tests/tests_HTMLparser_htmlParseComment.c

#include "unity/unity.h"
#include <libxml/HTMLparser.h>

#include <stdlib.h>
#include <string.h>

/* Wrapper for the static function under test (provided in the module) */
void test_htmlParseComment(htmlParserCtxtPtr ctxt, int bogus);

/* Capture structure for SAX comment callback */
typedef struct {
    int count;
    char *last;
} CommentCapture;

static void free_capture(CommentCapture *cap) {
    if (cap->last) {
        free(cap->last);
        cap->last = NULL;
    }
    cap->count = 0;
}

static void comment_cb(void *ctx, const xmlChar *value) {
    CommentCapture *cap = (CommentCapture *)ctx;
    cap->count++;
    /* Copy the comment text into an owned buffer (ASCII-safe) */
    const char *val = (const char *)value;
    size_t n = strlen(val);
    free_capture(cap);
    cap->last = (char *)malloc(n + 1);
    if (cap->last) {
        memcpy(cap->last, val, n + 1);
    }
}

/* Helper to create a parser context with given input buffer */
static htmlParserCtxtPtr make_ctxt(const char *data) {
    if (data == NULL) data = "";
    int len = (int)strlen(data);
    htmlParserCtxtPtr ctxt = htmlCreateMemoryParserCtxt(data, len);
    return ctxt;
}

void setUp(void) {
    /* Setup code here, or leave empty */
}

void tearDown(void) {
    /* Cleanup code here, or leave empty */
}

/* Bogus comment: content is read until '>', which is then skipped */
void test_htmlParseComment_bogus_basic(void) {
    CommentCapture cap = {0, NULL};
    htmlParserCtxtPtr ctxt = make_ctxt("hello>");
    TEST_ASSERT_NOT_NULL(ctxt);

    xmlSAXHandler sax;
    memset(&sax, 0, sizeof(sax));
    sax.comment = comment_cb;

    ctxt->sax = &sax;
    ctxt->userData = &cap;
    ctxt->disableSAX = 0;

    test_htmlParseComment(ctxt, 1);

    TEST_ASSERT_EQUAL_INT(1, cap.count);
    TEST_ASSERT_NOT_NULL(cap.last);
    TEST_ASSERT_EQUAL_STRING("hello", cap.last);

    free_capture(&cap);
    htmlFreeParserCtxt(ctxt);
}

/* Bogus comment without closing '>' should still emit collected content */
void test_htmlParseComment_bogus_no_terminator(void) {
    CommentCapture cap = {0, NULL};
    htmlParserCtxtPtr ctxt = make_ctxt("noend");
    TEST_ASSERT_NOT_NULL(ctxt);

    xmlSAXHandler sax;
    memset(&sax, 0, sizeof(sax));
    sax.comment = comment_cb;

    ctxt->sax = &sax;
    ctxt->userData = &cap;
    ctxt->disableSAX = 0;

    test_htmlParseComment(ctxt, 1);

    TEST_ASSERT_EQUAL_INT(1, cap.count);
    TEST_ASSERT_NOT_NULL(cap.last);
    TEST_ASSERT_EQUAL_STRING("noend", cap.last);

    free_capture(&cap);
    htmlFreeParserCtxt(ctxt);
}

/* Non-bogus: immediate '>' produces an empty comment */
void test_htmlParseComment_nonbogus_immediate_gt(void) {
    CommentCapture cap = {0, NULL};
    htmlParserCtxtPtr ctxt = make_ctxt(">");
    TEST_ASSERT_NOT_NULL(ctxt);

    xmlSAXHandler sax;
    memset(&sax, 0, sizeof(sax));
    sax.comment = comment_cb;

    ctxt->sax = &sax;
    ctxt->userData = &cap;
    ctxt->disableSAX = 0;

    test_htmlParseComment(ctxt, 0);

