HIVE数仓数据血缘分析工具-SQL解析 - thomas0yang的专栏 - 博客频道 - CSDN.NET http://blog.csdn.net/thomas0yang/article/details/49449723
sql解析代码_bak - 下载频道 - CSDN.NET http://download.csdn.net/download/thomas0yang/9369949
一、数仓经常会碰到的几类问题: 1、两个数据报表进行对比,结果差异很大,需要人工核对分析指标的维度信息,比如从头分析数据指标从哪里来,处理条件是什么,最后才能分析出问题原因。 2、基础数据表因某种原因需要修改字段,需要评估其对数仓的影响,费时费力,然后在做方案。
二、问题分析: 数据源长途跋涉,经过大量的处理和组件来传递,呈现在业务用户面前,对数据进行回溯其实很难。元数据回溯在有效决策、策略制定、差异分析等过程中很重要。这两类问题都属于数据血缘分析问题,第一类叫做数据回溯、第二类叫做影响分析,是数据回溯的逆向。
三、解决方法: 自己实现了一套基于Hive数仓的数据血缘分析工具,来完成各个数据表、字段之间的关系梳理,进而解决上面两个问题。
工具主要目标:解析计算脚本中的HQL语句,分析得到输入输出表、输入输出字段和相应的处理条件,进行分析展现。
实现思路:对AST深度优先遍历,遇到操作的token则判断当前的操作,遇到子句则压栈当前处理,处理子句。子句处理完,栈弹出。处理字句的过程中,遇到子查询就保存当前子查询的信息,判断与其父查询的关系,最终形成树形结构; 遇到字段或者条件处理则记录当前的字段和条件信息、组成Block,嵌套调用。
关键点解析: 1、遇到TOK_TAB或TOK_TABREF则判断出当前操作的表 2、压栈判断是否是join,判断join条件 3、定义数据结构Block,遇到在where\select\join时获得其下相应的字段和条件,组成Block 4、定义数据结构ColLine,遇到TOK_SUBQUERY保存当前的子查询信息,供父查询使用 5、定义数据结构ColLine,遇到TOK_UNION结束时,合并并截断当前的列信息 6、遇到select 或者未明确指出的字段,查询元数据进行辅助分析 7、解析结果进行相关校验
代码如下: Block类
package com.xiaoju.products.parse;import java.util.ArrayList;import java.util.HashMap;import java.util.HashSet;import java.util.List;import java.util.Map;import java.util.Set;import java.util.Stack;import java.util.Map.Entry;import java.util.LinkedHashSet;import org.antlr.runtime.tree.Tree;import org.apache.hadoop.hive.ql.parse.ASTNode;import org.apache.hadoop.hive.ql.parse.BaseSemanticAnalyzer;import org.apache.hadoop.hive.ql.parse.HiveParser;import org.apache.hadoop.hive.ql.parse.ParseDriver;import com.xiaoju.products.bean.Block;import com.xiaoju.products.bean.ColLine;import com.xiaoju.products.bean.QueryTree;import com.xiaoju.products.exception.SQLParseException;import com.xiaoju.products.exception.UnSupportedException;import com.xiaoju.products.util.Check;import com.xiaoju.products.util.MetaCache;import com.xiaoju.products.util.NumberUtil;import com.xiaoju.products.util.ParseUtil;import com.xiaoju.products.util.PropertyFileUtil;/** * hive sql解析类 * * 目的:实现HQL的语句解析,分析出输入输出表、字段和相应的处理条件。为字段级别的数据血缘提供基础。 * 重点:获取SELECT操作中的表和列的相关操作。其他操作这判断到字段级别。 * 实现思路:对AST深度优先遍历,遇到操作的token则判断当前的操作,遇到子句则压栈当前处理,处理子句。子句处理完,栈弹出。 * 处理字句的过程中,遇到子查询就保存当前子查询的信息,判断与其父查询的关系,最终形成树形结构; * 遇到字段或者条件处理则记录当前的字段和条件信息、组成Block,嵌套调用。 * 关键点解析 * 1、遇到TOK_TAB或TOK_TABREF则判断出当前操作的表 * 2、压栈判断是否是join,判断join条件 * 3、定义数据结构Block,遇到在where\select\join时获得其下相应的字段和条件,组成Block * 4、定义数据结构ColLine,遇到TOK_SUBQUERY保存当前的子查询信息,供父查询使用 * 5、定义数据结构ColLine,遇到TOK_UNION结束时,合并并截断当前的列信息 * 6、遇到select 或者未明确指出的字段,查询元数据进行辅助分析 * 7、解析结果进行相关校验 * 试用范围: * 1、支持标准SQL * 2、不支持transform using script * * @author yangyangthomas * /public class LineParser { private static final String SPLIT_DOT = "."; private static final String SPLIT_COMMA = ","; private static final String SPLIT_AND = "&"; private static final String TOK_EOF = "<EOF>"; private static final String CON_WHERE = "WHERE:"; private static final String TOK_TMP_FILE = "TOK_TMP_FILE"; private Map<String /table/, List<String/column/>> dbMap = new HashMap<String, List<String>>(); private List<QueryTree> queryTreeList = new ArrayList<QueryTree>(); //子查询树形关系保存 private Stack<Set<String>> conditionsStack = new Stack<Set<String>>(); private Stack<List<ColLine>> colsStack = new Stack<List<ColLine>>(); private Map<String, List<ColLine>> resultQueryMap = new HashMap<String, List<ColLine>>(); private Set<String> conditions = new HashSet<String>(); //where or join 条件缓存 private List<ColLine> cols = new ArrayList<ColLine>(); //一个子查询内的列缓存 private Stack<String> tableNameStack = new Stack<String>(); private Stack<Boolean> joinStack = new Stack<Boolean>(); private Stack<ASTNode> joinOnStack = new Stack<ASTNode>(); private Map<String, QueryTree> queryMap = new HashMap<String, QueryTree>(); private boolean joinClause = false; private ASTNode joinOn = null; private String nowQueryDB = "default"; //hive的默认库 private boolean isCreateTable = false; //结果 private List<ColLine> colLines = new ArrayList<ColLine>(); private Set<String> outputTables = new HashSet<String>(); private Set<String> inputTables = new