debug info through the lr parser

1 files changed, 196 insertions, 71 deletions

diff --git a/demos/sample-files/lbp-skeleton.c b/demos/sample-files/lbp-skeleton.c
index bf7bdca..1ee54d8 100644
--- a/demos/sample-files/lbp-skeleton.c
+++ b/demos/sample-files/lbp-skeleton.c
@@ -3,12 +3,20 @@
 #include <string.h>
 #include <stdint.h>
 #include <ctype.h>
+#include <stdarg.h>
 
-// TODO: - lr parser is bad for debugging
-//       - deal with errors
+// TODO: - lr parser is bad for debugging (now its better)
+//       - deal with errors (the token queue for example)!!!!
+//       - debuginfo in the token that gets propagaded through the
+//         stack_items and lr_parse returns a generic errorinfo
+//         with user-implemented compilation error messages
+//       - ast should show the specific operation (match, assignment, etc)
+
+#define MIN(a, b) ((a) > (b) ? (b) : (a))
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
 
 #define INPUT_CAP 4096
-#define ARENA_CAP 8192 //4096
+#define ARENA_CAP 4096 * 3
 
 #define ARENA_IMPLEMENTATION
 #include "util/arena.h"
@@ -51,40 +59,75 @@ static void xfree(void *ptr) { (void)ptr; return; }
 
 #include "util/list.h"
 
-struct ast_strlist { char *str; struct list_head list; };
+struct datatype {
+    int typeflags;
+    // TODO: add ptr
+    char *iden;
+    struct list_head *function_exprlist;
+};
+
+struct strlist { char *str; struct list_head list; };
+
+// TODO:
+// - type of each each exprlist, be it assignement, match, function call, etc
+// - revise expr types, etc
+
+// struct ast_exprlist {
+//     enum { AST_OP_ASSIGNMENT, AST_OP_TYPE, AST_OP_CALL } type;
+//     struct list_head *exprlist;
+// }
+
+// struct ast_exprlist {
+//     enum { AST_OP_ASSIGNMENT, AST_OP_IMPLEMENT, AST_OP_MATCH, AST_OP_EQUAL, AST_OP_CALL } type;
+//     struct list_head *exprlist;
+// };
+
+// struct ast_expr {
+//     enum { AST_LITERAL, AST_IDENTIFIER, AST_DECLARATION, AST_SUB } type;
+
+//     struct datatype datatype;
+//     struct list_head list;
+// }
 
 struct ast_expr {
-    enum { AST_NUMBER, AST_VARIABLE, AST_FIELDLIST, AST_DECLARATION, AST_DEFINITION, AST_OPERATION, AST_PARENLIST } type;
+    enum { AST_NUMBER, AST_VARIABLE, AST_DECLARATION, AST_FIELDLIST, AST_OPERATION, AST_PARENLIST, AST_BRACELIST } type;
     union {
         int number;
         struct ast_vrbl { int is_atom; char *iden; } variable;
         struct ast_fiel { struct ast_vrbl variable; struct list_head *fields_strlist; } fieldlist;
-        struct ast_decl { struct ast_vrbl variable; int typed; } declaration;
-        struct ast_defn { struct ast_decl declartion; struct list_head *block_exprlist; } definition;
+
         struct ast_oprn { enum { AST_OP_AND, AST_OP_OR } optype; struct list_head *left_exprlist; struct list_head *right_exprlist; } operation;
-        struct list_head *paren_exprlist;
+        struct list_head *exprlist;
     };
+
+    struct datatype datatype;
     struct list_head list;
 };
 
-#define NEW(t, ...)  ((struct ast_##t){__VA_ARGS__})
-#define g_NEW(t, ...) ((stack_item){.t = (struct ast_##t){__VA_ARGS__}})
+#define AST(t, ...)  ((struct ast_##t){__VA_ARGS__})
+#define g_AST(t, ...) ((stack_item){.t = {__VA_ARGS__}})
 
-#define LST(v) ({ typeof(v) *r = xalloc(sizeof(v)); *r = v; LIST_EMPTY(&r->list); &r->list; })
-#define g_LST(v) ((stack_item){.list = LST(v)})
+#define NEW(t, ...)  ((struct t){__VA_ARGS__})
+#define g_NEW(t, ...) ((stack_item){.t = {__VA_ARGS__}})
 
