add combinators h_skip, h_seek, h_tell

src/SConscript

@@ -52,7 +52,8 @@ parsers = ['parsers/%s.c'%s for s in
             'unimplemented',
             'whitespace',
             'xor',
-            'value']]
+            'value',
+            'seek']]
 
 backends = ['backends/%s.c' % s for s in
             ['packrat', 'llk', 'regex', 'glr', 'lalr', 'lr', 'lr0']]

src/bitreader.c

@@ -108,3 +108,77 @@ int64_t h_read_bits(HInputStream* state, int count, char signed_p) {
   out <<= final_shift;
   return (out ^ msb) - msb; // perform sign extension
 }
+
+void h_skip_bits(HInputStream* stream, size_t count) {
+  size_t left;
+
+  if (count == 0)
+    return;
+
+  if (stream->overrun)
+    return;
+
+  if (stream->index == stream->length) {
+    stream->overrun = true;
+    return;
+  }
+
+  // consume from a partial byte?
+  left = 8 - stream->bit_offset - stream->margin;
+  if (count < left) {
+    stream->bit_offset += count;
+    return;
+  }
+  if (left < 8) {
+    stream->index += 1;
+    stream->bit_offset = 0;
+    stream->margin = 0;
+    count -= left;
+  }
+  assert(stream->bit_offset == 0);
+  assert(stream->margin == 0);
+
+  // consume full bytes
+  left = stream->length - stream->index;
+  if (count / 8 <= left) {
+    stream->index += count / 8;
+    count = count % 8;
+  } else {
+    stream->index = stream->length;
+    stream->overrun = true;
+    return;
+  }
+  assert(count < 8);
+
+  // final partial byte
+  if (count > 0 && stream->index == stream->length)
+    stream->overrun = true;
+  else
+    stream->bit_offset = count;
+}
+
+void h_seek_bits(HInputStream* stream, size_t pos) {
+  size_t pos_index = pos / 8;
+  size_t pos_offset = pos % 8;
+
+  /* seek within the current byte? */
+  if (pos_index == stream->index) {
+    stream->bit_offset = pos_offset;
+    return;
+  }
+
+  stream->margin = 0;
+
+  /* seek past the end? */
+  if ((pos_index > stream->length) ||
+      (pos_index == stream->length && pos_offset > 0)) {
+    stream->index = stream->length;
+    stream->bit_offset = 0;
+    stream->overrun = true;
+    return;
+  }
+
+  stream->index = pos_index;
+  stream->bit_offset = pos_offset;
+  stream->margin = 0;
+}

src/hammer.h

@@ -716,6 +716,32 @@ HAMMER_FN_DECL(HParser*, h_get_value, const char* name);
  */
 HAMMER_FN_DECL(HParser*, h_bind, const HParser *p, HContinuation k, void *env);
 
+/**
+ * This parser skips 'n' bits of input.
+ *
+ * Result: None. The HParseResult exists but its AST is NULL.
+ */
+HAMMER_FN_DECL(HParser*, h_skip, size_t n);
+
+/**
+ * The HParser equivalent of fseek(), 'h_seek' modifies the parser's input
+ * position.  Note that contrary to 'fseek', offsets are in bits, not bytes.
+ * The 'whence' argument uses the same values and semantics: SEEK_SET,
+ * SEEK_CUR, SEEK_END.
+ *
+ * Fails if the new input position would be negative or past the end of input.
+ *
+ * Result: TT_UINT. The new input position.
+ */
+HAMMER_FN_DECL(HParser*, h_seek, ssize_t offset, int whence);
+
+/**
+ * Report the current position in bits. Consumes no input.
+ *
+ * Result: TT_UINT. The current input position.
+ */
+HAMMER_FN_DECL_NOARG(HParser*, h_tell);
+
 /**
  * Free the memory allocated to an HParseResult when it is no longer needed.
  */

src/internal.h

@@ -327,9 +327,16 @@ extern HParserBackendVTable h__glr_backend_vtable;
 // TODO(thequux): Set symbol visibility for these functions so that they aren't exported.
 
