replaced leg-based parser with handmade parser for templates

yay! no leg parser needed anymore. parsers still needs some work and
error handling, though.

5 files changed, 302 insertions, 213 deletions

diff --git a/src/main.c b/src/main.c
index d05c8d6..4e72b6e 100644
--- a/src/main.c
+++ b/src/main.c
@@ -13,19 +13,13 @@
 #include <stdio.h>
 
 #include "source-parser.h"
+#include "template-parser.h"
 #include <string.h>
 
 
 int
 main(int argc, char **argv)
 {
-    const char *a =
-        "\n  \nBOLA : guda\n\t\n\n\n\n"
-        "CHUNDA: asd\n"
-        "----\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
index bd96b8e..1664afe 100644
--- a/src/output.c
+++ b/src/output.c
@@ -32,7 +32,7 @@ blogc_parser_syntax_error(const char *name, const char *src, size_t src_len,
         current++;
     }
 
-    fprintf(stderr, "%s parser error: syntax error near \"%s\"\n", name,
+    fprintf(stderr, "%s parser error: failed to parse input near \"%s\".\n", name,
         msg->str);
 
     b_string_free(msg, true);
diff --git a/src/template-grammar.leg b/src/template-grammar.leg
deleted file mode 100644
index 2425582..0000000
--- a/src/template-grammar.leg
+++ /dev/null
@@ -1,200 +0,0 @@
-#
-# 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.
-#
-
-%{
-
-#include <stdio.h>
-#include "utils/utils.h"
-#include "template-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_slist_t *stmts = NULL;
-static int if_count = 0;
-static int block_count = 0;
-static const char *charbuf = NULL;
-
-
-static void
-blogc_template_if_stmt(const char *value)
-{
-    if (block_count <= 0) {
-        fprintf(stderr, "Syntax error: {" "%% if ... %%" "} statement before "
-            "any {" "%% block ... %%" "} statement\n");
-        exit(1);
-    }
-    blogc_template_stmt_t *stmt = malloc(sizeof(blogc_template_stmt_t));
-    stmt->value = b_strdup(value);
-    stmt->type = BLOGC_TEMPLATE_IF_STMT;
-    stmts = b_slist_append(stmts, stmt);
-    if_count++;
-}
-
-
-static void
-blogc_template_else_stmt(void)
-{
-    if (if_count <= 0) {
-        fprintf(stderr, "Syntax error: {" "%% else %%" "} statement without "
-            "any open {" "%% if ... %%" "} statement\n");
-        exit(1);
-    }
-    blogc_template_stmt_t *stmt = malloc(sizeof(blogc_template_stmt_t));
-    stmt->value = NULL;
-    stmt->type = BLOGC_TEMPLATE_ELSE_STMT;
-    stmts = b_slist_append(stmts, stmt);
-}
-
-
-static void
-blogc_template_endif_stmt(void)
-{
-    if (if_count-- <= 0) {
-        fprintf(stderr, "Syntax error: {" "%% endif %%" "} statement before "
-            "any {" "%% if ... %%" "} statement\n");
-        exit(1);
-    }
-    blogc_template_stmt_t *stmt = malloc(sizeof(blogc_template_stmt_t));
-    stmt->value = NULL;
-    stmt->type = BLOGC_TEMPLATE_ENDIF_STMT;
-    stmts = b_slist_append(stmts, stmt);
-}
-
-
-static void
-blogc_template_block_stmt(const char *value)
-{
-    if (block_count > 0) {
-        fprintf(stderr, "Syntax error: {" "%% block %%" "} statements "
-            "can't be nested\n");
-        exit(1);
-    }
-    blogc_template_stmt_t *stmt = malloc(sizeof(blogc_template_stmt_t));
-    stmt->value = b_strdup(value);
-    stmt->type = BLOGC_TEMPLATE_BLOCK_STMT;
-    stmts = b_slist_append(stmts, stmt);
-    block_count++;
-}
-
-
-static void
-blogc_template_endblock_stmt(void)
-{
-    if (block_count-- <= 0) {
-        fprintf(stderr, "Syntax error: {" "%% endblock %%" "} statement before "
-            "any {" "%% block ... %%" "} statement\n");
-        exit(1);
-    }
-    blogc_template_stmt_t *stmt = malloc(sizeof(blogc_template_stmt_t));
-    stmt->value = NULL;
-    stmt->type = BLOGC_TEMPLATE_ENDBLOCK_STMT;
-    stmts = b_slist_append(stmts, stmt);
-}
-
-
-static void
-blogc_template_variable_stmt(const char *value)
-{
-    if (block_count <= 0) {
-        fprintf(stderr, "Syntax error: {{ ... }} statement before "
-            "any {" "%% block ... %%" "} statement\n");
-        exit(1);
-    }
-    blogc_template_stmt_t *stmt = malloc(sizeof(blogc_template_stmt_t));
-    stmt->value = b_strdup(value);
