Parse and normalize all URL's inside Dillo. More...

#include "d_size.h"
#include "../dlib/dlib.h"

Include dependency graph for url.h:

This graph shows which files directly or indirectly include this file:

Classes
struct	DilloUrl

Macros
#define	URL_HTTP_PORT 80

#define	URL_HTTPS_PORT 443

#define	URL_HOST_ERROR -1

#define	URL_HOST_NAME 0

#define	URL_HOST_IPV4 1

#define	URL_HOST_IPV6 2

#define	URL_Get (1 << 0)

#define	URL_Post (1 << 1)

#define	URL_E2EQuery (1 << 5)

#define	URL_ReloadPage (1 << 7)

#define	URL_ReloadFromCache (1 << 8)

#define	URL_IgnoreScroll (1 << 9)

#define	URL_SpamSafe (1 << 10)

#define	URL_MultipartEnc (1 << 11)

#define	URL_SCHEME_(u) (u)->scheme

#define	URL_AUTHORITY_(u) (u)->authority

#define	URL_PATH_(u) (u)->path

#define	URL_QUERY_(u) (u)->query

#define	URL_FRAGMENT_(u) (u)->fragment

#define	URL_HOST_(u) a_Url_hostname(u)

#define	URL_STR_(u) a_Url_str(u)

#define	URL_DATA_(u) (u)->data

#define	URL_PORT_(u) (URL_HOST(u), (u)->port)

#define	URL_FLAGS_(u) (u)->flags

#define	URL_ILLEGAL_CHARS_(u) (u)->illegal_chars

#define	URL_ILLEGAL_CHARS_SPC_(u) (u)->illegal_chars_spc

#define	NPTR2STR(p) ((p) ? (p) : "")

#define	URL_SCHEME(u) NPTR2STR(URL_SCHEME_(u))

#define	URL_AUTHORITY(u) NPTR2STR(URL_AUTHORITY_(u))

#define	URL_PATH(u) NPTR2STR(URL_PATH_(u))

#define	URL_QUERY(u) NPTR2STR(URL_QUERY_(u))

#define	URL_FRAGMENT(u) NPTR2STR(URL_FRAGMENT_(u))

#define	URL_HOST(u) NPTR2STR(URL_HOST_(u))

#define	URL_STR(u) NPTR2STR(URL_STR_(u))

#define	URL_DATA(u) URL_DATA_(u)

#define	URL_PORT(u) URL_PORT_(u)

#define	URL_FLAGS(u) URL_FLAGS_(u)

#define	URL_ILLEGAL_CHARS(u) URL_ILLEGAL_CHARS_(u)

#define	URL_ILLEGAL_CHARS_SPC(u) URL_ILLEGAL_CHARS_SPC_(u)

Functions
DilloUrl *	a_Url_new (const char url_str, const char base_url)
	Transform (and resolve) an URL string into the respective DilloURL.

void	a_Url_free (DilloUrl *u)
	Free a DilloUrl.

char *	a_Url_str (const DilloUrl *url)
	Return the url as a string.

const char *	a_Url_hostname (const DilloUrl *u)
	Return the hostname as a string.

DilloUrl *	a_Url_dup (const DilloUrl *u)
	Duplicate a Url structure.

int	a_Url_cmp (const DilloUrl A, const DilloUrl B)
	Compare two Url's to check if they're the same, or which one is bigger.

void	a_Url_set_flags (DilloUrl *u, int flags)
	Set DilloUrl flags.

void	a_Url_set_data (DilloUrl u, Dstr *data)
	Set DilloUrl data (like POST info, etc.)

void	a_Url_set_ismap_coords (DilloUrl u, char coord_str)
	Set DilloUrl ismap coordinates.

char *	a_Url_decode_hex_str (const char *str)
	Parse possible hexadecimal octets in the URI path.

char *	a_Url_encode_hex_str (const char *str)
	Urlencode 'str'.

char *	a_Url_string_strip_delimiters (const char *str)
	RFC-3986 suggests this stripping when "importing" URLs from other media.

int	a_Url_host_type (const char *host)
	What type of host is this?

bool_t	a_Url_same_organization (const DilloUrl u1, const DilloUrl u2)

Detailed Description

Parse and normalize all URL's inside Dillo.