-void ast_vrbl_print(struct ast_vrbl *vrbl);
-void ast_fiel_print(struct ast_fiel *fiel);
-void ast_decl_print(struct ast_decl *decl);
-void ast_defn_print(struct ast_defn *defn);
-void ast_oprn_print(struct ast_oprn *oprn);
-void ast_expr_print(struct ast_expr *expr);
-void ast_exprlist_print(struct list_head *list);
+#define LST(v) ({ typeof(v) *r = malloc(sizeof(v)); *r = v; LIST_EMPTY(&r->list); &r->list; })
+#define g_LST(v) ((stack_item){.list = LST(v)})
 
-void ast_exprlist_free(struct list_head *list);
+// void ast_decide_datatype(struct list_head *list);
+void ast_print(struct list_head *list);
+void ast_free(struct list_head *list);
 
 // generated
 #include "bin/lbp.h"
+
+enum { DATATYPE_INT = 0, DATATYPE_STRUCT, DATATYPE_ENUM };
+enum { SUBTYPE_FUNCTION = 1 << 0, SUBTYPE_LITTLE   = 1 << 1,
+       SUBTYPE_BIG      = 1 << 2, SUBTYPE_NATIVE   = 1 << 3 };
+
+#define TYPEFLAGS(type, subtype) (((subtype) << 3) | (type))
+#define flags2type(flags)    ((flags) & 7)
+#define flags2subtype(flags) ((flags) >> 3)
+
 #include "bin/lbp.c"
 
 #include "util/dict.h"
@@ -107,32 +150,31 @@ static uint8_t dict_lowercase_char_to_bit[256] = {
     ['y'] = 26, ['z'] = 27, [ 0 ] = 1,  [' '] = 1
 };
 
+static inline char *substring(char *str, size_t sub_end);
+static inline char *linestart(char *strstart, char *pos);
+static inline size_t tillch(char *str, size_t len, char ch);
+#define strdup(...) _strdup(__VA_ARGS__)
+static inline char *_strdup(char *str);
+
 #include "parts/toklist.h"
 struct token {
     symbol s;
     stack_item v;
 };
 #define TOKEN_INIT(sym, val) (struct token){ .s = sym, .v = val }
-static void print_token(struct token *t);
-
 symbol token_sym(struct token *t) { return t->s; }
 intptr_t token_val(struct token *t) { return (intptr_t)&t->v; }
+static void print_token(struct token *t);
 
-static char *input;
 static size_t line = 1;
 static size_t active_region;
 static char *next_token(char *str);
 
-static inline char *substring(char *str, size_t sub_end);
-static inline char *linestart(char *strstart, char *pos);
-static inline size_t tillch(char *str, size_t len, char ch);
-#define strdup(...) _strdup(__VA_ARGS__)
-static inline char *_strdup(char *str);
-
-
 #include "util/queue.h"
 QUEUE_GENERATE(tokbuf, struct token, 16)
 
+static char *input;
+
 struct token *toklist_eat()
 {
     static struct token t;
@@ -147,13 +189,22 @@ struct token *toklist_peek() {
     return &t;
 }
 
+struct debuginfo {
+    size_t line;
+    size_t active_region;
+    char *end_ptr;
+};
+#define debuginfo_merge(out, ...) _debuginfo_merge(out, __VA_ARGS__, NULL);
+void _debuginfo_merge(struct debuginfo *out, ...);
+
+void errmsg_print(char *filename, char *input_buf, struct debuginfo *debuginfo, char *message);
+
 // #define _LR_PARSER_DEBUG
 #include "lr-parser.c"
 
 int main(void)
 {
-    char *filename = "stdin";
-
+    static char *filename = "stdin";
     static char input_buf[INPUT_CAP];
     if(fread(input_buf, INPUT_CAP, 1, stdin) == INPUT_CAP) {
         fprintf(stderr, "INPUT_CAP reached\n");
@@ -174,26 +225,17 @@ int main(void)
     //     if(token_sym(tok) == END_INPUT) break;
     // } return 0;
 
