start of grammar parsing

5 files changed, 324 insertions, 70 deletions

diff --git a/demos/generate-parser.c b/demos/generate-parser.c
index 23201fa..48fa48c 100644
--- a/demos/generate-parser.c
+++ b/demos/generate-parser.c
@@ -5,12 +5,13 @@
 #include <unistd.h> // getopt
 #include <assert.h>
 
-#define DEFUALT_PATH "./bin"
-#define DEFUALT_TYPE "lalr-table"
 #define DEFAULT_OUTPUT "bin/a"
+#define DEFUALT_MODPATH "./bin"
+#define DEFUALT_TYPE "lalr-table"
 
 #include "parts/symbol.h"
 size_t total_symbols;
+char **symbol_to_str;
 int (*symbol_is_terminal)(symbol s);
 int (*symbol_is_input_end)(symbol s);
 int (*symbol_is_valid)(symbol s);
@@ -42,8 +43,7 @@ void (*table_free)();
 
 #include "util-tables.c"
 
-
-void *xdlsym(void *handle, char *sym)
+static void *xdlsym(void *handle, char *sym)
 {
     void *r = dlsym(handle, sym);
     if(!r) {
@@ -58,18 +58,18 @@ void *xdlsym(void *handle, char *sym)
     var = *(typeof(&var))xdlsym(handle, #var)
 
 
-char *modpath(char *name)
+static char *modpath(char *name)
 {
     static char fullpath[128];
 
     // TODO: search the GENERATE_PARSER_PATH env var
-    char *path = DEFUALT_PATH;
+    char *path = DEFUALT_MODPATH;
 
     assert(snprintf(fullpath, 128, "%s/%s.so", path, name) < 128);
     return fullpath;
 }
 
-char *add_extension(char *str, char *ext)
+static char *add_extension(char *str, char *ext)
 {
     static char full[128];
     assert((strlen(str) + strlen(ext) + 1) <= 128);
@@ -77,7 +77,7 @@ char *add_extension(char *str, char *ext)
     return strcat(full, ext);
 }
 
-void set_stdout(char *filename)
+static void set_stdout(char *filename)
 {
     if(!filename) filename = "/dev/tty";
     assert(freopen(filename, "w", stdout));
@@ -114,6 +114,7 @@ int main(int argc, char **argv)
     GET_VARIABLE(table_free, table_handle);
 
     GET_VARIABLE(total_symbols, def_handle);
+    GET_VARIABLE(symbol_to_str, def_handle);
     GET_VARIABLE(symbol_is_terminal, def_handle);
     GET_VARIABLE(symbol_is_input_end, def_handle);
     GET_VARIABLE(symbol_is_valid, def_handle);
@@ -132,8 +133,13 @@ int main(int argc, char **argv)
         goto cleanup;
     }
 
+    table_print();
+
     set_stdout(add_extension(output_path, ".c"));
     printf("size_t total_symbols = %zu;\n", total_symbols);
+    printf("char **symbol_to_string = (char *([])){\n");
+    for(size_t i = 0; i < total_symbols; i++) printf("\"%s\", ", symbol_to_str[i]);
+    printf("};\n");
     printf("IMPLEMENT_FUNCPTR(int, symbol_is_valid, (symbol s)) {return s < total_symbols;}\n");
 
     printf("struct production _grammar[] = {\n");
@@ -157,19 +163,33 @@ int main(int argc, char **argv)
 
     for(size_t i = 0; i < total_productions; i++) {
         printf("#define A(n) (*(stack_head-3*%zu+3*n-1))\n", grammar[i].nRHS-1);
-        printf("int __prod%zu_action(int *stack_head)\n", i);
-        printf("{ int v;\n");
-        printf(semantic_action_str[i]);
+        printf("intptr_t __prod%zu_action(intmax_t *stack_head)\n", i);
+        printf("{ intptr_t v;\n");
+        puts(semantic_action_str[i]);
         printf("return v; }\n");
         printf("#undef A\n");
     }
 
