WIP

grate.unittests/Basic/Infrastructure/PostgreSQL/Statement_Splitting/StatementSplitter_.cs

@@ -1,4 +1,5 @@
-using FluentAssertions;
+using System.Linq;
+using FluentAssertions;
 using grate.Infrastructure;
 using grate.Migration;
 using Microsoft.Extensions.Logging.Abstractions;
@@ -36,5 +37,107 @@ USING btree
 
  batches.Should().HaveCount(4);
  }
+
+ [Test]
+ public void Ignores_semicolons_in_backslash_escaped_strings()
+ {
+ var original = @"
+DO E'
+BEGIN
+ IF NOT EXISTS(
+ SELECT schema_name
+ FROM information_schema.schemata
+ WHERE schema_name = \'random\'
+ )
+ THEN
+ EXECUTE \'CREATE SCHEMA random\';
+ END IF;
+END
+';
+";
+ var batches = Splitter.Split(original);
+
+ batches.Should().HaveCount(1);
+ }
+
+ [TestCase("$$")]
+ [TestCase("$sometag$")]
+ public void Ignores_semicolons_in_dollar_quoted_strings(string tag)
+ {
+ var original = @$"
+DO 
+{tag}
+BEGIN
+ IF NOT EXISTS(
+ SELECT schema_name
+ FROM information_schema.schemata
+ WHERE schema_name = 'random'
+ )
+ THEN
+ EXECUTE 'CREATE SCHEMA random';
+ END IF;
+END
+{tag};
+";
+ var batches = Splitter.Split(original);
+ batches.Should().HaveCount(1);
+ }
+
+ [Test]
+ public void Splits_on_semicolon_after_single_quotes_when_there_is_another_semicolon_in_the_quote()
+ {
+ var original = @"SELECT 1 WHERE whatnot = '; ' ; MOO";
+ var batches = Splitter.Split(original);
+ batches.Should().HaveCount(2);
+
+ batches.First().Should().Be("SELECT 1 WHERE whatnot = '; ' ");
+ batches.Last().Should().Be(" MOO");
+ }
+
+ [Test]
+ public void Ignores_semicolon_in_single_quotes_when_there_is_no_other_semicolon()
+ {
+ var original = @"SELECT 1 WHERE whatnot = '; ' ;;";
+ var batches = Splitter.Split(original);
+ batches.Should().HaveCount(1);
+ }
+
+ [Test]
+ public void Ignores_semicolon_in_strings_with_directly_appended_operator()
+ {
+ var original = @"
+CREATE VIEW wgs.wgs_pages_branches AS
+ SELECT branches.instrumentid,
+ branches.plabel AS from_plabel,
+ CASE
+ WHEN (NOT (branches.branch ~~* '%?%'::text)) THEN ""substring""(branches.branch, (""position""(branches.branch, '->'::text) + 3), length(branches.branch))
+ ELSE NULL::text
+ END AS to_plabel,
+ CASE
+ WHEN (branches.branch ~~* '?%'::text) THEN branches.branch
+ WHEN (branches.branch ~~* '%?%'::text) THEN ""substring""(branches.branch, (""position""(branches.branch, '->'::text) + 2), length(branches.branch))
+ ELSE NULL::text
+ END AS to_other,
+ branches.rank AS condition_rank,
+ CASE
+ WHEN (""substring""(branches.branch, 0, ""position""(branches.branch, '->'::text)) ~~* 'ELSE'::text) THEN 'TRUE'::text
+ ELSE ""substring""(branches.branch, 0, ""position""(branches.branch, '->'::text))
+ END AS condition
+ FROM ( SELECT foo.instrumentid,
+ foo.plabel,
+ foo.branch,
+ row_number() OVER (PARTITION BY foo.instrumentid, foo.plabel ORDER BY foo.rankgroup) AS rank
+ FROM ( SELECT wgs_pages.instrumentid,
+ wgs_pages.plabel,
+ unnest(string_to_array(wgs_pages.nextbranch, ';'::text)) AS branch,
+ row_number() OVER (ORDER BY wgs_pages.instrumentid, wgs_pages.plabel) AS rankgroup
+ FROM wgs.wgs_pages) foo) branches;
+";
+
+ var batches = Splitter.Split(original);
+ batches.Should().HaveCount(1);
+
+ }
+
 
 }

grate/Infrastructure/PostgreSqlSyntax.cs

@@ -6,11 +6,13 @@ public string StatementSeparatorRegex
  {
  get
  {
- const string strings = @"(?<KEEP1>'[^']*')";
+ const string strings = @"(?<KEEP1>'([^']|\'\')*')";
+ const string backslashEscapedSrings = @"(?<KEEP1>E(?<!\\)('[\S\s]*?(?<!\\)'))";
+ const string dollarQuotedStrings = @"(?<KEEP1>\$(?'tag'\w*)\$[\S\s]*?\$\k'tag'\$)";
  const string dashComments = @"(?<KEEP1>--.*$)";
  const string starComments = @"(?<KEEP1>/\*[\S\s]*?\*/)";
  const string separator = @"(?<KEEP1>.*)(?<BATCHSPLITTER>;)(?<KEEP2>.*)";
- return strings + "|" + dashComments + "|" + starComments + "|" + separator;
+ return strings + "|" + backslashEscapedSrings + "|" + dollarQuotedStrings + "|" + dashComments + "|" + starComments + "|" + separator;
  }
  }