ezxml.c

开放源码实时操作系统源码.

    return &root->xml;
}

// a wrapper for ezxml_parse_fd that accepts a file name
ezxml_t ezxml_parse_file(const char *file)
{
    int fd = open(file, O_RDONLY, 0);
    ezxml_t xml = ezxml_parse_fd(fd);
    
    if (fd >= 0) close(fd);
    return xml;
}
#endif // CYGPKG_IO_FILEIO

// Encodes ampersand sequences appending the results to *dst, reallocating *dst
// if length excedes max. a is non-zero for attribute encoding. Returns *dst
char *ezxml_ampencode(const char *s, size_t len, char **dst, size_t *dlen,
                      size_t *max, short a)
{
    const char *e;
    
    for (e = s + len; s != e; s++) {
        while (*dlen + 10 > *max) *dst = realloc(*dst, *max += EZXML_BUFSIZE);

        switch (*s) {
        case '\0': return *dst;
        case '&': *dlen += sprintf(*dst + *dlen, "&"); break;
        case '<': *dlen += sprintf(*dst + *dlen, "&lt;"); break;
        case '>': *dlen += sprintf(*dst + *dlen, "&gt;"); break;
        case '"': *dlen += sprintf(*dst + *dlen, (a) ? "&quot;" : "\""); break;
        case '\n': *dlen += sprintf(*dst + *dlen, (a) ? "&#xA;" : "\n"); break;
        case '\t': *dlen += sprintf(*dst + *dlen, (a) ? "&#x9;" : "\t"); break;
        case '\r': *dlen += sprintf(*dst + *dlen, "&#xD;"); break;
        default: (*dst)[(*dlen)++] = *s;
        }
    }
    return *dst;
}

// Recursively converts each tag to xml appending it to *s. Reallocates *s if
// its length excedes max. start is the location of the previous tag in the
// parent tag's character content. Returns *s.
char *ezxml_toxml_r(ezxml_t xml, char **s, size_t *len, size_t *max,
                    size_t start, char ***attr)
{
    int i, j;
    char *txt = (xml->parent) ? xml->parent->txt : "";
    size_t off = 0;

    // parent character content up to this tag
    *s = ezxml_ampencode(txt + start, xml->off - start, s, len, max, 0);

    while (*len + strlen(xml->name) + 4 > *max) // reallocate s
        *s = realloc(*s, *max += EZXML_BUFSIZE);

    *len += sprintf(*s + *len, "<%s", xml->name); // open tag
    for (i = 0; xml->attr[i]; i += 2) { // tag attributes
        if (ezxml_attr(xml, xml->attr[i]) != xml->attr[i + 1]) continue;
        while (*len + strlen(xml->attr[i]) + 7 > *max) // reallocate s
            *s = realloc(*s, *max += EZXML_BUFSIZE);

        *len += sprintf(*s + *len, " %s=\"", xml->attr[i]);
        ezxml_ampencode(xml->attr[i + 1], -1, s, len, max, 1);
        *len += sprintf(*s + *len, "\"");
    }

    for (i = 0; attr[i] && strcmp(attr[i][0], xml->name); i++);
    for (j = 1; attr[i] && attr[i][j]; j += 3) { // default attributes
        if (! attr[i][j + 1] || ezxml_attr(xml, attr[i][j]) != attr[i][j + 1])
            continue; // skip duplicates and non-values
        while (*len + strlen(attr[i][j]) + 7 > *max) // reallocate s
            *s = realloc(*s, *max += EZXML_BUFSIZE);

        *len += sprintf(*s + *len, " %s=\"", attr[i][j]);
        ezxml_ampencode(attr[i][j + 1], -1, s, len, max, 1);
        *len += sprintf(*s + *len, "\"");
    }
    *len += sprintf(*s + *len, ">");

    *s = (xml->child) ? ezxml_toxml_r(xml->child, s, len, max, 0, attr) //child
                      : ezxml_ampencode(xml->txt, -1, s, len, max, 0);  //data
    
    while (*len + strlen(xml->name) + 4 > *max) // reallocate s
        *s = realloc(*s, *max += EZXML_BUFSIZE);

    *len += sprintf(*s + *len, "</%s>", xml->name); // close tag

    while (txt[off] && off < xml->off) off++; // make sure off is within bounds
    return (xml->ordered) ? ezxml_toxml_r(xml->ordered, s, len, max, off, attr)
                          : ezxml_ampencode(txt + off, -1, s, len, max, 0);
}