-    printf("typedef int (*semantic_action_fn)(int *stack_head);\n");
+    printf("typedef intptr_t (*semantic_action_fn)(intmax_t *stack_head);\n");
 
     printf("semantic_action_fn *semantic_actions = (semantic_action_fn[]){\n");
     for(size_t i = 0; i < total_productions; i++)
         printf("__prod%zu_action, ", i);
     printf("};");
+
+    set_stdout(add_extension(output_path, ".h"));
+    printf("#ifndef GENERATED_H\n");
+    printf("#define GENERATED_H\n");
+    printf("#include \"parts/symbol.h\"\n");
+    printf("enum symbol {\n");
+    for(size_t i = 0; i < total_symbols; i++) printf("%s, ", symbol_to_str[i]);
+    printf("};\n");
+    printf("#include \"parts/grammar.h\"\n");
+    printf("#include \"parts/table.h\"\n");
+    printf("#include <stdint.h>\n");
+    printf("typedef intptr_t (*semantic_action_fn)(intmax_t *stack_head);\n");
+    printf("extern semantic_action_fn *semantic_actions;\n");
+    printf("#endif\n");
     set_stdout(NULL);
 
 cleanup:
@@ -189,10 +209,10 @@ void precedence_tables_fill()
 
     for(size_t i = 0; i < nprecedence_defs; i++)
         for(size_t j = 0; j < precedence_defs[i].nlist; j++)
-            if(precedence_defs[i].flag >= 0)
+            if(precedence_defs[i].list[j] >= 0)
                 precedence_symbol[precedence_defs[i].list[j]] = PRECEDENCE_SET(precedence_defs[i].flag, i+1);
             else
-                precedence_production[precedence_defs[i].list[j]] = PRECEDENCE_SET(~precedence_defs[i].flag, i+1);
+                precedence_production[~precedence_defs[i].list[j]] = PRECEDENCE_SET(precedence_defs[i].flag, i+1);
 
     for(size_t i = 0; i < total_productions; i++) {
         if(precedence_production[i]) continue;
diff --git a/demos/sample-files/calc-defs.c b/demos/sample-files/calc-defs.c
index 7321a88..103b69e 100644
--- a/demos/sample-files/calc-defs.c
+++ b/demos/sample-files/calc-defs.c
@@ -1,17 +1,19 @@
-#include <stddef.h> // size_t
+#include "util/util.h"
+#define SYMBOLS(X)                                    \
+    X(PLUS) X(MINUS) X(TIMES) X(MORE) X(LESS) X(EQUA) \
+    X(LPAREN) X(RPAREN)                               \
+    X(QMARK) X(COLON)                                 \
+    X(NUM) X(END_INPUT)                               \
+                                                      \
+    X(EP) X(E)                                        \
+    X(SYMBOLS_END)
 
 #include "parts/symbol.h"
-enum symbol {
-    PLUS, MINUS, TIMES,
-    LPAREN, RPAREN,
-    NUM, END_INPUT,
-
-    EP, E, T,
-    SYMBOLS_END,
-};
-
+enum symbol { SYMBOLS(X_TO_ENUM) };
 size_t total_symbols = SYMBOLS_END;
 
+extern char **symbol_to_str = (char *([])){ SYMBOLS(X_TO_STR) };
+
 IMPLEMENT_FUNCPTR(int, symbol_is_terminal, (symbol s)) { return s < EP; }
 IMPLEMENT_FUNCPTR(int, symbol_is_input_end, (symbol s)) { return s == END_INPUT; }
 IMPLEMENT_FUNCPTR(int, symbol_is_valid, (symbol s)) { return s < SYMBOLS_END; }
@@ -24,6 +26,12 @@ static struct production _grammar[] = {
     PROD(E, -->, E, MINUS, E),
     PROD(E, -->, E, TIMES, E),
     PROD(E, -->, LPAREN, E, RPAREN),
+    PROD(E, -->, MINUS, E),
+    PROD(E, -->, E, QMARK, E, COLON, E),
+    PROD(E, -->, E, QMARK, E),
+    PROD(E, -->, E, MORE, E),
+    PROD(E, -->, E, LESS, E),
+    PROD(E, -->, E, EQUA, E),
     PROD(E, -->, NUM),
 };
 
@@ -37,20 +45,29 @@ char **semantic_action_str = (char *([])){
     "v = A(0) - A(2);",
     "v = A(0) * A(2);",
     "v = A(1);",
+    "v = - A(1);",
+    "v = A(0) ? A(2) : A(4);",
+    "v = A(0) ? A(2) : 0;",
+    "v = A(0) > A(1);",
+    "v = A(0) < A(1);",
+    "v = A(0) = A(1);",
     "v = A(0);",
 };
 