Definition in file url.h.

Macro Definition Documentation

◆ NPTR2STR

#define NPTR2STR ( p ) ((p) ? (p) : "")

Definition at line 69 of file url.h.

◆ URL_AUTHORITY

#define URL_AUTHORITY ( u ) NPTR2STR(URL_AUTHORITY_(u))

Definition at line 71 of file url.h.

◆ URL_AUTHORITY_

#define URL_AUTHORITY_ ( u ) (u)->authority

Definition at line 50 of file url.h.

◆ URL_DATA

#define URL_DATA ( u ) URL_DATA_(u)

Definition at line 77 of file url.h.

◆ URL_DATA_

#define URL_DATA_ ( u ) (u)->data

Definition at line 57 of file url.h.

◆ URL_E2EQuery

#define URL_E2EQuery (1 << 5)

Definition at line 35 of file url.h.

◆ URL_FLAGS

#define URL_FLAGS ( u ) URL_FLAGS_(u)

Definition at line 79 of file url.h.

◆ URL_FLAGS_

#define URL_FLAGS_ ( u ) (u)->flags

Definition at line 60 of file url.h.

◆ URL_FRAGMENT

#define URL_FRAGMENT ( u ) NPTR2STR(URL_FRAGMENT_(u))

Definition at line 74 of file url.h.

◆ URL_FRAGMENT_

#define URL_FRAGMENT_ ( u ) (u)->fragment

Definition at line 53 of file url.h.

◆ URL_Get

#define URL_Get (1 << 0)

Definition at line 32 of file url.h.

◆ URL_HOST

#define URL_HOST ( u ) NPTR2STR(URL_HOST_(u))

Definition at line 75 of file url.h.

◆ URL_HOST_

#define URL_HOST_ ( u ) a_Url_hostname(u)

Definition at line 54 of file url.h.

◆ URL_HOST_ERROR

#define URL_HOST_ERROR -1

Definition at line 23 of file url.h.

◆ URL_HOST_IPV4

#define URL_HOST_IPV4 1

Definition at line 25 of file url.h.

◆ URL_HOST_IPV6

#define URL_HOST_IPV6 2

Definition at line 26 of file url.h.

◆ URL_HOST_NAME

#define URL_HOST_NAME 0

Definition at line 24 of file url.h.

◆ URL_HTTP_PORT

#define URL_HTTP_PORT 80

Definition at line 19 of file url.h.

◆ URL_HTTPS_PORT

#define URL_HTTPS_PORT 443

Definition at line 20 of file url.h.

◆ URL_IgnoreScroll

#define URL_IgnoreScroll (1 << 9)

Definition at line 39 of file url.h.

◆ URL_ILLEGAL_CHARS

#define URL_ILLEGAL_CHARS ( u ) URL_ILLEGAL_CHARS_(u)

Definition at line 80 of file url.h.

◆ URL_ILLEGAL_CHARS_

#define URL_ILLEGAL_CHARS_ ( u ) (u)->illegal_chars

Definition at line 61 of file url.h.

◆ URL_ILLEGAL_CHARS_SPC

#define URL_ILLEGAL_CHARS_SPC ( u ) URL_ILLEGAL_CHARS_SPC_(u)

Definition at line 81 of file url.h.

◆ URL_ILLEGAL_CHARS_SPC_

#define URL_ILLEGAL_CHARS_SPC_ ( u ) (u)->illegal_chars_spc

Definition at line 62 of file url.h.

◆ URL_MultipartEnc

#define URL_MultipartEnc (1 << 11)

Definition at line 42 of file url.h.

◆ URL_PATH

#define URL_PATH ( u ) NPTR2STR(URL_PATH_(u))

Definition at line 72 of file url.h.

◆ URL_PATH_

#define URL_PATH_ ( u ) (u)->path

Definition at line 51 of file url.h.

◆ URL_PORT

#define URL_PORT ( u ) URL_PORT_(u)

Definition at line 78 of file url.h.