-    stmt->type = BLOGC_TEMPLATE_VARIABLE_STMT;
-    stmts = b_slist_append(stmts, stmt);
-}
-
-
-static void
-blogc_template_content_stmt(const char *value)
-{
-    blogc_template_stmt_t *stmt = malloc(sizeof(blogc_template_stmt_t));
-    stmt->value = b_strdup(value);
-    stmt->type = BLOGC_TEMPLATE_CONTENT_STMT;
-    stmts = b_slist_append(stmts, stmt);
-}
-
-%}
-
-page = if | else | endif | block | endblock | print | content | 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
-
-# Conditionals
-if_open = '{%' -
-if_close = - '%}'
-if = if_open 'if' ' '+ < id > if_close  { blogc_template_if_stmt(yytext); }
-else = if_open 'else' if_close          { blogc_template_else_stmt(); }
-endif = if_open 'endif' if_close        { blogc_template_endif_stmt(); }
-
-# Blocks
-block_open = '{%' -
-block_close = - '%}'
-block_name = ( 'single_source' | 'multiple_sources_once' | 'multiple_sources' )
-block = block_open 'block' ' '+ < block_name > block_close  { blogc_template_block_stmt(yytext); }
-endblock = block_open 'endblock' block_close                { blogc_template_endblock_stmt(); }
-
-# Print calls
-print_open = '{{' -
-print_close = - '}}'
-print_var = < id >                      { blogc_template_variable_stmt(yytext); }
-print = print_open print_var print_close
-
-# Generic content
-content = < ( !eof !if_open !if_close !block_open !block_close !print_open !print_close . )+ >
-    { blogc_template_content_stmt(yytext); }
-
-%%
-
-
-void
-blogc_template_free_stmts(b_slist_t *stmts)
-{
-    for (b_slist_t *tmp = stmts; tmp != NULL; tmp = tmp->next) {
-        blogc_template_stmt_t *data = tmp->data;
-        free(data->value);
-        free(data);
-    }
-    b_slist_free(stmts);
-}
-
-
-b_slist_t*
-blogc_template_parse(const char *tmpl)
-{
-    if_count = 0;
-    block_count = 0;
-    charbuf = tmpl;
-    while(yyparse());
-    if (if_count != 0) {
-        fprintf(stderr, "Syntax error: You left %d open {" "%% if ... %%" "} statements.\n", if_count);
-        exit(1);
-    }
-    if (block_count != 0) {
-        fprintf(stderr, "Syntax error: You left %d open {" "%% block ... %%" "} statements.\n", block_count);
-        exit(1);
-    }
-    b_slist_t *rv = stmts;
-    charbuf = NULL;
-    stmts = NULL;
-    return rv;
-}
diff --git a/src/template-parser.c b/src/template-parser.c
new file mode 100644
index 0000000..c5286ea
--- /dev/null
+++ b/src/template-parser.c
@@ -0,0 +1,296 @@
+/*
+ * 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 <string.h>
+
+#include "utils/utils.h"
+#include "template-parser.h"
+#include "output.h"
+
+
+typedef enum {
+    TEMPLATE_START = 1,
+    TEMPLATE_OPEN_BRACKET,
+    TEMPLATE_BLOCK_START,
+    TEMPLATE_BLOCK_TYPE,
+    TEMPLATE_BLOCK_BLOCK_TYPE_START,
+    TEMPLATE_BLOCK_BLOCK_TYPE,
+    TEMPLATE_BLOCK_IF_VARIABLE_START,
+    TEMPLATE_BLOCK_IF_VARIABLE,
+    TEMPLATE_BLOCK_END,
+    TEMPLATE_VARIABLE_START,
+    TEMPLATE_VARIABLE,
+    TEMPLATE_VARIABLE_END,
+    TEMPLATE_CLOSE_BRACKET,
+} blogc_template_parser_state_t;
+
+
+b_slist_t*
+blogc_template_parse(const char *src, size_t src_len)
+{
+    size_t current = 0;
+    size_t start = 0;
+    size_t end = 0;
+    size_t remaining = 0;
+
+    bool error = false;
+    char *tmp = NULL;
+
+    bool open_block = false;
+    unsigned int if_count = 0;
+
+    b_slist_t *stmts = NULL;
+    blogc_template_stmt_t *stmt = NULL;
+
+    blogc_template_parser_state_t state = TEMPLATE_START;
+    blogc_template_stmt_type_t type = BLOGC_TEMPLATE_CONTENT_STMT;
+
+    while (current < src_len) {
+        char c = src[current];
+        bool last = current == src_len - 1;
+
+        switch (state) {
+
+            case TEMPLATE_START:
+                if (last) {
+                    stmt = malloc(sizeof(blogc_template_stmt_t));
+                    stmt->type = type;
+                    stmt->value = b_strndup(src + start, src_len - start);
+                    stmts = b_slist_append(stmts, stmt);
+                    stmt = NULL;
+                }
+                if (c == '{') {
