| 1 |
// Site Mapper |
| 2 |
// |
| 3 |
// Douglas Thrift |
| 4 |
// |
| 5 |
// $Id$ |
| 6 |
|
| 7 |
#include "SiteMapper.hpp" |
| 8 |
#include "Matcher.hpp" |
| 9 |
#include "Page.hpp" |
| 10 |
|
| 11 |
string program; |
| 12 |
bool debug = false; |
| 13 |
|
| 14 |
int main(int argc, char* argv[]) |
| 15 |
{ |
| 16 |
program = argv[0]; |
| 17 |
|
| 18 |
string siteIndex, siteMap; |
| 19 |
|
| 20 |
for (int index = 1; index < argc; index++) |
| 21 |
{ |
| 22 |
string arg(argv[index]); |
| 23 |
Matcher matcher; |
| 24 |
|
| 25 |
if (arg == matcher("^-index=(.*)$")) |
| 26 |
{ |
| 27 |
siteIndex = matcher[1]; |
| 28 |
} |
| 29 |
else if (arg == matcher("^-map=(.*)$")) |
| 30 |
{ |
| 31 |
siteMap = matcher[1]; |
| 32 |
} |
| 33 |
else if (arg == "-D") |
| 34 |
{ |
| 35 |
if (!debug) debug = true; |
| 36 |
} |
| 37 |
} |
| 38 |
|
| 39 |
if (siteIndex != "" && siteMap != "") |
| 40 |
{ |
| 41 |
XMLPlatformUtils::Initialize(); |
| 42 |
XPathEvaluator::initialize(); |
| 43 |
|
| 44 |
SiteMapper mapper(siteIndex, siteMap); |
| 45 |
|
| 46 |
XPathEvaluator::terminate(); |
| 47 |
XMLPlatformUtils::Terminate(); |
| 48 |
} |
| 49 |
else |
| 50 |
{ |
| 51 |
cout << "Usage: " << program << " -index=index -map=map [-D]\n"; |
| 52 |
} |
| 53 |
|
| 54 |
return 0; |
| 55 |
} |
| 56 |
|
| 57 |
SiteMapper::SiteMapper(const string& siteIndex, const string& siteMap) |
| 58 |
{ |
| 59 |
oldMap(siteMap); |
| 60 |
newIndex(siteIndex); |
| 61 |
newMap(siteMap); |
| 62 |
} |
| 63 |
|
| 64 |
void SiteMapper::oldMap(const string& siteMap) |
| 65 |
{ |
| 66 |
support.setParserLiaison(&liaison); |
| 67 |
|
| 68 |
XalanDOMString file(siteMap.c_str()); |
| 69 |
LocalFileInputSource source(file.c_str()); |
| 70 |
|
| 71 |
XalanDocument* document = liaison.parseXMLStream(source); |
| 72 |
|
| 73 |
if (document == 0) return; |
| 74 |
|
| 75 |
XalanNode* list = evaluator.selectSingleNode(support, document, |
| 76 |
XalanDOMString("/page/section/list").c_str()); |
| 77 |
|
| 78 |
if (list == 0) return; |
| 79 |
|
| 80 |
item = evaluator.createXPath(XalanDOMString("item").c_str()); |
| 81 |
address = evaluator.createXPath(XalanDOMString("link/@address").c_str()); |
| 82 |
link = evaluator.createXPath(XalanDOMString("link").c_str()); |
| 83 |
list_ = evaluator.createXPath(XalanDOMString("list").c_str()); |
| 84 |
|
| 85 |
oldMap(pages, list); |
| 86 |
|
| 87 |
evaluator.destroyXPath(item); |
| 88 |
evaluator.destroyXPath(address); |
| 89 |
evaluator.destroyXPath(link); |
| 90 |
evaluator.destroyXPath(list_); |
| 91 |
} |
| 92 |
|
| 93 |
void SiteMapper::oldMap(vector<Page>& pages, XalanNode* list_) |
| 94 |
{ |
| 95 |
NodeRefList nodes = evaluator.selectNodeList(support, list_, *item); |
| 96 |
|
| 97 |
for (int index = 0; index < nodes.getLength(); index++) |
| 98 |
{ |
| 99 |
XalanNode* node = nodes.item(index); |
| 100 |
ostringstream url, title; |
| 101 |
|
| 102 |
url << evaluator.evaluate(support, node, *address)->str(); |
| 103 |
title << evaluator.evaluate(support, node, *link)->str(); |
| 104 |
|
| 105 |
Page page(url.str(), title.str()); |
| 106 |
XalanNode* list = evaluator.selectSingleNode(support, node, |
| 107 |
*(this->list_)); |
| 108 |
|
| 109 |
if (list != 0) oldMap(page.getChildren(), list); |
| 110 |
|
| 111 |
pages.push_back(page); |
| 112 |
} |
| 113 |
} |
| 114 |
|
| 115 |
void SiteMapper::newIndex(const string& siteIndex) |
| 116 |
{ |
| 117 |
XalanDOMString file(siteIndex.c_str()); |
| 118 |
