Public Types | Public Member Functions | Data Fields | Static Public Attributes

HtmlParser Class Reference

#include <Update.h>

Collaboration diagram for HtmlParser:
Collaboration graph
[legend]

Public Types

enum  scan_state_t {
  SCAN_INIT = 1, SCAN_START, FIND_TAG_START, COPY_TAG,
  IGNORE_COMMENT_START, IGNORE_COMMENT, FIND_ATTR_START, COPY_ATTR,
  FIND_ATTR_VALUE_DELIMITER, FIND_ATTR_VALUE_START, COPY_ATTR_VALUE, VALIDATE_ENTRY,
  VALIDATE_ENTRY_RESTART, RESUME_ATTR_VALUE_SCAN, TERMINATE_COPY_ATTR_VALUE
}
enum  { MAX_TAG_NAME_LENGTH = 1024, MAX_ATTR_NAME_LENGTH = 1024 }

Public Member Functions

 HtmlParser ()
 ~HtmlParser ()
void Init (char *url, struct html_tag *allowed_html_tags, struct html_tag *allowed_html_attrs=NULL)
int ParseHtml (IOBufferReader *, char **, char **)
int ScanHtmlForURL (IOBufferReader *, char **, char **)
int AllowTagAttrValue ()
int ValidProtoScheme (char *)
int ValidSupportedProtoScheme (char *)
int ExtractURL (char **, char **)
int ConstructURL (char **, char **)
DynArray< char > * MakeURL (char *, char *, int, int)
DynArray< char > * PrependString (const char *, int, char *, int)

Data Fields

bool _attr_matched
char * _url
char * _comment_end_ptr
scan_state_t _scan_state
DynArray< char > _tag
DynArray< char > _attr
DynArray< char > _attr_value
intptr_t _attr_value_hash_char_index
unsigned char _attr_value_quoted
DynArray< char > _html_doc_base
DynArray< char > _result
struct html_tag * allowable_html_tags
struct html_tag * allowable_html_attrs

Static Public Attributes

static char default_zero_char = '\0'

Detailed Description

Definition at line 382 of file Update.h.

Member Enumeration Documentation

anonymous enum
Enumerator:
MAX_TAG_NAME_LENGTH 
MAX_ATTR_NAME_LENGTH 

Definition at line 407 of file Update.h.

enum HtmlParser::scan_state_t
Enumerator:
SCAN_INIT 
SCAN_START 
FIND_TAG_START 
COPY_TAG 
IGNORE_COMMENT_START 
IGNORE_COMMENT 
FIND_ATTR_START 
COPY_ATTR 
FIND_ATTR_VALUE_DELIMITER 
FIND_ATTR_VALUE_START 
COPY_ATTR_VALUE 
VALIDATE_ENTRY 
VALIDATE_ENTRY_RESTART 
RESUME_ATTR_VALUE_SCAN 
TERMINATE_COPY_ATTR_VALUE 

Definition at line 388 of file Update.h.

Constructor & Destructor Documentation

HtmlParser::HtmlParser (  )  [inline]

Definition at line 413 of file Update.h.

HtmlParser::~HtmlParser (  )  [inline]

Definition at line 422 of file Update.h.

Member Function Documentation

int HtmlParser::AllowTagAttrValue (  ) 

Definition at line 2133 of file Update.cc.

References _attr, _attr_matched, _attr_value, _tag, allowable_html_attrs, allowable_html_tags, html_tag::attr, DynArray< T >::length(), and html_tag::tag.

Referenced by ScanHtmlForURL().

int HtmlParser::ConstructURL ( char **  url,
char **  url_end 
)

Definition at line 2265 of file Update.cc.

References _html_doc_base, _result, _url, DynArray< T >::clear(), DynArray< T >::detach(), DynArray< T >::length(), MakeURL(), PrependString(), ValidProtoScheme(), and ValidSupportedProtoScheme().

Referenced by ParseHtml().

int HtmlParser::ExtractURL ( char **  url,
char **  url_end 
)

Definition at line 2192 of file Update.cc.

References _attr, _attr_value, _attr_value_hash_char_index, _html_doc_base, _tag, DynArray< T >::clear(), ParseRules::is_digit(), DynArray< T >::length(), and DynArray< T >::set_length().

Referenced by ScanHtmlForURL().

void HtmlParser::Init ( char *  url,
struct html_tag *  allowed_html_tags,
struct html_tag *  allowed_html_attrs = NULL 
) [inline]

Definition at line 425 of file Update.h.

References _attr_matched, _url, allowable_html_attrs, and allowable_html_tags.

Referenced by RecursiveHttpGet::Init(), and PrefetchTransform::PrefetchTransform().

