.gitignore

@@ -54,3 +54,4 @@ dkms.conf
 shortcodes.c
 shortcodes
 tests
+.cgreen*

Makefile

@@ -1,16 +1,13 @@
-CC=tcc
-all: tests
-run: shortcodes
-	./shortcodes
+CC=gcc
+all: test
 shortcodes.c: shortcodes.rl
 	ragel -G2 shortcodes.rl -o shortcodes.c
-shortcodes: shortcodes.c
-	$(CC) shortcodes.c -g -o shortcodes
-tests: shortcodes.c shortcodes.h tests.c
-	$(CC) tests.c shortcodes.c -lbg -lcgreen -g -o tests
-test: tests
-	./tests
+tests.so: shortcodes.c tests.c
+	$(CC) -fPIC -shared -g -o $@ $^ -lbg -lcgreen
 clean:
-	rm -f shortcodes shortcodes.c
-
-.PHONY: test run
+	rm -f shortcodes.c *.o *.so tests
+test: tests.so
+	cgreen-runner $^
+debug:
+	cgreen-debug tests.so
+.PHONY: test debug

TODO.md

@@ -2,7 +2,6 @@
 
 ## Things That May Get Done
 
-* Produce decent error structures
 * Inform tag type (% or <) in result
 * Handle other kinds of quotes
 * Handle escaping quote characters
@@ -11,5 +10,6 @@
 * Add self-closing shortcodes `{{< foo />}}` 
   (also, how TF do they work?)
 * Add inline shortcodes `{{< time.inline >}}{{ now }}{{< /time.inline >}}`
-
+* Start integrating with Crystal
+* ~~Produce decent error structures~~
 * ~~Detect mismatched start/end like {{% foo >}}~~ 

shortcodes.h

@@ -1,12 +1,23 @@
-#ifndef SHORTCODES_H
-#define SHORTCODES_H
+#pragma once
+
+// A chunk is a reference to a piece of string
+// from "start" relative to its start
+// and goes on for len characters.
 struct chunk
 {
-    int start, len;
+    unsigned int start, len;
 };
-
 typedef struct chunk chunk;
 
+// An error
+struct sc_error
+{
+    unsigned int position;
+    unsigned int code;
+};
+typedef struct sc_error sc_error;
+
+// Describes a parsed shortcode
 struct shortcode
 {
     chunk whole;
@@ -15,9 +26,34 @@ struct shortcode
     char matching;
     chunk argnames[100];
     chunk argvals[100];
-    int argcount;
+    unsigned int argcount;
 };
 typedef struct shortcode shortcode;
 
-shortcode *parse(char *);
-#endif
+struct sc_result
+{
+    shortcode sc[100];
+    unsigned int sccount;
+    sc_error errors[10];
+    unsigned int errcount;
+};
+typedef struct sc_result sc_result;
+
+// Error codes
+
+/* You are closing the wrong shortcode.
+Example:
+
+{{% foo %}}  {{% /bar %}}
+*/
+#define ERR_MISMATCHED_CLOSING_TAG 1
+
+/* You are using mismatched brackets.
+Example:
+
+{{% foo >}}
+*/
+#define ERR_MISMATCHED_BRACKET 2
+
+
+sc_result parse(char *);

shortcodes.rl

@@ -1,7 +1,6 @@
 #include <string.h>
 #include "shortcodes.h"
 
-
 %%{
   machine shortcode;
 
@@ -59,6 +58,9 @@
     // Since it's mismatched, remove the name
     sc_list[c_sc].name.start = 0;
     sc_list[c_sc].name.len=0;
+    result.errors[result.errcount].position = p-start-2;
+    result.errors[result.errcount].code = ERR_MISMATCHED_BRACKET;
+    result.errcount++;
   };
 
   shortcode = ((start_p content end_p) | (start_b content end_b))
@@ -85,27 +87,30 @@
 