LocalFileInputSource source(file.c_str()); |
| 119 |
|
| 120 |
XalanDocument* document = liaison.parseXMLStream(source); |
| 121 |
|
| 122 |
if (document == 0) return; |
| 123 |
|
| 124 |
address = evaluator.createXPath(XalanDOMString("address").c_str()); |
| 125 |
port = evaluator.createXPath(XalanDOMString("port").c_str()); |
| 126 |
path = evaluator.createXPath(XalanDOMString("path").c_str()); |
| 127 |
title = evaluator.createXPath(XalanDOMString("title").c_str()); |
| 128 |
|
| 129 |
NodeRefList nodes = evaluator.selectNodeList(support, document, |
| 130 |
XalanDOMString("/index/page").c_str()); |
| 131 |
|
| 132 |
for (int index = 0; index < nodes.getLength(); index++) |
| 133 |
{ |
| 134 |
XalanNode* node = nodes.item(index); |
| 135 |
ostringstream address; |
| 136 |
|
| 137 |
address << evaluator.evaluate(support, node, *(this->address))->str(); |
| 138 |
|
| 139 |
double port = evaluator.evaluate(support, node, *(this->port))->num(); |
| 140 |
|
| 141 |
if (port >= 0 && port <= 65535) |
| 142 |
{ |
| 143 |
address << ':' << int(port); |
| 144 |
} |
| 145 |
|
| 146 |
ostringstream path, title; |
| 147 |
|
| 148 |
path << evaluator.evaluate(support, node, *(this->path))->str(); |
| 149 |
title << evaluator.evaluate(support, node, *(this->title))->str(); |
| 150 |
|
| 151 |
Page page(address.str(), path.str(), title.str()); |
| 152 |
Matcher matcher; |
| 153 |
|
| 154 |
if (page == matcher(string("^Douglas\\sThrift's\\sWebsite\\s\\|\\sDou") |
| 155 |
+ "glas\\sThrift's\\sBlog:\\s(.+)$")) |
| 156 |
{ |
| 157 |
if (Matcher("^\\w+\\s\\d\\d\\d\\d\\sArchives$") == matcher[1]) |
| 158 |
{ |
| 159 |
page.setTitle(matcher[1]); |
| 160 |
|
| 161 |
if (newIndex(pages, page)) continue; |
| 162 |
} |
| 163 |
else continue; |
| 164 |
} |
| 165 |
else if (page == matcher("^Douglas\\sThrift's.+Website\\s\\|\\s(.+)$")) |
| 166 |
{ |
| 167 |
page.setTitle(matcher[1]); |
| 168 |
|
| 169 |
if (newIndex(pages, page)) continue; |
| 170 |
} |
| 171 |
else continue; |
| 172 |
|
| 173 |
cerr << page.getTitle() << ' ' << page.getChildOf() << '\n'; |
| 174 |
// newPages.insert(page); |
| 175 |
} |
| 176 |
|
| 177 |
evaluator.destroyXPath(address); |
| 178 |
evaluator.destroyXPath(port); |
| 179 |
evaluator.destroyXPath(path); |
| 180 |
evaluator.destroyXPath(title); |
| 181 |
} |
| 182 |
|
| 183 |
bool SiteMapper::newIndex(vector<Page>& pages, Page& page) |
| 184 |
{ |
| 185 |
for (unsigned index = 0; index < pages.size(); index++) |
| 186 |
{ |
| 187 |
if (pages[index] == page.getAddress()) |
| 188 |
{ |
| 189 |
Matcher matcher; |
| 190 |
|
| 191 |
if (pages[index] == page) |
| 192 |
{ |
| 193 |
page.setChildren(pages[index].getChildren()); |
| 194 |
|
| 195 |
pages[index] = page; |
| 196 |
|
| 197 |
return true; |
| 198 |
} |
| 199 |
else if (matcher('^' + pages[index].getPath()) == page) |
| 200 |
{ |
| 201 |
page.setChildOf(matcher[0]); |
| 202 |
|
| 203 |
if (matcher('^' + pages[index].getTitle() + "\\s\\|\\s(.+)$") |
| 204 |
== page) |
| 205 |
{ |
| 206 |
page.setTitle(matcher[1]); |
| 207 |
} |
| 208 |
|
| 209 |
return newIndex(pages[index].getChildren(), page); |
| 210 |
} |
| 211 |
} |
| 212 |
} |
| 213 |
|
| 214 |
cerr << "Here!\n"; |
| 215 |
|
| 216 |
return false; |
| 217 |
} |
| 218 |
|
| 219 |
void SiteMapper::newMap(const string& siteMap) |
| 220 |
{ |
| 221 |
// |
| 222 |
|
| 223 |
for (unsigned index = 0; index < pages.size(); index++) |
| 224 |
{ |
| 225 |
cout << pages[index] << '\n'; |
| 226 |
} |
| 227 |
} |