HashSet<String>(); private List<ColLine> tmpColLines = new ArrayList<ColLine>(); private Set<String> tmpOutputTables = new HashSet<String>(); private Set<String> tmpInputTables = new HashSet<String>(); public List<ColLine> getColLines() { return colLines; } public Set<String> getOutputTables() { return outputTables; } public Set<String> getInputTables() { return inputTables; } private void parseIteral(ASTNode ast) { prepareToParseCurrentNodeAndChilds(ast); parseChildNodes(ast); parseCurrentNode(ast); endParseCurrentNode(ast); } /** * 解析当前节点 * @param ast * @param set * @return / private void parseCurrentNode(ASTNode ast){ if (ast.getToken() != null) { switch (ast.getToken().getType()) { case HiveParser.TOK_CREATETABLE: //outputtable isCreateTable = true; String tableOut = fillDB(BaseSemanticAnalyzer.getUnescapedName((ASTNode) ast.getChild(0))); tmpOutputTables.add(tableOut); MetaCache.getInstance().init(tableOut); //初始化数据,供以后使用 break; case HiveParser.TOK_TAB:// outputTable String tableTab = BaseSemanticAnalyzer.getUnescapedName((ASTNode) ast.getChild(0)); String tableOut2 = fillDB(tableTab); tmpOutputTables.add(tableOut2); MetaCache.getInstance().init(tableOut2); //初始化数据,供以后使用 break; case HiveParser.TOK_TABREF:// inputTable ASTNode tabTree = (ASTNode) ast.getChild(0); String tableInFull = fillDB((tabTree.getChildCount() == 1) ? BaseSemanticAnalyzer.getUnescapedName((ASTNode) tabTree.getChild(0)) : BaseSemanticAnalyzer.getUnescapedName((ASTNode) tabTree.getChild(0)) + SPLIT_DOT + BaseSemanticAnalyzer.getUnescapedName((ASTNode) tabTree.getChild(1)) ); String tableIn = tableInFull.substring(tableInFull.indexOf(SPLIT_DOT) + 1); tmpInputTables.add(tableInFull); MetaCache.getInstance().init(tableInFull); //初始化数据,供以后使用 queryMap.clear(); String alia = null; if (ast.getChild(1) != null) { //(TOK_TABREF (TOK_TABNAME detail usersequence_client) c) alia = ast.getChild(1).getText().toLowerCase(); QueryTree qt = new QueryTree(); qt.setCurrent(alia); qt.getTableSet().add(tableInFull); QueryTree pTree = getSubQueryParent(ast); qt.setpId(pTree.getpId()); qt.setParent(pTree.getParent()); queryTreeList.add(qt); if (joinClause && ast.getParent() == joinOn) { // TOK_SUBQUERY join TOK_TABREF ,此处的TOK_SUBQUERY信息不应该清楚 for (QueryTree entry : queryTreeList) { //当前的查询范围 if (qt.getParent().equals(entry.getParent())) { queryMap.put(entry.getCurrent(), entry); } } } else { queryMap.put(qt.getCurrent(), qt); } } else { alia = tableIn.toLowerCase(); QueryTree qt = new QueryTree(); qt.setCurrent(alia); qt.getTableSet().add(tableInFull); QueryTree pTree = getSubQueryParent(ast); qt.setpId(pTree.getpId()); qt.setParent(pTree.getParent()); queryTreeList.add(qt); if (joinClause && ast.getParent() == joinOn) { for (QueryTree entry : queryTreeList) { if (qt.getParent().equals(entry.getParent())) { queryMap.put(entry.getCurrent(), entry); } } } else { queryMap.put(qt.getCurrent(), qt); //此处检查查询 select app.t1.c1,t1.c1 from t1 的情况 queryMap.put(tableInFull.toLowerCase(), qt); } } break; case HiveParser.TOK_SUBQUERY: if (ast.getChildCount() == 2) { String tableAlias = BaseSemanticAnalyzer.unescapeIdentifier(ast.getChild(1).getText()); String aliaReal = ""; if(aliaReal.length() !=0){ aliaReal = aliaReal.substring(0, aliaReal.length()-1); } QueryTree qt = new QueryTree(); qt.setCurrent(tableAlias.toLowerCase()); qt.setColLineList(generateColLineList(cols, conditions)); QueryTree pTree = getSubQueryParent(ast); qt.setId(generateTreeId(ast)); qt.setpId(pTree.getpId()); qt.setParent(pTree.getParent()); qt.setChildList(getSubQueryChilds(qt.getId())); if (Check.notEmpty(qt.getChildList())) { for (QueryTree cqt : qt.getChildList()) { qt.getTableSet().addAll(cqt.getTableSet()); queryTreeList.remove(cqt); // 移除子节点信息 } } queryTreeList.add(qt); cols.clear(); queryMap.clear(); for (QueryTree _qt : queryTreeList) { if (qt.getParent().equals( _qt.getParent())) { //当前子查询才保存 queryMap.put(_qt.getCurrent(), _qt); } } } break; case HiveParser.TOK_SELEXPR: //输入输出字段的处理 /* * (TOK_DESTINATION (TOK_DIR TOK_TMP_FILE)) * (TOK_SELECT (TOK_SELEXPR TOK_ALLCOLREF)) * * (TOK_DESTINATION (TOK_DIR TOK_TMP_FILE)) * (TOK_SELECT * (TOK_SELEXPR (. (TOK_TABLE_OR_COL p) datekey) datekey) * (TOK_SELEXPR (TOK_TABLE_OR_COL datekey)) * (TOK_SELEXPR (TOK_FUNCTIONDI count (. (TOK_TABLE_OR_COL base) userid)) buyer_count)) * (TOK_SELEXPR (TOK_FUNCTION when (> (. (TOK_TABLE_OR_COL base) userid) 5) (. (TOK_TABLE_OR_COL base) clienttype) (> (. (TOK_TABLE_OR_COL base) userid) 1) (+ (. (TOK_TABLE_OR_COL base) datekey) 5) (+ (. (TOK_TABLE_OR_COL base) clienttype) 1)) bbbaaa) / //解析需要插入的表 Tree tok_insert = ast.getParent().getParent(); Tree child = tok_insert.getChild(0).getChild(0); String tName = BaseSemanticAnalyzer.getUnescapedName((ASTNode) child.getChild(0)); String destTable = TOK_TMP_FILE.equals(tName) ? TOK_TMP_FILE : fillDB(tName); //select a.,* from t1 和 select * from (select c1 as a,c2 from t1) t 的情况 if (ast.getChild(0).getType() == HiveParser.TOK_ALLCOLREF) { String tableOrAlias = ""; if (ast.getChild(0).getChild(0) != null) { tableOrAlias = ast.getChild(0).getChild(0).getChild(0).getText(); } String[] result = getTableAndAlia(tableOrAlias); String _alia = result[1]; boolean isSub = false; //处理嵌套select * 的情况 if (Check.notEmpty(_alia)) { for (String string : _alia.split(SPLIT_AND)) { //迭代循环的时候查询 QueryTree qt = queryMap.get(string.toLowerCase()); if (null != qt) { List<ColLine> colLineList = qt.getColLineList(); if (Check.notEmpty(colLineList)) { isSub = true; for (ColLine colLine : colLineList) { cols.add(colLine); } } } } } if (!isSub) { //处理直接select * 的情况 String nowTable = result[0]; String[] tableArr = nowTable.split(SPLIT_AND); //fact.test&test2 for (String tables : tableArr) { String[] split = tables.split("\."); if (split.length > 2) { throw new SQLParseException("parse table:" + nowTable); } List<String> colByTab = MetaCache.getInstance().getColumnByDBAndTable(tables); for (String column : colByTab) { Set<String> fromNameSet = new LinkedHashSet<String>(); fromNameSet.add(tables + SPLIT_DOT + column); ColLine cl = new ColLine(column, tables + SPLIT_DOT + column, fromNameSet, new LinkedHashSet<String>() , destTable, column); cols.add(cl); } } } } else { Block bk = getBlockIteral((ASTNode)ast.getChild(0)); String toNameParse = getToNameParse(ast, bk); Set<String> fromNameSet = filterData(bk.getColSet()); ColLine cl = new ColLine(toNameParse, bk.getCondition(), fromNameSet, new LinkedHashSet<String>() , destTable, ""); cols.add(cl); } break; case HiveParser.TOK_WHERE: //3、过滤条件的处理select类 conditions.add(CON_WHERE + getBlockIteral((ASTNode) ast.getChild(0)).getCondition()); break; default: /** * (or * (> (. (TOK_TABLE_OR_COL p) orderid) (. (TOK_TABLE_OR_COL c) orderid)) * (and (= (. (TOK_TABLE_OR_COL p) a) (. (TOK_TABLE_OR_COL c) b)) * (= (. (TOK_TABLE_OR_COL p) aaa) (. (TOK_TABLE_OR_COL c) bbb)))) / //1、过滤条件的处理join类 if (joinOn != null && joinOn.getTokenStartIndex() == ast.getTokenStartIndex() && joinOn.getTokenStopIndex() == ast.getTokenStopIndex()) { ASTNode astCon = (ASTNode)ast.getChild(2); conditions.add(ast.getText().substring(4) + ":" + getBlockIteral(astCon).getCondition()); break; } } } } /* * 查找当前节点的父子查询节点 * @param ast / private QueryTree getSubQueryParent(Tree ast) { Tree _tree = ast; QueryTree qt = new QueryTree(); while(!(_tree = _tree.getParent()).isNil()){ if(_tree.getType() == HiveParser.TOK_SUBQUERY){ qt.setpId(generateTreeId(_tree)); qt.setParent(BaseSemanticAnalyzer.getUnescapedName((ASTNode)_tree.getChild(1))); return qt; } } qt.setpId(-1); qt.setParent("NIL"); return qt; } private int generateTreeId(Tree tree) { return tree.getTokenStartIndex() + tree.getTokenStopIndex(); } /* * 查找当前节点的子子查询节点(索引) * @param ast / private List<QueryTree> getSubQueryChilds(int id) { List<QueryTree> list = new ArrayList<QueryTree>(); for (int i = 0; i < queryTreeList.size(); i++) { QueryTree qt = queryTreeList.get(i); if (id == qt.getpId()) { list.add(qt); } } return list; } /* * 获得要解析的名称 * @param ast * @param bk * @return / private String getToNameParse(ASTNode ast, Block bk) { String alia = ""; Tree child = ast.getChild(0); if (ast.getChild(1) != null) { //有别名 ip as alia alia = ast.getChild(1).getText(); } else if (child.getType() == HiveParser.DOT //没有别名 a.ip && child.getChild(0).getType() == HiveParser.TOK_TABLE_OR_COL && child.getChild(0).getChildCount() == 1 && child.getChild(1).getType() == HiveParser.Identifier) { alia = BaseSemanticAnalyzer.unescapeIdentifier(child.getChild(1).getText()); } else if (child.getType() == HiveParser.TOK_TABLE_OR_COL //没有别名 ip && child.getChildCount() == 1 && child.getChild(0).getType() == HiveParser.Identifier) { alia = BaseSemanticAnalyzer.unescapeIdentifier(child.getChild(0).getText()); } return alia; } /* * 获得解析的块,主要应用在WHERE、JOIN和SELECT端 * 如: <p>where a=1 * <p>t1 join t2 on t1.col1=t2.col1 and t1.col2=123 * <p>select count(distinct col1) from t1 * @param ast * @return / private Block getBlockIteral(ASTNode ast) { if (ast.getType() == HiveParser.KW_OR ||ast.getType() == HiveParser.KW_AND) { Block bk1 = getBlockIteral((ASTNode)ast.getChild(0)); Block bk2 = getBlockIteral((ASTNode)ast.getChild(1)); bk1.getColSet().addAll(bk2.getColSet()); bk1.setCondition("(" + bk1.getCondition() + " " + ast.getText() + " " + bk2.getCondition() + ")"); return bk1; } else if (ast.getType() == HiveParser.NOTEQUAL //判断条件 > < like in || ast.getType() == HiveParser.EQUAL || ast.getType() == HiveParser.LESSTHAN || ast.getType() == HiveParser.LESSTHANOREQUALTO || ast.getType() == HiveParser.GREATERTHAN || ast.getType() == HiveParser.GREATERTHANOREQUALTO || ast.getType() == HiveParser.KW_LIKE || ast.getType() == HiveParser.DIVIDE || ast.getType() == HiveParser.PLUS || ast.getType() == HiveParser.MINUS || ast.getType() == HiveParser.STAR || ast.getType() == HiveParser.MOD || ast.getType() == HiveParser.AMPERSAND || ast.getType() == HiveParser.TILDE || ast.getType() == HiveParser.BITWISEOR || ast.getType() == HiveParser.BITWISEXOR) { Block bk1 = getBlockIteral((ASTNode)ast.getChild(0)); if (ast.getChild(1) == null) { // -1 bk1.setCondition(ast.getText() + bk1.getCondition()); } else { Block bk2 = getBlockIteral((ASTNode)ast.getChild(1)); bk1.getColSet().addAll(bk2.getColSet()); bk1.setCondition(bk1.getCondition() + " " + ast.getText() + " " + bk2.getCondition()); } return bk1; } else if (ast.getType() == HiveParser.TOK_FUNCTIONDI) { Block col = getBlockIteral((ASTNode) ast.getChild(1)); String condition = ast.getChild(0).getText(); col.setCondition(condition + "(distinct (" + col.getCondition() +"))"); return col; } else if (ast.getType() == HiveParser.TOK_FUNCTION){ String fun = ast.getChild(0).getText(); Block col = ast.getChild(1) == null ? new Block() : getBlockIteral((ASTNode) ast.getChild(1)); if ("when".equalsIgnoreCase(fun)) { col.setCondition(getWhenCondition(ast)); Set<Block> processChilds = processChilds(ast, 1); col.getColSet().addAll(bkToCols(col, processChilds)); return col; } else if("IN".equalsIgnoreCase(fun)) { col.setCondition(col.getCondition() + " in (" + blockCondToString(processChilds(ast, 2)) + ")"); return col; } else if("TOK_ISNOTNULL".equalsIgnoreCase(fun) //isnull isnotnull || "TOK_ISNULL".equalsIgnoreCase(fun)){ col.setCondition(col.getCondition() + " " + fun.toLowerCase().substring(4)); return col; } else if("BETWEEN".equalsIgnoreCase(fun)){ col.setCondition(getBlockIteral((ASTNode) ast.getChild(2)).getCondition() + " between " + getBlockIteral((ASTNode) ast.getChild(3)).getCondition() + " and " + getBlockIteral((ASTNode) ast.getChild(4)).getCondition()); return col; } Set<Block> processChilds = processChilds(ast, 1); col.getColSet().addAll(bkToCols(col, processChilds)); col.setCondition(fun +"("+ blockCondToString(processChilds) + ")"); return col; } else if(ast.getType() == HiveParser.LSQUARE){ //map,array Block column = getBlockIteral((ASTNode) ast.getChild(0)); Block key = getBlockIteral((ASTNode) ast.getChild(1)); column.setCondition(column.getCondition() +"["+ key.getCondition() + "]"); return