   matched_shortcode = (shortcode any* closing_shortcode)
   @ {
-    sc_list[c_sc-1].matching = 1;
-    sc_list[c_sc-1].whole.len = p-start-sc_list[c_sc-1].whole.start + 1;
     if (
         sc_list[c_sc-1].name.len != sc_list[c_sc].name.len ||
         strncmp(
           start + sc_list[c_sc-1].name.start,
           start + sc_list[c_sc].name.start,
           sc_list[c_sc-1].name.len) !=0) 
-     {
-      return NULL;
+    {
+      result.errors[result.errcount].position = 
+          sc_list[c_sc].whole.start;
+      result.errors[result.errcount].code = ERR_MISMATCHED_CLOSING_TAG;
+      result.errcount++;
+    } else {
+      // The previous shortcode is matching (mark it)
+      sc_list[c_sc-1].matching = 1;
+      sc_list[c_sc-1].whole.len = p-start-sc_list[c_sc-1].whole.start + 1;
     }
-    // Reuse this shortcode entry for next one
-    sc_list[c_sc].name.start = 0;
-    sc_list[c_sc].name.len=0;
+    // Do NOT increase c_sc
 };
 
   main := (any* (shortcode | matched_shortcode | mismatched))*;
 }%%
 
 
-shortcode *parse(char *input) {
+sc_result parse(char *input) {
 
   %%write data;
   char *eof, *ts, *te = 0;
@@ -115,24 +120,16 @@ shortcode *parse(char *input) {
   char *p = input;
   char *pe = p + strlen(input);
 
-  shortcode sc_list[1000]; 
+  sc_result result;
+  shortcode *sc_list = result.sc; 
   int c_sc = 0;
 
-  int sc_start = 0;
-  int sc_end = 0;
-
   char *mark = p;
   char *data_mark = p;
 
   %% write init;
   %% write exec;
 
-  return sc_list;
+  result.sccount = c_sc;
+  return result;
 }
-
-// int main(int argc, char **argv) {
-//     parse(
-// "bbb{{% sarasa sar1 sar2 \"sar3\" %}}ccc"
-// "{{< c1  arg2  >}}foobar{{% /c1%}}aaa{{% sarasa name=\"pepe\" %}}");
-//     return 0;
-// };

tests.c

@@ -6,10 +6,11 @@ Describe(parse);
 BeforeEach(parse) {}
 AfterEach(parse) {}
 
-shortcode *result;
+sc_result result;
 str s;
 
-chunk_s(char *buffer, chunk c){
+void chunk_s(char *buffer, chunk c)
+{
     str_copyb(&s, buffer + c.start, c.len);
 }
 
@@ -18,7 +19,8 @@ Ensure(parse, empty_string)
     char *input = "";
     result = parse(input);
     // This means no shortcodes in it
-    assert_that(result[0].name.len, is_equal_to(0));
+    assert_that(result.sccount, is_equal_to(0));
+    assert_that(result.errcount, is_equal_to(0));
 }
 
 Ensure(parse, simple_shortcode)
@@ -26,24 +28,41 @@ Ensure(parse, simple_shortcode)
     char *input = "foobar {{% shortcode %}}blah";
     result = parse(input);
     // Only 1 shortcode
-    assert_that(result[1].name.len, is_equal_to(0));
+    assert_that(result.sccount, is_equal_to(1));
 
     // It's a simple one called shortcode, no args
-    chunk_s(input, result[0].name);
+    chunk_s(input, result.sc[0].name);
     assert_that(s.s, is_equal_to_string("shortcode"));
-    assert_that(result[0].matching, is_equal_to(0));
-    assert_that(result[0].argcount, is_equal_to(0));
+    assert_that(result.sc[0].matching, is_equal_to(0));
+    assert_that(result.sc[0].argcount, is_equal_to(0));
     // The whole shortcode is the whole thing
-    chunk_s(input, result[0].whole);
+    chunk_s(input, result.sc[0].whole);
     assert_that(s.s, is_equal_to_string("{{% shortcode %}}"));
+    assert_that(result.errcount, is_equal_to(0));
+}
+
+Ensure(parse, mismatched_tags)
+{
+    char *input = "foobar {{% shortcode %}}blah{{% /foo %}}";
+    result = parse(input);
+    // One shortcode, one error
+    assert_that(result.sccount, is_equal_to(1));
+    assert_that(result.errcount, is_equal_to(1));
+    assert_that(result.errors[0].code, is_equal_to(ERR_MISMATCHED_CLOSING_TAG));
+    str_copyb(&s, input + result.errors[0].position, 8);
+    assert_that(s.s, is_equal_to_string("{{% /foo"));
 }
 
