treewide: implement default lstrip_blocks and trim_blocks behaviour

Signed-off-by: Jo-Philipp Wich <jo@mein.io>

12 files changed, 65 insertions, 35 deletions

diff --git a/ast.h b/ast.h
index 9840ad4..4c5318c 100644
--- a/ast.h
+++ b/ast.h
@@ -78,6 +78,8 @@ struct ut_state {
 	uint8_t semicolon_emitted:1;
 	uint8_t start_tag_seen:1;
 	uint8_t srand_called:1;
+	uint8_t trim_blocks:1;
+	uint8_t lstrip_blocks:1;
 	size_t off;
 	enum ut_block_type blocktype;
 	struct {
diff --git a/lexer.c b/lexer.c
index 0220421..0ce9643 100644
--- a/lexer.c
+++ b/lexer.c
@@ -58,6 +58,7 @@ static const struct token tokens[] = {
 	{ T_ASRIGHT,	">>=",   3 },
 	{ T_LEXP,		"{{-",   3 },
 	{ T_REXP,		"-}}",   3 },
+	{ T_LSTM,		"{%+",   3 },
 	{ T_LSTM,		"{%-",   3 },
 	{ T_RSTM,		"-%}",   3 },
 	{ T_AND,		"&&",    2 },
@@ -731,9 +732,16 @@ ut_get_token(struct ut_state *s, const char *input, int *mlen)
 				s->off += *mlen;
 
 				/* strip whitespace before block */
-				if (p[2] == '-')
+				if (p[2] == '-') {
 					while (p > o && isspace(p[-1]))
 						p--;
+				}
+
+				/* lstrip */
+				else if (s->lstrip_blocks && s->blocktype == UT_BLOCK_STATEMENT && p[2] != '+') {
+					while (p > o && p[-1] != '\n' && isspace(p[-1]))
+						p--;
+				}
 
 				if (p == o)
 					return 0;
@@ -789,9 +797,6 @@ ut_get_token(struct ut_state *s, const char *input, int *mlen)
 					*mlen = 0;
 				}
 				else {
-					s->semicolon_emitted = false;
-					s->blocktype = UT_BLOCK_NONE;
-
 					/* strip whitespace after block */
 					if (*p == '-') {
 						while (isspace(p[3])) {
@@ -799,6 +804,13 @@ ut_get_token(struct ut_state *s, const char *input, int *mlen)
 							p++;
 						}
 					}
+					else if (s->blocktype == UT_BLOCK_STATEMENT &&
+					         s->trim_blocks && p[2] == '\n') {
+						(*mlen)++;
+					}
+
+					s->semicolon_emitted = false;
+					s->blocktype = UT_BLOCK_NONE;
 				}
 			}
 
diff --git a/main.c b/main.c
index 334d66b..68b39c1 100644
--- a/main.c
+++ b/main.c
@@ -38,11 +38,13 @@ print_usage(char *app)
 {
 	printf(
 	"== Usage ==\n\n"
-	"  # %s [-d] {-i <file> | -s \"utpl script...\"}\n"
+	"  # %s [-d] [-l] [-r] {-i <file> | -s \"utpl script...\"}\n"
 	"  -h, --help	Print this help\n"
 	"  -i file	Specify an utpl script to parse\n"
 	"  -s \"utpl script...\"	Specify an utpl code fragment to parse\n"
-	"  -d Instead of executing the script, dump the resulting AST as dot\n",
+	"  -d Instead of executing the script, dump the resulting AST as dot\n"
+	"  -l Do not strip leading block whitespace\n"
+	"  -r Do not trim trailing block newlines\n",
 		app);
 }
 
@@ -136,9 +138,8 @@ static void dump(struct ut_state *s, uint32_t off, int level) {
 #endif /* NDEBUG */
 
 static enum ut_error_type
-parse(const char *source, bool dumponly)
+parse(struct ut_state *state, const char *source, bool dumponly)
 {
-	struct ut_state *state = calloc(1, sizeof(*state));
 	enum ut_error_type err;
 	char *msg;
 
@@ -173,8 +174,9 @@ parse(const char *source, bool dumponly)
 int
 main(int argc, char **argv)
 {
-	size_t rlen, tlen = 0;
+	struct ut_state *state;
 	bool dumponly = false;
+	size_t rlen, tlen = 0;
 	char buf[1024], *tmp;
 	char *source = NULL;
 	FILE *input = NULL;
@@ -186,7 +188,17 @@ main(int argc, char **argv)
 		goto out;
 	}
 