column; } else { return parseBlock(ast); } } private Set<String> bkToCols(Block col, Set<Block> processChilds) { Set<String> set = new LinkedHashSet<String>(processChilds.size()); for (Block colLine : processChilds) { if (Check.notEmpty(colLine.getColSet())) { set.addAll(colLine.getColSet()); } } return set; } private String blockCondToString(Set<Block> processChilds) { StringBuilder sb = new StringBuilder(); for (Block colLine : processChilds) { sb.append(colLine.getCondition()).append(SPLIT_COMMA); } if (sb.length()>0) { sb.setLength(sb.length()-1); } return sb.toString(); } /* * 解析when条件 * @param ast * @return case when c1>100 then col1 when c1>0 col2 else col3 end / private String getWhenCondition(ASTNode ast) { int cnt = ast.getChildCount(); StringBuilder sb = new StringBuilder(); for (int i = 1; i < cnt; i++) { String condition = getBlockIteral((ASTNode)ast.getChild(i)).getCondition(); if (i == 1) { sb.append("(case when " + condition); } else if (i == cnt-1) { //else sb.append(" else " + condition + " end)"); } else if (i % 2 == 0){ //then sb.append(" then " + condition); } else { sb.append(" when " + condition); } } return sb.toString(); } /* * 保存subQuery查询别名和字段信息 * @param sqlIndex * @param tableAlias / private void putResultQueryMap(int sqlIndex, String tableAlias) { List<ColLine> list = generateColLineList(cols, conditions); String key = sqlIndex == 0 ? tableAlias : tableAlias + sqlIndex; //没有重名的情况就不用标记 resultQueryMap.put(key, list); } private List<ColLine> generateColLineList(List<ColLine> cols, Set<String> conditions) { List<ColLine> list = new ArrayList<ColLine>(); for (ColLine entry : cols) { entry.getConditionSet().addAll(conditions); list.add(ParseUtil.cloneColLine(entry)); } return list; } /* * 判断正常列, * 正常:a as col, a * 异常:1 ,'a' //数字、字符等作为列名 / private boolean notNormalCol(String column) { return Check.isEmpty(column) || NumberUtil.isNumeric(column) || (column.startsWith(""") && column.endsWith(""")) || (column.startsWith("'") && column.endsWith("'")); } /* * 从指定索引位置开始解析子树 * @param ast * @param startIndex 开始索引 * @param isSimple 是否简写 * @param withCond 是否包含条件 * @return / private Set<Block> processChilds(ASTNode ast,int startIndex) { int cnt = ast.getChildCount(); Set<Block> set = new LinkedHashSet<Block>(); for (int i = startIndex; i < cnt; i++) { Block bk = getBlockIteral((ASTNode) ast.getChild(i)); if (Check.notEmpty(bk.getCondition()) || Check.notEmpty(bk.getColSet())){ set.add(bk); } } return set; } /* * 解析获得列名或者字符数字等和条件 * @param ast * @param isSimple * @return / private Block parseBlock(ASTNode ast) { if (ast.getType() == HiveParser.DOT && ast.getChild(0).getType() == HiveParser.TOK_TABLE_OR_COL && ast.getChild(0).getChildCount() == 1 && ast.getChild(1).getType() == HiveParser.Identifier) { String column = BaseSemanticAnalyzer.unescapeIdentifier(ast.getChild(1).getText()); String alia = BaseSemanticAnalyzer.unescapeIdentifier(ast.getChild(0).getChild(0).getText()); return getBlock(column, alia); } else if (ast.getType() == HiveParser.TOK_TABLE_OR_COL && ast.getChildCount() == 1 && ast.getChild(0).getType() == HiveParser.Identifier) { String column = ast.getChild(0).getText(); return getBlock(column, null); } else if (ast.getType() == HiveParser.Number || ast.getType() == HiveParser.StringLiteral || ast.getType() == HiveParser.Identifier) { Block bk = new Block(); bk.setCondition(ast.getText()); bk.getColSet().add(ast.getText()); return bk; } return new Block(); } /* * 根据列名和别名获得块信息 * @param column * @param alia * @param tree 当前子查询下的别名可以使用 * @return / private Block getBlock(String column, String alia) { String[] result = getTableAndAlia(alia); String tableArray = result[0]; String _alia = result[1]; for (String string : _alia.split(SPLIT_AND)) { //迭代循环的时候查询 QueryTree qt = queryMap.get(string.toLowerCase()); if (Check.notEmpty(column)) { for (ColLine colLine : qt.getColLineList()) { if (column.equalsIgnoreCase(colLine.getToNameParse())) { Block bk = new