Save HTML document with dependencies implementation. More...

#include <assert.h>
#include <errno.h>
#include <stdio.h>
#include <string.h>
#include <strings.h>
#include <sys/types.h>
#include <dom/dom.h>
#include "utils/config.h"
#include "utils/regex.h"
#include "utils/corestrings.h"
#include "utils/log.h"
#include "utils/nsurl.h"
#include "utils/utf8.h"
#include "utils/utils.h"
#include "utils/file.h"
#include "utils/messages.h"
#include "utils/ascii.h"
#include "netsurf/content.h"
#include "content/hlcache.h"
#include "css/css.h"
#include "html/box.h"
#include "html/html_save.h"
#include "html/html.h"
#include "netsurf/misc.h"
#include "desktop/gui_internal.h"
#include "desktop/save_complete.h"

Include dependency graph for save_complete.c:

Data Structures
struct	save_complete_entry
	An entry in save_complete_list. More...

struct	save_complete_ctx

Typedefs
typedef struct save_complete_entry	save_complete_entry
	An entry in save_complete_list. More...

typedef struct save_complete_ctx	save_complete_ctx

Enumerations
enum	save_complete_event_type { EVENT_ENTER , EVENT_LEAVE }

Functions
static nserror	save_complete_save_html (save_complete_ctx ctx, hlcache_handle c, bool index)
	Save an HTML page with all dependencies, recursing through imported pages. More...

static nserror	save_complete_save_imported_sheets (save_complete_ctx ctx, struct nscss_import imports, uint32_t import_count)

static void	save_complete_ctx_initialise (save_complete_ctx ctx, const char path, save_complete_set_type_cb set_type)

static void	save_complete_ctx_finalise (save_complete_ctx *ctx)

static nserror	save_complete_ctx_add_content (save_complete_ctx ctx, struct hlcache_handle content)

static struct hlcache_handle *	save_complete_ctx_find_content (save_complete_ctx ctx, const nsurl url)
	find handle to content for url More...

static bool	save_complete_ctx_has_content (save_complete_ctx ctx, struct hlcache_handle content)

static nserror	save_complete_save_buffer (save_complete_ctx ctx, const char leafname, const uint8_t data, size_t data_len, lwc_string mime_type)

static int	snregexec (regex_t preg, const char string, size_t stringlen, size_t nmatch, regmatch_t pmatch[], int eflags)
	perform a posix regexec on a string without a null terminator More...

static uint8_t *	save_complete_rewrite_stylesheet_urls (save_complete_ctx ctx, const uint8_t source, size_t size, const nsurl base, size_t osize)
	Rewrite stylesheet @import rules for save complete. More...

static nserror	save_complete_save_stylesheet (save_complete_ctx ctx, hlcache_handle css)

static nserror	save_complete_save_html_stylesheet (save_complete_ctx ctx, struct html_stylesheet sheet)

static nserror	save_complete_save_html_stylesheets (save_complete_ctx ctx, hlcache_handle c)

static nserror	save_complete_save_html_object (save_complete_ctx ctx, hlcache_handle obj)

static nserror	save_complete_save_html_objects (save_complete_ctx ctx, hlcache_handle c)

static bool	save_complete_libdom_treewalk (dom_node root, bool(callback)(dom_node node, save_complete_event_type event_type, void ctx), void *ctx)

static bool	save_complete_rewrite_url_value (save_complete_ctx ctx, const char value, size_t value_len)

static bool	save_complete_write_value (save_complete_ctx ctx, const char value, size_t value_len)

static bool	save_complete_handle_attr_value (save_complete_ctx ctx, dom_string node_name, dom_string attr_name, dom_string attr_value)

static bool	save_complete_handle_attr (save_complete_ctx ctx, dom_string node_name, dom_attr *attr)

static bool	save_complete_handle_attrs (save_complete_ctx ctx, dom_string node_name, dom_namednodemap *attrs)

static bool	save_complete_handle_element (save_complete_ctx ctx, dom_node node, save_complete_event_type event_type)

static bool	save_complete_node_handler (dom_node node, save_complete_event_type event_type, void ctxin)

static nserror	save_complete_save_html_document (save_complete_ctx ctx, hlcache_handle c, bool index)

static nserror	save_complete_inventory (save_complete_ctx *ctx)
	Create the inventory file listing original URLs. More...

static nserror	regcomp_wrapper (regex_t preg, const char regex, int cflags)
	Compile a regular expression, handling errors. More...

void	save_complete_init (void)
	Initialise save complete module. More...

nserror	save_complete_finalise (void)
	Finalise save complete module. More...

nserror	save_complete (hlcache_handle c, const char path, save_complete_set_type_cb set_type)
	Save an HTML page with all dependencies. More...

