From af336048f7f053f1c27e9c87530b2300ab20b015 Mon Sep 17 00:00:00 2001
From: Dan Goodliffe <dan@randomdan.homeip.net>
Date: Mon, 4 Dec 2017 20:32:55 +0000
Subject: Assorted improvements to he SQL lexer to better handle whitespace and
 comments mixed into statements

---
 libdbpp/sqlParse.ll                          | 38 +++++++++++++++++----------
 libdbpp/unittests/Jamfile.jam                |  4 +++
 libdbpp/unittests/commentsMixedIn.sql        |  7 +++++
 libdbpp/unittests/indentedBlockComment.sql   |  5 ++++
 libdbpp/unittests/indentedOneLineComment.sql |  5 ++++
 libdbpp/unittests/indentedStatement.sql      |  5 ++++
 libdbpp/unittests/testParse.cpp              | 39 ++++++++++++++++++++++++++++
 7 files changed, 89 insertions(+), 14 deletions(-)
 create mode 100644 libdbpp/unittests/commentsMixedIn.sql
 create mode 100644 libdbpp/unittests/indentedBlockComment.sql
 create mode 100644 libdbpp/unittests/indentedOneLineComment.sql
 create mode 100644 libdbpp/unittests/indentedStatement.sql

diff --git a/libdbpp/sqlParse.ll b/libdbpp/sqlParse.ll
index f9c1525..835ce20 100644
--- a/libdbpp/sqlParse.ll
+++ b/libdbpp/sqlParse.ll
@@ -14,9 +14,9 @@
 
 space			[ \t\n\r\f]
 non_newline [^\r\n]
-mcomment_start "/*"
-mcomment_stop "*/"
-comment			("--"{non_newline}*)
+mcomment_start ("/*"{space}*)
+mcomment_stop ({space}*"*/")
+lcomment_start	({space}*"--"{space}*)
 other .
 term ;
 any ({other}|{space})
@@ -28,41 +28,54 @@ dolq_cont [A-Za-z\200-\377_0-9]
 dollarquote \$({dolq_start}{dolq_cont}*)?\$
 scriptdir "$SCRIPTDIR"
 
-%x COMMENT
+%x MCOMMENT
+%x LCOMMENT
 %x STATEMENT
 %x QUOTE
 %x DOLLARQUOTE
 
 %%
 
+<STATEMENT>{mcomment_start} {
+	yy_push_state(MCOMMENT);
+}
+
 {mcomment_start} {
-	comment += YYText();
-	yy_push_state(COMMENT);
+	yy_push_state(MCOMMENT);
 }
 
-<COMMENT>{mcomment_stop} {
-	comment += YYText();
+<STATEMENT>{lcomment_start} {
+	yy_push_state(LCOMMENT);
+}
+
+{lcomment_start} {
+	yy_push_state(LCOMMENT);
+}
+
+<MCOMMENT>{mcomment_stop} {
 	Comment(comment);
 	comment.clear();
 	yy_pop_state();
 }
 
-<COMMENT>{any} {
+<MCOMMENT>{any} {
 	comment += YYText();
 }
 
-<COMMENT><<EOF>> {
+<MCOMMENT><<EOF>> {
 	throw SqlParseException("Unterminated comment", yylineno);
 }
 
-{comment} {
+<LCOMMENT>{non_newline}* {
 	Comment(YYText());
+  yy_pop_state();
 }
 
 <INITIAL>{term} {
 	// Random terminator
 }
 
+{space} { }
 {other} {
 	statement += YYText();
 	yy_push_state(STATEMENT);
@@ -126,6 +139,3 @@ scriptdir "$SCRIPTDIR"
 	statement += YYText();
 }
 
