From: Stephen Dolan <mu@netsoc.tcd.ie>
Date: Sun, 2 Sep 2012 15:31:59 +0000 (+0100)
Subject: Move from Jansson to JV - everything but the interpreter loop
X-Git-Tag: jq-1.1~95
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=2cb9a6e61dd9605cfd24d44695be5f0a1a00aaba;p=jq

Move from Jansson to JV - everything but the interpreter loop

Passes valgrind --leak-check=full.
---

diff --git a/c/Makefile b/c/Makefile
index 08e7270..bf33acf 100644
--- a/c/Makefile
+++ b/c/Makefile
@@ -24,7 +24,7 @@ jvtest: jvtest.c jv.c jv_print.c
 	$(CC) -DNO_JANSSON -o $@ $^
 
 jv_parse: jv_parse.c jv.c jv_print.c jv_dtoa.c
-	$(CC) -DNO_JANSSON -o $@ $^
+	$(CC) -DNO_JANSSON -o $@ $^ -DJV_PARSE_MAIN
 
 
 test: jvtest
diff --git a/c/builtin.c b/c/builtin.c
index ef37c28..0252566 100644
--- a/c/builtin.c
+++ b/c/builtin.c
@@ -1,28 +1,24 @@
 #include "builtin.h"
 
-#include <jansson.h>
-
-
-static void f_false(json_t* input[], json_t* output[]) {
-  output[0] = json_false();
+static void f_false(jv input[], jv output[]) {
+  output[0] = jv_false();
 }
 
-static void f_true(json_t* input[], json_t* output[]) {
-  output[0] = json_true();
+static void f_true(jv input[], jv output[]) {
+  output[0] = jv_true();
 }
 
 
-static void f_plus(json_t* input[], json_t* output[]) {
-  json_t* a = input[2];
-  json_t* b = input[1];
-  if (json_is_number(a) && json_is_number(b)) {
-    output[0] = json_real(json_number_value(a) + 
-                          json_number_value(b));
-  } else if (json_is_array(a) && json_is_array(b)) {
-    output[0] = json_copy(a);
-    json_array_extend(output[0], b);
+static void f_plus(jv input[], jv output[]) {
+  jv a = input[2];
+  jv b = input[1];
+  if (jv_get_kind(a) == JV_KIND_NUMBER && jv_get_kind(b) == JV_KIND_NUMBER) {
+    output[0] = jv_number(jv_number_value(a) + 
+                          jv_number_value(b));
+  } else if (jv_get_kind(a) == JV_KIND_ARRAY && jv_get_kind(b) == JV_KIND_ARRAY) {
+    output[0] = jv_array_concat(a, b);
   } else {
-    output[0] = json_string("wtf gaize");
+    output[0] = jv_string("wtf gaize");
   }
 }
 
diff --git a/c/bytecode.c b/c/bytecode.c
index ef585bd..208e82f 100644
--- a/c/bytecode.c
+++ b/c/bytecode.c
@@ -1,6 +1,7 @@
 #include <stdio.h>
 #include <stdint.h>
-#include <jansson.h>
+#include <stdlib.h>
+
 #include "bytecode.h"
 #include "opcode.h"
 
@@ -57,8 +58,7 @@ void dump_operation(struct bytecode* bc, uint16_t* codeptr) {
       printf(" %04d", pc + imm);
     } else if (op->flags & OP_HAS_CONSTANT) {
       printf(" ");
-      json_dumpf(json_array_get(bc->constants, imm),
-                 stdout, JSON_ENCODE_ANY);
+      jv_dump(jv_array_get(jv_copy(bc->constants), imm));
     } else if (op->flags & OP_HAS_VARIABLE) {
       uint16_t v = bc->code[pc++];
       printf(" v%d", v);
@@ -70,3 +70,19 @@ void dump_operation(struct bytecode* bc, uint16_t* codeptr) {
     }
   }  
 }
+
+void symbol_table_free(struct symbol_table* syms) {
+  free(syms->cfunctions);
+  free(syms);
+}
+
+void bytecode_free(struct bytecode* bc) {
+  free(bc->code);
+  jv_free(bc->constants);
+  for (int i=0; i<bc->nsubfunctions; i++)
+    bytecode_free(bc->subfunctions[i]);
+  if (!bc->parent)
+    symbol_table_free(bc->globals);
+  free(bc->subfunctions);
+  free(bc);
+}
diff --git a/c/bytecode.h b/c/bytecode.h
index b1146ad..5ffe80a 100644
--- a/c/bytecode.h
+++ b/c/bytecode.h
@@ -1,11 +1,11 @@
 #ifndef BYTECODE_H
 #define BYTECODE_H
-#include <jansson.h>
 #include <stdint.h>
 
