replaced leg-based parser with handmade parser for source files

12 files changed, 283 insertions, 140 deletions

diff --git a/.gitignore b/.gitignore
index b7c5f8e..92256ca 100644
--- a/.gitignore
+++ b/.gitignore
@@ -40,12 +40,11 @@ Makefile.in
 /blogc
 
 # tests
-/tests/check_source_grammar
+/tests/check_source_parser
 /tests/check_template_grammar
 /tests/check_utils
 
 # leg generated source
-/src/source-grammar.c
 /src/template-grammar.c
 
 # tarballs
diff --git a/Makefile.am b/Makefile.am
index b1548bc..3138d25 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -14,7 +14,6 @@ AM_DISTCHECK_CONFIGURE_FLAGS = \
 EXTRA_DIST = \
 	autogen.sh \
 	README.md \
-	src/source-grammar.leg \
 	src/template-grammar.leg \
 	$(NULL)
 
@@ -22,7 +21,8 @@ CLEANFILES = \
 	$(NULL)
 
 noinst_HEADERS = \
-	src/source-grammar.h \
+	src/output.h \
+	src/source-parser.h \
 	src/template-grammar.h \
 	src/utils/utils.h \
 	$(NULL)
@@ -43,7 +43,8 @@ check_PROGRAMS = \
 
 
 libblogc_la_SOURCES = \
-	src/source-grammar.c \
+	src/output.c \
+	src/source-parser.c \
 	src/template-grammar.c \
 	src/utils/slist.c \
 	src/utils/strings.c \
@@ -92,24 +93,24 @@ endif
 if USE_CMOCKA
 
 check_PROGRAMS += \
-	tests/check_source_grammar \
+	tests/check_source_parser \
 	tests/check_template_grammar \
 	tests/check_utils \
 	$(NULL)
 
-tests_check_source_grammar_SOURCES = \
-	tests/check_source_grammar.c \
+tests_check_source_parser_SOURCES = \
+	tests/check_source_parser.c \
 	$(NULL)
 
-tests_check_source_grammar_CFLAGS = \
+tests_check_source_parser_CFLAGS = \
 	$(CMOCKA_CFLAGS) \
 	$(NULL)
 
-tests_check_source_grammar_LDFLAGS = \
+tests_check_source_parser_LDFLAGS = \
 	-no-install \
 	$(NULL)
 
-tests_check_source_grammar_LDADD = \
+tests_check_source_parser_LDADD = \
 	$(CMOCKA_LIBS) \
 	libblogc.la \
 	$(NULL)
diff --git a/src/main.c b/src/main.c
index 3dbf325..d05c8d6 100644
--- a/src/main.c
+++ b/src/main.c
@@ -12,17 +12,19 @@
 