 int64_t h_read_bits(HInputStream* state, int count, char signed_p);
+void h_skip_bits(HInputStream* state, size_t count);
+void h_seek_bits(HInputStream* state, size_t pos);
 static inline size_t h_input_stream_pos(HInputStream* state) {
+  assert(state->index < SIZE_MAX / 8);
   return state->index * 8 + state->bit_offset + state->margin;
 }
+static inline size_t h_input_stream_length(HInputStream *state) {
+  assert(state->length <= SIZE_MAX / 8);
+  return state->length * 8;
+}
 // need to decide if we want to make this public. 
 HParseResult* h_do_parse(const HParser* parser, HParseState *state);
 void put_cached(HParseState *ps, const HParser *p, HParseResult *cached);

src/parsers/seek.c

@@ -0,0 +1,118 @@
+#include "parser_internal.h"
+
+typedef struct {
+  ssize_t offset;
+  int whence;
+} HSeek;
+
+static HParseResult *parse_skip(void *env, HParseState *state)
+{
+  size_t n = (uintptr_t)env;
+
+  h_skip_bits(&state->input_stream, n);
+  return make_result(state->arena, NULL);
+}
+
+static HParseResult *parse_seek(void *env, HParseState *state)
+{
+  HSeek *s = (HSeek *)env;
+  HInputStream *stream = &state->input_stream;
+  size_t pos;
+
+  /* determine base position */
+  switch (s->whence) {
+  case SEEK_SET:
+    pos = 0;
+    break;
+  case SEEK_END:
+    pos = h_input_stream_length(stream);
+    break;
+  case SEEK_CUR:
+    pos = h_input_stream_pos(stream);
+    break;
+  default:
+    return NULL;	/* invalid argument */
+  }
+
+  /* calculate target position and do basic overflow checks */
+  if (s->offset < 0 && (size_t)(- s->offset) > pos)
+    return NULL;	/* underflow */
+  if (s->offset > 0 && SIZE_MAX - s->offset < pos)
+    return NULL;	/* overflow */
+  pos += s->offset;
+
+  /* perform the seek and check for overrun */
+  h_seek_bits(stream, pos);
+  if (stream->overrun)
+    return NULL;
+
+  HParsedToken *tok = a_new(HParsedToken, 1);
+  tok->token_type = TT_UINT;
+  tok->uint = pos;
+  return make_result(state->arena, tok);
+}
+
+static HParseResult *parse_tell(void *env, HParseState *state)
+{
+  HParsedToken *tok = a_new(HParsedToken, 1);
+  tok->token_type = TT_UINT;
+  tok->uint = h_input_stream_pos(&state->input_stream);
+  return make_result(state->arena, tok);
+}
+
+static const HParserVtable skip_vt = {
+  .parse = parse_skip,
+  .isValidRegular = h_false,
+  .isValidCF = h_false,
+  .compile_to_rvm = h_not_regular,
+  .higher = false,
+};
+
+static const HParserVtable seek_vt = {
+  .parse = parse_seek,
+  .isValidRegular = h_false,
+  .isValidCF = h_false,
+  .compile_to_rvm = h_not_regular,
+  .higher = false,
+};
+
+static const HParserVtable tell_vt = {
+  .parse = parse_tell,
+  .isValidRegular = h_false,
+  .isValidCF = h_false,
+  .compile_to_rvm = h_not_regular,
+  .higher = false,
+};
+
+HParser* h_skip(size_t n)
+{
+  return h_skip__m(&system_allocator, n);
+}
+
+HParser *h_skip__m(HAllocator* mm__, size_t n)
+{
+  return h_new_parser(mm__, &skip_vt, (void *)n);
+}
+
+HParser* h_seek(ssize_t offset, int whence)
+{
+  return h_seek__m(&system_allocator, offset, whence);
+}
+
+HParser *h_seek__m(HAllocator* mm__, ssize_t offset, int whence)
+{
+  HSeek *env = h_new(HSeek, 1);
+  env->offset = offset;
+  env->whence = whence;
+  return h_new_parser(mm__, &seek_vt, env);
+}
+
+HParser *h_tell()
+{
+  return h_tell__m(&system_allocator);
+}
+
+HParser *h_tell__m(HAllocator* mm__)
+{
+  return h_new_parser(mm__, &tell_vt, NULL);
+}

src/t_parser.c

@@ -743,6 +743,70 @@ static void test_bind(gconstpointer backend) {
   g_check_parse_failed(p, be, "272{", 4);
 }
 