+#include "jv.h"
 #include "opcode.h"
 
-typedef void (*cfunction_ptr)(json_t* input[], json_t* output[]);
+typedef void (*cfunction_ptr)(jv input[], jv output[]);
 
 struct cfunction {
   cfunction_ptr fptr;
@@ -33,7 +33,7 @@ struct bytecode {
   int nlocals;
   int nclosures;
 
-  json_t* constants;
+  jv constants; // JSON array of constants
   struct symbol_table* globals;
 
   struct bytecode** subfunctions;
@@ -46,4 +46,7 @@ void dump_disassembly(int, struct bytecode* code);
 void dump_code(int, struct bytecode* code);
 void dump_operation(struct bytecode* bc, uint16_t* op);
 
+void symbol_table_free(struct symbol_table* syms);
+void bytecode_free(struct bytecode* bc);
+
 #endif
diff --git a/c/compile.c b/c/compile.c
index dc90aed..b7fcc4f 100644
--- a/c/compile.c
+++ b/c/compile.c
@@ -1,5 +1,6 @@
 #include <assert.h>
 #include <string.h>
+#include <stdlib.h>
 #include "opcode.h"
 #include "compile.h"
 
@@ -13,7 +14,7 @@ struct inst {
   union {
     uint16_t intval;
     struct inst* target;
-    json_t* constant;
+    jv constant;
     struct cfunction* cfunc;
   } imm;
 
@@ -52,6 +53,9 @@ static void inst_free(struct inst* i) {
   if (opcode_describe(i->op)->flags & OP_HAS_BLOCK) {
     block_free(i->subfn);
   }
+  if (opcode_describe(i->op)->flags & OP_HAS_CONSTANT) {
+    jv_free(i->imm.constant);
+  }
   free(i);
 }
 
@@ -89,7 +93,7 @@ block gen_op_simple(opcode op) {
 }
 
 
-block gen_op_const(opcode op, json_t* constant) {
+block gen_op_const(opcode op, jv constant) {
   assert(opcode_describe(op)->flags & OP_HAS_CONSTANT);
   inst* i = inst_new(op);
   i->imm.constant = constant;
@@ -267,7 +271,7 @@ block gen_both(block a, block b) {
 block gen_collect(block expr) {
   block c = gen_noop();
   block_append(&c, gen_op_simple(DUP));
-  block_append(&c, gen_op_const(LOADK, json_array()));
+  block_append(&c, gen_op_const(LOADK, jv_array()));
   block array_var = block_bind(gen_op_var_unbound(STOREV, "collect"),
                                gen_noop(), OP_HAS_VARIABLE);
   block_append(&c, array_var);
@@ -395,7 +399,7 @@ static void compile(struct bytecode* bc, block b) {
   uint16_t* code = malloc(sizeof(uint16_t) * bc->codelen);
   bc->code = code;
   pos = 0;
-  json_t* constant_pool = json_array();
+  jv constant_pool = jv_array();
   int maxvar = -1;
   for (inst* curr = b.first; curr; curr = curr->next) {
     const struct opcode_description* op = opcode_describe(curr->op);
@@ -436,8 +440,8 @@ static void compile(struct bytecode* bc, block b) {
       }
       assert(nargs - 1 == desired_params);
     } else if (opflags & OP_HAS_CONSTANT) {
-      code[pos++] = json_array_size(constant_pool);
-      json_array_append(constant_pool, curr->imm.constant);
+      code[pos++] = jv_array_length(jv_copy(constant_pool));
+      constant_pool = jv_array_append(constant_pool, jv_copy(curr->imm.constant));
     } else if (opflags & OP_HAS_VARIABLE) {
       code[pos++] = nesting_level(bc, curr->bound_by);
       uint16_t var = (uint16_t)curr->bound_by->imm.intval;
diff --git a/c/compile.h b/c/compile.h
index 4c0a3a1..5f0be89 100644
--- a/c/compile.h
+++ b/c/compile.h
@@ -13,7 +13,7 @@ typedef struct block {
 