◆ URL_PORT_

#define URL_PORT_ ( u ) (URL_HOST(u), (u)->port)

Definition at line 59 of file url.h.

◆ URL_Post

#define URL_Post (1 << 1)

Definition at line 33 of file url.h.

◆ URL_QUERY

#define URL_QUERY ( u ) NPTR2STR(URL_QUERY_(u))

Definition at line 73 of file url.h.

◆ URL_QUERY_

#define URL_QUERY_ ( u ) (u)->query

Definition at line 52 of file url.h.

◆ URL_ReloadFromCache

#define URL_ReloadFromCache (1 << 8)

Definition at line 37 of file url.h.

◆ URL_ReloadPage

#define URL_ReloadPage (1 << 7)

Definition at line 36 of file url.h.

◆ URL_SCHEME

#define URL_SCHEME ( u ) NPTR2STR(URL_SCHEME_(u))

Definition at line 70 of file url.h.

◆ URL_SCHEME_

#define URL_SCHEME_ ( u ) (u)->scheme

Definition at line 49 of file url.h.

◆ URL_SpamSafe

#define URL_SpamSafe (1 << 10)

Definition at line 40 of file url.h.

◆ URL_STR

#define URL_STR ( u ) NPTR2STR(URL_STR_(u))

Definition at line 76 of file url.h.

◆ URL_STR_

#define URL_STR_ ( u ) a_Url_str(u)

Definition at line 55 of file url.h.

Function Documentation

◆ a_Url_cmp()

int a_Url_cmp	(	const DilloUrl *	A,
		const DilloUrl *	B
	)

Compare two Url's to check if they're the same, or which one is bigger.

The fields which are compared here are: <scheme>, <authority>, <path>, <query> and <data> Other fields are left for the caller to check

Return value: 0 if equal, > 0 if A > B, < 0 if A < B.

Note: this function defines a sorting order different from strcmp!

Definition at line 506 of file url.c.

References DilloUrl::authority, DilloUrl::data, dReturn_val_if_fail, dStr_cmp(), DilloUrl::path, DilloUrl::query, DilloUrl::scheme, URL_STR_FIELD_CMP, and URL_STR_FIELD_I_CMP.

Referenced by a_Bw_add_url(), a_Bw_get_url_doc(), a_Capi_conn_abort_by_url(), a_History_add_url(), a_History_get_title_by_url(), a_History_set_title_by_url(), a_Nav_cancel_expect_if_eq(), a_Nav_push(), Cache_entry_by_url_cmp(), Cache_entry_cmp(), Dicache_entry_cmp(), Html_tag_open_meta(), DilloHtml::loadImages(), and Nav_open_url().

◆ a_Url_decode_hex_str()

char * a_Url_decode_hex_str ( const char * str )

Parse possible hexadecimal octets in the URI path.

Returns a new allocated string.

Definition at line 586 of file url.c.

References dNew, dRealloc(), dStrdup(), and Url_decode_hex_octet().

◆ a_Url_dup()

DilloUrl * a_Url_dup ( const DilloUrl * u )

Duplicate a Url structure.

Definition at line 477 of file url.c.

References DilloUrl::data, dReturn_val_if_fail, dStr_append_l(), dStr_new(), dStr_sized_new(), DilloUrl::flags, DilloUrl::illegal_chars, DilloUrl::illegal_chars_spc, DilloUrl::ismap_url_len, DilloUrl::port, URL_DATA, Url_object_new(), URL_STR, URL_STR_, and DilloUrl::url_string.

Referenced by a_Bw_add_url(), a_Bw_expect(), a_Capi_set_vsource_url(), a_History_add_url(), a_Http_init(), a_Menu_bugmeter_popup(), a_Menu_form_popup(), a_Menu_image_popup(), a_Menu_link_popup(), a_Menu_page_popup(), a_Nav_expect_done(), a_Nav_redirection0(), a_Web_new(), DilloHtml::addCssUrl(), Auth_do_auth_dialog(), Cache_auth_entry(), Cache_entry_init(), Cache_process_queue(), Capi_conn_new(), Dicache_add_entry(), DilloHtml::DilloHtml(), Html_tag_open_form(), Html_tag_open_isindex(), Http_get(), Nav_reload_callback(), Nav_repush(), StyleEngine::StyleEngine(), Tls_conn_new(), and Tls_conn_new().