 #include <stdio.h>
 
-#include "source-grammar.h"
+#include "source-parser.h"
+#include <string.h>
 
 
 int
 main(int argc, char **argv)
 {
-    blogc_source_t *t = blogc_source_parse(
-        "\n  \nBOLA: guda\n\t\n\n\n\n"
+    const char *a =
+        "\n  \nBOLA : guda\n\t\n\n\n\n"
         "CHUNDA: asd\n"
         "----\n"
-        "{% block single_source %}\nbola\n\nzas\n");
+        "{% block single_source %}\nbola\n\nzas\n";
+    blogc_source_t *t = blogc_source_parse(a, strlen(a));
     printf("%s\n", t->content);
     printf("Hello, World!\n");
     return 0;
diff --git a/src/output.c b/src/output.c
new file mode 100644
index 0000000..bd96b8e
--- /dev/null
+++ b/src/output.c
@@ -0,0 +1,39 @@
+/*
+ * blogc: A blog compiler.
+ * Copyright (C) 2015 Rafael G. Martins <rafael@rafaelmartins.eng.br>
+ *
+ * This program can be distributed under the terms of the BSD License.
+ * See the file COPYING.
+ */
+
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif /* HAVE_CONFIG_H */
+
+#include <stdio.h>
+#include "utils/utils.h"
+#include "output.h"
+
+
+void
+blogc_parser_syntax_error(const char *name, const char *src, size_t src_len,
+    size_t current)
+{
+    b_string_t *msg = b_string_new();
+
+    while (current < src_len) {
+        char c = src[current];
+
+        if (c == '\r' || c == '\n')
+            break;
+
+        b_string_append_c(msg, c);
+
+        current++;
+    }
+
+    fprintf(stderr, "%s parser error: syntax error near \"%s\"\n", name,
+        msg->str);
+
+    b_string_free(msg, true);
+}
diff --git a/src/output.h b/src/output.h
new file mode 100644
index 0000000..b2f3cf3
--- /dev/null
+++ b/src/output.h
@@ -0,0 +1,17 @@
+/*
+ * blogc: A blog compiler.
+ * Copyright (C) 2015 Rafael G. Martins <rafael@rafaelmartins.eng.br>
+ *
+ * This program can be distributed under the terms of the LGPL-2 License.
+ * See the file COPYING.
+ */
+
+#ifndef _OUTPUT_H
+#define _OUTPUT_H
+
+#include <stdlib.h>
+
+void blogc_parser_syntax_error(const char *name, const char *src,
+    size_t src_len, size_t current);
+
+#endif /* _OUTPUT_H */
diff --git a/src/source-grammar.leg b/src/source-grammar.leg
deleted file mode 100644
index faab9a4..0000000
--- a/src/source-grammar.leg
+++ /dev/null
@@ -1,116 +0,0 @@
-#
-# blogc: A balde compiler.
-# Copyright (C) 2015 Rafael G. Martins <rafael@rafaelmartins.eng.br>
-#
-# This program can be distributed under the terms of the BSD License.
-# See the file COPYING.
-#
-
-%{
-
-#include <stdio.h>
-#include "utils/utils.h"
-#include "source-grammar.h"
-
-#define YY_INPUT(buf, result, max_size)                          \
-{                                                                \
-    int yyc = (charbuf && *charbuf != '\0') ? *charbuf++ : EOF;  \
-    result = (EOF == yyc) ? 0 : (*buf = yyc, 1);                 \
-}
-
-
-static b_trie_t *config = NULL;
-static const char *charbuf = NULL;
-static char *key = NULL;
-static char *content = NULL;
-
-
-static void
-blogc_source_config_key(const char *value)
-{
-    if (key != NULL) {
-        fprintf(stderr, "Syntax error: configuration key already set: %s\n", key);
-        exit(1);
-    }
-    key = b_strdup(value);
-}
-
-
-static void
-blogc_source_config_value(const char *value)
-{
-    if (key == NULL) {
-        fprintf(stderr, "Syntax error: configuration value without a key: %s\n", value);
-        exit(1);
-    }
-    b_trie_insert(config, key, b_str_strip(b_strdup(value)));
-    free(key);
-    key = NULL;
-}
-
-
-static void
-blogc_source_content(const char *value)
-{
-    if (content != NULL) {
-        fprintf(stderr, "Syntax error: content set twice\n");
-        exit(1);
-    }
-    content = b_strdup(value);
-}
-
-%}
-
-page = ( ( - eol )* ( config ( - eol )* )+ '----' '-'* eol content eof ) | anything
-    { fprintf(stderr, "Syntax error near: %s\n", yytext); exit(1); }
-
-# Useful rules
-eol = '\n' | '\r\n' | '\r'
-eof = !.
-- = [\t ]*
-id = [A-Z][A-Z0-9_]*
-anything = < ( !eol . )* > eol
-
-# Configuration
-config_key = < id >                     { blogc_source_config_key(yytext); }
-config_value = < anything >             { blogc_source_config_value(yytext); }
-config = config_key ':' ' '+ config_value
-
-# Generic content
-content = < ( !eof . )+ >               { blogc_source_content(yytext); }
-
-%%
-
-
-void
-blogc_source_free(blogc_source_t *source)
-{
-    if (source == NULL)
-        return;
-    free(source->content);
-    b_trie_free(source->config);
-    free(source);
-}
-
-
-static void
-blogc_source_config_free(void *ptr)
-{
-    free(ptr);
-}
-
-
-blogc_source_t*
-blogc_source_parse(const char *tmpl)
-{
-    charbuf = tmpl;
-    config = b_trie_new(blogc_source_config_free);
-    key = NULL;
-    content = NULL;
-    while(yyparse());
-    blogc_source_t *rv = malloc(sizeof(blogc_source_t));
-    rv->content = content;
-    rv->config = config;
-    charbuf = NULL;
-    return rv;
-}
diff --git a/src/source-parser.c b/src/source-parser.c
new file mode 100644
index 0000000..ad7e77d
--- /dev/null
+++ b/src/source-parser.c
@@ -0,0 +1,147 @@
+/*
+ * blogc: A blog compiler.
+ * Copyright (C) 2015 Rafael G. Martins <rafael@rafaelmartins.eng.br>
+ *
+ * This program can be distributed under the terms of the BSD License.
+ * See the file COPYING.
+ */
+
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif /* HAVE_CONFIG_H */
+
+#include <stdbool.h>
+
+#include "utils/utils.h"
+#include "source-parser.h"
+#include "output.h"
+
+
+typedef enum {
+    SOURCE_START = 1,
+    SOURCE_CONFIG_KEY,
+    SOURCE_CONFIG_VALUE_START,
+    SOURCE_CONFIG_VALUE,
+    SOURCE_SEPARATOR,
+    SOURCE_CONTENT_START,
+    SOURCE_CONTENT,
+} blogc_source_parser_state_t;
+
+
+blogc_source_t*
+blogc_source_parse(const char *src, size_t src_len)
+{
+    size_t current = 0;
+    size_t start = 0;
+
+    bool error = false;
+    char *key = NULL;
+    char *tmp = NULL;
+    b_trie_t *config = b_trie_new(free);
+    char *content = NULL;
+
+    blogc_source_parser_state_t state = SOURCE_START;
+
+    while (current < src_len) {
+        char c = src[current];
+
+        switch (state) {
+
+            case SOURCE_START:
+                if (c == ' ' || c == '\t' || c == '\n' || c == '\r')
+                    break;
+                if (c >= 'A' && c <= 'Z') {
+                    state = SOURCE_CONFIG_KEY;
+                    start = current;
+                    break;
+                }
+                if (c == '-') {
+                    state = SOURCE_SEPARATOR;
+                    break;
+                }
+                error = true;
+                break;
+
+            case SOURCE_CONFIG_KEY:
+                if ((c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || c == '_')
+                    break;
+                if (c == ':') {
+                    key = b_strndup(src + start, current - start);
+                    state = SOURCE_CONFIG_VALUE_START;
+                    break;
+                }
+                error = true;
+                break;
+
+            case SOURCE_CONFIG_VALUE_START:
+                if (c != '\n' && c != '\r') {
+                    state = SOURCE_CONFIG_VALUE;
+                    start = current;
+                    break;
+                }
+                error = true;
+                break;
+
+            case SOURCE_CONFIG_VALUE:
+                if (c == '\n' || c == '\r') {
+                    tmp = b_strndup(src + start, current - start);
+                    b_trie_insert(config, key, b_strdup(b_str_strip(tmp)));
+                    free(tmp);
+                    free(key);
+                    key = NULL;
+                    state = SOURCE_START;
+                }
+                break;
+
+            case SOURCE_SEPARATOR:
+                if (c == '-')
+                    break;
+                if (c == '\n' || c == '\r') {
+                    state = SOURCE_CONTENT_START;
+                    break;
+                }
+                error = true;
+                break;
+
+            case SOURCE_CONTENT_START:
+                start = current;
+                state = SOURCE_CONTENT;
+                break;
+
+            case SOURCE_CONTENT:
+                if (current == (src_len - 1))
+                    content = b_strndup(src + start, src_len - start);
+                break;
+        }
+
+        if (error)
+            break;
+
+        current++;
+    }
+
+    if (error) {
+        free(key);
+        free(content);
+        b_trie_free(config);
+        blogc_parser_syntax_error("source", src, src_len, current);
+        return NULL;
+    }
+
+    blogc_source_t *rv = malloc(sizeof(blogc_source_t));
+    rv->config = config;
+    rv->content = content;
+
+    return rv;
+}
+
+
+void
+blogc_source_free(blogc_source_t *source)
+{
+    if (source == NULL)
+        return;
+    free(source->content);
+    b_trie_free(source->config);
+    free(source);
+}
diff --git a/src/source-grammar.h b/src/source-parser.h
index 540c21b..6f41c14 100644
--- a/src/source-grammar.h
+++ b/src/source-parser.h
@@ -6,9 +6,10 @@
  * See the file COPYING.
  */
 