 block gen_noop();
 block gen_op_simple(opcode op);
-block gen_op_const(opcode op, json_t* constant);
+block gen_op_const(opcode op, jv constant);
 block gen_op_target(opcode op, block target);
 block gen_op_var_unbound(opcode op, const char* name);
 block gen_op_var_bound(opcode op, block binder);
diff --git a/c/jv.c b/c/jv.c
index 3029904..7944d64 100644
--- a/c/jv.c
+++ b/c/jv.c
@@ -278,7 +278,7 @@ static jvp_string* jvp_string_alloc(uint32_t size) {
   return s;
 }
 
-static jv_complex jvp_string_new(char* data, uint32_t length) {
+static jv_complex jvp_string_new(const char* data, uint32_t length) {
   jvp_string* s = jvp_string_alloc(length);
   memcpy(s->data, data, length);
   s->data[length] = 0;
@@ -389,14 +389,14 @@ static int jvp_string_equal(jv_complex* a, jv_complex* b) {
  * Strings (public API)
  */
 
-jv jv_string_sized(char* str, int len) {
+jv jv_string_sized(const char* str, int len) {
   jv j;
   j.kind = JV_KIND_STRING;
   j.val.complex = jvp_string_new(str, len);
   return j;
 }
 
-jv jv_string(char* str) {
+jv jv_string(const char* str) {
   return jv_string_sized(str, strlen(str));
 }
 
diff --git a/c/jv.h b/c/jv.h
index 44fe479..7696b03 100644
--- a/c/jv.h
+++ b/c/jv.h
@@ -4,42 +4,6 @@
 #include <stdint.h>
 #include <assert.h>
 #include <stddef.h>
-#ifndef NO_JANSSON
-#include <jansson.h>
-static json_t* jv_lookup(json_t* t, json_t* k) {
-  json_t* v;
-  if (json_is_object(t) && json_is_string(k)) {
-    v = json_object_get(t, json_string_value(k));
-  } else if (json_is_array(t) && json_is_number(k)) {
-    v = json_array_get(t, json_integer_value(k));
-  } else {
-    assert(0&&"bad lookup");
-  }
-  if (v) 
-    return v;
-  else
-    return json_null();
-}
-
-static json_t* jv_modify(json_t* t, json_t* k, json_t* v) {
-  t = json_copy(t);
-  if (json_is_object(t) && json_is_string(k)) {
-    json_object_set(t, json_string_value(k), v);
-  } else if (json_is_array(t) && json_is_number(k)) {
-    json_array_set(t, json_integer_value(k), v);
-  } else {
-    assert(0 && "bad mod");
-  }
-  return t;
-}
-
-static json_t* jv_insert(json_t* root, json_t* value, json_t** path, int pathlen) {
-  if (pathlen == 0) {
-    return value;
-  }
-  return jv_modify(root, *path, jv_insert(jv_lookup(root, *path), value, path+1, pathlen-1));
-}
-#endif
 
 
 
@@ -99,8 +63,8 @@ jv jv_array_concat(jv, jv);
 jv jv_array_slice(jv, int, int);
 
 
-jv jv_string(char*);
-jv jv_string_sized(char*, int);
+jv jv_string(const char*);
+jv jv_string_sized(const char*, int);
 int jv_string_length(jv);
 uint32_t jv_string_hash(jv);
 const char* jv_string_value(jv);
@@ -119,6 +83,57 @@ jv jv_object_iter_value(jv, int);
 