 #include "parts/precedence.h"
-
 struct precedence_def {
     int flag;
     int *list;
     size_t nlist;
 };
 #define PREC(f, ...) {f, (int[]){__VA_ARGS__}, sizeof((int[]){__VA_ARGS__})/sizeof(int)}
+#define USE_PROD(n) (~(n))
 struct precedence_def _precedence_defs[] = {
+    PREC(0, QMARK),
+    PREC(0, USE_PROD(7)),
+    PREC(PRECEDENCE_LEFT_ASSOC, MORE, LESS, EQUA),
     PREC(PRECEDENCE_LEFT_ASSOC, MINUS, PLUS),
-    PREC(PRECEDENCE_LEFT_ASSOC, TIMES),
+    PREC(PRECEDENCE_LEFT_ASSOC, TIMES, USE_PROD(5)),
     PREC(PRECEDENCE_LEFT_ASSOC, LPAREN, RPAREN),
 };
 
diff --git a/demos/sample-files/calc-skeleton.c b/demos/sample-files/calc-skeleton.c
index b0cbf00..6e5d2d5 100644
--- a/demos/sample-files/calc-skeleton.c
+++ b/demos/sample-files/calc-skeleton.c
@@ -1,30 +1,57 @@
 #include <stdio.h>
 #include <string.h>
+#include <stdint.h>
 #include <ctype.h>
 
-#include "lr-parser.c"
-#include "bin/a.c" // generated
-
-// these should come from a generated
-// header file by the parser generator
-#include "parts/symbol.h"
-enum symbol {
-    PLUS, MINUS, TIMES,
-    LPAREN, RPAREN,
-    NUM, END_INPUT,
-
-    EP, E, T,
-    SYMBOLS_END,
-};
+// generated
+#include "bin/calc.h"
+#include "bin/calc.c"
 
+#include "parts/toklist.h"
 static struct token {
     symbol s;
     int v;
 } tok;
 
+static char *next_token(char *str);
+
+symbol token_sym(struct token *t) { return t->s; }
+intptr_t token_val(struct token *t) { return (intptr_t)t->v; }
+
+static char *input;
+
+struct token *toklist_eat()
+{
+    static struct token t;
+    t = tok;
+    input = next_token(input);
+    return &t;
+}
+
+struct token *toklist_peek() { return &tok; }
+
+#include "lr-parser.c"
+
+int main(int argc, char **argv)
+{
+    if(argc != 2) return 1;
+
+    input = next_token(argv[1]);
+
+    intptr_t value;
+    if(lr_parser(&value)) return 1;
+
+    printf("INPUT: '%s'\n", argv[1]);
+    printf("OUTPUT: %jd\n", value);
+
+    return 0;
+}
+
+// LEXER
+
 static inline int issep(char c)
 {
-    return isspace(c) || c == '\0' || c == '(' || c == ')' || c == '+' || c == '-' || c == '*';;
+    return isspace(c) || c == '\0' || c == '(' || c == ')' || c == '+' || c == '-' || c == '*' || c == '>' || c == '<' || c == '=' || c == '?' || c == ':';
 }
 
 static inline int tillsep(char *str)
@@ -59,6 +86,11 @@ static char *next_token(char *str)
             case '-': tok.s = MINUS; break;
             case '+': tok.s = PLUS; break;
             case '*': tok.s = TIMES; break;
+            case '>': tok.s = MORE; break;
+            case '<': tok.s = LESS; break;
+            case '=': tok.s = EQUA; break;
+            case '?': tok.s = QMARK; break;
+            case ':': tok.s = COLON; break;
             }
         } else if(c0 >= '0' && c0 <= '9') { // num
             tok.s = NUM;
@@ -68,32 +100,3 @@ static char *next_token(char *str)
 
