tests/tests_pcre2_compile_class_compile_class_binary_tight.c

#include "unity/unity.h"
#define PCRE2_CODE_UNIT_WIDTH 8
#include "pcre2.h"

/* We need internal ECLASS tokens/types and eclass_op_info. */
#include "pcre2_compile.h"

#include <string.h>
#include <stdint.h>

/* The wrapper is defined in the module, but is not declared in a header. */
extern BOOL test_compile_class_binary_tight();

void setUp(void) {
  /* nothing */
}

void tearDown(void) {
  /* nothing */
}

/* Helper: assert bitmap is all ones (0xFF..). */
static void assert_bits_all_ones(const eclass_op_info *op)
{
  for (int i = 0; i < 8; i++) {
    TEST_ASSERT_EQUAL_HEX32(0xFFFFFFFFu, op->bits.classwords[i]);
  }
}

/* Helper: assert bitmap is all zeros. */
static void assert_bits_all_zeros(const eclass_op_info *op)
{
  for (int i = 0; i < 8; i++) {
    TEST_ASSERT_EQUAL_HEX32(0u, op->bits.classwords[i]);
  }
}

/* Case 1: single operand (META_CLASS_EMPTY_NOT) -> ECL_ANY with all-ones bitmap. */
void test_compile_class_binary_tight_single_any(void)
{
  /* Tokens: [META_CLASS_EMPTY_NOT] then end. */
  uint32_t tokens[] = { META_CLASS_EMPTY_NOT, META_CLASS_END };
  uint32_t *ptr = tokens;
  uint32_t *pptr = ptr;
  PCRE2_UCHAR codebuf[16];
  PCRE2_UCHAR *code = codebuf;
  eclass_op_info op_info;
  memset(&op_info, 0, sizeof(op_info));

  BOOL ok = test_compile_class_binary_tight(NULL, /*negated=*/FALSE,
                                            &pptr, &code, &op_info,
                                            /*lengthptr=*/NULL);
  TEST_ASSERT_TRUE(ok);

  /* One byte written: ECL_ANY. */
  TEST_ASSERT_EQUAL_PTR(codebuf + 1, code);
  TEST_ASSERT_EQUAL_UINT8(ECL_ANY, codebuf[0]);
  TEST_ASSERT_EQUAL_UINT8(ECL_ANY, op_info.op_single_type);
  TEST_ASSERT_EQUAL_UINT(1, op_info.length);
  assert_bits_all_ones(&op_info);

  /* pptr advanced to META_CLASS_END */
  TEST_ASSERT_EQUAL_PTR(&tokens[1], pptr);
  TEST_ASSERT_EQUAL_UINT(META_CLASS_END, *pptr);
}

/* Case 2: ANY && ANY -> ANY (no operator emitted; RHS dropped). */
void test_compile_class_binary_tight_and_any_any(void)
{
  uint32_t tokens[] = { META_CLASS_EMPTY_NOT, META_ECLASS_AND, META_CLASS_EMPTY_NOT, META_CLASS_END };
  uint32_t *pptr = tokens;
  PCRE2_UCHAR codebuf[16];
  PCRE2_UCHAR *code = codebuf;
  eclass_op_info op_info;
  memset(&op_info, 0, sizeof(op_info));

  BOOL ok = test_compile_class_binary_tight(NULL, FALSE, &pptr, &code, &op_info, NULL);
  TEST_ASSERT_TRUE(ok);

  /* Still a single byte: ECL_ANY. */
  TEST_ASSERT_EQUAL_PTR(codebuf + 1, code);
  TEST_ASSERT_EQUAL_UINT8(ECL_ANY, codebuf[0]);
  TEST_ASSERT_EQUAL_UINT8(ECL_ANY, op_info.op_single_type);
  TEST_ASSERT_EQUAL_UINT(1, op_info.length);
  assert_bits_all_ones(&op_info);

  /* pptr should now point to META_CLASS_END */
  TEST_ASSERT_EQUAL_PTR(&tokens[3], pptr);
  TEST_ASSERT_EQUAL_UINT(META_CLASS_END, *pptr);
}

/* Case 3: NONE && ANY -> NONE. Use META_CLASS_EMPTY (NONE) and META_CLASS_EMPTY_NOT (ANY). */
void test_compile_class_binary_tight_and_none_any(void)
{
  uint32_t tokens[] = { META_CLASS_EMPTY, META_ECLASS_AND, META_CLASS_EMPTY_NOT, META_CLASS_END };
  uint32_t *pptr = tokens;
  PCRE2_UCHAR codebuf[16];
  PCRE2_UCHAR *code = codebuf;
  eclass_op_info op_info;
  memset(&op_info, 0, sizeof(op_info));