+                    end = current;
+                    state = TEMPLATE_OPEN_BRACKET;
+                }
+                break;
+
+            case TEMPLATE_OPEN_BRACKET:
+                if (c == '%' || c == '{') {
+                    if (c == '%')
+                        state = TEMPLATE_BLOCK_START;
+                    else
+                        state = TEMPLATE_VARIABLE_START;
+                    if (end > start) {
+                        stmt = malloc(sizeof(blogc_template_stmt_t));
+                        stmt->type = type;
+                        stmt->value = b_strndup(src + start, end - start);
+                        stmts = b_slist_append(stmts, stmt);
+                        stmt = NULL;
+                    }
+                }
+                break;
+
+            case TEMPLATE_BLOCK_START:
+                if (c == ' ')
+                    break;
+                if (c >= 'a' && c <= 'z') {
+                    state = TEMPLATE_BLOCK_TYPE;
+                    start = current;
+                    break;
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_BLOCK_TYPE:
+                if (c >= 'a' && c <= 'z')
+                    break;
+                if (c == ' ') {
+                    if (0 == strncmp("block", src + start, current - start)) {
+                        if (!open_block) {
+                            state = TEMPLATE_BLOCK_BLOCK_TYPE_START;
+                            type = BLOGC_TEMPLATE_BLOCK_STMT;
+                            start = current;
+                            open_block = true;
+                            break;
+                        }
+                    }
+                    else if (0 == strncmp("endblock", src + start, current - start)) {
+                        if (open_block) {
+                            state = TEMPLATE_BLOCK_END;
+                            type = BLOGC_TEMPLATE_ENDBLOCK_STMT;
+                            open_block = false;
+                            break;
+                        }
+                    }
+                    else if (0 == strncmp("if", src + start, current - start)) {
+                        if (open_block) {
+                            state = TEMPLATE_BLOCK_IF_VARIABLE_START;
+                            type = BLOGC_TEMPLATE_IF_STMT;
+                            start = current;
+                            if_count++;
+                            break;
+                        }
+                    }
+                    else if (0 == strncmp("endif", src + start, current - start)) {
+                        if (open_block) {
+                            if (if_count > 0) {
+                                state = TEMPLATE_BLOCK_END;
+                                type = BLOGC_TEMPLATE_ENDIF_STMT;
+                                if_count--;
+                                break;
+                            }
+                        }
+                    }
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_BLOCK_BLOCK_TYPE_START:
+                if (c == ' ')
+                    break;
+                if (c >= 'a' && c <= 'z') {
+                    state = TEMPLATE_BLOCK_BLOCK_TYPE;
+                    start = current;
+                    break;
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_BLOCK_BLOCK_TYPE:
+                if ((c >= 'a' && c <= 'z') || c == '_')
+                    break;
+                if (c == ' ') {
+                    if ((0 == strncmp("single_source", src + start, current - start)) ||
+                        (0 == strncmp("multiple_sources", src + start, current - start)) ||
+                        (0 == strncmp("multiple_sources_once", src + start, current - start)))
+                    {
+                        end = current;
+                        state = TEMPLATE_BLOCK_END;
+                        break;
+                    }
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_BLOCK_IF_VARIABLE_START:
+                if (c == ' ')
+                    break;
+                if (c >= 'A' && c <= 'Z') {