     return str+off;
 }
-
-static char *input;
-
-symbol token_sym(struct token *t) { return t->s; }
-int token_val(struct token *t) { return t->v; }
-
-struct token *toklist_eat()
-{
-    static struct token t;
-    t = tok;
-    input = next_token(input);
-    return &t;
-}
-struct token *toklist_peek() { return &tok; }
-
-int main(int argc, char **argv)
-{
-    if(argc != 2) return 1;
-
-    input = next_token(argv[1]);
-
-    int value;
-    if(lr_parser(&value)) return 1;
-
-    printf("INPUT: '%s'\n", argv[1]);
-    printf("OUTPUT: %d\n", value);
-
-    return 0;
-}
diff --git a/demos/sample-files/gram-defs.c b/demos/sample-files/gram-defs.c
new file mode 100644
index 0000000..733a866
--- /dev/null
+++ b/demos/sample-files/gram-defs.c
@@ -0,0 +1,65 @@
+#include "util/util.h"
+#define SYMBOLS(X)                                  \
+    X(COLON) X(PIPE) X(SEMICOL) X(DOT)              \
+    X(D_LEFT) X(D_RIGHT) X(D_TERMINAL) X(D_NONTERM) \
+    X(IDEN) X(NUM) X(ACTION) X(END_INPUT)           \
+                                                    \
+    X(Sp) X(S) X(Slist) X(Prod) X(Prec)             \
+    X(Prodlist) X(Idenlist) X(IorN) X(IorNlist)     \
+    X(SYMBOLS_END)                                  \
+
+#include "parts/symbol.h"
+enum symbol { SYMBOLS(X_TO_ENUM) };
+size_t total_symbols = SYMBOLS_END;
+
+char **symbol_to_str = (char *([])){ SYMBOLS(X_TO_STR) };
+
+IMPLEMENT_FUNCPTR(int, symbol_is_terminal, (symbol s)) { return s < Sp; }
+IMPLEMENT_FUNCPTR(int, symbol_is_input_end, (symbol s)) { return s == END_INPUT; }
+IMPLEMENT_FUNCPTR(int, symbol_is_valid, (symbol s)) { return s < SYMBOLS_END; }
+
+#include "parts/grammar.h"
+#define PROD(LHS, _, ...) {LHS, (symbol[]){__VA_ARGS__}, sizeof((symbol[]){__VA_ARGS__})/sizeof(symbol)}
+#define GRAMMAR_ACTION_DEF(X)                                          \
+    X(PROD(Sp, ->, Slist, END_INPUT), "v = 0;")                        \
+    X(PROD(Slist, -->, S, SEMICOL, Slist), "v = 0;")                   \
+    X(PROD(Slist, -->, S, DOT), "v = 0;")                              \
+    X(PROD(S, -->, Prod), "v = 0;")                                    \
+    X(PROD(S, -->, Prec), "v = 0;")                                    \
+    X(PROD(Idenlist, -->, IDEN, Idenlist), "v = 0;")                   \
+    X(PROD(Idenlist, -->, IDEN), "v = 0;")                             \
+    X(PROD(Prod, -->, IDEN, COLON, Prodlist), "v = 0;")                \
+    X(PROD(Prodlist, -->, Idenlist, ACTION, PIPE, Prodlist), "printf(\"ACTION: '%s'\\n\", A(1));") \
+    X(PROD(Prodlist, -->, Idenlist, ACTION), "printf(\"ACTION: '%s'\\n\", A(1));") \
+    X(PROD(Prec, -->, D_TERMINAL, Idenlist), "v = 0;")                 \
+    X(PROD(Prec, -->, D_NONTERM,  Idenlist), "v = 0;")                 \
+    X(PROD(Prec, -->, D_LEFT,  IorNlist), "v = 0;")                    \
+    X(PROD(Prec, -->, D_RIGHT, IorNlist), "v = 0;")                    \
+    X(PROD(IorNlist,  -->, IorN, IorNlist), "v = 0;")                  \
+    X(PROD(IorNlist,  -->, IorN), "v = 0;")                            \
+    X(PROD(IorN, -->, IDEN), "v = 0;")                                 \
+    X(PROD(IorN, -->, NUM), "v = 0;")
+
+#define X_GRAMMAR(G, A) G,
+#define X_ACTION(G, A) A,
+
+static struct production _grammar[] = {
+    GRAMMAR_ACTION_DEF(X_GRAMMAR)
+};
+
+struct production *grammar = _grammar;
+size_t total_productions = sizeof(_grammar) / sizeof(*_grammar);
+
+// #include "???.h"
+char **semantic_action_str = (char *([])){
+    GRAMMAR_ACTION_DEF(X_ACTION)
+};
+
+#include "parts/precedence.h"
+struct precedence_def {
+    int flag;
+    int *list;
+    size_t nlist;
+};
+struct precedence_def *precedence_defs = NULL;
+size_t nprecedence_defs = 0;
diff --git a/demos/sample-files/gram-skeleton.c b/demos/sample-files/gram-skeleton.c