-#ifndef _SOURCE_GRAMAR_H
-#define _SOURCE_GRAMAR_H
+#ifndef _SOURCE_PARSER_H
+#define _SOURCE_PARSER_H
 
+#include <stdlib.h>
 #include "utils/utils.h"
 
 typedef struct {
@@ -16,7 +17,7 @@ typedef struct {
     char *content;
 } blogc_source_t;
 
-blogc_source_t* blogc_source_parse(const char *tmpl);
+blogc_source_t* blogc_source_parse(const char *src, size_t src_len);
 void blogc_source_free(blogc_source_t *source);
 
-#endif /* _SOURCE_GRAMAR_H */
+#endif /* _SOURCE_PARSER_H */
diff --git a/src/utils/trie.c b/src/utils/trie.c
index f447860..b92573f 100644
--- a/src/utils/trie.c
+++ b/src/utils/trie.c
@@ -38,6 +38,8 @@ b_trie_free_node(b_trie_t *trie, b_trie_node_t *node)
 void
 b_trie_free(b_trie_t *trie)
 {
+    if (trie == NULL)
+        return;
     b_trie_free_node(trie, trie->root);
     free(trie);
 }
diff --git a/src/utils/utils.h b/src/utils/utils.h
index 20259d8..55b9c59 100644
--- a/src/utils/utils.h
+++ b/src/utils/utils.h
@@ -10,6 +10,7 @@
 #define _UTILS_UTILS_H
 