  BOOL ok = test_compile_class_binary_tight(NULL, FALSE, &pptr, &code, &op_info, NULL);
  TEST_ASSERT_TRUE(ok);

  TEST_ASSERT_EQUAL_PTR(codebuf + 1, code);
  TEST_ASSERT_EQUAL_UINT8(ECL_NONE, codebuf[0]);
  TEST_ASSERT_EQUAL_UINT8(ECL_NONE, op_info.op_single_type);
  TEST_ASSERT_EQUAL_UINT(1, op_info.length);
  assert_bits_all_zeros(&op_info);

  TEST_ASSERT_EQUAL_PTR(&tokens[3], pptr);
  TEST_ASSERT_EQUAL_UINT(META_CLASS_END, *pptr);
}

/* Case 4: negated context flips AND to OR semantics.
   Input: EMPTY && EMPTY, negated=true
   EMPTY -> NONE when un-negated; with negated=true, each operand becomes ANY.
   ANY || ANY -> ANY. */
void test_compile_class_binary_tight_negated_transforms_and_to_or(void)
{
  uint32_t tokens[] = { META_CLASS_EMPTY, META_ECLASS_AND, META_CLASS_EMPTY, META_CLASS_END };
  uint32_t *pptr = tokens;
  PCRE2_UCHAR codebuf[16];
  PCRE2_UCHAR *code = codebuf;
  eclass_op_info op_info;
  memset(&op_info, 0, sizeof(op_info));

  BOOL ok = test_compile_class_binary_tight(NULL, /*negated=*/TRUE, &pptr, &code, &op_info, NULL);
  TEST_ASSERT_TRUE(ok);

  TEST_ASSERT_EQUAL_PTR(codebuf + 1, code);
  TEST_ASSERT_EQUAL_UINT8(ECL_ANY, codebuf[0]);
  TEST_ASSERT_EQUAL_UINT8(ECL_ANY, op_info.op_single_type);
  TEST_ASSERT_EQUAL_UINT(1, op_info.length);
  assert_bits_all_ones(&op_info);

  TEST_ASSERT_EQUAL_PTR(&tokens[3], pptr);
  TEST_ASSERT_EQUAL_UINT(META_CLASS_END, *pptr);
}

/* Case 5: length-only mode. Ensure *lengthptr increases by operand bytes and no code is written. */
void test_compile_class_binary_tight_length_only_mode(void)
{
  uint32_t tokens[] = { META_CLASS_EMPTY_NOT, META_ECLASS_AND, META_CLASS_EMPTY_NOT, META_CLASS_END };
  uint32_t *pptr = tokens;
  PCRE2_UCHAR codebuf[16];
  memset(codebuf, 0xA5, sizeof(codebuf));
  PCRE2_UCHAR *code = codebuf; /* Should remain unchanged in length-only mode. */
  eclass_op_info op_info;
  memset(&op_info, 0, sizeof(op_info));
  PCRE2_SIZE length = 0;

  BOOL ok = test_compile_class_binary_tight(NULL, FALSE, &pptr, &code, &op_info, &length);
  TEST_ASSERT_TRUE(ok);

  /* Two operands each contribute 1 unit; no operator emitted after folding. */
  TEST_ASSERT_EQUAL_UINT(2, length);
  /* Code pointer should be unchanged when lengthptr != NULL. */
  TEST_ASSERT_EQUAL_PTR(codebuf, code);
  /* op_info still reflects the folded result (ANY, all ones). */
  TEST_ASSERT_EQUAL_UINT8(ECL_ANY, op_info.op_single_type);
  assert_bits_all_ones(&op_info);

  TEST_ASSERT_EQUAL_PTR(&tokens[3], pptr);
  TEST_ASSERT_EQUAL_UINT(META_CLASS_END, *pptr);
}

int main(void)
{
  UNITY_BEGIN();
  RUN_TEST(test_compile_class_binary_tight_single_any);
  RUN_TEST(test_compile_class_binary_tight_and_any_any);
  RUN_TEST(test_compile_class_binary_tight_and_none_any);
  RUN_TEST(test_compile_class_binary_tight_negated_transforms_and_to_or);
  RUN_TEST(test_compile_class_binary_tight_length_only_mode);
  return UNITY_END();
}