diff --git a/src/adql/parser/ADQLParser.java b/src/adql/parser/ADQLParser.java
index ce4f24d3bb90e57a924248e56ff015d16dcfa4e9..704d43b0adf8103419f8b2ee3ae9bca2feea7a8f 100644
--- a/src/adql/parser/ADQLParser.java
+++ b/src/adql/parser/ADQLParser.java
@@ -67,7 +67,7 @@ import adql.translator.TranslationException;
* @see ADQLQueryFactory
*
* @author Grégory Mantelet (CDS;ARI) - gmantele@ari.uni-heidelberg.de
-* @version 1.4 (04/2017)
+* @version 1.4 (09/2017)
*/
public class ADQLParser implements ADQLParserConstants {
@@ -295,7 +295,11 @@ public class ADQLParser implements ADQLParserConstants {
public final ADQLQuery parseQuery() throws ParseException{
stackQuery.clear();
query = null;
- return Query();
+ try{
+ return Query();
+ }catch(TokenMgrError tme){
+ throw new ParseException(tme.getMessage(), new TextPosition(tme.getErrorLine(), tme.getErrorColumn()));
+ }
}
/**
@@ -313,7 +317,11 @@ public class ADQLParser implements ADQLParserConstants {
stackQuery.clear();
query = null;
ReInit(new java.io.ByteArrayInputStream(q.getBytes()));
- return Query();
+ try{
+ return Query();
+ }catch(TokenMgrError tme){
+ throw new ParseException(tme.getMessage(), new TextPosition(tme.getErrorLine(), tme.getErrorColumn()));
+ }
}
/**
@@ -331,7 +339,11 @@ public class ADQLParser implements ADQLParserConstants {
stackQuery.clear();
query = null;
ReInit(stream);
- return Query();
+ try{
+ return Query();
+ }catch(TokenMgrError tme){
+ throw new ParseException(tme.getMessage(), new TextPosition(tme.getErrorLine(), tme.getErrorColumn()));
+ }
}
public final void setDebug(boolean debug){
@@ -3912,82 +3924,6 @@ public class ADQLParser implements ADQLParserConstants {
}
}
- private boolean jj_3R_59(){
- if (jj_3R_24())
- return true;
- return false;
- }
-
- private boolean jj_3R_58(){
- if (jj_scan_token(LEFT_PAR))
- return true;
- if (jj_3R_42())
- return true;
- if (jj_scan_token(RIGHT_PAR))
- return true;
- return false;
- }
-
- private boolean jj_3R_57(){
- if (jj_3R_27())
- return true;
- return false;
- }
-
- private boolean jj_3R_124(){
- if (jj_scan_token(CIRCLE))
- return true;
- if (jj_scan_token(LEFT_PAR))
- return true;
- if (jj_3R_134())
- return true;
- if (jj_scan_token(COMMA))
- return true;
- if (jj_3R_135())
- return true;
- if (jj_scan_token(COMMA))
- return true;
- if (jj_3R_101())
- return true;
- if (jj_scan_token(RIGHT_PAR))
- return true;
- return false;
- }
-
- private boolean jj_3R_44(){
- if (jj_scan_token(SELECT))
- return true;
- return false;
- }
-
- private boolean jj_3R_56(){
- if (jj_3R_101())
- return true;
- return false;
- }
-
- private boolean jj_3R_123(){
- if (jj_scan_token(CENTROID))
- return true;
- if (jj_scan_token(LEFT_PAR))
- return true;
- if (jj_3R_113())
- return true;
- if (jj_scan_token(RIGHT_PAR))
- return true;
- return false;
- }
-
- private boolean jj_3R_117(){
- if (jj_scan_token(LEFT_PAR))
- return true;
- if (jj_3R_27())
- return true;
- if (jj_scan_token(RIGHT_PAR))
- return true;
- return false;
- }
-
private boolean jj_3R_16(){
if (jj_scan_token(LEFT_PAR))
return true;
@@ -5612,6 +5548,82 @@ public class ADQLParser implements ADQLParserConstants {
return false;
}
+ private boolean jj_3R_59(){
+ if (jj_3R_24())
+ return true;
+ return false;
+ }
+
