tests/tests_HTMLparser_htmlEncodeEntities.c

#include "unity/unity.h"
#include <libxml/HTMLparser.h>
#include <libxml/parser.h>
#include <string.h>
#include <stdlib.h>
#include <stdio.h>

void setUp(void) {
    xmlInitParser();
}
void tearDown(void) {
    xmlCleanupParser();
}

/* Helper to compare output buffer to expected C string.
   Ensures null-termination locally without relying on function behavior. */
static void assert_output_equals(const unsigned char *out, int outlen, const char *expected) {
    size_t explen = strlen(expected);
    TEST_ASSERT_EQUAL_UINT32_MESSAGE((uint32_t)explen, (uint32_t)outlen, "Output length mismatch");
    char *tmp = (char *)malloc(outlen + 1);
    TEST_ASSERT_NOT_NULL(tmp);
    memcpy(tmp, out, outlen);
    tmp[outlen] = '\0';
    TEST_ASSERT_EQUAL_STRING(expected, tmp);
    free(tmp);
}

/* 1) Null argument checks */
void test_htmlEncodeEntities_null_args(void) {
    unsigned char outbuf[32];
    int outlen = (int)sizeof(outbuf);
    const unsigned char inbuf[] = "test";
    int inlen = 4;
    int rc;

    rc = htmlEncodeEntities(NULL, &outlen, inbuf, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(-1, rc);

    outlen = (int)sizeof(outbuf); inlen = 4;
    rc = htmlEncodeEntities(outbuf, NULL, inbuf, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(-1, rc);

    outlen = (int)sizeof(outbuf); inlen = 4;
    rc = htmlEncodeEntities(outbuf, &outlen, NULL, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(-1, rc);

    outlen = (int)sizeof(outbuf); inlen = 4;
    rc = htmlEncodeEntities(outbuf, &outlen, inbuf, NULL, 0);
    TEST_ASSERT_EQUAL_INT(-1, rc);
}

/* 2) ASCII passthrough without special chars */
void test_htmlEncodeEntities_ascii_passthrough(void) {
    const char *in = "Hello world!";
    int inlen = (int)strlen(in);
    unsigned char outbuf[128];
    int outlen = (int)sizeof(outbuf);

    int rc = htmlEncodeEntities(outbuf, &outlen, (const unsigned char*)in, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(0, rc);
    TEST_ASSERT_EQUAL_INT((int)strlen(in), outlen);
    TEST_ASSERT_EQUAL_INT((int)strlen(in), inlen);
    assert_output_equals(outbuf, outlen, in);
}

/* 3) Special ASCII chars &, <, > */
void test_htmlEncodeEntities_special_ascii_entities(void) {
    const char *in = "&<>";
    int inlen = (int)strlen(in);
    unsigned char outbuf[64];
    int outlen = (int)sizeof(outbuf);

    int rc = htmlEncodeEntities(outbuf, &outlen, (const unsigned char*)in, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(0, rc);
    assert_output_equals(outbuf, outlen, "&amp;&lt;&gt;");
    TEST_ASSERT_EQUAL_INT(3, inlen);
}

/* 4) Quotes handling with quoteChar = '\"' */
void test_htmlEncodeEntities_double_quote_encoded(void) {
    const char inbytes[] = {'\'', '\"'};
    int inlen = 2;
    unsigned char outbuf[64];
    int outlen = (int)sizeof(outbuf);
    int rc = htmlEncodeEntities(outbuf, &outlen, (const unsigned char*)inbytes, &inlen, '\"');
    TEST_ASSERT_EQUAL_INT(0, rc);
    assert_output_equals(outbuf, outlen, "'&quot;");
    TEST_ASSERT_EQUAL_INT(2, inlen);
}

/* 5) Single quote encoded when quoteChar = '\'' (either &apos; or numeric) */
void test_htmlEncodeEntities_single_quote_encoded(void) {
    const char inbytes[] = {'\''};
    int inlen = 1;
    unsigned char outbuf[64];
    int outlen = (int)sizeof(outbuf);
    int rc = htmlEncodeEntities(outbuf, &outlen, (const unsigned char*)inbytes, &inlen, '\'');
    TEST_ASSERT_EQUAL_INT(0, rc);

    /* Convert output to C string */
    char tmp[64];
    memcpy(tmp, outbuf, outlen);
    tmp[outlen] = '\0';

    /* Accept either &apos; or &#39; depending on entity table */
    int is_apos = (strcmp(tmp, "&apos;") == 0);
    int is_num = (strcmp(tmp, "&#39;") == 0);
    TEST_ASSERT_TRUE_MESSAGE(is_apos || is_num, "Expected &apos; or &#39; for single quote");
    TEST_ASSERT_EQUAL_INT(1, inlen);
}

/* 6) Non-ASCII with named entity: © -> &copy; */
void test_htmlEncodeEntities_named_entity_copyright(void) {
    const unsigned char inbytes[] = {0xC2, 0xA9}; /* U+00A9 © */
    int inlen = 2;
    unsigned char outbuf[64];
    int outlen = (int)sizeof(outbuf);

    int rc = htmlEncodeEntities(outbuf, &outlen, inbytes, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(0, rc);
    assert_output_equals(outbuf, outlen, "&copy;");
    TEST_ASSERT_EQUAL_INT(2, inlen);
}