 Ensure(parse, mismatched_brackets)
 {
     char *input = "foobar {{% shortcode >}}blah";
     result = parse(input);
-    // No shortcodes
-    assert_that(result[0].name.len, is_equal_to(0));
+    // No shortcodes, 1 error
+    assert_that(result.sccount, is_equal_to(0));
+    assert_that(result.errcount, is_equal_to(1));
+    assert_that(result.errors[0].code, is_equal_to(ERR_MISMATCHED_BRACKET));
+    str_copyb(&s, input + result.errors[0].position, 3);
+    assert_that(s.s, is_equal_to_string(">}}"));
 }
 
 Ensure(parse, mismatched_brackets_inside_data_are_ok)
@@ -51,11 +70,12 @@ Ensure(parse, mismatched_brackets_inside_data_are_ok)
     char *input = "foobar {{% sc %}} >}}blah {{% /sc %}} ";
     result = parse(input);
     // 1 shortcode
-    assert_that(result[1].name.len, is_equal_to(0));
-    chunk_s(input, result[0].whole);
+    assert_that(result.sccount, is_equal_to(1));
+    chunk_s(input, result.sc[0].whole);
     assert_that(s.s, is_equal_to_string("{{% sc %}} >}}blah {{% /sc %}}"));
-    chunk_s(input, result[0].data);
+    chunk_s(input, result.sc[0].data);
     assert_that(s.s, is_equal_to_string(" >}}blah "));
+    assert_that(result.errcount, is_equal_to(0));
 }
 
 Ensure(parse, mismatched_brackets_in_qval_are_ok)
@@ -63,11 +83,12 @@ Ensure(parse, mismatched_brackets_in_qval_are_ok)
     char *input = "foobar {{% sc  \">}}blah\" %}} {{% /sc %}}";
     result = parse(input);
     // 1 shortcode
-    assert_that(result[1].name.len, is_equal_to(0));
-    chunk_s(input, result[0].whole);
+    assert_that(result.sccount, is_equal_to(1));
+    chunk_s(input, result.sc[0].whole);
     assert_that(s.s, is_equal_to_string("{{% sc  \">}}blah\" %}} {{% /sc %}}"));
-    chunk_s(input, result[0].argvals[0]);
+    chunk_s(input, result.sc[0].argvals[0]);
     assert_that(s.s, is_equal_to_string(">}}blah"));
+    assert_that(result.errcount, is_equal_to(0));
 }
 
 Ensure(parse, inner_spaces_optional)
@@ -75,16 +96,17 @@ Ensure(parse, inner_spaces_optional)
     char *input = "foobar {{%    shortcode%}}blah";
     result = parse(input);
     // Only 1 shortcode
-    assert_that(result[1].name.len, is_equal_to(0));
+    assert_that(result.sccount, is_equal_to(1));
 
     // It's a simple one called shortcode, no args
-    chunk_s(input, result[0].name);
+    chunk_s(input, result.sc[0].name);
     assert_that(s.s, is_equal_to_string("shortcode"));
-    assert_that(result[0].matching, is_equal_to(0));
-    assert_that(result[0].argcount, is_equal_to(0));
+    assert_that(result.sc[0].matching, is_equal_to(0));
+    assert_that(result.sc[0].argcount, is_equal_to(0));
     // The whole shortcode is the whole thing
-    chunk_s(input, result[0].whole);
+    chunk_s(input, result.sc[0].whole);
     assert_that(s.s, is_equal_to_string("{{%    shortcode%}}"));
+    assert_that(result.errcount, is_equal_to(0));
 }
 
 Ensure(parse, name_can_be_path)
@@ -92,13 +114,14 @@ Ensure(parse, name_can_be_path)
     char *input = "foobar {{% shortcode/foo/bar %}}blah";
     result = parse(input);
     // Only 1 shortcode
-    assert_that(result[1].name.len, is_equal_to(0));
+    assert_that(result.sccount, is_equal_to(1));
 
