122 lines
3.2 KiB
C++
122 lines
3.2 KiB
C++
// _____ _ _____
|
|
// |_ _|____ _| ||_ _|_ __
|
|
// | |/ _ \ \/ / __|| | \ \ / /
|
|
// | | __/> <| |_ | | \ V /
|
|
// |_|\___/_/\_\\__||_| \_/
|
|
// Author: Love Billenius <lovebillenius@disroot.org>
|
|
// License: GPL-3
|
|
|
|
#include "Page.hpp"
|
|
|
|
#include <iostream>
|
|
#include <format>
|
|
|
|
#include "stringutil.hpp"
|
|
#include "cpr/cpr.h"
|
|
#include "libxml/HTMLparser.h"
|
|
#include "libxml/xpath.h"
|
|
|
|
|
|
Page::Page(const uint_fast8_t number): number(number), subpages(fetchSubpages()) {
|
|
}
|
|
|
|
Page Page::operator--(int) const {
|
|
return Page(number - 1);
|
|
}
|
|
|
|
Page Page::operator++(int) const {
|
|
return Page(number + 1);
|
|
}
|
|
|
|
Page &Page::operator-=(int) {
|
|
number--;
|
|
refresh();
|
|
return *this;
|
|
}
|
|
|
|
std::string Page::str() const {
|
|
std::string ret;
|
|
|
|
for (const std::string &pageText: subpages) {
|
|
std::istringstream stream(pageText);
|
|
std::string line;
|
|
while (std::getline(stream, line))
|
|
ret += line + "\n";
|
|
|
|
string_utils::limitConsecutiveWhitespace(ret, MAX_WHITESPACE);
|
|
string_utils::removeTrailingWhitespace(ret);
|
|
}
|
|
return ret;
|
|
}
|
|
|
|
Page &Page::operator+=(int) {
|
|
number++;
|
|
refresh();
|
|
return *this;
|
|
}
|
|
|
|
bool Page::refresh() {
|
|
std::vector<std::string> newSubpages = fetchSubpages();
|
|
const bool replace = !contentEquals(newSubpages);
|
|
if (replace)
|
|
subpages = newSubpages;
|
|
|
|
return replace;
|
|
}
|
|
|
|
std::string Page::url() const {
|
|
return std::format("https://www.svt.se/svttext/web/pages/{}.html", number);
|
|
}
|
|
|
|
std::vector<std::string> Page::fetchSubpages() const {
|
|
const cpr::Response response = cpr::Get(cpr::Url{url()});
|
|
if (response.status_code / 100 != 2)
|
|
throw std::runtime_error("Page not found");
|
|
const htmlDocPtr doc = htmlReadMemory(response.text.c_str(), response.text.size(), nullptr, nullptr,
|
|
HTML_PARSE_RECOVER | HTML_PARSE_NOERROR | HTML_PARSE_NOWARNING);
|
|
if (doc == nullptr)
|
|
throw std::runtime_error("Could not parse HTML.");
|
|
|
|
const auto xpathCtx = xmlXPathNewContext(doc);
|
|
if (!xpathCtx) {
|
|
xmlFreeDoc(doc);
|
|
throw std::runtime_error("Could not create XPath context.");
|
|
}
|
|
const auto xpathObj = xmlXPathEvalExpression(
|
|
reinterpret_cast<const xmlChar *>("//div[contains(@class,'Content_screenreaderOnly')]"), xpathCtx);
|
|
|
|
if (!xpathObj) {
|
|
xmlXPathFreeContext(xpathCtx);
|
|
xmlFreeDoc(doc);
|
|
throw std::runtime_error("Could not evaluate XPath expression.");
|
|
}
|
|
|
|
std::vector<std::string> pages;
|
|
if (const xmlNodeSetPtr nodes = xpathObj->nodesetval) {
|
|
for (int i = 0; i < nodes->nodeNr; ++i) {
|
|
xmlChar *content = xmlNodeGetContent(nodes->nodeTab[i]);
|
|
if (!content)
|
|
continue;
|
|
pages.emplace_back(reinterpret_cast<const char *>(content));
|
|
xmlFree(content);
|
|
}
|
|
}
|
|
|
|
xmlXPathFreeObject(xpathObj);
|
|
xmlXPathFreeContext(xpathCtx);
|
|
xmlFreeDoc(doc);
|
|
|
|
return pages;
|
|
}
|
|
|
|
bool Page::contentEquals(const std::vector<std::string> &subpagesOther) const {
|
|
if (subpagesOther.size() != subpages.size())
|
|
return false;
|
|
|
|
for (std::size_t i = 0; i < subpages.size(); i++)
|
|
if (subpages[i] != subpagesOther[i])
|
|
return false;
|
|
|
|
return true;
|
|
}
|