More error handling - locations of refs to undefined symbols.

c/compile.c

@@ -3,7 +3,7 @@
 #include <stdlib.h>
 #include "opcode.h"
 #include "compile.h"
-
+#include "locfile.h"
 
 struct inst {
   struct inst* next;
@@ -18,6 +18,8 @@ struct inst {
     struct cfunction* cfunc;
   } imm;
 
+  location source;
+
   // Binding
   // An instruction requiring binding (for parameters/variables)
   // is in one of three states:
@@ -45,6 +47,7 @@ static inst* inst_new(opcode op) {
   i->bound_by = 0;
   i->symbol = 0;
   i->subfn = gen_noop();
+  i->source = UNKNOWN_LOCATION;
   return i;
 }
 
@@ -82,6 +85,16 @@ static inst* block_take(block* b) {
   return i;
 }
 
+block gen_location(location loc, block b) {
+  for (inst* i = b.first; i; i = i->next) {
+    if (i->source.start == UNKNOWN_LOCATION.start &&
+        i->source.end == UNKNOWN_LOCATION.end) {
+      i->source = loc;
+    }
+  }
+  return b;
+}
+
 block gen_noop() {
   block b = {0,0};
   return b;
@@ -429,7 +442,8 @@ static int count_cfunctions(block b) {
   return n;
 }
 
-static void compile(struct bytecode* bc, block b) {
+static int compile(struct locfile* locations, struct bytecode* bc, block b) {
+  int errors = 0;
   int pos = 0;
   int var_frame_idx = 0;
   bc->nsubfunctions = 0;
@@ -442,7 +456,10 @@ static void compile(struct bytecode* bc, block b) {
 
     int opflags = opcode_describe(curr->op)->flags;
     if (opflags & OP_HAS_BINDING) {
-      assert(curr->bound_by && "unbound term");
+      if (!curr->bound_by) {
+        locfile_locate(locations, curr->source, "error: %s is not defined", curr->symbol);
+        errors++;
+      }
     }
     if ((opflags & OP_HAS_VARIABLE) &&
         curr->bound_by == curr) {
@@ -472,7 +489,7 @@ static void compile(struct bytecode* bc, block b) {
       bc->subfunctions[curr->imm.intval] = subfn;
       subfn->globals = bc->globals;
       subfn->parent = bc;
-      compile(subfn, curr->subfn);
+      errors += compile(locations, subfn, curr->subfn);
     }
   } else {
     bc->subfunctions = 0;
@@ -483,7 +500,7 @@ static void compile(struct bytecode* bc, block b) {
   pos = 0;
   jv constant_pool = jv_array();
   int maxvar = -1;
-  for (inst* curr = b.first; curr; curr = curr->next) {
+  if (!errors) for (inst* curr = b.first; curr; curr = curr->next) {
     const struct opcode_description* op = opcode_describe(curr->op);
     if (op->length == 0)
       continue;
@@ -551,18 +568,25 @@ static void compile(struct bytecode* bc, block b) {
   }
   bc->constants = constant_pool;
   bc->nlocals = maxvar + 2; // FIXME: frames of size zero?
+  return errors;
 }
 
-struct bytecode* block_compile(block b) {
+int block_compile(block b, struct locfile* locations, struct bytecode** out) {
   struct bytecode* bc = malloc(sizeof(struct bytecode));
   bc->parent = 0;
   bc->globals = malloc(sizeof(struct symbol_table));
   int ncfunc = count_cfunctions(b);
   bc->globals->ncfunctions = 0;
   bc->globals->cfunctions = malloc(sizeof(struct cfunction) * ncfunc);
-  compile(bc, b);
+  int nerrors = compile(locations, bc, b);
   assert(bc->globals->ncfunctions == ncfunc);
-  return bc;
+  if (nerrors > 0) {
+    bytecode_free(bc);
+    *out = 0;
+  } else {
+    *out = bc;
+  }
+  return nerrors;
 }
 
 void block_free(block b) {

c/compile.h

@@ -1,6 +1,7 @@
 #include <stdint.h>
 #include "bytecode.h"
 #include "opcode.h"
+#include "locfile.h"
 
 struct inst;
 typedef struct inst inst;
@@ -11,6 +12,8 @@ typedef struct block {
   inst* last;
 } block;
 