-	while ((opt = getopt(argc, argv, "dhi:s:")) != -1)
+	state = calloc(1, sizeof(*state));
+
+	if (!state) {
+		rv = UT_ERROR_OUT_OF_MEMORY;
+		goto out;
+	}
+
+	state->lstrip_blocks = 1;
+	state->trim_blocks = 1;
+
+	while ((opt = getopt(argc, argv, "dhlri:s:")) != -1)
 	{
 		switch (opt) {
 		case 'h':
@@ -208,6 +220,14 @@ main(int argc, char **argv)
 			dumponly = true;
 			break;
 
+		case 'l':
+			state->lstrip_blocks = 0;
+			break;
+
+		case 'r':
+			state->trim_blocks = 0;
+			break;
+
 		case 's':
 			source = optarg;
 			break;
@@ -240,7 +260,7 @@ main(int argc, char **argv)
 		}
 	}
 
-	rv = source ? parse(source, dumponly) : 0;
+	rv = source ? parse(state, source, dumponly) : 0;
 
 out:
 	if (input && input != stdin)
diff --git a/tests/00_syntax/04_statement_blocks b/tests/00_syntax/04_statement_blocks
index faafd9a..920ed71 100644
--- a/tests/00_syntax/04_statement_blocks
+++ b/tests/00_syntax/04_statement_blocks
@@ -8,11 +8,13 @@ Alternatively print it: {%
 -- End --
 
 -- Testcase --
-The result of 3 * 7 is {% print(3 * 7) %}.
-A statement block may contain multiple statements: {%
+The result of 3 * 7 is {%+ print(3 * 7) %}.
+A statement block may contain multiple statements: {%+
 	print("Hello ");
 	print("World");
 %}
+
 To escape the start tag, output it as string expression: {{ "{%" }}
-Alternatively print it: {% print("{%") %}
+Alternatively print it: {%+ print("{%") %}
+
 -- End --
diff --git a/tests/00_syntax/07_embedded_single_line_comments b/tests/00_syntax/07_embedded_single_line_comments
index 8c5fd4a..43f188c 100644
--- a/tests/00_syntax/07_embedded_single_line_comments
+++ b/tests/00_syntax/07_embedded_single_line_comments
@@ -11,10 +11,11 @@ Statement blocks may use C++ comments too: Test Another test.
 The result of 5 + 9 is {{ // The block end tag is ignored: }}
 // And the expression block continues here!
 5 + 9 }}.
-Statement blocks may use C++ comments too: {%
+Statement blocks may use C++ comments too: {%+
 	print("Test"); // A comment.
 
 	// Another comment.
 	print(" Another test.");
 %}
+
 -- End --
diff --git a/tests/00_syntax/08_embedded_multi_line_comments b/tests/00_syntax/08_embedded_multi_line_comments
index b27554f..75aba5f 100644
--- a/tests/00_syntax/08_embedded_multi_line_comments
+++ b/tests/00_syntax/08_embedded_multi_line_comments
@@ -12,7 +12,7 @@ Statement blocks may use C comments too: Test Another test. The final test.
 
 -- Testcase --
 The result of 12 - 4 is {{ /* A comment before */ 12 - /* or even within */ 4 /* or after an expression */ }}.
-Statement blocks may use C comments too: {%
+Statement blocks may use C comments too: {%+
 	print("Test"); /* A comment. */
 
 	/* Another comment. */
@@ -20,4 +20,5 @@ Statement blocks may use C comments too: {%
 
 	print(/* A comment within */ " The final test.");
 %}
+
 -- End --
diff --git a/tests/00_syntax/12_block_whitespace_control b/tests/00_syntax/12_block_whitespace_control
index 23f7380..911171c 100644
--- a/tests/00_syntax/12_block_whitespace_control
+++ b/tests/00_syntax/12_block_whitespace_control
@@ -22,7 +22,7 @@ expected too.This is after the block.
 -- Testcase --
 Whitespace control applies to all block types:
 Comment before: | {#- test #} |, after: | {#- test #} |, both: | {#- test -#} |
-Statement before: | {%- print("test") %} |, after: | {% print("test") -%} |, both: | {%- print("test") -%} |
+Statement before: | {%- print("test") %} |, after: | {%+ print("test") -%} |, both: | {%- print("test") -%} |
 Expression before: | {{- "test" }} |, after: | {{ "test" -}} |, both: | {{- "test" -}} |
 
 By default whitespace {{ "around a block" }} is retained.
@@ -37,6 +37,7 @@ Stripping works across multiple lines as well:
     print("test")
 %}
 
+
 Likewise, stripping over multiple lines of trailing whitespace works as
 expected too.
 
diff --git a/tests/00_syntax/15_function_declarations b/tests/00_syntax/15_function_declarations
index 846f379..cb391a4 100644
--- a/tests/00_syntax/15_function_declarations
+++ b/tests/00_syntax/15_function_declarations
@@ -17,9 +17,9 @@ function() { ... }
 function test_fn(a, b) { ... }
 function test2_fn(a, b) { ... }
 