     // It's a simple one called shortcode, no args
-    chunk_s(input, result[0].name);
+    chunk_s(input, result.sc[0].name);
     assert_that(s.s, is_equal_to_string("shortcode/foo/bar"));
-    assert_that(result[0].matching, is_equal_to(0));
-    assert_that(result[0].argcount, is_equal_to(0));
+    assert_that(result.sc[0].matching, is_equal_to(0));
+    assert_that(result.sc[0].argcount, is_equal_to(0));
+    assert_that(result.errcount, is_equal_to(0));
 }
 
 Ensure(parse, multiple_shortcodes)
@@ -106,24 +129,25 @@ Ensure(parse, multiple_shortcodes)
     char *input = "foobar {{% shortcode %}}blah {{<sc2 >}}blahblah";
     result = parse(input);
     // 2 shortcodes
-    assert_that(result[2].name.len, is_equal_to(0));
+    assert_that(result.sccount, is_equal_to(2));
 
     // It's a simple one called shortcode, no args
-    chunk_s(input, result[0].name);
+    chunk_s(input, result.sc[0].name);
     assert_that(s.s, is_equal_to_string("shortcode"));
-    assert_that(result[0].matching, is_equal_to(0));
-    assert_that(result[0].argcount, is_equal_to(0));
+    assert_that(result.sc[0].matching, is_equal_to(0));
+    assert_that(result.sc[0].argcount, is_equal_to(0));
     // The whole shortcode is the whole thing
-    chunk_s(input, result[0].whole);
+    chunk_s(input, result.sc[0].whole);
     assert_that(s.s, is_equal_to_string("{{% shortcode %}}"));
     // It's a simple one called sc2, no args
-    chunk_s(input, result[1].name);
+    chunk_s(input, result.sc[1].name);
     assert_that(s.s, is_equal_to_string("sc2"));
-    assert_that(result[1].matching, is_equal_to(0));
-    assert_that(result[1].argcount, is_equal_to(0));
+    assert_that(result.sc[1].matching, is_equal_to(0));
+    assert_that(result.sc[1].argcount, is_equal_to(0));
     // The whole shortcode is the whole thing
-    chunk_s(input, result[1].whole);
+    chunk_s(input, result.sc[1].whole);
     assert_that(s.s, is_equal_to_string("{{<sc2 >}}"));
+    assert_that(result.errcount, is_equal_to(0));
 }
 
 Ensure(parse, matching_shortcode)
@@ -132,19 +156,20 @@ Ensure(parse, matching_shortcode)
     result = parse(input);
 
     // Only 1 shortcode
-    assert_that(result[1].name.len, is_equal_to(0));
+    assert_that(result.sccount, is_equal_to(1));
 
     // It's a matching one called shortcode, no args
-    chunk_s(input, result[0].name);
+    chunk_s(input, result.sc[0].name);
     assert_that(s.s, is_equal_to_string("shortcode"));
-    assert_that(result[0].matching, is_equal_to(1));
-    assert_that(result[0].argcount, is_equal_to(0));
+    assert_that(result.sc[0].matching, is_equal_to(1));
+    assert_that(result.sc[0].argcount, is_equal_to(0));
     // data is the stuff between the shortcode tags
-    chunk_s(input, result[0].data);
+    chunk_s(input, result.sc[0].data);
     assert_that(s.s, is_equal_to_string("foo bar"));
     // The whole shortcode is the whole thing
-    chunk_s(input, result[0].whole);
+    chunk_s(input, result.sc[0].whole);
     assert_that(s.s, is_equal_to_string("{{% shortcode %}}foo bar{{% /shortcode %}}"));
+    assert_that(result.errcount, is_equal_to(0));
 }
 
 Ensure(parse, shortcode_args)
@@ -152,69 +177,50 @@ Ensure(parse, shortcode_args)
     char *input = "foobar {{% shortcode foo \"bar\" 42 bat=v1 baz=\"v2\" %}}blah";
     result = parse(input);
     // Only 1 shortcode
-    assert_that(result[1].name.len, is_equal_to(0));
+    assert_that(result.sccount, is_equal_to(1));
 