+block gen_location(location, block);
+
 block gen_noop();
 block gen_op_simple(opcode op);
 block gen_op_const(opcode op, jv constant);
@@ -39,7 +42,7 @@ void block_append(block* b, block b2);
 block block_join(block a, block b);
 block block_bind(block binder, block body, int bindflags);
 
-struct bytecode* block_compile(block);
+int block_compile(block, struct locfile*, struct bytecode**);
 
 void block_free(block);
 

c/locfile.h

@@ -1,10 +1,15 @@
+#ifndef _LOCFILE_H
+#define _LOCFILE_H
 #include <stdlib.h>
 #include <stdio.h>
 #include <assert.h>
+#include <stdarg.h>
 typedef struct {
   int start, end;
 } location;
 
+static const location UNKNOWN_LOCATION = {-1, -1};
+
 struct locfile {
   const char* data;
   int length;
@@ -48,15 +53,26 @@ static int locfile_line_length(struct locfile* l, int line) {
   return l->linemap[line+1] - l->linemap[line];
 }
 
-static void locfile_locate(struct locfile* l, location loc) {
+static void locfile_locate(struct locfile* l, location loc, const char* fmt, ...) {
+  va_list fmtargs;
+  va_start(fmtargs, fmt);
+  vfprintf(stderr, fmt, fmtargs);
+  va_end(fmtargs);
+  fprintf(stderr, "\n");
+  if (loc.start == -1) {
+    fprintf(stderr, "<unknown location>\n");
+    return;
+  }
   int startline = locfile_get_line(l, loc.start);
   int offset = l->linemap[startline];
-  printf("%.*s\n", locfile_line_length(l, startline), l->data + offset);
-  printf("%*s", loc.start - offset, "");
+  fprintf(stderr, "%.*s\n", locfile_line_length(l, startline), l->data + offset);
+  fprintf(stderr, "%*s", loc.start - offset, "");
   for (int i = loc.start; 
        i < loc.end && i < offset + locfile_line_length(l, startline);
        i++){
-    printf("^");
+    fprintf(stderr, "^");
   }
-  printf("\n");
+  fprintf(stderr, "\n");
 }
+
+#endif

c/main.c

@@ -1,15 +1,36 @@
 #include <stdio.h>
+#include <string.h>
 #include "compile.h"
 #include "parser.tab.h"
 #include "builtin.h"
 #include "jv.h"
+#include "locfile.h"
 
-int compile(const char* str, block* answer);
+int jq_parse(struct locfile* source, block* answer);
 
 void jq_init(struct bytecode* bc, jv value);
 jv jq_next();
 void jq_teardown();
 
+struct bytecode* jq_compile(const char* str) {
+  struct locfile locations;
+  locfile_init(&locations, str, strlen(str));
+  block program;
+  struct bytecode* bc = 0;
+  int nerrors = jq_parse(&locations, &program);
+  if (nerrors == 0) {
+    block_append(&program, block_join(gen_op_simple(YIELD), gen_op_simple(BACKTRACK)));
+    program = gen_cbinding(&builtins, program);
+    nerrors = block_compile(program, &locations, &bc);
+    block_free(program);
+  }
+  if (nerrors) {
+    fprintf(stderr, "%d compile %s\n", nerrors, nerrors > 1 ? "errors" : "error");
+  }
+  locfile_free(&locations);
+  return bc;
+}
+
 