+ private boolean jj_3R_58(){
+ if (jj_scan_token(LEFT_PAR))
+ return true;
+ if (jj_3R_42())
+ return true;
+ if (jj_scan_token(RIGHT_PAR))
+ return true;
+ return false;
+ }
+
+ private boolean jj_3R_57(){
+ if (jj_3R_27())
+ return true;
+ return false;
+ }
+
+ private boolean jj_3R_124(){
+ if (jj_scan_token(CIRCLE))
+ return true;
+ if (jj_scan_token(LEFT_PAR))
+ return true;
+ if (jj_3R_134())
+ return true;
+ if (jj_scan_token(COMMA))
+ return true;
+ if (jj_3R_135())
+ return true;
+ if (jj_scan_token(COMMA))
+ return true;
+ if (jj_3R_101())
+ return true;
+ if (jj_scan_token(RIGHT_PAR))
+ return true;
+ return false;
+ }
+
+ private boolean jj_3R_44(){
+ if (jj_scan_token(SELECT))
+ return true;
+ return false;
+ }
+
+ private boolean jj_3R_56(){
+ if (jj_3R_101())
+ return true;
+ return false;
+ }
+
+ private boolean jj_3R_123(){
+ if (jj_scan_token(CENTROID))
+ return true;
+ if (jj_scan_token(LEFT_PAR))
+ return true;
+ if (jj_3R_113())
+ return true;
+ if (jj_scan_token(RIGHT_PAR))
+ return true;
+ return false;
+ }
+
+ private boolean jj_3R_117(){
+ if (jj_scan_token(LEFT_PAR))
+ return true;
+ if (jj_3R_27())
+ return true;
+ if (jj_scan_token(RIGHT_PAR))
+ return true;
+ return false;
+ }
+
/** Generated Token Manager. */
public ADQLParserTokenManager token_source;
SimpleCharStream jj_input_stream;
diff --git a/src/adql/parser/ParseException.java b/src/adql/parser/ParseException.java
index 108993a7ccccd9dfe17c2e84f088499cd734b6fe..2751dcda98e31b93bc4285bcb35884cb65c7dc8f 100644
--- a/src/adql/parser/ParseException.java
+++ b/src/adql/parser/ParseException.java
@@ -2,12 +2,13 @@
/* JavaCCOptions:KEEP_LINE_COL=null
*
* Modified by Grégory Mantelet (CDS), on March 2017
- * Modification: several small modifications.
- * /!\ DO NOT RE-GENERATE THIS FILE /!\
- * In case of re-generation, replace it by
- * ParseException.java.backup (but maybe after a diff
- * in case of significant modifications have been done
- * by a new version of JavaCC).
+ * Modifications:
+ * - several small modifications.
+ *
+ * /!\ DO NOT RE-GENERATE THIS FILE /!\
+ * In case of re-generation, replace it by ParseException.java.backup (but maybe
+ * after a diff in case of significant modifications have been done by a new
+ * version of JavaCC).
*/
package adql.parser;
diff --git a/src/adql/parser/ParseException.java.backup b/src/adql/parser/ParseException.java.backup
index 108993a7ccccd9dfe17c2e84f088499cd734b6fe..2751dcda98e31b93bc4285bcb35884cb65c7dc8f 100644
--- a/src/adql/parser/ParseException.java.backup
+++ b/src/adql/parser/ParseException.java.backup
@@ -2,12 +2,13 @@
/* JavaCCOptions:KEEP_LINE_COL=null
*
* Modified by Grégory Mantelet (CDS), on March 2017
- * Modification: several small modifications.
- * /!\ DO NOT RE-GENERATE THIS FILE /!\
- * In case of re-generation, replace it by
- * ParseException.java.backup (but maybe after a diff
- * in case of significant modifications have been done
- * by a new version of JavaCC).
+ * Modifications:
+ * - several small modifications.
+ *
+ * /!\ DO NOT RE-GENERATE THIS FILE /!\
+ * In case of re-generation, replace it by ParseException.java.backup (but maybe
+ * after a diff in case of significant modifications have been done by a new
+ * version of JavaCC).