DynArray< char > * HtmlParser::MakeURL ( char *  url,
char *  sub,
int  subsize,
int  relative_url 
)

Definition at line 2366 of file Update.cc.

References default_zero_char, and DynArray< T >::length().

Referenced by ConstructURL().

int HtmlParser::ParseHtml ( IOBufferReader *  r,
char **  url,
char **  url_end 
)

Definition at line 1794 of file Update.cc.

References ConstructURL(), and ScanHtmlForURL().

Referenced by PrefetchTransform::parse_data(), and RecursiveHttpGet::RecursiveHttpGetEvent().

DynArray< char > * HtmlParser::PrependString ( const char *  pre,
int  presize,
char *  sub,
int  subsize 
)

Definition at line 2441 of file Update.cc.

References default_zero_char, and DynArray< T >::length().

Referenced by ConstructURL().

int HtmlParser::ScanHtmlForURL ( IOBufferReader *  r,
char **  url,
char **  url_end 
)

Definition at line 1809 of file Update.cc.

References _attr, _attr_matched, _attr_value, _attr_value_hash_char_index, _attr_value_quoted, _comment_end_ptr, _scan_state, _tag, AllowTagAttrValue(), DynArray< T >::clear(), COPY_ATTR, COPY_ATTR_VALUE, COPY_TAG, ExtractURL(), FIND_ATTR_START, FIND_ATTR_VALUE_DELIMITER, FIND_ATTR_VALUE_START, FIND_TAG_START, HTML_COMMENT_END, HTML_COMMENT_TAG, IGNORE_COMMENT, IGNORE_COMMENT_START, ink_release_assert, DynArray< T >::length(), IOBufferReader::read(), RESUME_ATTR_VALUE_SCAN, SCAN_INIT, SCAN_START, TERMINATE_COPY_ATTR_VALUE, VALIDATE_ENTRY, and VALIDATE_ENTRY_RESTART.

Referenced by ParseHtml().

int HtmlParser::ValidProtoScheme ( char *  p  ) 

Definition at line 2168 of file Update.cc.

References html_tag::tag, and schemes_descriptor::tag.

Referenced by ConstructURL().

int HtmlParser::ValidSupportedProtoScheme ( char *  p  ) 

Definition at line 2180 of file Update.cc.

References html_tag::tag, and schemes_descriptor::tag.

Referenced by ConstructURL().

Field Documentation

DynArray<char> HtmlParser::_attr

Definition at line 447 of file Update.h.

Referenced by AllowTagAttrValue(), ExtractURL(), and ScanHtmlForURL().

bool HtmlParser::_attr_matched

Definition at line 441 of file Update.h.

Referenced by AllowTagAttrValue(), Init(), and ScanHtmlForURL().

DynArray<char> HtmlParser::_attr_value

Definition at line 448 of file Update.h.

Referenced by AllowTagAttrValue(), ExtractURL(), and ScanHtmlForURL().

intptr_t HtmlParser::_attr_value_hash_char_index

Definition at line 449 of file Update.h.

Referenced by ExtractURL(), and ScanHtmlForURL().

unsigned char HtmlParser::_attr_value_quoted

Definition at line 450 of file Update.h.

Referenced by ScanHtmlForURL().

char* HtmlParser::_comment_end_ptr

Definition at line 444 of file Update.h.

Referenced by ScanHtmlForURL().

DynArray<char> HtmlParser::_html_doc_base

Definition at line 451 of file Update.h.

Referenced by ConstructURL(), and ExtractURL().

DynArray<char> HtmlParser::_result

Definition at line 452 of file Update.h.

Referenced by ConstructURL().

scan_state_t HtmlParser::_scan_state

Definition at line 445 of file Update.h.

Referenced by ScanHtmlForURL().

DynArray<char> HtmlParser::_tag

Definition at line 446 of file Update.h.

Referenced by AllowTagAttrValue(), ExtractURL(), and ScanHtmlForURL().

char* HtmlParser::_url

Definition at line 443 of file Update.h.

Referenced by ConstructURL(), RecursiveHttpGet::ExitEventHandler(), Init(), RecursiveHttpGet::Init(), and RecursiveHttpGet::RecursiveHttpGetEvent().

struct html_tag* HtmlParser::allowable_html_attrs

Definition at line 455 of file Update.h.

Referenced by AllowTagAttrValue(), and Init().

struct html_tag* HtmlParser::allowable_html_tags

Definition at line 454 of file Update.h.

Referenced by AllowTagAttrValue(), and Init().

char HtmlParser::default_zero_char = '\0' [static]

Definition at line 386 of file Update.h.

Referenced by MakeURL(), and PrependString().

The documentation for this class was generated from the following files:
Generated by  doxygen 1.7.1