+                    state = TEMPLATE_BLOCK_IF_VARIABLE;
+                    start = current;
+                    break;
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_BLOCK_IF_VARIABLE:
+                if ((c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || c == '_')
+                    break;
+                if (c == ' ') {
+                    end = current;
+                    state = TEMPLATE_BLOCK_END;
+                    break;
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_BLOCK_END:
+                if (c == ' ')
+                    break;
+                if (c == '%') {
+                    state = TEMPLATE_CLOSE_BRACKET;
+                    break;
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_VARIABLE_START:
+                if (c == ' ')
+                    break;
+                if (c >= 'A' && c <= 'Z') {
+                    if (open_block) {
+                        state = TEMPLATE_VARIABLE;
+                        type = BLOGC_TEMPLATE_VARIABLE_STMT;
+                        start = current;
+                        break;
+                    }
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_VARIABLE:
+                if ((c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || c == '_')
+                    break;
+                if (c == ' ') {
+                    end = current;
+                    state = TEMPLATE_VARIABLE_END;
+                    break;
+                }
+                if (c == '}') {
+                    end = current;
+                    state = TEMPLATE_CLOSE_BRACKET;
+                    break;
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_VARIABLE_END:
+                if (c == ' ')
+                    break;
+                if (c == '}') {
+                    state = TEMPLATE_CLOSE_BRACKET;
+                    break;
+                }
+                error = true;
+                break;
+
+            case TEMPLATE_CLOSE_BRACKET:
+                if (c == '}') {
+                    stmt = malloc(sizeof(blogc_template_stmt_t));
+                    stmt->type = type;
+                    stmt->value = NULL;
+                    if (end > start)
+                        stmt->value = b_strndup(src + start, end - start);
+                    stmts = b_slist_append(stmts, stmt);
+                    stmt = NULL;
+                    state = TEMPLATE_START;
+                    type = BLOGC_TEMPLATE_CONTENT_STMT;
+                    start = current + 1;
+                    break;
+                }
+                error = true;
+                break;
+
+        }
+
+        if (error)
+            break;
+
+        current++;
+    }
+
+    if (error) {
+        if (stmt != NULL) {
+            free(stmt->value);
+            free(stmt);
+        }
+        blogc_template_free_stmts(stmts);
+        blogc_parser_syntax_error("template", src, src_len, current);
+        return NULL;
+    }
+
+    return stmts;
+}
+
+
+void
+blogc_template_free_stmts(b_slist_t *stmts)
+{
+    for (b_slist_t *tmp = stmts; tmp != NULL; tmp = tmp->next) {
+        blogc_template_stmt_t *data = tmp->data;
+        free(data->value);
+        free(data);
+    }
+    b_slist_free(stmts);
+}
diff --git a/src/template-grammar.h b/src/template-parser.h
index 310acb5..1f43fe8 100644
--- a/src/template-grammar.h
+++ b/src/template-parser.h
@@ -6,14 +6,13 @@
  * See the file COPYING.
  */
 
-#ifndef _TEMPLATE_GRAMMAR_H
-#define _TEMPLATE_GRAMMAR_H
+#ifndef _TEMPLATE_PARSER_H
+#define _TEMPLATE_PARSER_H
 
 #include "utils/utils.h"
 
 typedef enum {
-    BLOGC_TEMPLATE_IF_STMT,
-    BLOGC_TEMPLATE_ELSE_STMT,
+    BLOGC_TEMPLATE_IF_STMT = 1,
     BLOGC_TEMPLATE_ENDIF_STMT,
     BLOGC_TEMPLATE_BLOCK_STMT,
     BLOGC_TEMPLATE_ENDBLOCK_STMT,
@@ -26,7 +25,7 @@ typedef struct {
     char *value;
 } blogc_template_stmt_t;
 
-b_slist_t* blogc_template_parse(const char *tmpl);
+b_slist_t* blogc_template_parse(const char *src, size_t src_len);
 void blogc_template_free_stmts(b_slist_t *stmts);
 
 #endif /* _TEMPLATE_GRAMMAR_H */