Block(); bk.setCondition(colLine.getColCondition()); bk.setColSet(ParseUtil.cloneSet(colLine.getFromNameSet())); return bk; } } } } String _realTable = tableArray; int cnt = 0; //匹配字段和元数据字段相同数目,如果有多个匹配,即此sql有二义性 for (String tables : tableArray.split(SPLIT_AND)) { //初始化的时候查询数据库对应表 String[] split = tables.split("\."); if (split.length > 2) { throw new SQLParseException("parse table:" + tables); } List<String> colByTab = MetaCache.getInstance().getColumnByDBAndTable(tables); for (String col : colByTab) { if (column.equalsIgnoreCase(col)) { _realTable = tables; cnt++; } } }// if (cnt == 0) { //此类没有找到的检查在Validater类中检查// } if (cnt > 1) { //二义性检查 throw new SQLParseException("SQL is ambiguity, column: " + column + " tables:" + tableArray); } Block bk = new Block(); bk.setCondition(_realTable + SPLIT_DOT + column); bk.getColSet().add(_realTable + SPLIT_DOT + column); return bk; } /* * 过滤掉无用的列:如col1,123,'2013',col2 ==>> col1,col2 * @param col * @return / private Set<String> filterData(Set<String> colSet){ Set<String> set = new LinkedHashSet<String>(); for (String string : colSet) { if (!notNormalCol(string)) { set.add(string); } } return set; } /* * 解析所有子节点 * @param ast * @return / private void parseChildNodes(ASTNode ast){ int numCh = ast.getChildCount(); if (numCh > 0) { for (int num = 0; num < numCh; num++) { ASTNode child = (ASTNode) ast.getChild(num); parseIteral(child); } } } /* * 准备解析当前节点 * @param ast / private void prepareToParseCurrentNodeAndChilds(ASTNode ast){ if (ast.getToken() != null) { switch (ast.getToken().getType()) { case HiveParser.TOK_SWITCHDATABASE: System.out.println("nowQueryDB changed " + nowQueryDB+ " to " +ast.getChild(0).getText()); nowQueryDB = ast.getChild(0).getText(); break; case HiveParser.TOK_TRANSFORM: throw new UnSupportedException("no support transform using clause"); case HiveParser.TOK_RIGHTOUTERJOIN: case HiveParser.TOK_LEFTOUTERJOIN: case HiveParser.TOK_JOIN: case HiveParser.TOK_LEFTSEMIJOIN: case HiveParser.TOK_MAPJOIN: case HiveParser.TOK_FULLOUTERJOIN: case HiveParser.TOK_UNIQUEJOIN: joinStack.push(joinClause); joinClause = true; joinOnStack.push(joinOn); joinOn = ast; break; } } } /* * 结束解析当前节点 * @param ast / private void endParseCurrentNode(ASTNode ast){ if (ast.getToken() != null) { Tree parent = ast.getParent(); switch (ast.getToken().getType()) { //join 从句结束,跳出join case HiveParser.TOK_RIGHTOUTERJOIN: case HiveParser.TOK_LEFTOUTERJOIN: case HiveParser.TOK_JOIN: case HiveParser.TOK_LEFTSEMIJOIN: case HiveParser.TOK_MAPJOIN: case HiveParser.TOK_FULLOUTERJOIN: case HiveParser.TOK_UNIQUEJOIN: joinClause = joinStack.pop(); joinOn = joinOnStack.pop(); break; case HiveParser.TOK_QUERY: processUnionStack(ast, parent); //union的子节点 case HiveParser.TOK_INSERT: case HiveParser.TOK_SELECT: break; case HiveParser.TOK_UNION: //合并union字段信息 mergeUnionCols(); processUnionStack(ast, parent); //union的子节点 break; } } } private void mergeUnionCols() { validateUnion(cols); int size = cols.size(); int colNum = size / 2; List<ColLine> list = new ArrayList<ColLine>(colNum); for (int i = 0; i < colNum; i++) { //合并字段 ColLine col = cols.get(i); for (int j = i + colNum; j < size; j = j + colNum) { ColLine col2 = cols.get(j); list.add(col2); if (notNormalCol(col.getToNameParse()) && !notNormalCol(col2.getToNameParse())) { col.setToNameParse(col2.getToNameParse()); } col.getFromNameSet().addAll(col2.getFromNameSet()); col.setColCondition(col.getColCondition() + SPLIT_AND + col2.getColCondition()); Set<String> conditionSet = ParseUtil.cloneSet(col.getConditionSet()); conditionSet.addAll(col2.getConditionSet()); conditionSet.addAll(conditions); col.getConditionSet().addAll(conditionSet); } } cols.removeAll(list); //移除已经合并的数据 } private void processUnionStack(ASTNode ast, Tree parent) { boolean isNeedAdd = parent.getType() == HiveParser.TOK_UNION; if (isNeedAdd) { if (parent.getChild(0) == ast && parent.getChild(1) != null) {//有弟节点(是第一节点) //压栈 conditionsStack.push(ParseUtil.cloneSet(conditions)); conditions.clear(); colsStack.push(ParseUtil.cloneList(cols)); cols.clear(); } else { //无弟节点(是第二节点) //出栈 if (!conditionsStack.isEmpty()) { conditions.addAll(conditionsStack.pop()); } if (!colsStack.isEmpty()) { cols.addAll(0, colsStack.pop()); } } } } private void parseAST(ASTNode ast) { parseIteral(ast); } public void parse(String sqlAll) throws Exception{ if (Check.isEmpty(sqlAll)) { return; } startParseAll(); //清空最终结果集 int i = 0; //当前是第几个sql for (String sql : sqlAll.split("(?