-<*>[ \t\r\n\f] {
-}
-
diff --git a/libdbpp/unittests/Jamfile.jam b/libdbpp/unittests/Jamfile.jam
index d0500f0..83e1ad4 100644
--- a/libdbpp/unittests/Jamfile.jam
+++ b/libdbpp/unittests/Jamfile.jam
@@ -29,6 +29,10 @@ run
 	<library>..//dbppcore
 	<library>..//adhocutil
 	<library>boost_utf
+	<dependency>commentsMixedIn.sql
+	<dependency>indentedStatement.sql
+	<dependency>indentedOneLineComment.sql
+	<dependency>indentedBlockComment.sql
 	<dependency>parseTest.sql
 	<dependency>unterminatedComment.sql
 	<dependency>unterminatedString.sql
diff --git a/libdbpp/unittests/commentsMixedIn.sql b/libdbpp/unittests/commentsMixedIn.sql
new file mode 100644
index 0000000..1766309
--- /dev/null
+++ b/libdbpp/unittests/commentsMixedIn.sql
@@ -0,0 +1,7 @@
+CREATE TABLE foo(/*
+										Foo contains test things
+										*/
+		id int, -- Every table deserves an Id, right?
+		timestamp time stamp -- And a timestamp
+		);
+
diff --git a/libdbpp/unittests/indentedBlockComment.sql b/libdbpp/unittests/indentedBlockComment.sql
new file mode 100644
index 0000000..45294a0
--- /dev/null
+++ b/libdbpp/unittests/indentedBlockComment.sql
@@ -0,0 +1,5 @@
+
+
+     /* Some comment text
+			*/
+
diff --git a/libdbpp/unittests/indentedOneLineComment.sql b/libdbpp/unittests/indentedOneLineComment.sql
new file mode 100644
index 0000000..d72bb31
--- /dev/null
+++ b/libdbpp/unittests/indentedOneLineComment.sql
@@ -0,0 +1,5 @@
+
+
+     -- Some comment text
+
+
diff --git a/libdbpp/unittests/indentedStatement.sql b/libdbpp/unittests/indentedStatement.sql
new file mode 100644
index 0000000..4273f35
--- /dev/null
+++ b/libdbpp/unittests/indentedStatement.sql
@@ -0,0 +1,5 @@
+
+
+    SELECT 1;
+
+
diff --git a/libdbpp/unittests/testParse.cpp b/libdbpp/unittests/testParse.cpp
index 1272f9b..093cc7d 100644
--- a/libdbpp/unittests/testParse.cpp
+++ b/libdbpp/unittests/testParse.cpp
@@ -78,6 +78,45 @@ BOOST_AUTO_TEST_CASE( parseStringParse )
 	BOOST_REQUIRE_EQUAL("INSERT INTO name(t, i) VALUES('fancy string '' \\' \\r \\n', 7)", p.executed[1]);
 }
 
+BOOST_AUTO_TEST_CASE( indentedStatement )
+{
+	RecordingParser p(rootDir / "indentedStatement.sql");
+	p.Execute();
+	BOOST_REQUIRE_EQUAL(1, p.executed.size());
+	BOOST_REQUIRE_EQUAL("SELECT 1", p.executed[0]);
+	BOOST_REQUIRE(p.comments.empty());
+}
+
+BOOST_AUTO_TEST_CASE( indentedOneLineComment )
+{
+	RecordingParser p(rootDir / "indentedOneLineComment.sql");
+	p.Execute();
+	BOOST_REQUIRE_EQUAL(1, p.comments.size());
+	BOOST_REQUIRE_EQUAL("Some comment text", p.comments[0]);
+	BOOST_REQUIRE(p.executed.empty());
+}
+
+BOOST_AUTO_TEST_CASE( indentedBlockComment )
+{
+	RecordingParser p(rootDir / "indentedBlockComment.sql");
+	p.Execute();
+	BOOST_REQUIRE_EQUAL(1, p.comments.size());
+	BOOST_REQUIRE_EQUAL("Some comment text", p.comments[0]);
+	BOOST_REQUIRE(p.executed.empty());
+}
+
+BOOST_AUTO_TEST_CASE( commentsMixedIn )
+{
+	RecordingParser p(rootDir / "commentsMixedIn.sql");
+	p.Execute();
+	BOOST_REQUIRE_EQUAL(1, p.executed.size());
+	BOOST_REQUIRE_EQUAL("CREATE TABLE foo(\n\t\tid int,\n\t\ttimestamp time stamp\n\t\t)", p.executed[0]);
+	BOOST_REQUIRE_EQUAL(3, p.comments.size());
+	BOOST_REQUIRE_EQUAL("Foo contains test things", p.comments[0]);
+	BOOST_REQUIRE_EQUAL("Every table deserves an Id, right?", p.comments[1]);
+	BOOST_REQUIRE_EQUAL("And a timestamp", p.comments[2]);
+}
+
 BOOST_AUTO_TEST_CASE( parseUnterminateComment )
 {
 	assertFail(rootDir / "unterminatedComment.sql");
-- 
cgit v1.2.3