-
 A function declaration using the alternative syntax:
 The function was called with arguments 123 and 456.
+
 -- End --
 
 -- Testcase --
@@ -50,10 +50,8 @@ The function was called with arguments 123 and 456.
 %}
 
 A function declaration using the alternative syntax:
-
-{%- function test3_fn(a, b): %}
+{% function test3_fn(a, b): %}
 The function was called with arguments {{ a }} and {{ b }}.
-{%- endfunction -%}
-
+{% endfunction %}
 {{ test3_fn(123, 456) }}
 -- End --
diff --git a/tests/00_syntax/16_for_loop b/tests/00_syntax/16_for_loop
index 61e86a1..4c233bc 100644
--- a/tests/00_syntax/16_for_loop
+++ b/tests/00_syntax/16_for_loop
@@ -20,7 +20,6 @@ Iteration 7
 Iteration 8
 Iteration 9
 
-
 If the loop body consists of only one statement, the curly braces
 may be omitted:
 Iteration 0
@@ -34,7 +33,6 @@ Iteration 7
 Iteration 8
 Iteration 9
 
-
 Any of the init-, test- and increment expressions may be omitted.
 
 Loop without initialization statement:
@@ -42,30 +40,25 @@ Iteration null
 Iteration 1
 Iteration 2
 
-
 Loop without test statement:
 Iteration 0
 Iteration 1
 Iteration 2
 
-
 Loop without init-, test- or increment statement:
 Iteration 1
 Iteration 2
 Iteration 3
 
-
 For-in loop enumerating object properties:
 Key: foo
 Key: bar
 
-
 For-in loop enumerating array items:
 Item: one
 Item: two
 Item: three
 
-
 A counting for-loop using the alternative syntax:
 Iteration 0
 Iteration 1
@@ -78,12 +71,10 @@ Iteration 7
 Iteration 8
 Iteration 9
 
-
 A for-in loop using the alternative syntax:
 Item 123
 Item 456
 Item 789
-
 -- End --
 
 -- Testcase --
diff --git a/tests/00_syntax/17_while_loop b/tests/00_syntax/17_while_loop
index 6d6dc4a..4dfaccc 100644
--- a/tests/00_syntax/17_while_loop
+++ b/tests/00_syntax/17_while_loop
@@ -18,7 +18,6 @@ Iteration 7
 Iteration 8
 Iteration 9
 
-
 If the loop body consists of only one statement, the curly braces
 may be omitted:
 Iteration 0
@@ -32,7 +31,6 @@ Iteration 7
 Iteration 8
 Iteration 9
 
-
 A counting while-loop using the alternative syntax:
 Iteration null
 Iteration 1
@@ -44,7 +42,6 @@ Iteration 6
 Iteration 7
 Iteration 8
 Iteration 9
-
 -- End --
 
 -- Testcase --
diff --git a/tests/00_syntax/18_if_condition b/tests/00_syntax/18_if_condition
index 0148466..9601f11 100644
--- a/tests/00_syntax/18_if_condition
+++ b/tests/00_syntax/18_if_condition
@@ -40,6 +40,7 @@ This should print "one":
 	}
 %}
 
+
 This should print "two":
 {%
 	x = 1;
@@ -52,6 +53,7 @@ This should print "two":
 	}
 %}
 
+
 Multiple conditions can be used by chaining if/else statements:
 {%
 	x = 2;
@@ -70,6 +72,7 @@ Multiple conditions can be used by chaining if/else statements:
 	}
 %}
 
+
 If the conditional block consists of only one statement, the curly
 braces may be omitted:
 {%
@@ -81,6 +84,7 @@ braces may be omitted:
 		print("two");
 %}
 
+
 An if-condition using the alternative syntax:
 {% if (x == 1): -%}
 Variable x was set to one.
@@ -88,6 +92,7 @@ Variable x was set to one.
 Variable x has another value.
 {% endif %}
 
+
 Ternary expressions function similar to if/else statements but
 only allow for a single expression in the true and false branches:
 {%
@@ -96,4 +101,5 @@ only allow for a single expression in the true and false branches:
 
 	print(s);
 %}
+
 -- End --
diff --git a/tests/01_arithmetic/04_inc_dec b/tests/01_arithmetic/04_inc_dec
index 6647ced..5a048fe 100644
--- a/tests/01_arithmetic/04_inc_dec
+++ b/tests/01_arithmetic/04_inc_dec
@@ -28,7 +28,6 @@ Incrementing a non-numeric value will convert it to a number:
 3.5
 2
 NaN
-
 -- End --
 
 -- Testcase --