| 4 |
|
// |
| 5 |
|
// $Id$ |
| 6 |
|
|
| 7 |
< |
#include "Matcher.hpp" |
| 8 |
< |
#include "Page.hpp" |
| 7 |
> |
#include "SiteMapper.hpp" |
| 8 |
> |
#include "Matcher/Matcher.hpp" |
| 9 |
|
|
| 10 |
< |
Page::Page(const string& address, const string& path, const string& title) |
| 10 |
> |
void Page::setUrl(const ext::String& url) |
| 11 |
|
{ |
| 12 |
< |
setAddress(address); |
| 13 |
< |
setPath(path); |
| 14 |
< |
setTitle(title); |
| 12 |
> |
api::Cerr << PCRE_MULTILINE << " = " << Matcher::defaults << ios::NewLine; |
| 13 |
|
|
| 14 |
< |
tab = 0; |
| 17 |
< |
} |
| 18 |
< |
|
| 19 |
< |
Page::Page(const string& url, const string& title) |
| 20 |
< |
{ |
| 21 |
< |
setUrl(url); |
| 22 |
< |
setTitle(title); |
| 23 |
< |
|
| 24 |
< |
tab = 0; |
| 25 |
< |
} |
| 14 |
> |
Matcher matcher("^http://(.+)(/.*)?$", (PCRE_UNGREEDY | PCRE_DOTALL)); |
| 15 |
|
|
| 16 |
< |
void Page::setUrl(const string& url) |
| 28 |
< |
{ |
| 29 |
< |
Matcher matcher("^http://(.+)(/.*)?$"); |
| 16 |
> |
api::Cerr << (PCRE_UNGREEDY | PCRE_DOTALL) << " = " << matcher.options << ios::NewLine; |
| 17 |
|
|
| 18 |
|
if (url == matcher) |
| 19 |
|
{ |
| 20 |
|
address = matcher[1]; |
| 21 |
|
|
| 22 |
< |
if (matcher.size() > 2) |
| 36 |
< |
{ |
| 37 |
< |
path = matcher[2]; |
| 38 |
< |
} |
| 39 |
< |
else |
| 40 |
< |
{ |
| 41 |
< |
path = '/'; |
| 42 |
< |
} |
| 22 |
> |
if (matcher.size() > 2) path = matcher[2]; else path = "/"; |
| 23 |
|
} |
| 24 |
|
else |
| 25 |
|
{ |
| 26 |
< |
cerr << program << ": Page.setUrl(" << url << ") failure.\n"; |
| 26 |
> |
api::Cerr << program << ": Page.setUrl(" << url << ") failure.\n"; |
| 27 |
|
|
| 28 |
< |
exit(1); |
| 28 |
> |
throw; |
| 29 |
|
} |
| 30 |
|
} |
| 31 |
|
|
| 32 |
< |
bool Page::operator==(const string& thing) |
| 32 |
> |
bool Page::operator==(const ext::String& thing) |
| 33 |
|
{ |
| 34 |
< |
if (address == thing) |
| 55 |
< |
{ |
| 56 |
< |
return true; |
| 57 |
< |
} |
| 58 |
< |
else if (path == thing) |
| 59 |
< |
{ |
| 60 |
< |
return true; |
| 61 |
< |
} |
| 62 |
< |
else if (title == thing) |
| 63 |
< |
{ |
| 64 |
< |
return true; |
| 65 |
< |
} |
| 66 |
< |
|
| 67 |
< |
return false; |
| 34 |
> |
return address == thing || path == thing || title == thing; |
| 35 |
|
} |
| 36 |
|
|
| 37 |
|
bool Page::operator==(Matcher& matcher) |
| 38 |
|
{ |
| 39 |
< |
if (address == matcher) |
| 73 |
< |
{ |
| 74 |
< |
return true; |
| 75 |
< |
} |
| 76 |
< |
else if (path == matcher) |
| 77 |
< |
{ |
| 78 |
< |
return true; |
| 79 |
< |
} |
| 80 |
< |
else if (title == matcher) |
| 81 |
< |
{ |
| 82 |
< |
return true; |
| 83 |
< |
} |
| 84 |
< |
|
| 85 |
< |
return false; |
| 39 |
> |
return address == matcher || path == matcher || title == matcher; |
| 40 |
|
} |
| 41 |
|
|
| 42 |
|
bool Page::operator==(const Page& page) const |
| 43 |
|
{ |
| 44 |
|
if (address == page.address) |
| 45 |
|
{ |
| 46 |
< |
if (path == page.path || title == page.title) return true; |
| 46 |
> |
return path == page.path || title == page.title; |
| 47 |
|
} |
| 48 |
|
|
| 49 |
|
return false; |
| 50 |
|
} |
| 51 |
|
|
| 52 |
< |
bool Page::operator<(const Page& page) const |
| 52 |
> |
xml::TextWriter& operator<<(xml::TextWriter& xml, Page& page) |
| 53 |
|
{ |
| 54 |
< |
if (address == page.address) |
| 101 |
< |
{ |
| 102 |
< |
return path < page.path; |
| 103 |
< |
} |
| 54 |
> |
xml::ScopeElement item(xml, "item"); |
| 55 |
|
|
| 56 |
< |
return address < page.address; |
| 57 |
< |
} |
| 56 |
> |
xml.OpenElement("link"); |
| 57 |
> |
xml.SetAttribute("address", page.getUrl()); |
| 58 |
> |
xml.OutputText(page.title); |
| 59 |
> |
xml.CloseElement(); |
| 60 |
|
|
| 61 |
< |
bool Page::operator>(const Page& page) const |
| 109 |
< |
{ |
| 110 |
< |
if (address == page.address) |
| 61 |
> |
if (!page.children.IsEmpty()) |
| 62 |
|
{ |
| 63 |
< |
return path > page.path; |
| 113 |
< |
} |
| 63 |
> |
xml::ScopeElement list(xml, "list"); |
| 64 |
|
|
| 65 |
< |
return address > page.address; |
| 116 |
< |
} |
| 117 |
< |
|
| 118 |
< |
ostream& operator<<(ostream& output, Page& page) |
| 119 |
< |
{ |
| 120 |
< |
string tab(page.tab, '\t'); |
| 121 |
< |
|
| 122 |
< |
output << tab << "<item><link address=\"" << page.getUrl() << "\">" |
| 123 |
< |
<< page.title << "</link>\n"; |
| 124 |
< |
|
| 125 |
< |
if (!page.children.empty()) |
| 126 |
< |
{ |
| 127 |
< |
output << tab << "\t<list>\n"; |
| 128 |
< |
|
| 129 |
< |
for (unsigned index = 0; index < page.children.size(); index++) |
| 130 |
< |
{ |
| 131 |
< |
output << page.children[index](page.tab + 1) << '\n'; |
| 132 |
< |
} |
| 133 |
< |
|
| 134 |
< |
output << tab << "\t</list>\n"; |
| 65 |
> |
_mforeach (ext::Vector<Page>, child, page.children) xml << *child; |
| 66 |
|
} |
| 67 |
|
|
| 68 |
< |
output << tab << "</item>"; |
| 138 |
< |
|
| 139 |
< |
return output; |
| 68 |
> |
return xml; |
| 69 |
|
} |