◆ a_Url_encode_hex_str()

char * a_Url_encode_hex_str ( const char * str )

Urlencode 'str'.

-RL :: According to the RFC 1738, only alphanumerics, the special characters "$-_.+!*'(),", and reserved characters ";/?:@=&" used for their reserved purposes may be used unencoded within a URL. We'll escape everything but alphanumeric and "-_.*" (as lynx). –Jcid

Note: the content type "application/x-www-form-urlencoded" is used: i.e., ' ' -> '+' and '
' -> CR LF (see HTML 4.01, Sec. 17.13.4)

Definition at line 620 of file url.c.

References d_isascii, dIsalnum, dNew, and HEX.

Referenced by Menu_bugmeter_validate(), and UIcmd_make_search_str().

◆ a_Url_free()

void a_Url_free ( DilloUrl * url )

Free a DilloUrl.

Do nothing if the argument is NULL

Definition at line 208 of file url.c.

References DilloUrl::authority, DilloUrl::buffer, DilloUrl::data, dFree(), dStr_free(), DilloUrl::hostname, TRUE, and DilloUrl::url_string.

Referenced by a_Bw_cancel_expect(), a_Bw_cleanup(), a_Bw_expect(), a_Bw_free(), a_Cache_init(), a_Capi_set_vsource_url(), a_Dicache_freeall(), a_History_freeall(), a_Http_freeall(), a_Menu_bugmeter_popup(), a_Menu_file_popup(), a_Menu_form_popup(), a_Menu_image_popup(), a_Menu_link_popup(), a_Menu_page_popup(), a_Nav_expect_done(), a_Nav_redirection0(), a_Prefs_freeall(), a_UIcmd_book(), a_UIcmd_open_file(), a_UIcmd_open_urlstr(), a_UIcmd_view_page_source(), a_Web_free(), StyleEngine::apply(), Auth_do_auth_dialog(), Cache_auth_callback(), Cache_entry_free(), Cache_process_queue(), Cache_redirect(), Cache_savelink_cb(), Capi_conn_unref(), Dicache_remove(), Html_tag_content_img(), Html_tag_content_map(), Html_tag_open_base(), Html_tag_open_form(), Html_tag_open_isindex(), Html_tag_open_link(), Html_tag_open_meta(), Html_tag_open_object(), Http_socket_free(), main(), Nav_redirection0_callback(), Nav_reload_callback(), Nav_repush(), CssParser::parseImport(), parseOption(), CssParser::parseUrl(), Tls_close_by_key(), Tls_close_by_key(), Url_resolve_relative(), DilloHtml::~DilloHtml(), and StyleEngine::~StyleEngine().

◆ a_Url_host_type()

int a_Url_host_type ( const char * host )

What type of host is this?

Definition at line 683 of file url.c.

References URL_HOST_ERROR, URL_HOST_IPV4, URL_HOST_IPV6, and URL_HOST_NAME.

Referenced by a_Tls_openssl_connect(), Cache_parse_header(), Tls_cert_authorities_print_summary(), and Url_host_find_public_suffix().

◆ a_Url_hostname()

const char * a_Url_hostname ( const DilloUrl * u )

Return the hostname as a string.

(initializing 'hostname' and 'port' fields if necessary) Note: a similar approach can be taken for user:password auth.

Definition at line 98 of file url.c.

References DilloUrl::authority, dStrAsciiCasecmp(), dStrndup(), DilloUrl::hostname, DilloUrl::port, DilloUrl::scheme, URL_HTTP_PORT, and URL_HTTPS_PORT.

Referenced by a_Url_new().

◆ a_Url_new()

DilloUrl * a_Url_new	(	const char *	url_str,
		const char *	base_url
	)

Transform (and resolve) an URL string into the respective DilloURL.

