bool xml_open_file(char *path)
{
int sz;
FILE *file;
struct stat sb;
// get file size
if (stat(path,&sb)!=0) return(FALSE);
sz=(int)sb.st_size;
// open file
file=fopen(path,"rb");
if (file==NULL) return(FALSE);
xml_dataptr=(char*)malloc(sz);
if (xml_dataptr==NULL) {
fclose(file);
return(FALSE);
}
fread(xml_dataptr,1,sz,file);
fclose(file);
xml_filesz=sz;
xml_buffersz=xml_buffer_chunk_sz;
xml_taglistptr=NULL;
return(xml_decode());
}
static bool decode(const char *probe, const char *exp)
{
std::string res;
xml_decode(probe, res);
if (res != exp) {
g_warning("want: %s, got %s", exp, res.c_str());
return false;
}
return true;
}
static void html2result(const char *p, ParseResult &result)
{
LinksPosList links_list;
std::string res;
const char *tag, *next;
std::string name;
std::string::size_type cur_pos;
int i;
struct ReplaceTag {
const char *match_;
int match_len_;
const char *replace_;
int char_len_;
};
static const ReplaceTag replace_arr[] = {
{ "b>", 2, "<b>", 0 },
{ "/b>", 3, "</b>", 0 },
{ "big>", 4, "<big>", 0},
{ "/big>", 5, "</big>", 0},
{ "i>", 2, "<i>", 0 },
{ "/i>", 3, "</i>", 0 },
{ "s>", 2, "<s>", 0 },
{ "/s>", 3, "</s>", 0 },
{ "sub>", 4, "<sub>", 0 },
{ "/sub>", 5, "</sub>", 0},
{ "sup>", 4, "<sup>", 0},
{ "/sup>", 5, "</sup>", 0},
{ "small>", 6, "<small>", 0},
{ "/small>", 7, "</small>", 0},
{ "tt>", 3, "<tt>", 0},
{ "/tt>", 4, "</tt>", 0},
{ "u>", 2, "<u>", 0 },
{ "/u>", 3, "</u>", 0 },
{ "br>", 3, "\n", 1 },
{ "nl>", 3, "", 0 },
{ "hr>", 3, "\n<span foreground=\"gray\"><s> </s></span>\n", 7 },
{ "/font>", 6, "</span>", 0 },
{ NULL, 0, NULL },
};
for (cur_pos = 0; *p && (tag = strchr(p, '<')) != NULL;) {
std::string chunk(p, tag - p);
size_t pango_len;
std::string pango;
html_topango(chunk, pango, pango_len);
res += pango;
cur_pos += pango_len;
p = tag;
for (i = 0; replace_arr[i].match_; ++i)
if (strncasecmp(replace_arr[i].match_, p + 1,
replace_arr[i].match_len_) == 0) {
res += replace_arr[i].replace_;
p += 1 + replace_arr[i].match_len_;
cur_pos += replace_arr[i].char_len_;
goto cycle_end;
}
if (strncasecmp(p+1, "font ", 5)==0) {
next = strchr(p, '>');
if (!next) {
++p;
continue;
}
res += "<span";
name.assign(p + 6, next - (p + 6));
const char *p1 = strcasestr(name.c_str(), "face=");
if (p1) {
p1 += sizeof("face=") -1 +1;
const char *p2 = p1;
while (true) {
if (*p2 == '\0') {
p2 = NULL;
break;
}
if (*p2 == '\'' || *p2 == '"')
break;
p2++;
}
if (p2) {
std::string face(p1, p2-p1);
res += " face=\"";
res += face;
res += "\"";
}
}
p1 = strcasestr(name.c_str(), "color=");
if (p1) {
p1 += sizeof("color=") -1;
if (*p1 == '\'' || *p1 == '\"')
p1++;
const char *p2 = p1;
while (true) {
if (*p2 == '\0') {
p2 = NULL;
break;
}
if (*p2 == '\'' || *p2 == '"' || *p2 == ' ' || *p2 == '>')
break;
p2++;
}
if (p2) {
std::string color(p1, p2-p1);
if (pango_color_parse(NULL, color.c_str())) {
res += " foreground=\"";
res += color;
res += "\"";
}
}
}
res += ">";
p = next + 1;
} else if ((*(p + 1) == 'a' || *(p + 1) == 'A') && *(p + 2) == ' ') {
next = strchr(p, '>');
if (!next) {
p++;
continue;
}
p+=3;
name.assign(p, next - p);
const char *p1 = strcasestr(name.c_str(), "href=");
std::string link;
if (p1) {
p1 += sizeof("href=") -1 +1;
const char *p2 = p1;
while (true) {
if (*p2 == '\0') {
p2 = NULL;
break;
}
if (*p2 == '\'' || *p2 == '"')
break;
p2++;
}
if (p2) {
link.assign(p1, p2-p1);
}
}
p = next + 1;
next = strcasestr(p, "</a>");
if (!next) {
continue;
}
res += "<span foreground=\"lightblue\" underline=\"single\">";
std::string::size_type link_len = next - p;
std::string chunk(p, link_len);
html_topango(chunk, pango, pango_len);
links_list.push_back(LinkDesc(cur_pos, pango_len, link));
res += pango;
cur_pos += pango_len;
res += "</span>";
p = next + sizeof("</a>") - 1;
} else if (strncasecmp(p+1, "ref>", 4)==0) {
next = strcasestr(p, "</ref>");
if (!next) {
p++;
continue;
}
p+=5;
res += "<span foreground=\"lightblue\" underline=\"single\">";
std::string::size_type link_len = next - p;
std::string chunk(p, link_len);
html_topango(chunk, pango, pango_len);
std::string xml_enc;
xml_decode(chunk.c_str(), xml_enc);
std::string link;
link = "query://";
link += xml_enc;
links_list.push_back(LinkDesc(cur_pos, pango_len, link));
res += pango;
cur_pos += pango_len;
res += "</span>";
p = next + sizeof("</ref>") - 1;
} else if (strncasecmp(p+1, "img ", 4)==0) {
next = strchr(p+5, '>');
if (!next) {
p++;
continue;
}
name.assign(p+5, next - (p+5));
p = next + 1;
const char *p1 = strcasestr(name.c_str(), "src=");
std::string src;
if (p1) {
p1 += sizeof("src=") -1 +1;
const char *p2 = p1;
while (true) {
if (*p2 == '\0') {
p2 = NULL;
break;
}
if (*p2 == '\'' || *p2 == '"')
break;
p2++;
}
if (p2) {
src.assign(p1, p2-p1);
}
}
if (!src.empty()) {
ParseResultItem item;
item.type = ParseResultItemType_link;
item.link = new ParseResultLinkItem;
item.link->pango = res;
item.link->links_list = links_list;
result.item_list.push_back(item);
res.clear();
cur_pos = 0;
links_list.clear();
item.type = ParseResultItemType_res;
item.res = new ParseResultResItem;
item.res->type = "image";
int n = src.length();
if (src[0]==0x1e && src[n-1]==0x1f) {
item.res->key.assign(src.c_str()+1, n-2);
} else {
item.res->key = src;
}
result.item_list.push_back(item);
}
} else {
next = strchr(p+1, '>');
if (!next) {
p++;
res += "<";
cur_pos++;
continue;
}
p = next + 1;
}
cycle_end:
;
}
res += p;
ParseResultItem item;
item.type = ParseResultItemType_link;
item.link = new ParseResultLinkItem;
item.link->pango = res;
item.link->links_list = links_list;
result.item_list.push_back(item);
}