Variables
static regex_t	save_complete_import_re

Detailed Description

Save HTML document with dependencies implementation.

Definition in file save_complete.c.

Typedef Documentation

◆ save_complete_ctx

typedef struct save_complete_ctx save_complete_ctx

◆ save_complete_entry

typedef struct save_complete_entry save_complete_entry

An entry in save_complete_list.

Enumeration Type Documentation

◆ save_complete_event_type

enum save_complete_event_type

Enumerator
EVENT_ENTER
EVENT_LEAVE

Definition at line 72 of file save_complete.c.

Function Documentation

◆ regcomp_wrapper()

static nserror regcomp_wrapper	(	regex_t *	preg,
		const char *	regex,
		int	cflags
	)

static

Compile a regular expression, handling errors.

Parameters as for regcomp(), see man regex.

Definition at line 1273 of file save_complete.c.

References NSERROR_INIT_FAILED, NSERROR_OK, NSLOG, regcomp(), and regerror().

Referenced by save_complete_init().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete()

nserror save_complete	(	struct hlcache_handle *	c,
		const char *	path,
		save_complete_set_type_cb	set_type
	)

Save an HTML page with all dependencies.

Parameters

c	CONTENT_HTML to save
path	Native path to directory to save in to (must exist)
set_type	Callback to set type of a file, or NULL

Returns: NSERROR_OK on success else error code

Definition at line 1330 of file save_complete.c.

References NSERROR_OK, path(), result, save_complete_ctx_finalise(), save_complete_ctx_initialise(), save_complete_inventory(), and save_complete_save_html().

Referenced by ami_file_save(), menu_save_page(), ro_gui_save_complete(), and savepage_button_clicked_cb().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_ctx_add_content()

static nserror save_complete_ctx_add_content	(	save_complete_ctx *	ctx,
		struct hlcache_handle *	content
	)

static

Definition at line 103 of file save_complete.c.

References save_complete_entry::content, save_complete_ctx::list, save_complete_entry::next, NSERROR_NOMEM, and NSERROR_OK.

Referenced by save_complete_save_html_object(), and save_complete_save_stylesheet().

Here is the caller graph for this function:

◆ save_complete_ctx_finalise()

static void save_complete_ctx_finalise ( save_complete_ctx * ctx )

static

Definition at line 91 of file save_complete.c.

References save_complete_ctx::list, and save_complete_entry::next.

Referenced by save_complete().

Here is the caller graph for this function:

◆ save_complete_ctx_find_content()

static struct hlcache_handle * save_complete_ctx_find_content	(	save_complete_ctx *	ctx,
		const nsurl *	url
	)

static

find handle to content for url

Parameters

ctx	The save context
url	The url to find content handle for

Returns: The content handle or NULL if not found.

Definition at line 128 of file save_complete.c.

References hlcache_entry::content, hlcache_handle::entry, hlcache_handle_get_url(), save_complete_ctx::list, hlcache_entry::next, nsurl_compare(), and NSURL_COMPLETE.

Referenced by save_complete_rewrite_stylesheet_urls(), save_complete_rewrite_url_value(), save_complete_save_html_object(), and save_complete_save_stylesheet().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_ctx_has_content()

static bool save_complete_ctx_has_content	(	save_complete_ctx *	ctx,
		struct hlcache_handle *	content
	)

static

Definition at line 145 of file save_complete.c.

References hlcache_entry::content, hlcache_handle::entry, hlcache_handle_get_content(), save_complete_ctx::list, and hlcache_entry::next.

Referenced by save_complete_save_html().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_ctx_initialise()

static void save_complete_ctx_initialise	(	save_complete_ctx *	ctx,
		const char *	path,
		save_complete_set_type_cb	set_type
	)

static

Definition at line 83 of file save_complete.c.

References save_complete_ctx::list, save_complete_ctx::path, path(), and save_complete_ctx::set_type.

Referenced by save_complete().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_finalise()

nserror save_complete_finalise ( void )

Finalise save complete module.

Definition at line 1322 of file save_complete.c.

References NSERROR_OK, regfree(), and save_complete_import_re.

Referenced by nsgtk_finalise().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_handle_attr()

static bool save_complete_handle_attr	(	save_complete_ctx *	ctx,
		dom_string *	node_name,
		dom_attr *	attr
	)

static

Definition at line 786 of file save_complete.c.

References save_complete_ctx::fp, node_name(), and save_complete_handle_attr_value().