+static void test_skip(gconstpointer backend) {
+  HParserBackend be = (HParserBackend)GPOINTER_TO_INT(backend);
+  const HParser *p, *p_le, *p_be;
+
+  p = h_sequence(h_ch('a'), h_skip(32), h_ch('f'), NULL);
+  g_check_parse_match(p, be, "abcdef", 6, "(u0x61 u0x66)");
+  g_check_parse_failed(p, be, "abcdex", 6);
+  g_check_parse_failed(p, be, "abc", 3);
+
+  p = h_sequence(h_ch('a'), h_skip(32), h_end_p(), NULL);
+  g_check_parse_match(p, be, "abcde", 5, "(u0x61)");
+
+  p = h_sequence(h_ch('a'), h_skip(3), h_ch('\0'), h_skip(5), h_ch('b'), NULL);
+  g_check_parse_match(p, be, "a\xe0\x1f\x62", 4, "(u0x61 u0 u0x62)"); // big-endian
+  p_le = h_with_endianness(BYTE_LITTLE_ENDIAN|BIT_LITTLE_ENDIAN, p);
+  p_be = h_with_endianness(BYTE_LITTLE_ENDIAN|BIT_BIG_ENDIAN, p);
+  g_check_parse_match(p_be, be, "a\xe0\x1f\x62", 4, "(u0x61 u0 u0x62)");
+  g_check_parse_match(p_le, be, "a\x07\xf8\x62", 4, "(u0x61 u0 u0x62)");
+
+  p = h_sequence(h_ch('a'), h_skip(3), h_ch('\0'), h_skip(5), h_end_p(), NULL);
+  g_check_parse_match(p, be, "a\xe0\x1f", 3, "(u0x61 u0)"); // big-endian
+}
+
+static void test_tell(gconstpointer backend) {
+  HParserBackend be = (HParserBackend)GPOINTER_TO_INT(backend);
+  const HParser *p;
+
+  p = h_sequence(h_ch('a'), h_ch('b'), h_tell(), h_end_p(), NULL);
+  g_check_parse_match(p, be, "ab", 2, "(u0x61 u0x62 u0x10)");
+  g_check_parse_failed(p, be, "abc", 1);
+  g_check_parse_failed(p, be, "a", 1);
+}
+
+static void test_seek(gconstpointer backend) {
+  HParserBackend be = (HParserBackend)GPOINTER_TO_INT(backend);
+  const HParser *p;
+
+  p = h_sequence(h_ch('a'), h_seek(40, SEEK_SET), h_ch('f'), NULL);
+  g_check_parse_match(p, be, "abcdef", 6, "(u0x61 u0x28 u0x66)");
+  g_check_parse_failed(p, be, "abcdex", 6);
+  g_check_parse_failed(p, be, "abc", 3);
+
+  p = h_sequence(h_ch('a'), h_seek(40, SEEK_SET), h_end_p(), NULL);
+  g_check_parse_match(p, be, "abcde", 5, "(u0x61 u0x28)");
+  g_check_parse_failed(p, be, "abcdex", 6);
+  g_check_parse_failed(p, be, "abc", 3);
+
+  p = h_sequence(h_ch('a'), h_seek(0, SEEK_END), h_end_p(), NULL);
+  g_check_parse_match(p, be, "abcde", 5, "(u0x61 u0x28)");
+  g_check_parse_match(p, be, "abc", 3, "(u0x61 u0x18)");
+
+  p = h_sequence(h_ch('a'), h_seek(-16, SEEK_END), h_ch('x'), NULL);
+  g_check_parse_match(p, be, "abcdxy", 6, "(u0x61 u0x20 u0x78)");
+  g_check_parse_match(p, be, "abxy", 4, "(u0x61 u0x10 u0x78)");
+  g_check_parse_failed(p, be, "abc", 3);
+  g_check_parse_failed(p, be, "x", 1);
+
+  p = h_sequence(h_ch('a'), h_seek(32, SEEK_CUR), h_ch('f'), NULL);
+  g_check_parse_match(p, be, "abcdef", 6, "(u0x61 u0x28 u0x66)");
+  g_check_parse_failed(p, be, "xbcdef", 6);
+  g_check_parse_failed(p, be, "abcdex", 6);
+  g_check_parse_failed(p, be, "abc", 3);
+}
+
 void register_parser_tests(void) {
   g_test_add_data_func("/core/parser/packrat/token", GINT_TO_POINTER(PB_PACKRAT), test_token);
   g_test_add_data_func("/core/parser/packrat/ch", GINT_TO_POINTER(PB_PACKRAT), test_ch);
@@ -795,6 +859,9 @@ void register_parser_tests(void) {
   g_test_add_data_func("/core/parser/packrat/bind", GINT_TO_POINTER(PB_PACKRAT), test_bind);
   g_test_add_data_func("/core/parser/packrat/result_length", GINT_TO_POINTER(PB_PACKRAT), test_result_length);
   //g_test_add_data_func("/core/parser/packrat/token_position", GINT_TO_POINTER(PB_PACKRAT), test_token_position);
+  g_test_add_data_func("/core/parser/packrat/skip", GINT_TO_POINTER(PB_PACKRAT), test_skip);
+  g_test_add_data_func("/core/parser/packrat/seek", GINT_TO_POINTER(PB_PACKRAT), test_seek);
+  g_test_add_data_func("/core/parser/packrat/tell", GINT_TO_POINTER(PB_PACKRAT), test_tell);
 
   g_test_add_data_func("/core/parser/llk/token", GINT_TO_POINTER(PB_LLk), test_token);
   g_test_add_data_func("/core/parser/llk/ch", GINT_TO_POINTER(PB_LLk), test_ch);