスト-進め_測試"))
+        assert("zhongyao zhongda" == PinyinUtil.toPinyin("重要 重大"))
+        assert(PinyinUtil.toPinyin("奇偶") == "jiou")
+        assert(PinyinUtil.toPinyin("奇怪奇异") == "qiguaiqiyi")
+        assert(PinyinUtil.toPinyin("屏风 屏障 屏蔽") == "pingfeng pingzhang pingbi")
+        assert(PinyinUtil.toPinyin("屏息 屏气") == "bingxi bingqi")
+    }
+}

src/backend/ci/core/process/biz-base/build.gradle.kts

@@ -43,5 +43,4 @@ dependencies {
     api("mysql:mysql-connector-java")
     implementation("com.github.ben-manes.caffeine:caffeine")
     testImplementation(project(":core:common:common-test"))
-    api("com.hankcs:hanlp")
 }

src/backend/ci/core/process/biz-base/src/main/kotlin/com/tencent/devops/process/engine/dao/PipelineInfoDao.kt

@@ -27,11 +27,10 @@
 
 package com.tencent.devops.process.engine.dao
 
-import com.hankcs.hanlp.HanLP
-import com.hankcs.hanlp.dictionary.py.Pinyin
 import com.tencent.devops.common.api.util.timestampmilli
 import com.tencent.devops.common.pipeline.enums.ChannelCode
 import com.tencent.devops.common.pipeline.pojo.BuildNo
+import com.tencent.devops.common.util.PinyinUtil
 import com.tencent.devops.model.process.Tables.T_PIPELINE_INFO
 import com.tencent.devops.model.process.tables.records.TPipelineInfoRecord
 import com.tencent.devops.process.engine.pojo.PipelineInfo
@@ -605,11 +604,7 @@ class PipelineInfoDao {
     }
 
     private fun nameToPinyin(pipelineName: String): String {
-        return HanLP.convertToPinyinList(pipelineName).asSequence().mapIndexed { index, it ->
-            // 不属于中文没有拼音
-            if (Pinyin.none5 == it) {
-                pipelineName[index].toString()
-            } else it.pinyinWithoutTone
-        }.joinToString("")
+        // 数据库字段长度1300
+        return PinyinUtil.toPinyin(pipelineName).take(1300)
     }
 }