/* 7) Non-ASCII without named entity: 🙂 -> &#128578; */
void test_htmlEncodeEntities_numeric_entity_smiley(void) {
    const unsigned char inbytes[] = {0xF0, 0x9F, 0x99, 0x82}; /* U+1F642 🙂 */
    int inlen = 4;
    unsigned char outbuf[64];
    int outlen = (int)sizeof(outbuf);

    int rc = htmlEncodeEntities(outbuf, &outlen, inbytes, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(0, rc);
    assert_output_equals(outbuf, outlen, "&#128578;");
    TEST_ASSERT_EQUAL_INT(4, inlen);
}

/* 8) Output buffer too small for any progress on entity */
void test_htmlEncodeEntities_output_too_small_no_progress(void) {
    const char *in = "&";
    int inlen = 1;
    unsigned char outbuf[1];
    int outlen = 1;

    int rc = htmlEncodeEntities(outbuf, &outlen, (const unsigned char*)in, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(0, rc);
    TEST_ASSERT_EQUAL_INT(0, outlen);
    TEST_ASSERT_EQUAL_INT(0, inlen);
}

/* 9) Partial progress: enough for 'A' but not for '&' entity */
void test_htmlEncodeEntities_partial_progress_due_to_output_limit(void) {
    const char *in = "A&B";
    int inlen = 3;
    unsigned char outbuf[1]; /* fits only 'A' */
    int outlen = 1;

    int rc = htmlEncodeEntities(outbuf, &outlen, (const unsigned char*)in, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(0, rc);
    TEST_ASSERT_EQUAL_INT(1, outlen);
    TEST_ASSERT_EQUAL_INT(1, inlen);
    TEST_ASSERT_EQUAL_UINT8('A', outbuf[0]);
}

/* 10) Malformed UTF-8: trailing byte in leading position (0x80) */
void test_htmlEncodeEntities_malformed_trailing_in_leading_pos(void) {
    const unsigned char inbytes[] = {0x80};
    int inlen = 1;
    unsigned char outbuf[16];
    int outlen = (int)sizeof(outbuf);

    int rc = htmlEncodeEntities(outbuf, &outlen, inbytes, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(-2, rc);
    TEST_ASSERT_EQUAL_INT(0, outlen);
    TEST_ASSERT_EQUAL_INT(0, inlen);
}

/* 11) Truncated multibyte sequence at end: 'A' then 0xC2 (incomplete) */
void test_htmlEncodeEntities_truncated_multibyte_at_end(void) {
    const unsigned char inbytes[] = {'A', 0xC2};
    int inlen = 2;
    unsigned char outbuf[16];
    int outlen = (int)sizeof(outbuf);

    int rc = htmlEncodeEntities(outbuf, &outlen, inbytes, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(0, rc);
    TEST_ASSERT_EQUAL_INT(1, outlen);
    TEST_ASSERT_EQUAL_INT(1, inlen);
    TEST_ASSERT_EQUAL_UINT8('A', outbuf[0]);
}

/* 12) Invalid trailing byte in multibyte sequence: 'A' 0xC2 'A' */
void test_htmlEncodeEntities_invalid_trailing_byte(void) {
    const unsigned char inbytes[] = {'A', 0xC2, 'A'};
    int inlen = 3;
    unsigned char outbuf[16];
    int outlen = (int)sizeof(outbuf);

    int rc = htmlEncodeEntities(outbuf, &outlen, inbytes, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(-2, rc);
    TEST_ASSERT_EQUAL_INT(1, outlen); /* only 'A' copied before error */
    TEST_ASSERT_EQUAL_INT(1, inlen);  /* only first 'A' consumed */
}

/* 13) Mixed content: pass ASCII, encode specials and non-ASCII */
void test_htmlEncodeEntities_mixed_content(void) {
    /* "A&©<B>🙂" */
    const unsigned char inbytes[] = {'A', '&', 0xC2, 0xA9, '<', 'B', '>', 0xF0, 0x9F, 0x99, 0x82};
    int inlen = (int)sizeof(inbytes);
    unsigned char outbuf[128];
    int outlen = (int)sizeof(outbuf);

    int rc = htmlEncodeEntities(outbuf, &outlen, inbytes, &inlen, 0);
    TEST_ASSERT_EQUAL_INT(0, rc);
    assert_output_equals(outbuf, outlen, "A&amp;&copy;&lt;B&gt;&#128578;");
    TEST_ASSERT_EQUAL_INT((int)sizeof(inbytes), inlen);
}

int main(void) {
    UNITY_BEGIN();
    RUN_TEST(test_htmlEncodeEntities_null_args);
    RUN_TEST(test_htmlEncodeEntities_ascii_passthrough);
    RUN_TEST(test_htmlEncodeEntities_special_ascii_entities);
    RUN_TEST(test_htmlEncodeEntities_double_quote_encoded);
    RUN_TEST(test_htmlEncodeEntities_single_quote_encoded);
    RUN_TEST(test_htmlEncodeEntities_named_entity_copyright);
    RUN_TEST(test_htmlEncodeEntities_numeric_entity_smiley);
    RUN_TEST(test_htmlEncodeEntities_output_too_small_no_progress);
    RUN_TEST(test_htmlEncodeEntities_partial_progress_due_to_output_limit);
    RUN_TEST(test_htmlEncodeEntities_malformed_trailing_in_leading_pos);
    RUN_TEST(test_htmlEncodeEntities_truncated_multibyte_at_end);
    RUN_TEST(test_htmlEncodeEntities_invalid_trailing_byte);
    RUN_TEST(test_htmlEncodeEntities_mixed_content);
    return UNITY_END();
}