// converts an ezxml structure back to xml, returning it as a string that must
// be freed
char *ezxml_toxml(ezxml_t xml)
{
    ezxml_t p = (xml) ? xml->parent : NULL, o = (xml) ? xml->ordered : NULL;
    ezxml_root_t root = (ezxml_root_t)xml;
    size_t len = 0, max = EZXML_BUFSIZE;
    char *s = strcpy(malloc(max), ""), *t, *n;
    int i, j, k;

    if (! xml || ! xml->name) return realloc(s, len + 1);
    while (root->xml.parent) root = (ezxml_root_t)root->xml.parent; // root tag

    for (i = 0; ! p && root->pi[i]; i++) { // pre-root processing instructions
        for (k = 2; root->pi[i][k - 1]; k++);
        for (j = 1; (n = root->pi[i][j]); j++) {
            if (root->pi[i][k][j - 1] == '>') continue; // not pre-root
            while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max)
                s = realloc(s, max += EZXML_BUFSIZE);
            len += sprintf(s + len, "<?%s%s%s?>\n", t, *n ? " " : "", n);
        }
    }

    xml->parent = xml->ordered = NULL;
    s = ezxml_toxml_r(xml, &s, &len, &max, 0, root->attr);
    xml->parent = p;
    xml->ordered = o;

    for (i = 0; ! p && root->pi[i]; i++) { // post-root processing instructions
        for (k = 2; root->pi[i][k - 1]; k++);
        for (j = 1; (n = root->pi[i][j]); j++) {
            if (root->pi[i][k][j - 1] == '<') continue; // not post-root
            while (len + strlen(t = root->pi[i][0]) + strlen(n) + 7 > max)
                s = realloc(s, max += EZXML_BUFSIZE);
            len += sprintf(s + len, "\n<?%s%s%s?>", t, *n ? " " : "", n);
        }
    }
    return realloc(s, len + 1);
}

// free the memory allocated for the ezxml structure
void ezxml_free(ezxml_t xml)
{
    ezxml_root_t root = (ezxml_root_t)xml;
    int i, j;
    char **a, *s;

    if (! xml) return;
    ezxml_free(xml->child);
    ezxml_free(xml->ordered);

    if (! xml->parent) { // free root tag allocations
        for (i = 10; root->ent[i]; i += 2) // 0 - 9 are default entites (<>&"')
            if ((s = root->ent[i + 1]) < root->s || s > root->e) free(s);
        free(root->ent); // free list of general entities

        for (i = 0; (a = root->attr[i]); i++) {
            for (j = 1; a[j++]; j += 2) // free malloced attribute values
                if (a[j] && (a[j] < root->s || a[j] > root->e)) free(a[j]);
            free(a);
        }
        if (root->attr[0]) free(root->attr); // free default attribute list

        for (i = 0; root->pi[i]; i++) {
            for (j = 1; root->pi[i][j]; j++);
            free(root->pi[i][j + 1]);
            free(root->pi[i]);
        }            
        if (root->pi[0]) free(root->pi); // free processing instructions

        if (root->len == -1) free(root->m); // malloced xml data
#ifndef EZXML_NOMMAP
        else if (root->len) munmap(root->m, root->len); // mem mapped xml data
#endif // EZXML_NOMMAP
        if (root->u) free(root->u); // utf8 conversion
    }

    ezxml_free_attr(xml->attr); // tag attributes
    if ((xml->flags & EZXML_TXTM)) free(xml->txt); // character content
    if ((xml->flags & EZXML_NAMEM)) free(xml->name); // tag name
    free(xml);
}

// return parser error message or empty string if none
const char *ezxml_error(ezxml_t xml)
{
    while (xml && xml->parent) xml = xml->parent; // find root tag
    return (xml) ? ((ezxml_root_t)xml)->err : "";
}

// returns a new empty ezxml structure with the given root tag name
ezxml_t ezxml_new(const char *name)
{
    static char *ent[] = { "lt;", "&#60;", "gt;", "&#62;", "quot;", "&#34;",
                           "apos;", "&#39;", "amp;", "&#38;", NULL };
    ezxml_root_t root = (ezxml_root_t)memset(malloc(sizeof(struct ezxml_root)), 
                                             '\0', sizeof(struct ezxml_root));
    root->xml.name = (char *)name;
    root->cur = &root->xml;
    strcpy(root->err, root->xml.txt = "");
    root->ent = memcpy(malloc(sizeof(ent)), ent, sizeof(ent));
    root->attr = root->pi = (char ***)(root->xml.attr = EZXML_NIL);
    return &root->xml;
}

// Adds a child tag. off is the offset of the child tag relative to the start
// of the parent tag's character content. returns the child tag
ezxml_t ezxml_add_child(ezxml_t xml, const char *name, size_t off)
{
    ezxml_t cur, head, child;

    if (! xml) return NULL;
    child = (ezxml_t)memset(malloc(sizeof(struct ezxml)), '\0',
                            sizeof(struct ezxml));
    child->name = (char *)name;
    child->attr = EZXML_NIL;
    child->off = off;
    child->parent = xml;
    child->txt = "";

    if ((head = xml->child)) { // already have sub tags
        if (head->off <= off) { // not first subtag
            for (cur = head; cur->ordered && cur->ordered->off <= off;
                 cur = cur->ordered);
            child->ordered = cur->ordered;
            cur->ordered = child;
        }
        else { // first subtag
            child->ordered = head;
            xml->child = child;
        }

        for (cur = head; cur->sibling && strcmp(cur->name, name);
             cur = cur->sibling); // find tag type
        if (! strcmp(cur->name, name) && cur->off <= off) { //not first of type
            while (cur->next && cur->next->off <= off) cur = cur->next;
            child->next = cur->next;
            cur->next = child;
        }
        else { // first tag of this type
            if (cur->off > off) child->next = cur; // not only tag of this type
            for (cur = head; cur->sibling && cur->sibling->off <= off;
                 cur = cur->sibling);
            child->sibling = cur->sibling;
            cur->sibling = child;
        }
    }
    else xml->child = child; // only sub tag
    
    return child;
}

// sets the character content for the given tag and returns the tag
ezxml_t ezxml_set_txt(ezxml_t xml, const char *txt)
{
    if (! xml) return NULL;
    if (xml->flags & EZXML_TXTM) free(xml->txt); // existing txt was malloced
    xml->flags &= ~EZXML_TXTM;
    xml->txt = (char *)txt;
    return xml;
}