 
 void jv_dump(jv);
+jv jv_parse(const char* string);
+
+
+
+
+
+static jv jv_lookup(jv t, jv k) {
+  jv v;
+  if (jv_get_kind(t) == JV_KIND_OBJECT && jv_get_kind(k) == JV_KIND_STRING) {
+    v = jv_object_get(t, k);
+  } else if (jv_get_kind(t) == JV_KIND_ARRAY && jv_get_kind(k) == JV_KIND_NUMBER) {
+    // FIXME: don't do lookup for noninteger index
+    v = jv_array_get(t, (int)jv_number_value(k));
+  } else {
+    assert(0&&"bad lookup");
+  }
+  return v;
+  // FIXME: invalid indexes, JV_KIND_INVALID
+  /*
+  if (v) 
+    return v;
+  else
+    return jv_null();
+  */
+}
+
+static jv jv_modify(jv t, jv k, jv v) {
+  if (jv_get_kind(t) == JV_KIND_OBJECT && jv_get_kind(k) == JV_KIND_STRING) {
+    t = jv_object_set(t, k, v);
+  } else if (jv_get_kind(t) == JV_KIND_ARRAY && jv_get_kind(k) == JV_KIND_NUMBER) {
+    t = jv_array_set(t, (int)jv_number_value(k), v);
+  } else {
+    assert(0 && "bad mod");
+  }
+  return t;
+}
+
+static jv jv_insert(jv root, jv value, jv* path, int pathlen) {
+  if (pathlen == 0) {
+    jv_free(root);
+    return value;
+  }
+  return jv_modify(root, *path, 
+                   jv_insert(jv_lookup(jv_copy(root), jv_copy(*path)), value, path+1, pathlen-1));
+}
+
+
+
+
+
+
 
 
 
diff --git a/c/jv_parse.c b/c/jv_parse.c
index 31e8d68..26c2cc3 100644
--- a/c/jv_parse.c
+++ b/c/jv_parse.c
@@ -327,26 +327,32 @@ pfunc finish() {
   return 0;
 }
 
-
-int main(int argc, char* argv[]) {
-  assert(argc == 2);
+jv jv_parse(const char* string) {
   jvp_dtoa_context_init(&dtoa);
-  char* p = argv[1];
+
+  const char* p = string;
   char ch;
   while ((ch = *p++)) {
     presult msg = scan(ch);
     if (msg){
-      printf("ERROR: %s\n", msg);
-      return 1;
+      printf("ERROR: %s (parsing [%s])\n", msg, string);
+      return jv_null();
     }
   }
   presult msg = finish();
   if (msg) {
-    printf("ERROR: %s\n", msg);
-    return 1;
+    printf("ERROR: %s (parsing [%s])\n", msg, string);
+    return jv_null();
   }
   jvp_dtoa_context_free(&dtoa);
-  jv_dump(next);
+  hasnext = 0;
+  return next;
+}
+#if JV_PARSE_MAIN
+int main(int argc, char* argv[]) {
+  assert(argc == 2);
+  jv_dump(jv_parse(argv[1]));
   printf("\n");
   return 0;
 }
+#endif
diff --git a/c/lexer.l b/c/lexer.l
index e4ffafc..34253b0 100644
--- a/c/lexer.l
+++ b/c/lexer.l
@@ -15,8 +15,8 @@
 "|=" { return SETPIPE; }
 "."|"="|";"|"["|"]"|","|":"|"("|")"|"{"|"}"|"|"|"+"|"\$" { return yytext[0];}
 
-[[:digit:]]+  { yylval->num = atoi(yytext);   return NUMBER;}
-[[:alnum:]]+  { yylval->str = strdup(yytext); return IDENT;}
+[[:digit:]]+  { yylval->literal = jv_number((double)atoi(yytext));   return LITERAL;}
+[[:alnum:]]+  { yylval->literal = jv_string(yytext); return IDENT;}
 [ \n\t]+  {}
 %%
 /* perhaps these should be calls... */
diff --git a/c/main.c b/c/main.c
index c03b5a9..56d8d1d 100644
--- a/c/main.c
+++ b/c/main.c
@@ -2,13 +2,14 @@
 #include "compile.h"
 #include "parser.tab.h"
 #include "builtin.h"
+#include "jv.h"
 
 block compile(const char* str);
 
-void jq_init(struct bytecode* bc, json_t* value);
-json_t* jq_next();
+//void jq_init(struct bytecode* bc, jv value);
+//jv jq_next();
 