     // The whole shortcode is the whole thing
-    chunk_s(input, result[0].whole);
+    chunk_s(input, result.sc[0].whole);
     assert_that(s.s, is_equal_to_string("{{% shortcode foo \"bar\" 42 bat=v1 baz=\"v2\" %}}"));
 
     // Name is shortcode
-    chunk_s(input, result[0].name);
+    chunk_s(input, result.sc[0].name);
     assert_that(s.s, is_equal_to_string("shortcode"));
-    assert_that(result[0].matching, is_equal_to(0));
+    assert_that(result.sc[0].matching, is_equal_to(0));
 
     // Has 5 args
-    assert_that(result[0].argcount, is_equal_to(5));
+    assert_that(result.sc[0].argcount, is_equal_to(5));
 
     // Arg1 is foo, no name
-    assert_that(result[0].argnames[0].len, is_equal_to(0));
-    chunk_s(input, result[0].argvals[0]);
+    assert_that(result.sc[0].argnames[0].len, is_equal_to(0));
+    chunk_s(input, result.sc[0].argvals[0]);
     assert_that(s.s, is_equal_to_string("foo"));
     // Arg2 is bar, no name
-    assert_that(result[0].argnames[1].len, is_equal_to(0));
-    chunk_s(input, result[0].argvals[1]);
+    assert_that(result.sc[0].argnames[1].len, is_equal_to(0));
+    chunk_s(input, result.sc[0].argvals[1]);
     assert_that(s.s, is_equal_to_string("bar"));
     // Arg3 is 42, no name
-    assert_that(result[0].argnames[2].len, is_equal_to(0));
-    chunk_s(input, result[0].argvals[2]);
+    assert_that(result.sc[0].argnames[2].len, is_equal_to(0));
+    chunk_s(input, result.sc[0].argvals[2]);
     assert_that(s.s, is_equal_to_string("42"));
     // Arg4 is bat=v1
-    chunk_s(input, result[0].argnames[3]);
+    chunk_s(input, result.sc[0].argnames[3]);
     assert_that(s.s, is_equal_to_string("bat"));
-    chunk_s(input, result[0].argvals[3]);
+    chunk_s(input, result.sc[0].argvals[3]);
     assert_that(s.s, is_equal_to_string("v1"));
     // Arg5 is baz=v2
-    chunk_s(input, result[0].argnames[4]);
+    chunk_s(input, result.sc[0].argnames[4]);
     assert_that(s.s, is_equal_to_string("baz"));
-    chunk_s(input, result[0].argvals[4]);
+    chunk_s(input, result.sc[0].argvals[4]);
     assert_that(s.s, is_equal_to_string("v2"));
+    assert_that(result.errcount, is_equal_to(0));
 }
 
-Ensure(parse, escaped_shortcode)
-{
-    char *input = "foobar \\{{% shortcode %}}";
-    result = parse(input);
-    // No shortcodes
-    assert_that(result[0].name.len, is_equal_to(0));
-}
-
-
-int main(int argc, char **argv)
-{
-    str_init(&s);
-    TestSuite *suite = create_test_suite();
-    add_test_with_context(suite, parse, empty_string);
-    add_test_with_context(suite, parse, simple_shortcode);
-    add_test_with_context(suite, parse, mismatched_brackets);
-    add_test_with_context(suite, parse, mismatched_brackets_inside_data_are_ok);
-    add_test_with_context(suite, parse, mismatched_brackets_in_qval_are_ok);
-    add_test_with_context(suite, parse, name_can_be_path);
-    add_test_with_context(suite, parse, inner_spaces_optional);
-    add_test_with_context(suite, parse, multiple_shortcodes);
-    add_test_with_context(suite, parse, matching_shortcode);
-    add_test_with_context(suite, parse, shortcode_args);
-
-    // Bugs
-    // add_test_with_context(suite, parse, escaped_shortcode);
-    return run_test_suite(suite, create_text_reporter());
-}
+// BUG?
+// Ensure(parse, escaped_shortcode)
+// {
+//     char *input = "foobar \\{{% shortcode %}}";
+//     result = parse(input);
+//     // No shortcodes
+//     assert_that(result.sc[0].name.len, is_equal_to(0));
+// }