*/
package adql.parser;
diff --git a/src/adql/parser/TokenMgrError.java b/src/adql/parser/TokenMgrError.java
index 60cb4686111c3b65f4ad36640e338d94949d231e..fa5d8aaa60363bea7b3f223dac1488421a804663 100644
--- a/src/adql/parser/TokenMgrError.java
+++ b/src/adql/parser/TokenMgrError.java
@@ -1,13 +1,17 @@
/* Generated By:JavaCC: Do not edit this line. TokenMgrError.java Version 6.0 */
/* JavaCCOptions:
*
- * Modified by Grégory Mantelet (CDS), on April 2017
- * Modification: addition of position (line and column) in the original text.
- * /!\ DO NOT RE-GENERATE THIS FILE /!\
- * In case of re-generation, replace it by
- * TokenMgrError.java.backup (but maybe after a diff
- * in case of significant modifications have been done
- * by a new version of JavaCC).
+ * Modified by Grégory Mantelet (CDS), on Sept. 2017
+ * Modifications:
+ * - addition of position (line and column) in the original text
+ * - adapt the error message so that being more explicit for humans
+ * and display the incorrect character as a character instead of an
+ * integer value
+ *
+ * /!\ DO NOT RE-GENERATE THIS FILE /!\
+ * In case of re-generation, replace it by TokenMgrError.java.backup (but maybe
+ * after a diff in case of significant modifications have been done by a new
+ * version of JavaCC).
*/
package adql.parser;
@@ -118,7 +122,7 @@ public class TokenMgrError extends Error {
* Note: You can customize the lexical error message by modifying this method.
*/
protected static String LexicalError(boolean EOFSeen, int lexState, int errorLine, int errorColumn, String errorAfter, char curChar){
- return ("Lexical error at line " + errorLine + ", column " + errorColumn + ". Encountered: " + (EOFSeen ? "<EOF> " : ("\"" + addEscapes(String.valueOf(curChar)) + "\"") + " (" + (int)curChar + "), ") + "after : \"" + addEscapes(errorAfter) + "\"");
+ return ("Incorrect character encountered at l." + errorLine + ", c." + errorColumn + ": " + (EOFSeen ? "<EOF> " : ("\"" + addEscapes(String.valueOf(curChar)) + "\"") + " ('" + curChar + "'), ") + "after : \"" + addEscapes(errorAfter) + "\"");
}
/**
diff --git a/src/adql/parser/TokenMgrError.java.backup b/src/adql/parser/TokenMgrError.java.backup
index 60cb4686111c3b65f4ad36640e338d94949d231e..61b4a9c384d64f169734169ee09dd7d17cef9fc4 100644
--- a/src/adql/parser/TokenMgrError.java.backup
+++ b/src/adql/parser/TokenMgrError.java.backup
@@ -1,13 +1,15 @@
/* Generated By:JavaCC: Do not edit this line. TokenMgrError.java Version 6.0 */
/* JavaCCOptions:
*
- * Modified by Grégory Mantelet (CDS), on April 2017
- * Modification: addition of position (line and column) in the original text.
- * /!\ DO NOT RE-GENERATE THIS FILE /!\
- * In case of re-generation, replace it by
- * TokenMgrError.java.backup (but maybe after a diff
- * in case of significant modifications have been done
- * by a new version of JavaCC).
+ * Modified by Grégory Mantelet (CDS), on Sept. 2017
+ * Modifications:
+ * - addition of position (line and column) in the original text
+ * - display the incorrect character as a character instead of an integer
+ *
+ * /!\ DO NOT RE-GENERATE THIS FILE /!\
+ * In case of re-generation, replace it by TokenMgrError.java.backup (but maybe
+ * after a diff in case of significant modifications have been done by a new
+ * version of JavaCC).
*/
package adql.parser;
@@ -118,7 +120,7 @@ public class TokenMgrError extends Error {
* Note: You can customize the lexical error message by modifying this method.