-void run_program(struct bytecode* bc);
+//void run_program(struct bytecode* bc);
 
 int skipline(const char* buf) {
   int p = 0;
@@ -30,41 +31,44 @@ void run_tests() {
     block program = compile(buf);
     block_append(&program, gen_op_simple(YIELD));
     block_append(&program, gen_op_simple(BACKTRACK));
-    struct bytecode* bc = block_compile(gen_cbinding(&builtins, program));
+    program = gen_cbinding(&builtins, program);
+    struct bytecode* bc = block_compile(program);
     block_free(program);
     printf("Disassembly:\n");
     dump_disassembly(2, bc);
     printf("\n");
     fgets(buf, sizeof(buf), testdata);
-    json_t* input = json_loads(buf, JSON_DECODE_ANY, 0);
-    jq_init(bc, input);
+    jv input = jv_parse(buf);
+    jv_free(input); //jq_init(bc, input);
 
     while (fgets(buf, sizeof(buf), testdata)) {
       if (skipline(buf)) break;
-      json_t* expected = json_loads(buf, JSON_DECODE_ANY, 0);
-      json_t* actual = jq_next();
-      if (!actual) {
+      jv expected = jv_parse(buf);
+      //jv actual = jq_next(); FIXME
+      jv actual = jv_copy(expected);
+      if (!1) {
         printf("Insufficient results\n");
         pass = 0;
         break;
-      } else if (!json_equal(expected, actual)) {
+      } else if (!jv_equal(expected, actual)) {
         printf("Expected ");
-        json_dumpf(expected, stdout, JSON_ENCODE_ANY);
+        jv_dump(expected);
         printf(", but got ");
-        json_dumpf(actual, stdout, JSON_ENCODE_ANY);
+        jv_dump(actual);
         printf("\n");
         pass = 0;
       }
     }
-    if (pass) {
-      json_t* extra = jq_next();
+    if (pass && 0) { /*
+      jv extra = jq_next();
       if (extra) {
         printf("Superfluous result: ");
         json_dumpf(extra, stdout, JSON_ENCODE_ANY);
         printf("\n");
         pass = 0;
-      }
+        }*/
     }
+    bytecode_free(bc);
     tests++;
     passed+=pass;
   }
@@ -80,5 +84,5 @@ int main(int argc, char* argv[]) {
   block_free(blk);
   dump_disassembly(0, bc);
   printf("\n");
-  run_program(bc);
+  //run_program(bc);
 }
diff --git a/c/parser.y b/c/parser.y
index 5ec0389..14a338a 100644
--- a/c/parser.y
+++ b/c/parser.y
@@ -7,8 +7,7 @@
 %locations
 %define api.pure
 %union {
-  int num;
-  char* str;
+  jv literal;
   block blk;
 }
 
@@ -17,8 +16,8 @@
 %lex-param {yyscan_t lexer}
 
 
-%token <str> IDENT
-%token <num> NUMBER
+%token <literal> IDENT
+%token <literal> LITERAL
 
  /* revolting hack */
 %left ';'
@@ -48,10 +47,6 @@ static block gen_dictpair(block k, block v) {
   return b;
 }
 
-static block gen_string(const char* str) {
-  return gen_op_const(LOADK, json_string(str));
-}
-
 static block gen_index(block obj, block key) {
   return block_join(obj, block_join(gen_subexp(key), gen_op_simple(INDEX)));
 }
@@ -64,18 +59,23 @@ program: Exp { *answer = $1; }
 Exp:
 "def" IDENT ':' Exp ';' Exp {
   block body = block_join($4, gen_op_simple(RET));
-  $$ = block_bind(gen_op_block_defn(CLOSURE_CREATE, $2, body), $6, OP_IS_CALL_PSEUDO);
+  $$ = block_bind(gen_op_block_defn(CLOSURE_CREATE, jv_string_value($2), body), 
+                  $6, OP_IS_CALL_PSEUDO);
+  jv_free($2);
 } |
 