new file mode 100644
index 0000000..89ef6b4
--- /dev/null
+++ b/demos/sample-files/gram-skeleton.c
@@ -0,0 +1,149 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <ctype.h>
+
+#define ARENA_IMPLEMENTATION
+#include "util/arena.h"
+
+static char buf[1024];
+static struct arena_ctx global_arena = ARENA_CTX_INIT(buf, sizeof(buf));
+static void *xalloc(size_t sz) {
+    void *addr = arena_allocate(&global_arena, sz);
+    if(!addr) {
+        fprintf(stderr, "ERROR: Arena empty\n"); exit(1);
+    }
+
+    return addr;
+}
+
+// generated
+#include "bin/gram.h"
+#include "bin/gram.c"
+
+#include "parts/toklist.h"
+struct token {
+    symbol s;
+    intptr_t v;
+} tok;
+
+static char *next_token(char *str);
+
+symbol token_sym(struct token *t) { return t->s; }
+intptr_t token_val(struct token *t) { return t->v; }
+
+static char *input = (char []){
+    "-left B;"
+    "-right C;"
+    "-left D;"
+    ""
+    "A: B {a}"
+    " | C N {d}."
+};
+
+struct token *toklist_eat()
+{
+    static struct token t;
+    t = tok;
+    input = next_token(input);
+    return &t;
+}
+
+struct token *toklist_peek() { return &tok; }
+
+#include "lr-parser.c"
+
+int main(void)
+{
+    input = next_token(input);
+
+    intptr_t value;
+    if(lr_parser(&value)) {
+        return 1;
+    }
+
+    printf("OUTPUT: %jd\n", value);
+    return 0;
+}
+
+// STR UTIL
+
+#define strdup(...) _strdup(__VA_ARGS__)
+static inline char *_strdup(char *str)
+{
+    return memcpy(xalloc(strlen(str) + 1), str, strlen(str)+1);
+}
+
+static inline char *substring(char *str, size_t sub_end)
+{
+    static char sub[128];
+    if(sub_end+1 > sizeof(sub)) return NULL;
+
+    sub[sub_end] = '\0';
+    return memcpy(sub, str, sub_end);
+}
+
+// LEXER
+
+static inline int issep(char c)
+{
+    return isspace(c) || c == '\0' || c == ':' || c == '|' || c == ';' || c == '.' || c == '-' || c == '{';
+}
+
+static inline int tillsep(char *str)
+{
+    size_t i = 0;
+    while(!issep(str[i++]));
+    return i-1;
+}
+
+static char *next_token(char *str)
+{
+    if(!str) return str;
+
+    size_t off = 0;
+    char c0 = str[0];
+
+    if(c0 == '\0') tok.s = END_INPUT;
+    if(isspace(c0)) return next_token(str+1);
+    else {
+        off = tillsep(str);
+        if(off == 0) { // sep
+            switch(str[off++]) {
+            case ':': tok.s = COLON; break;
+            case '|': tok.s = PIPE; break;
+            case ';': tok.s = SEMICOL; break;
+            case '.': tok.s = DOT; break;
+            case '-':
+                off = tillsep(++str);
+                char *s = substring(str, off);
+                if(strcmp(s, "left") == 0) tok.s = D_LEFT;
+                else if(strcmp(s, "right") == 0) tok.s = D_RIGHT;
+                else if(strcmp(s, "terminal") == 0) tok.s = D_TERMINAL;
+                else if(strcmp(s, "nonterminal") == 0) tok.s = D_NONTERM;
+                else { fprintf(stderr, "ERROR: Unknown directive '-%s'\n", s); goto fail; }
+                break;
+            case '{':
+                for(int c = 1; c != 0; off++)
+                    if(str[off] == '\0') { fprintf(stderr, "ERROR: No closing '{'\n"); goto fail; }
+                    else if(str[off] == '{') c++;
+                    else if(str[off] == '}') c--;
+                tok.s = ACTION;
+                tok.v = (intptr_t)strdup(substring(str, off));
+                break;
+            }
+        } else if(isalpha(c0)) { // iden or named symbol
+            tok.s = IDEN;
+            tok.v = (intptr_t)strdup(substring(str, off));
+        } else if(c0 >= '0' && c0 <= '9') { // num
+            tok.s = NUM;
+            tok.v = (intptr_t)atoi(substring(str, off));
+        }
+    }
+
+    return str+off;
+
+fail:
+    tok.s = END_INPUT;
+    return NULL;
+}