-    stack_item value;
-    struct lr_errinfo *errinfo;
-    if((errinfo = lr_parser(&value))) {
-        char *l = linestart(input_buf, input);
-
-        fprintf(stderr, "%s:%zu:%zu: ERROR: %s\n", filename, line, input - l - active_region+1, lr_err_str(errinfo));
-
-        size_t indent = fprintf(stderr, "   %zu ", line);
-        fprintf(stderr, "| %s\n", substring(l, tillch(l, strlen(l), '\n')));
-
-        fprintf(stderr, "%*s| %*s", indent, "", input - l - active_region, "");
-        if(active_region == 0) active_region = 1;
-        fprintf(stderr, "^"); for(size_t i = 0; i < active_region-1; i++) fprintf(stderr, "~");
-
-        fprintf(stderr, "\n\n");
+    struct lr_parseinfo parseinfo;
+    void *value = lr_parser(&parseinfo);
+    if(parseinfo.type == LR_ABORTED) {
+        goto cleanup;
+    } else if(parseinfo.type) {
+        errmsg_print(filename, input_buf, NULL, lr_err_str(&parseinfo));
         goto cleanup;
     }
 
-    ast_exprlist_print(value.list);
-    // ast_exprlist_free(value.list);
+    ast_print(((stack_item *)value)->list);
+    ast_free(((stack_item *)value)->list);
 
 cleanup:
     dict_free(&types_dict);
@@ -213,6 +255,46 @@ static void print_token(struct token *tok)
     }
 }
 
+void _debuginfo_merge(struct debuginfo *out, ...)
+{
+    va_list ap;
+    va_start(ap, out);
+
+    *out =  *va_arg(ap, struct debuginfo *);
+
+    struct debuginfo *arg;
+    while((arg = va_arg(ap, struct debuginfo *))) {
+        intptr_t start = MIN((intptr_t)out->end_ptr - out->active_region,
+                             (intptr_t)arg->end_ptr - arg->active_region);
+        intptr_t end   = MAX((intptr_t)out->end_ptr, (intptr_t)arg->end_ptr);
+
+        out->active_region = end - start;
+        out->end_ptr = (char *)end;
+    }
+
+    va_end(ap);
+}
+
+void errmsg_print(char *filename, char *input_buf, struct debuginfo *debuginfo, char *message)
+{
+    if(!debuginfo) debuginfo = &(struct debuginfo){.line = line, .active_region = active_region, .end_ptr = input};
+
+    char *l = linestart(input_buf, debuginfo->end_ptr);
+
+    fprintf(stderr, "%s:%zu:%zu: ERROR: %s\n", filename, debuginfo->line,
+            debuginfo->end_ptr - l - debuginfo->active_region+1,
+            message);
+
+    int indent = fprintf(stderr, "   %zu ", debuginfo->line);
+    fprintf(stderr, "| %s\n", substring(l, tillch(l, strlen(l), '\n')));
+
+    fprintf(stderr, "%*s| %*s", indent, "", debuginfo->end_ptr - l - debuginfo->active_region, "");
+    if(active_region == 0) debuginfo->active_region = 1;
+    for(size_t i = 0; i < debuginfo->active_region; i++) fprintf(stderr, "~");
+
+    fprintf(stderr, "\n");
+}
+
 // STR UTIL
 
 #define strdup(...) _strdup(__VA_ARGS__)
@@ -336,9 +418,10 @@ static char *next_token(char *str)
                 sub_off = tillch(str + 1, off - 1, ':') + 1;
                 if(hasfield) tokbuf_enqueue(&TOKEN_INIT(COLON, { .num=0 }));
 
+                int skip = hasfield || str[0] == ':';
                 tokbuf_enqueue(
-                    &TOKEN_INIT((!hasfield && str[0] == ':') ? ATOM : IDEN,
-                                { .str = strdup(substring(str+hasfield, sub_off-hasfield))}));
+                    &TOKEN_INIT((skip && !hasfield) ? ATOM : IDEN,
+                                { .str = strdup(substring(str+skip, sub_off-skip))}));
             } while(hasfield = 1, str += sub_off, off -= sub_off, off > 0);
 
             return str;
@@ -357,6 +440,34 @@ fail:
 
