|
a/src/internfile/myhtmlparse.h |
|
b/src/internfile/myhtmlparse.h |
|
... |
|
... |
| 27 |
// not in all charsets and perhaps spans of all \xa0 should become a single
|
27 |
// not in all charsets and perhaps spans of all \xa0 should become a single
|
| 28 |
// \xa0?
|
28 |
// \xa0?
|
| 29 |
#define WHITESPACE " \t\n\r"
|
29 |
#define WHITESPACE " \t\n\r"
|
| 30 |
|
30 |
|
| 31 |
class MyHtmlParser : public HtmlParser {
|
31 |
class MyHtmlParser : public HtmlParser {
|
| 32 |
public:
|
32 |
public:
|
| 33 |
bool in_script_tag;
|
33 |
bool in_script_tag;
|
| 34 |
bool in_style_tag;
|
34 |
bool in_style_tag;
|
| 35 |
bool in_body_tag;
|
35 |
bool in_body_tag;
|
|
|
36 |
bool in_pre_tag;
|
| 36 |
bool pending_space;
|
37 |
bool pending_space;
|
| 37 |
string title, sample, keywords, dump, dmtime;
|
38 |
string title, sample, keywords, dump, dmtime;
|
| 38 |
string ocharset; // This is the charset our user thinks the doc was
|
39 |
string ocharset; // This is the charset our user thinks the doc was
|
| 39 |
string charset; // This is the charset it was supposedly converted to
|
40 |
string charset; // This is the charset it was supposedly converted to
|
| 40 |
string doccharset; // Set this to value of charset parameter in header
|
41 |
string doccharset; // Set this to value of charset parameter in header
|
| 41 |
bool indexing_allowed;
|
42 |
bool indexing_allowed;
|
| 42 |
void process_text(const string &text);
|
43 |
void process_text(const string &text);
|
| 43 |
void opening_tag(const string &tag, const map<string,string> &p);
|
44 |
void opening_tag(const string &tag, const map<string,string> &p);
|
| 44 |
void closing_tag(const string &tag);
|
45 |
void closing_tag(const string &tag);
|
| 45 |
void do_eof();
|
46 |
void do_eof();
|
| 46 |
MyHtmlParser() :
|
47 |
MyHtmlParser() :
|
| 47 |
in_script_tag(false),
|
48 |
in_script_tag(false),
|
| 48 |
in_style_tag(false),
|
49 |
in_style_tag(false),
|
| 49 |
in_body_tag(false),
|
50 |
in_body_tag(false),
|
|
|
51 |
in_pre_tag(false),
|
| 50 |
pending_space(false),
|
52 |
pending_space(false),
|
| 51 |
indexing_allowed(true) { }
|
53 |
indexing_allowed(true) { }
|
| 52 |
};
|
54 |
};
|