    TEST_ASSERT_EQUAL_INT(1, cap.count);
    TEST_ASSERT_NOT_NULL(cap.last);
    TEST_ASSERT_EQUAL_STRING("", cap.last);

    free_capture(&cap);
    htmlFreeParserCtxt(ctxt);
}

/* Non-bogus: immediate "->" also produces an empty comment */
void test_htmlParseComment_nonbogus_immediate_arrow(void) {
    CommentCapture cap = {0, NULL};
    htmlParserCtxtPtr ctxt = make_ctxt("->");
    TEST_ASSERT_NOT_NULL(ctxt);

    xmlSAXHandler sax;
    memset(&sax, 0, sizeof(sax));
    sax.comment = comment_cb;

    ctxt->sax = &sax;
    ctxt->userData = &cap;
    ctxt->disableSAX = 0;

    test_htmlParseComment(ctxt, 0);

    TEST_ASSERT_EQUAL_INT(1, cap.count);
    TEST_ASSERT_NOT_NULL(cap.last);
    TEST_ASSERT_EQUAL_STRING("", cap.last);

    free_capture(&cap);
    htmlFreeParserCtxt(ctxt);
}

/* Non-bogus: parse data until '-' is encountered; expect content before '-' */
void test_htmlParseComment_nonbogus_until_dash(void) {
    CommentCapture cap = {0, NULL};
    htmlParserCtxtPtr ctxt = make_ctxt("abc-def->");
    TEST_ASSERT_NOT_NULL(ctxt);

    xmlSAXHandler sax;
    memset(&sax, 0, sizeof(sax));
    sax.comment = comment_cb;

    ctxt->sax = &sax;
    ctxt->userData = &cap;
    ctxt->disableSAX = 0;

    test_htmlParseComment(ctxt, 0);

    TEST_ASSERT_EQUAL_INT(1, cap.count);
    TEST_ASSERT_NOT_NULL(cap.last);
    TEST_ASSERT_EQUAL_STRING("abc", cap.last);

    free_capture(&cap);
    htmlFreeParserCtxt(ctxt);
}

/* Ensure no callback is emitted when SAX is disabled */
void test_htmlParseComment_SAX_disabled(void) {
    CommentCapture cap = {0, NULL};
    htmlParserCtxtPtr ctxt = make_ctxt("ignored>");
    TEST_ASSERT_NOT_NULL(ctxt);

    xmlSAXHandler sax;
    memset(&sax, 0, sizeof(sax));
    sax.comment = comment_cb;

    ctxt->sax = &sax;
    ctxt->userData = &cap;
    ctxt->disableSAX = 1; /* Disable SAX */

    test_htmlParseComment(ctxt, 1);

    TEST_ASSERT_EQUAL_INT(0, cap.count);
    TEST_ASSERT_NULL(cap.last);

    free_capture(&cap);
    htmlFreeParserCtxt(ctxt);
}

/* Ensure safe behavior when ctxt->sax is NULL */
void test_htmlParseComment_null_sax(void) {
    CommentCapture cap = {0, NULL};
    htmlParserCtxtPtr ctxt = make_ctxt("hello>");
    TEST_ASSERT_NOT_NULL(ctxt);

    ctxt->sax = NULL;          /* No SAX handler */
    ctxt->userData = &cap;
    ctxt->disableSAX = 0;

    test_htmlParseComment(ctxt, 1);

    TEST_ASSERT_EQUAL_INT(0, cap.count);
    TEST_ASSERT_NULL(cap.last);

    free_capture(&cap);
    htmlFreeParserCtxt(ctxt);
}

int main(void) {
    /* Initialize libxml2 parser library */
    xmlInitParser();

    UNITY_BEGIN();

    RUN_TEST(test_htmlParseComment_bogus_basic);
    RUN_TEST(test_htmlParseComment_bogus_no_terminator);
    RUN_TEST(test_htmlParseComment_nonbogus_immediate_gt);
    RUN_TEST(test_htmlParseComment_nonbogus_immediate_arrow);
    RUN_TEST(test_htmlParseComment_nonbogus_until_dash);
    RUN_TEST(test_htmlParseComment_SAX_disabled);
    RUN_TEST(test_htmlParseComment_null_sax);

    int rc = UNITY_END();

    /* Cleanup libxml2 global state */
    xmlCleanupParser();

    return rc;
}