*/
protected static String LexicalError(boolean EOFSeen, int lexState, int errorLine, int errorColumn, String errorAfter, char curChar){
- return ("Lexical error at line " + errorLine + ", column " + errorColumn + ". Encountered: " + (EOFSeen ? "<EOF> " : ("\"" + addEscapes(String.valueOf(curChar)) + "\"") + " (" + (int)curChar + "), ") + "after : \"" + addEscapes(errorAfter) + "\"");
+ return ("Incorrect character encountered at l." + errorLine + ", c." + errorColumn + ": " + (EOFSeen ? "<EOF> " : ("\"" + addEscapes(String.valueOf(curChar)) + "\"") + " ('" + curChar + "'), ") + "after : \"" + addEscapes(errorAfter) + "\"");
}
/**
diff --git a/src/adql/parser/adqlGrammar.jj b/src/adql/parser/adqlGrammar.jj
index a4d2cb4bddd98546b95325fc5ab742d2f277183d..1de71d3864354d1c64c3219a35fb85e7c0856b65 100644
--- a/src/adql/parser/adqlGrammar.jj
+++ b/src/adql/parser/adqlGrammar.jj
@@ -26,7 +26,7 @@
* If the syntax is not conform to the ADQL definition an error message is printed else it will be the message "Correct syntax".
*
* Author: Grégory Mantelet (CDS;ARI) - gmantele@ari.uni-heidelberg.de
-* Version: 1.4 (04/2017)
+* Version: 1.4 (09/2017)
*/
/* ########### */
@@ -89,7 +89,7 @@ import adql.translator.TranslationException;
* @see ADQLQueryFactory
*
* @author Grégory Mantelet (CDS;ARI) - gmantele@ari.uni-heidelberg.de
-* @version 1.4 (04/2017)
+* @version 1.4 (09/2017)
*/
public class ADQLParser {
@@ -317,7 +317,11 @@ public class ADQLParser {
public final ADQLQuery parseQuery() throws ParseException {
stackQuery.clear();
query = null;
- return Query();
+ try {
+ return Query();
+ }catch(TokenMgrError tme) {
+ throw new ParseException(tme.getMessage(), new TextPosition(tme.getErrorLine(), tme.getErrorColumn()));
+ }
}
/**
@@ -335,7 +339,11 @@ public class ADQLParser {
stackQuery.clear();
query = null;
ReInit(new java.io.ByteArrayInputStream(q.getBytes()));
- return Query();
+ try {
+ return Query();
+ }catch(TokenMgrError tme) {
+ throw new ParseException(tme.getMessage(), new TextPosition(tme.getErrorLine(), tme.getErrorColumn()));
+ }
}
/**
@@ -353,7 +361,11 @@ public class ADQLParser {
stackQuery.clear();
query = null;
ReInit(stream);
- return Query();
+ try {
+ return Query();
+ }catch(TokenMgrError tme) {
+ throw new ParseException(tme.getMessage(), new TextPosition(tme.getErrorLine(), tme.getErrorColumn()));
+ }
}
public final void setDebug(boolean debug){
diff --git a/test/adql/parser/TestADQLParser.java b/test/adql/parser/TestADQLParser.java
index 9e6a0169e107bd917d8f0d26c884ffdd550badd5..4549cb83a2719327de2a2b6e26969df2f78fdb1e 100644
--- a/test/adql/parser/TestADQLParser.java
+++ b/test/adql/parser/TestADQLParser.java
@@ -158,4 +158,25 @@ public class TestADQLParser {
}
}
+ @Test
+ public void testIncorrectCharacter(){
+ /* An identifier must be written only with digits, an underscore or
+ * regular latin characters: */
+ try{
+ (new ADQLParser()).parseQuery("select grégory FROM aTable");
+ }catch(Throwable t){
+ assertEquals(ParseException.class, t.getClass());
+ assertEquals("Incorrect character encountered at l.1, c.10: \"\\u00e9\" ('é'), after : \"\"", t.getMessage());
+ }
+
+ // But in a string, delimited identifier or a comment, it is fine:
+ try{
+ (new ADQLParser()).parseQuery("select 'grégory' FROM aTable");
+ (new ADQLParser()).parseQuery("select \"grégory\" FROM aTable");
+ (new ADQLParser()).parseQuery("select * FROM aTable -- a comment by Grégory");
+ }catch(Throwable t){
+ fail("This error should never occurs because all these queries have an accentuated character but at a correct place.");
+ }
+ }
+
}