 // ast printing
 
+int ident = 0;
+#define INDENT() printf("%*s", ident*2, "");
+
+void ast_exprlist_print(struct list_head *list);
+
+void datatype_print(struct datatype *datatype)
+{
+    // TOOD: fix, very messy
+    switch(flags2type(datatype->typeflags)) {
+    case DATATYPE_INT: printf("int"); break;
+    case DATATYPE_ENUM: printf("enum"); break;
+    case DATATYPE_STRUCT: printf("struct"); break;
+    }
+
+    if(datatype->iden) printf("(%s)", datatype->iden);
+
+    for(int i = 0; i < 3; i++)
+        switch(flags2subtype(datatype->typeflags) & (1 << i)) {
+        case 0: break;
+        case SUBTYPE_FUNCTION:
+            printf("-function(");
+            ast_exprlist_print(datatype->function_exprlist);
+            printf(")");
+            break;
+        default: printf("-%d", i); break;
+        }
+}
+
 void ast_vrbl_print(struct ast_vrbl *vrbl)
 {
     printf("%s%s", vrbl->is_atom ? ":" : "", vrbl->iden);
@@ -365,25 +476,25 @@ void ast_vrbl_print(struct ast_vrbl *vrbl)
 void ast_fiel_print(struct ast_fiel *fiel)
 {
     ast_vrbl_print(&fiel->variable);
-    list_for_each_entry(struct ast_strlist, entry, list, fiel->fields_strlist) {
+    list_for_each_entry(struct strlist, entry, list, fiel->fields_strlist) {
         printf(":%s", entry->str);
     }
 }
 
-void ast_decl_print(struct ast_decl *decl)
-{
-    // TODO: implement
-}
-
-void ast_defn_print(struct ast_defn *defn)
+void ast_decl_print(struct ast_vrbl *vrbl, struct datatype *datatype)
 {
-    // TODO: implement
+    ast_vrbl_print(vrbl);
+    printf("/");
+    datatype_print(datatype);
 }
 
 void ast_oprn_print(struct ast_oprn *oprn)
 {
     ast_exprlist_print(oprn->left_exprlist);
+
     if(oprn->optype == AST_OP_AND) printf(",\n"); else printf(";\n");
+    INDENT();
+
     ast_exprlist_print(oprn->right_exprlist);
 }
 
@@ -396,21 +507,34 @@ void ast_exprlist_print(struct list_head *list)
         case AST_NUMBER: printf("%d", entry->number); break;
         case AST_VARIABLE: ast_vrbl_print(&entry->variable); break;
         case AST_FIELDLIST: ast_fiel_print(&entry->fieldlist); break;
-        case AST_DECLARATION: ast_decl_print(&entry->declaration); break;
+        case AST_DECLARATION: ast_decl_print(&entry->variable, &entry->datatype); break;
         case AST_OPERATION: ast_oprn_print(&entry->operation); break;
         case AST_PARENLIST:
-            printf("(");
-            ast_exprlist_print(entry->paren_exprlist);
-            printf(")");
+            printf("(\n"); ident++;
+            INDENT(); ast_exprlist_print(entry->exprlist);
+            printf("\n");  ident--;
+            INDENT(); printf(")");
+            break;
+        case AST_BRACELIST:
+            printf("{\n"); ident++;
+            INDENT(); ast_exprlist_print(entry->exprlist);
+            printf(".\n");  ident--;
+            INDENT(); printf("}");
             break;
         default: fprintf(stderr, "UNKNOWN TYPE: %d\n", entry->type);
         }
-        printf(" ");
+
+        if(entry->list.next) printf(" ");
     }
-    printf("\n");
 }
 
-void ast_exprlist_free(struct list_head *list)
+void ast_print(struct list_head *list)
+{
+    ast_exprlist_print(list);
+    printf(".\n");
+}
+
+void ast_free(struct list_head *list)
 {
     list_for_each_safe(l, list) {
         struct ast_expr *entry = list_entry(l, typeof(*entry), list);
@@ -420,18 +544,19 @@ void ast_exprlist_free(struct list_head *list)
         case AST_VARIABLE: break;
         case AST_FIELDLIST:
             list_for_each_safe(l, entry->fieldlist.fields_strlist)
-                free(list_entry(l, struct ast_strlist, list));
+                free(list_entry(l, struct strlist, list));
             break;
         case AST_DECLARATION: break;
-        case AST_DEFINITION:  break;
         case AST_OPERATION:
-            ast_exprlist_free(entry->operation.left_exprlist);
-            ast_exprlist_free(entry->operation.right_exprlist);
+            ast_free(entry->operation.left_exprlist);
+            ast_free(entry->operation.right_exprlist);
             break;
-        case AST_PARENLIST: ast_exprlist_free(entry->paren_exprlist); break;
+        case AST_BRACELIST:
+        case AST_PARENLIST: ast_free(entry->exprlist); break;
         default: fprintf(stderr, "UNKNOWN TYPE: %d\n", entry->type);
         }
 
+        ast_free(entry->datatype.function_exprlist);
         free(entry);
     }
 }