// Sets the given tag attribute or adds a new attribute if not found. A value
// of NULL will remove the specified attribute.
void ezxml_set_attr(ezxml_t xml, const char *name, const char *value)
{
    int l = 0, c;

    if (! xml) return;
    while (xml->attr[l] && strcmp(xml->attr[l], name)) l += 2;
    if (! xml->attr[l]) { // not found, add as new attribute
        if (! value) return; // nothing to do
        if (xml->attr == EZXML_NIL) { // first attribute
            xml->attr = malloc(4 * sizeof(char *));
            xml->attr[1] = strdup(""); // empty list of malloced names/vals
        }
        else xml->attr = realloc(xml->attr, (l + 4) * sizeof(char *));

        xml->attr[l] = (char *)name; // set attribute name
        xml->attr[l + 2] = NULL; // null terminate attribute list
        xml->attr[l + 3] = realloc(xml->attr[l + 1],
                                   (c = strlen(xml->attr[l + 1])) + 2);
        strcpy(xml->attr[l + 3] + c, " "); // set name/value as not malloced
        if (xml->flags & EZXML_DUP) xml->attr[l + 3][c] = EZXML_NAMEM;
    }
    else if (xml->flags & EZXML_DUP) free((char *)name); // name was strduped

    for (c = l; xml->attr[c]; c += 2); // find end of attribute list
    if (xml->attr[c + 1][l / 2] & EZXML_TXTM) free(xml->attr[l + 1]); //old val
    if (xml->flags & EZXML_DUP) xml->attr[c + 1][l / 2] |= EZXML_TXTM;
    else xml->attr[c + 1][l / 2] &= ~EZXML_TXTM;

    if (value) xml->attr[l + 1] = (char *)value; // set attribute value
    else { // remove attribute
        if (xml->attr[c + 1][l / 2] & EZXML_NAMEM) free(xml->attr[l]);
        memmove(xml->attr + l, xml->attr + l + 2, (c - l + 2) * sizeof(char*));
        xml->attr = realloc(xml->attr, (c + 2) * sizeof(char *));
        memmove(xml->attr[c + 1] + (l / 2), xml->attr[c + 1] + (l / 2) + 1,
                (c / 2) - (l / 2)); // fix list of which name/vals are malloced
    }
    xml->flags &= ~EZXML_DUP; // clear strdup() flag
}

// sets a flag for the given tag and returns the tag
ezxml_t ezxml_set_flag(ezxml_t xml, short flag)
{
    if (xml) xml->flags |= flag;
    return xml;
}

// removes a tag along with all its subtags
void ezxml_remove(ezxml_t xml)
{
    ezxml_t cur;

    if (! xml) return; // nothing to do
    if (xml->next) xml->next->sibling = xml->sibling; // patch sibling list

    if (xml->parent) { // not root tag
        cur = xml->parent->child; // find head of subtag list
        if (cur == xml) xml->parent->child = xml->ordered; // first subtag
        else { // not first subtag
            while (cur->ordered != xml) cur = cur->ordered;
            cur->ordered = cur->ordered->ordered; // patch ordered list

            cur = xml->parent->child; // go back to head of subtag list
            if (strcmp(cur->name, xml->name)) { // not in first sibling list
                while (strcmp(cur->sibling->name, xml->name))
                    cur = cur->sibling;
                if (cur->sibling == xml) { // first of a sibling list
                    cur->sibling = (xml->next) ? xml->next
                                               : cur->sibling->sibling;
                }
                else cur = cur->sibling; // not first of a sibling list
            }

            while (cur->next && cur->next != xml) cur = cur->next;
            if (cur->next) cur->next = cur->next->next; // patch next list
        }        
    }
    xml->ordered = NULL; // prevent ezxml_free() from clobbering ordered list
    ezxml_free(xml);
}

#ifdef EZXML_TEST // test harness
int main(int argc, char **argv)
{
    ezxml_t xml;
    char *s;
    int i;

    if (argc != 2) return fprintf(stderr, "usage: %s xmlfile\n", argv[0]);

    xml = ezxml_parse_file(argv[1]);
    printf("%s\n", (s = ezxml_toxml(xml)));
    free(s);
    i = fprintf(stderr, "%s", ezxml_error(xml));
    ezxml_free(xml);
    return (i) ? 1 : 0;
}
#endif // EZXML_TEST