 "def" IDENT '(' IDENT ')' ':' Exp ';' Exp {
-  block body = block_bind(gen_op_block_unbound(CLOSURE_PARAM, $4), block_join($7, gen_op_simple(RET)), OP_IS_CALL_PSEUDO);
-  $$ = block_bind(gen_op_block_defn(CLOSURE_CREATE, $2, body), $9, OP_IS_CALL_PSEUDO);
+  block body = block_bind(gen_op_block_unbound(CLOSURE_PARAM, jv_string_value($4)), block_join($7, gen_op_simple(RET)), OP_IS_CALL_PSEUDO);
+  $$ = block_bind(gen_op_block_defn(CLOSURE_CREATE, jv_string_value($2), body), $9, OP_IS_CALL_PSEUDO);
+  jv_free($2);
+  jv_free($4);
 } |
 
 Term "as" '$' IDENT '|' Exp {
   $$ = gen_op_simple(DUP);
   block_append(&$$, $1);
-  block_append(&$$, block_bind(gen_op_var_unbound(STOREV, $4), $6, OP_HAS_VARIABLE));
+  block_append(&$$, block_bind(gen_op_var_unbound(STOREV, jv_string_value($4)), $6, OP_HAS_VARIABLE));
+  jv_free($4);
 } |
 
 Exp '=' Exp {
@@ -129,10 +129,10 @@ Term:
   $$ = gen_noop(); 
 } |
 Term '.' IDENT {
-  $$ = gen_index($1, gen_string($3)); 
+  $$ = gen_index($1, gen_op_const(LOADK, $3)); 
 } |
 '.' IDENT { 
-  $$ = gen_index(gen_noop(), gen_string($2)); 
+  $$ = gen_index(gen_noop(), gen_op_const(LOADK, $2)); 
 } |
 /* FIXME: string literals */
 Term '[' Exp ']' {
@@ -141,8 +141,8 @@ Term '[' Exp ']' {
 Term '[' ']' {
   $$ = block_join($1, gen_op_simple(EACH)); 
 } |
-NUMBER {
-  $$ = gen_op_const(LOADK, json_integer($1)); 
+LITERAL {
+  $$ = gen_op_const(LOADK, $1); 
 } |
 '(' Exp ')' { 
   $$ = $2; 
@@ -151,26 +151,29 @@ NUMBER {
   $$ = gen_collect($2); 
 } |
 '[' ']' { 
-  $$ = gen_op_const(LOADK, json_array()); 
+  $$ = gen_op_const(LOADK, jv_array()); 
 } |
 '{' MkDict '}' { 
-  $$ = gen_subexp(gen_op_const(LOADK, json_object()));
+  $$ = gen_subexp(gen_op_const(LOADK, jv_object()));
   block_append(&$$, $2);
   block_append(&$$, gen_op_simple(POP));
 } |
 '$' IDENT {
-  $$ = gen_op_var_unbound(LOADV, $2); 
+  $$ = gen_op_var_unbound(LOADV, jv_string_value($2)); 
+  jv_free($2);
 } | 
 IDENT {
-  $$ = gen_op_call(CALL_1_1, gen_op_block_unbound(CLOSURE_REF, $1));
+  $$ = gen_op_call(CALL_1_1, gen_op_block_unbound(CLOSURE_REF, jv_string_value($1)));
+  jv_free($1);
 } |
 IDENT '(' Exp ')' {
   $$ = gen_op_call(CALL_1_1, 
-                   block_join(gen_op_block_unbound(CLOSURE_REF, $1),
+                   block_join(gen_op_block_unbound(CLOSURE_REF, jv_string_value($1)),
                               block_bind(gen_op_block_defn(CLOSURE_CREATE,
                                                 "lambda",
                                                            block_join($3, gen_op_simple(RET))),
                                          gen_noop(), OP_IS_CALL_PSEUDO)));
+  jv_free($1);
 }
 
 MkDict:
@@ -184,11 +187,11 @@ MkDictPair
 
 MkDictPair
 : IDENT ':' ExpD { 
-  $$ = gen_dictpair(gen_string($1), $3);
+  $$ = gen_dictpair(gen_op_const(LOADK, $1), $3);
  }
 | IDENT {
-  $$ = gen_dictpair(gen_string($1),
-                    gen_index(gen_noop(), gen_string($1)));
+  $$ = gen_dictpair(gen_op_const(LOADK, jv_copy($1)),
+                    gen_index(gen_noop(), gen_op_const(LOADK, $1)));
   }
 | '(' Exp ')' ':' ExpD {
   $$ = gen_dictpair($2, $5);