<!\\);")) { ParseDriver pd = new ParseDriver(); String trim = sql.toLowerCase().trim(); if (trim.startsWith("set") || trim.startsWith("add") || Check.isEmpty(trim)) { continue; } ASTNode ast = pd.parse(sql); if ("local".equals(PropertyFileUtil.getProperty("environment"))) { System.out.println(ast.toStringTree()); } prepareParse(); parseAST(ast); endParse(++i); } } /* * 清空上次处理的结果 / private void startParseAll() { colLines.clear(); outputTables.clear(); inputTables.clear(); } private void prepareParse() { isCreateTable = false; dbMap.clear(); queryMap.clear(); queryTreeList.clear(); //结果 tmpColLines.clear(); tmpOutputTables.clear(); tmpInputTables.clear(); conditionsStack.clear(); //where or join 条件缓存 colsStack.clear(); //一个子查询内的列缓存 resultQueryMap.clear(); conditions.clear(); //where or join 条件缓存 cols.clear(); //一个子查询内的列缓存 tableNameStack.clear(); joinStack.clear(); joinOnStack.clear(); joinClause = false; joinOn = null; } /* * 所有解析完毕之后的后期处理 / private void endParse(int sqlIndex) { putResultQueryMap(sqlIndex, TOK_EOF); putDBMap(); setColLineList(); setOutInputTableSet(); } /** * 设置输出表的字段对应关系 / private void setColLineList() { Map<String, List<ColLine>> map = new HashMap<String, List<ColLine>>(); for (Entry<String, List<ColLine>> entry : resultQueryMap.entrySet()) { if (entry.getKey().startsWith(TOK_EOF)) { List<ColLine> value = entry.getValue(); for (ColLine colLine : value) { List<ColLine> list = map.get(colLine.getToTable()); if (Check.isEmpty(list)) { list = new ArrayList<ColLine>(); map.put(colLine.getToTable(), list); } list.add(colLine); } } } for (Entry<String, List<ColLine>> entry : map.entrySet()) { String table = entry.getKey(); List<ColLine> pList = entry.getValue(); List<String> dList = dbMap.get(table); int metaSize = Check.isEmpty(dList) ? 0 : dList.size(); for (int i = 0; i < pList.size(); i++) { //按顺序插入对应的字段 ColLine clp = pList.get(i); String colName = null; if (i < metaSize) { colName = table + SPLIT_DOT + dList.get(i); } if (isCreateTable && TOK_TMP_FILE.equals(table)) { for (String string : tmpOutputTables) { table = string; } } ColLine colLine = new ColLine(clp.getToNameParse(), clp.getColCondition(), clp.getFromNameSet(), clp.getConditionSet(), table, colName); colLines.add(colLine); } } } /** * 设置输出表的字段对应关系 / private void setOutInputTableSet() { outputTables.addAll(ParseUtil.cloneSet(tmpOutputTables)); inputTables.addAll(ParseUtil.cloneSet(tmpInputTables)); } private void putDBMap() { for (String table : tmpOutputTables) { List<String> list = MetaCache.getInstance().getColumnByDBAndTable(table); dbMap.put(table, list); } } /* * 补全db信息 * table1 ==>> db1.table1 * db1.table1 ==>> db1.table1 * db2.t1&t2 ==>> db2.t1&db1.t2 * @param tables / private String fillDB(String nowTable) { if (Check.isEmpty(nowTable)) { return nowTable; } StringBuilder sb = new StringBuilder(); String[] tableArr = nowTable.split(SPLIT_AND); //fact.test&test2&test3 for (String tables : tableArr) { String[] split = tables.split("\" + SPLIT_DOT); if (split.length > 2) { System.out.println(tables); throw new SQLParseException("parse table:" + nowTable); } String db = split.length == 2 ? split[0] : nowQueryDB ; String table = split.length == 2 ? split[1] : split[0] ; sb.append(db).append(SPLIT_DOT).append(table).append(SPLIT_AND); } if (sb.length()>0) { sb.setLength(sb.length()-1); } return sb.toString(); } /* * 根据别名查询表明 * @param alia * @return / private String[] getTableAndAlia(String alia) { String _alia = Check.notEmpty(alia) ? alia : ParseUtil.collectionToString(queryMap.keySet(), SPLIT_AND, true) ; String[] result = {"" , _alia}; Set<String> tableSet = new HashSet<String>(); if (Check.notEmpty(_alia)) { String[] split = _alia.split(SPLIT_AND); for (String string : split) { //别名又分单独起的别名 和 表名,即 select a.col,table_name.col from table_name a if (tmpInputTables.contains(string) || tmpInputTables.contains(fillDB(string))) { tableSet.add(fillDB(string)); } else if (queryMap.containsKey(string.toLowerCase())) { tableSet.addAll(queryMap.get(string.toLowerCase()).getTableSet()); } } result[0] = ParseUtil.collectionToString(tableSet, SPLIT_AND, true); result[1] = _alia; } return result; } /* * 校验union * @param list */ private void validateUnion(List<ColLine> list){ int size = list.size(); if (size % 2 == 1) { throw new SQLParseException("union column number are different, size=" + size); } int colNum = size / 2; checkUnion(list, 0, colNum); checkUnion(list, colNum, size); } private void checkUnion(List<ColLine> list, int start, int end) { String tmp = null; for (int i = start; i < end; i++) { //合并字段 ColLine col = list.get(i); if (Check.isEmpty(tmp)) { tmp = col.getToTable(); } else if (!tmp.equals(col.getToTable())){ throw new SQLParseException("union column number/types are different,table1=" + tmp +",table2="+ col.getToTable()); } } }}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
另,附上一AST语法树格式化的Python脚本。
!/usr/bin/env python# -- coding: utf-8 --'''Created on 2012-5-20'''import sys# explain select key from kv mykv join test mytest on (mykv.key == mytest.id);original_str = """(TOK_QUERY (TOK_FROM (TOK_LEFTOUTERJOIN (TOK_SUBQUERY (TOK_QUERY (TOK_FROM (TOK_TABREF (TOK_TABNAME dim_city))) (TOK_INSERT (TOK_DESTINATION (TOK_DIR TOK_TMP_FILE)) (TOK_SELECT (TOK_SELEXPR (TOK_TABLE_OR_COL city_name)) (TOK_SELEXPR (TOK_TABLE_OR_COL city_id)) (TOK_SELEXPR (TOK_TABLE_OR_COL pt))) (TOK_WHERE (AND (= (TOK_TABLE_OR_COL pt) '$yesday') (= (TOK_TABLE_OR_COL level) 2))) (TOK_GROUPBY (TOK_TABLE_OR_COL city_name) (TOK_TABLE_OR_COL city_id) (TOK_TABLE_OR_COL pt)))) b) (TOK_SUBQUERY (TOK_QUERY (TOK_FROM (TOK_TABREF (TOK_TABNAME dw_dri_wide_sheet))) (TOK_INSERT (TOK_DESTINATION (TOK_DIR TOK_TMP_FILE)) (TOK_SELECT (TOK_SELEXPR (TOK_TABLE_OR_COL city_id)) (TOK_SELEXPR (TOK_TABLE_OR_COL pt)) (TOK_SELEXPR (TOK_FUNCTIONDI count (TOK_FUNCTION when (= (TOK_FUNCTION to_date (TOK_TABLE_OR_COL last_sucgrabord_time)) '$data_desc') (TOK_TABLE_OR_COL dri_id))) last1_dri_cnt) (TOK_SELEXPR (TOK_FUNCTIONDI count (TOK_FUNCTION when (and (> (TOK_FUNCTION to_date (TOK_TABLE_OR_COL last_sucgrabord_time)) (TOK_FUNCTION date_sub '$data_desc' 7)) (<= (TOK_FUNCTION to_date (TOK_TABLE_OR_COL last_sucgrabord_time)) '$data_desc')) (TOK_TABLE_OR_COL dri_id))) last7_dri_cnt)) (TOK_WHERE (and (= (TOK_TABLE_OR_COL pt) '$data_desc') (TOK_FUNCTION TOK_ISNOTNULL (TOK_TABLE_OR_COL last_sucgrabord_time)))) (TOK_GROUPBY (TOK_TABLE_OR_COL city_id) (TOK_TABLE_OR_COL pt)))) a) (= (. (TOK_TABLE_OR_COL a) city_id) (. (TOK_TABLE_OR_COL b) city_id)))) (TOK_INSERT (TOK_DESTINATION (TOK_TAB (TOK_TABNAME test kd_st_kpi_dri_active_day_city_bi))) (TOK_SELECT (TOK_SELEXPR (. (TOK_TABLE_OR_COL b) city_name)) (TOK_SELEXPR (. (TOK_TABLE_OR_COL b) city_id)) (TOK_SELEXPR (TOK_FUNCTION nvl (TOK_TABLE_OR_COL last1_dri_cnt) 0)) (TOK_SELEXPR (TOK_FUNCTION nvl (TOK_TABLE_OR_COL last7_dri_cnt) 0)) (TOK_SELEXPR (. (TOK_TABLE_OR_COL b) pt))))) <EOF>"""tmp_str = original_str.strip().replace('\n', '')def my_print(mystr): sys.stdout.write(mystr)def print_indent(indent_level): for i in range(indent_level): my_print(' ' * 4)indent_level = 0for char in tmp_str: if char == '(': # 如果是左括号,先换行,然后打印缩进+( my_print('\n') print_indent(indent_level) my_print(char) indent_level += 1 elif char == ')': # 如果是右括号,先打印),再换行,打印下一级别的缩进 indent_level -= 1 my_print(char) my_print('\n') print_indent(indent_level - 1) else: # 其他的直接打印出来 my_print(char)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
参考文章: http://tech.meituan.com/hive-sql-to-mapreduce.html http://www.cnblogs.com/drawwindows/p/4595771.html https://cwiki.apache.org/confluence/display/Hive/LanguageManual