本页主题:");
71 | static const string keyword_toptip_end(">[楼 主]");
72 | const pair& pair_tmp = fetchStringBetweenKeywords( webpage_txt,
73 | keyword_toptip_begin,
74 | keyword_toptip_end );
75 | string toptip = pair_tmp.first;
76 | if (toptip.empty()) {
77 | cerr << "ERROR! there is no toptip. " << endl;
78 | return(false);
79 | }
80 |
81 | // the list may be on the webpage at the same time
82 | static const vector> begin_and_end_keywords_list = { make_pair("![](\"",) keywords_seed_begin_list = { "http://www.jandown.com",
101 | "http://jandown.com",
102 | "http://www6.mimima.com",
103 | "http://mimima.com" };
104 |
105 | const auto body_pos = webpage_txt.find("");
106 | if (string::npos == body_pos) {
107 | //cerr << "warning! parseseedurl() cannot find the keyword \"\"" << endl;
108 | return(false);
109 | }
110 | const string& body = webpage_txt.substr(body_pos);
111 |
112 | for (const auto& e : keywords_seed_begin_list) {
113 | const string& keyword_seed_begin = e;
114 | static const string keyword_seed_end("\"");
115 |
116 | const pair& pair_tmp = fetchStringBetweenKeywords( body,
117 | keyword_seed_begin,
118 | keyword_seed_end );
119 | if (!pair_tmp.first.empty()) {
120 | seed_url = keyword_seed_begin + pair_tmp.first;
121 | return(true);
122 | }
123 | }
124 |
125 |
126 | return(false);
127 | }
128 |
129 | AichengTopicWebpage::AichengTopicWebpage (const string& url, const string& proxy_addr)
130 | : TopicWebpage(url, parsePicturesUrls, parseSeedUrl, proxy_addr, "gbk", "UTF-8")
131 | {
132 | ;
133 | }
134 |
135 | AichengTopicWebpage::~AichengTopicWebpage ()
136 | {
137 | ;
138 | }
139 |
140 |
--------------------------------------------------------------------------------
/src/lib/self/AichengTopicWebpage.h:
--------------------------------------------------------------------------------
1 | // last modified
2 |
3 | #pragma once
4 |
5 | #include
6 | #include
7 | #include "TopicWebpage.h"
8 |
9 | using std::string;
10 | using std::vector;
11 |
12 |
13 | class AichengTopicWebpage : public TopicWebpage
14 | {
15 | public:
16 | AichengTopicWebpage (const string& url, const string& proxy_addr);
17 | virtual ~AichengTopicWebpage ();
18 | };
19 |
20 |
--------------------------------------------------------------------------------
/src/lib/self/AichengTopicsListWebpage.cpp:
--------------------------------------------------------------------------------
1 | // last modified
2 |
3 | #include "AichengTopicsListWebpage.h"
4 | #include
5 | #include
6 | #include "../helper/Misc.h"
7 |
8 |
9 | using namespace std;
10 |
11 |
12 |
13 | static bool
14 | parseTitlesAndUrls ( const string& webpage_txt,
15 | const string& portal_url,
16 | vector>& titles_and_urls_list )
17 | {
18 | const unsigned size_back = titles_and_urls_list.size();
19 |
20 | const auto topics_list_txt_pos = webpage_txt.find(R"(style="border-top:0">普通主题 |
)");
21 | size_t keyword_topic_url_begin_pos = ((string::npos == topics_list_txt_pos) ? 0 : topics_list_txt_pos);
22 | size_t keyword_topic_url_end_pos = 0;
23 |
24 | while (true) {
25 | // parse topic URL
26 | static const string keyword_topic_url_begin("
12 | 