If URL = "http://dillo.sf.net:8080/index.html?long#part2" then the resulting DilloURL should be:

DilloURL = {
   url_string         = "http://dillo.sf.net:8080/index.html?long#part2"
   scheme             = "http"
   authority          = "dillo.sf.net:8080:
   path               = "/index.html"
   query              = "long"
   fragment           = "part2"
   hostname           = "dillo.sf.net"
   port               = 8080
   flags              = URL_Get
   data               = Dstr * ("")
   ismap_url_len      = 0
}

Return NULL if URL is badly formed.

Definition at line 371 of file url.c.

References _MSG, a_Hsts_require_https(), a_Url_hostname(), DilloUrl::authority, DilloUrl::data, dFree(), dNew, dStr_free(), dStr_new(), dStrAsciiCasecmp(), dStrconcat(), FALSE, HEX, DilloPrefs::http_force_https, DilloPrefs::http_strict_transport_security, DilloUrl::illegal_chars, DilloUrl::illegal_chars_spc, DilloUrl::port, prefs, DilloUrl::scheme, Dstr::str, TRUE, URL_HTTP_PORT, URL_HTTPS_PORT, Url_object_new(), Url_resolve_relative(), and DilloUrl::url_string.

Referenced by a_Cache_init(), a_Html_url_new(), a_Http_init(), a_Prefs_init(), a_UIcmd_book(), a_UIcmd_open_file(), a_UIcmd_open_urlstr(), a_UIcmd_view_page_source(), StyleEngine::apply(), Cache_parse_header(), Cache_redirect(), makeStartUrl(), parseOption(), and CssParser::parseUrl().

◆ a_Url_same_organization()

bool_t a_Url_same_organization	(	const DilloUrl *	u1,
		const DilloUrl *	u2
	)

Definition at line 798 of file url.c.

References dStrAsciiCasecmp(), FALSE, TRUE, URL_HOST, and Url_host_find_public_suffix().

Referenced by a_Cookies_get_query(), a_Domain_permit(), and Cache_parse_header().

◆ a_Url_set_data()

void a_Url_set_data	(	DilloUrl *	u,
		Dstr **	data
	)

Set DilloUrl data (like POST info, etc.)

Definition at line 536 of file url.c.

References DilloUrl::data, and dStr_free().

◆ a_Url_set_flags()

void a_Url_set_flags	(	DilloUrl *	u,
		int	flags
	)

Set DilloUrl flags.

Definition at line 527 of file url.c.

References DilloUrl::flags.

Referenced by a_Capi_open_url(), a_Nav_expect_done(), a_Nav_redirection0(), Cache_redirect(), Html_tag_open_base(), makeStartUrl(), Nav_reload_callback(), and Nav_repush().

◆ a_Url_set_ismap_coords()

void a_Url_set_ismap_coords	(	DilloUrl *	u,
		char *	coord_str
	)

Set DilloUrl ismap coordinates.

(this is optimized for not hogging the CPU)

Definition at line 549 of file url.c.

References dReturn_if_fail, dStr_append(), dStr_truncate(), DilloUrl::ismap_url_len, Dstr::len, DilloUrl::query, Dstr::str, URL_STR_, and DilloUrl::url_string.

Referenced by Html_set_link_coordinates().

◆ a_Url_str()

char * a_Url_str ( const DilloUrl * u )

Return the url as a string.

(initializing 'url_string' field if necessary)

Definition at line 66 of file url.c.

References DilloUrl::authority, dReturn_val_if_fail, dStr_sized_new(), dStr_sprintf(), DilloUrl::fragment, DilloUrl::path, DilloUrl::query, DilloUrl::scheme, Dstr::str, and DilloUrl::url_string.

Referenced by DilloHtml::HtmlLinkReceiver::click(), CssParser::parseImport(), and CssParser::parseUrl().

◆ a_Url_string_strip_delimiters()

char * a_Url_string_strip_delimiters ( const char * str )

RFC-3986 suggests this stripping when "importing" URLs from other media.

Strip: "URL:", enclosing < >, and embedded whitespace. (We also strip illegal chars: 00-1F and 7F-FF)

Definition at line 658 of file url.c.

References dStrdup().

Referenced by a_UIcmd_open_urlstr(), and makeStartUrl().

Classes

Macros

Functions