Improved links in the grammar doc

h2/src/docsrc/html/advanced.html

@@ -106,7 +106,7 @@ on the client side.
 <p>
 This database stores large LOB (CLOB and BLOB) objects as separate files.
 Small LOB objects are stored in-place, the threshold can be set using
-<a href="grammar.html#setmaxlengthinplacelob">MAX_LENGTH_INPLACE_LOB</a>,
+<a href="grammar.html#set_max_length_inplace_lob">MAX_LENGTH_INPLACE_LOB</a>,
 but there is still an overhead to use CLOB/BLOB. Because of this, BLOB and CLOB
 should never be used for columns with a maximum size below about 200 bytes.
 The best threshold depends on the use case; reading in-place objects is faster
@@ -116,7 +116,7 @@ that don't involve this column.
 
 <h3>Large Object Compression</h3>
 <p>
-CLOB and BLOB values can be compressed by using <a href="grammar.html#setcompresslob">SET COMPRESS_LOB</a>.
+CLOB and BLOB values can be compressed by using <a href="grammar.html#set_compress_lob">SET COMPRESS_LOB</a>.
 The LZF algorithm is faster but needs more disk space. By default compression is disabled, which usually speeds up write
 operations. If you store many large compressible values such as XML, HTML, text, and uncompressed binary files,
 then compressing can save a lot of disk space (sometimes more than 50%), and read operations may even be faster.

h2/src/main/org/h2/bnf/Bnf.java

@@ -91,8 +91,8 @@ public class Bnf {
                 continue;
             }
             String topic = StringUtils.toLowerEnglish(rs.getString("TOPIC").trim());
-            topic = StringUtils.replaceAll(topic, " ", "");
-            topic = StringUtils.replaceAll(topic, "_", "");
+            topic = StringUtils.replaceAll(topic, " ", "_");
+            // topic = StringUtils.replaceAll(topic, "_", "");
             syntax = rs.getString("SYNTAX").trim();
             currentTopic = section;
             if (section.startsWith("Function")) {