Merge pull request #10223 from dbeaver/sql-parser-blocks#5467

#5467 Fix for parsing inner blocks

Former-commit-id: fa213e7f

plugins/org.jkiss.dbeaver.ext.oracle/src/org/jkiss/dbeaver/ext/oracle/model/OracleSQLDialect.java

@@ -63,6 +63,11 @@ class OracleSQLDialect extends JDBCSQLDialect {
         "DECLARE",
     };
 
+    public static final String[] ORACLE_INNER_BLOCK_PREFIXES = new String[]{
+        "AS",
+        "IS",
+    };
+
     public static final String[] OTHER_TYPES_FUNCTIONS = {
         //functions without parentheses #8710
         "CURRENT_DATE",
@@ -350,6 +355,12 @@ class OracleSQLDialect extends JDBCSQLDialect {
         return ORACLE_BLOCK_HEADERS;
     }
 
+    @Nullable
+    @Override
+    public String[] getInnerBlockPrefixes() {
+        return ORACLE_INNER_BLOCK_PREFIXES;
+    }
+
     @NotNull
     @Override
     public String[] getExecuteKeywords() {

plugins/org.jkiss.dbeaver.ext.postgresql/src/org/jkiss/dbeaver/ext/postgresql/model/PostgreDialect.java

@@ -779,12 +779,6 @@ public class PostgreDialect extends JDBCSQLDialect implements TPRuleProvider {
         return BLOCK_BOUND_KEYWORDS;
     }
 
-    @Nullable
-    @Override
-    public String[] getBlockHeaderStrings() {
-        return new String[] { "DECLARE" };
-    }
-
     @NotNull
     @Override
     public String getTypeCastClause(DBSAttributeBase attribute, String expression) {

plugins/org.jkiss.dbeaver.model.sql/src/org/jkiss/dbeaver/model/sql/parser/SQLScriptParser.java

@@ -129,10 +129,7 @@ public class SQLScriptParser
                 }
 
                 if (tokenType == SQLTokenType.T_BLOCK_HEADER) {
-                    if (curBlock == null) {
-                        // Check for double block header, e.g. DO, DECLARE
-                        curBlock = new ScriptBlockInfo(curBlock, true);
-                    }
+                    curBlock = new ScriptBlockInfo(curBlock, true);
                     hasBlocks = true;
                 } else if (tokenType == SQLTokenType.T_BLOCK_TOGGLE) {
                     String togglePattern;
@@ -142,39 +139,28 @@ public class SQLScriptParser
                         log.warn(e);
                         togglePattern = "";
                     }
-                    // Second toggle pattern must be the same as first one.
-                    // Toggles can be nested (PostgreSQL) and we need to count only outer
-                    if (curBlock != null && curBlock.parent == null && togglePattern.equals(curBlock.togglePattern)) {
+
+                    if (curBlock != null && togglePattern.equals(curBlock.togglePattern)) {
                         curBlock = curBlock.parent;
-                    } else if (curBlock == null) {
-                        curBlock = new ScriptBlockInfo(curBlock, togglePattern);
                     } else {
-                        log.debug("Block toggle token inside another block. Can't process it");
+                        curBlock = new ScriptBlockInfo(curBlock, togglePattern);
+                        hasBlocks = true;
                     }
-                    hasBlocks = true;
                 } else if (tokenType == SQLTokenType.T_BLOCK_BEGIN) {
-                    if (curBlock == null || !curBlock.isHeader) {
-                        curBlock = new ScriptBlockInfo(curBlock, false);
-                    } else {
-                        curBlock.isHeader = false;
+                    // Drop header block if it is followed by a regular block and
+                    // that block is not preceded by the prefix e.g 'AS', because in many dialects
+                    // there's no direct header block terminators
+                    // like 'BEGIN ... END' but 'DECLARE ... BEGIN ... END'
+                    if (curBlock != null && curBlock.isHeader) {
+                        if (prevNotEmptyTokenType != SQLTokenType.T_KEYWORD || !ArrayUtils.contains(dialect.getInnerBlockPrefixes(), lastKeyword)) {
+                            curBlock = curBlock.parent;
+                        }
                     }
+                    curBlock = new ScriptBlockInfo(curBlock, false);
                     hasBlocks = true;
                 } else if (tokenType == SQLTokenType.T_BLOCK_END) {
-                    // Sometimes query contains END clause without BEGIN. E.g. CASE, IF, etc.
-                    // This END doesn't mean block
                     if (curBlock != null) {
-                        if (!CommonUtils.isEmpty(curBlock.togglePattern)) {
-                            // Block end inside of block toggle (#7460).
-                            // Actually it is a result of some wrong SQL parse (e.g. we didn't recognize block begin correctly).
-                            // However block toggle has higher priority. At the moment it is PostgreSQL specific.
-                            try {
-                                log.debug("Block end '" + document.get(tokenOffset, tokenLength) + "' inside of named block toggle '" + curBlock.togglePattern + "'. Ignore.");
-                            } catch (Throwable e) {
-                                log.debug(e);
-                            }
-                        } else {
-                            curBlock = curBlock.parent;
-                        }
+                        curBlock = curBlock.parent;
                     }
                 } else if (isDelimiter && curBlock != null) {
                     // Delimiter in some brackets or inside block. Ignore it.
@@ -285,6 +271,9 @@ public class SQLScriptParser
                         if (tokenType == SQLTokenType.T_DELIMITER) {
                             queryEndPos += tokenLength;
                         }
+                        if (curBlock != null) {
+                            log.warn("Found leftover blocks in script after parsing");
+                        }
                         // make script line
                         return new SQLQuery(
                             context.getDataSource(),

plugins/org.jkiss.dbeaver.model/src/org/jkiss/dbeaver/model/impl/sql/AbstractSQLDialect.java

@@ -292,6 +292,12 @@ public abstract class AbstractSQLDialect implements SQLDialect {
         return null;
     }
 
+    @Nullable
+    @Override
+    public String[] getInnerBlockPrefixes() {
+        return null;
+    }
+
     @Override
     public boolean validIdentifierStart(char c) {
         return Character.isLetter(c);

plugins/org.jkiss.dbeaver.model/src/org/jkiss/dbeaver/model/sql/SQLDialect.java

@@ -206,6 +206,14 @@ public interface SQLDialect {
     @Nullable
     String[] getBlockHeaderStrings();
 
+    /**
+     * Inner block prefixes strings.
+     * Determines if the block is a child of the header block.
+     * @return inner block prefixes or null (if not supported)
+     */
+    @Nullable
+    String[] getInnerBlockPrefixes();
+
     /**
      * Retrieves whether a catalog appears at the start of a fully qualified
      * table name.  If not, the catalog appears at the end.