 void run_program(struct bytecode* bc) {
 #if JQ_DEBUG
@@ -52,14 +73,8 @@ void run_tests() {
     if (skipline(buf)) continue;
     printf("Testing %s\n", buf);
     int pass = 1;
-    block program;
-    int nerrors = compile(buf, &program);
-    assert(nerrors == 0);
-    block_append(&program, gen_op_simple(YIELD));
-    block_append(&program, gen_op_simple(BACKTRACK));
-    program = gen_cbinding(&builtins, program);
-    struct bytecode* bc = block_compile(program);
-    block_free(program);
+    struct bytecode* bc = jq_compile(buf);
+    assert(bc);
     printf("Disassembly:\n");
     dump_disassembly(2, bc);
     printf("\n");
@@ -111,16 +126,8 @@ void run_tests() {
 
 int main(int argc, char* argv[]) {
   if (argc == 1) { run_tests(); return 0; }
-  block blk;
-  int nerrors = compile(argv[1], &blk);
-  if (nerrors > 0) {
-    printf("%d compile %s\n", nerrors, nerrors > 1 ? "errors" : "error");
-    return 1;
-  }
-  block_append(&blk, block_join(gen_op_simple(YIELD), gen_op_simple(BACKTRACK)));
-  blk = gen_cbinding(&builtins, blk);
-  struct bytecode* bc = block_compile(blk);
-  block_free(blk);
+  struct bytecode* bc = jq_compile(argv[1]);
+  if (!bc) return 1;
   run_program(bc);
   bytecode_free(bc);
   return 0;

c/parser.y

@@ -83,8 +83,7 @@
 void yyerror(YYLTYPE* loc, block* answer, int* errors, 
              struct locfile* locations, yyscan_t lexer, const char *s){
   (*errors)++;
-  printf("error: %s\n", s);
-  locfile_locate(locations, *loc);
+  locfile_locate(locations, *loc, "error: %s", s);
 }
 
 int yylex(YYSTYPE* yylval, YYLTYPE* yylloc, block* answer, int* errors, 
@@ -293,11 +292,11 @@ LITERAL {
   block_append(&$$, gen_op_simple(POP));
 } |
 '$' IDENT {
-  $$ = gen_op_var_unbound(LOADV, jv_string_value($2)); 
+  $$ = gen_location(@$, gen_op_var_unbound(LOADV, jv_string_value($2)));
   jv_free($2);
 } | 
 IDENT {
-  $$ = gen_op_call(CALL_1_1, gen_op_block_unbound(CLOSURE_REF, jv_string_value($1)));
+  $$ = gen_location(@$, gen_op_call(CALL_1_1, gen_op_block_unbound(CLOSURE_REF, jv_string_value($1))));
   jv_free($1);
 } |
 IDENT '(' Exp ')' {
@@ -307,6 +306,7 @@ IDENT '(' Exp ')' {
                                                 "lambda",
                                                            block_join($3, gen_op_simple(RET))),
                                          gen_noop(), OP_IS_CALL_PSEUDO)));
+  $$ = gen_location(@1, $$);
   jv_free($1);
 } |
 '(' error ')' { $$ = gen_noop(); } |
@@ -342,17 +342,14 @@ MkDictPair
 | '(' error ')' ':' ExpD { $$ = $5; }
 %%
 
-int compile(const char* str, block* answer) {
+int jq_parse(struct locfile* locations, block* answer) {
   yyscan_t scanner;
   YY_BUFFER_STATE buf;
   jq_yylex_init_extra(0, &scanner);
-  buf = jq_yy_scan_string(str, scanner);
+  buf = jq_yy_scan_bytes(locations->data, locations->length, scanner);
   int errors = 0;
-  struct locfile locations;
-  locfile_init(&locations, str, strlen(str));
   *answer = gen_noop();
-  yyparse(answer, &errors, &locations, scanner);
-  locfile_free(&locations);
+  yyparse(answer, &errors, locations, scanner);
   jq_yy_delete_buffer(buf, scanner);
   jq_yylex_destroy(scanner);
   if (errors > 0) {