 #include <stdbool.h>
+#include <stdlib.h>
 #include <stdarg.h>
 
 #define B_STRING_CHUNK_SIZE 128
diff --git a/tests/check_source_grammar.c b/tests/check_source_parser.c
index b2f581b..2f5880a 100644
--- a/tests/check_source_grammar.c
+++ b/tests/check_source_parser.c
@@ -14,19 +14,21 @@
 #include <stddef.h>
 #include <setjmp.h>
 #include <cmocka.h>
-#include "../src/source-grammar.h"
+#include <string.h>
+#include "../src/source-parser.h"
 
 
 static void
 test_source_parse(void **state)
 {
-    blogc_source_t *source = blogc_source_parse(
+    const char *a =
         "VAR1: asd asd\n"
         "VAR2: 123chunda\n"
         "----------\n"
         "# This is a test\n"
         "\n"
-        "bola\n");
+        "bola\n";
+    blogc_source_t *source = blogc_source_parse(a, strlen(a));
     assert_non_null(source);
     assert_int_equal(b_trie_size(source->config), 2);
     assert_string_equal(b_trie_lookup(source->config, "VAR1"), "asd asd");
@@ -42,7 +44,7 @@ test_source_parse(void **state)
 static void
 test_source_parse_with_spaces(void **state)
 {
-    blogc_source_t *source = blogc_source_parse(
+    const char *a =
         "\n  \n"
         "VAR1: chunda     \t   \n"
         "\n\n"
@@ -50,7 +52,8 @@ test_source_parse_with_spaces(void **state)
         "----------\n"
         "# This is a test\n"
         "\n"
-        "bola\n");
+        "bola\n";
+    blogc_source_t *source = blogc_source_parse(a, strlen(a));
     assert_non_null(source);
     assert_int_equal(b_trie_size(source->config), 2);
     assert_string_equal(b_trie_lookup(source->config, "VAR1"), "chunda");
diff --git a/tests/check_template_grammar.c b/tests/check_template_grammar.c
index ffad8fb..3b4dcca 100644
--- a/tests/check_template_grammar.c
+++ b/tests/check_template_grammar.c
@@ -88,11 +88,58 @@ test_template_parse(void **state)
 }
 
 
+static void
+test_template_parse_html(void **state)
+{
+    b_slist_t *stmts = blogc_template_parse(
+        "<html>\n"
+        "    <head>\n"
+        "        {% block single_source %}\n"
+        "        <title>My cool blog >> {{ TITLE }}</title>\n"
+        "        {% endblock %}\n"
+        "        {% block multiple_sources %}\n"
+        "        <title>My cool blog - Main page</title>\n"
+        "        {% endblock %}\n"
+        "    </head>\n"
+        "    <body>\n"
+        "        <h1>My cool blog</h1>\n"
+        "        {% block single_source %}\n"
+        "        <h2>{{ TITLE }}</h2>\n"
+        "        {% if DATE %}<h4>Published in: {{ DATE }}</h4>{% endif %}\n"
+        "        <pre>{{ CONTENT }}</pre>\n"
+        "        {% endblock %}\n"
+        "        {% block multiple_sources_once %}<ul>{% endblock %}\n"
+        "        {% block multiple_sources %}<p><a href=\"{{ FILENAME }}.html\">"
+        "{{ TITLE }}</a>{% if DATE %} - {{ DATE }}{% endif %}</p>{% endblock %}\n"
+        "        {% block multiple_sources_once %}</ul>{% endblock %}\n"
+        "    </body>\n"
+        "</html>\n");
+    assert_non_null(stmts);
+    blogc_assert_template_stmt(stmts, "<html>\n    <head>\n        ",
+        BLOGC_TEMPLATE_CONTENT_STMT);
+    blogc_assert_template_stmt(stmts->next, "single_source",
+        BLOGC_TEMPLATE_BLOCK_STMT);
+    blogc_assert_template_stmt(stmts->next->next,
+        "\n        <title>My cool blog >> ", BLOGC_TEMPLATE_CONTENT_STMT);
+    blogc_assert_template_stmt(stmts->next->next->next, "TITLE",
+        BLOGC_TEMPLATE_VARIABLE_STMT);
+    blogc_assert_template_stmt(stmts->next->next->next->next,
+        "</title>\n        ", BLOGC_TEMPLATE_CONTENT_STMT);
+
+
+
+
+
+    blogc_template_free_stmts(stmts);
+}
+
+
 int
 main(void)
 {
     const UnitTest tests[] = {
         unit_test(test_template_parse),
+        unit_test(test_template_parse_html),
     };
     return run_tests(tests);
 }