Referenced by save_complete_handle_attrs().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_handle_attr_value()

static bool save_complete_handle_attr_value	(	save_complete_ctx *	ctx,
		dom_string *	node_name,
		dom_string *	attr_name,
		dom_string *	attr_value
	)

static

We only need to consider the following cases:

Attribute: Elements:

1) data object 2) href a, area, link 3) src script, input, frame, iframe, img 4) background any (except those above)

Definition at line 697 of file save_complete.c.

References node_name(), save_complete_rewrite_url_value(), save_complete_write_value(), and SLEN.

Referenced by save_complete_handle_attr().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_handle_attrs()

static bool save_complete_handle_attrs	(	save_complete_ctx *	ctx,
		dom_string *	node_name,
		dom_namednodemap *	attrs
	)

static

Definition at line 832 of file save_complete.c.

References node_name(), and save_complete_handle_attr().

Referenced by save_complete_handle_element().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_handle_element()

static bool save_complete_handle_element	(	save_complete_ctx *	ctx,
		dom_node *	node,
		save_complete_event_type	event_type
	)

static

Definition at line 865 of file save_complete.c.

References save_complete_ctx::base, content_html_object::content, EVENT_ENTER, EVENT_LEAVE, save_complete_ctx::fp, save_complete_ctx::iter_state, save_complete_handle_attrs(), save_complete_rewrite_stylesheet_urls(), and SLEN.

Referenced by save_complete_node_handler().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_init()

void save_complete_init ( void )

Initialise save complete module.

Definition at line 1289 of file save_complete.c.

References REG_EXTENDED, REG_ICASE, regcomp_wrapper(), and save_complete_import_re.

Referenced by gui_init(), main(), menu_save_page(), and nsgtk_setup().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_inventory()

static nserror save_complete_inventory ( save_complete_ctx * ctx )

static

Create the inventory file listing original URLs.

Definition at line 1237 of file save_complete.c.

References save_complete_entry::content, hlcache_handle_get_url(), save_complete_ctx::list, netsurf_mkpath(), save_complete_entry::next, NSERROR_OK, NSERROR_SAVE_FAILED, NSLOG, nsurl_access(), and save_complete_ctx::path.

Referenced by save_complete().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_libdom_treewalk()

static bool save_complete_libdom_treewalk	(	dom_node *	root,
		bool()(dom_node node, save_complete_event_type event_type, void *ctx)	callback,
		void *	ctx
	)

static

Definition at line 540 of file save_complete.c.

References EVENT_ENTER, EVENT_LEAVE, content_html_object::next, and root.

Referenced by save_complete_save_html_document().

Here is the caller graph for this function:

◆ save_complete_node_handler()

static bool save_complete_node_handler	(	dom_node *	node,
		save_complete_event_type	event_type,
		void *	ctxin
	)

static

Definition at line 1017 of file save_complete.c.

References EVENT_ENTER, save_complete_ctx::fp, save_complete_ctx::iter_state, NSERROR_OK, NSLOG, save_complete_handle_element(), text(), type, and utf8_to_html().

Referenced by save_complete_save_html_document().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_rewrite_stylesheet_urls()

static uint8_t * save_complete_rewrite_stylesheet_urls	(	save_complete_ctx *	ctx,
		const uint8_t *	source,
		size_t	size,
		const nsurl *	base,
		size_t *	osize
	)

static

Rewrite stylesheet @import rules for save complete.

Parameters

ctx	Save complete context.
source	stylesheet source.
size	size of source.
base	url of stylesheet.
osize	updated with the size of the result.

Returns: converted source, or NULL on out of memory.

Definition at line 234 of file save_complete.c.

References ascii_to_lower(), base, NSERROR_NOMEM, nsurl_join(), nsurl_unref(), regmatch_t::rm_eo, regmatch_t::rm_so, save_complete_ctx_find_content(), save_complete_import_re, SLEN, snregexec(), and strndup().

Referenced by save_complete_handle_element(), and save_complete_save_stylesheet().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_rewrite_url_value()

static bool save_complete_rewrite_url_value	(	save_complete_ctx *	ctx,
		const char *	value,
		size_t	value_len
	)

static

Definition at line 635 of file save_complete.c.

References save_complete_ctx::base, content_html_object::content, save_complete_ctx::fp, NSERROR_NOMEM, NSERROR_OK, nsurl_access(), nsurl_join(), nsurl_length(), nsurl_unref(), save_complete_ctx_find_content(), and utf8_to_html().

Referenced by save_complete_handle_attr_value().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ save_complete_save_buffer()

static nserror save_complete_save_buffer	(	save_complete_ctx *	ctx,
		const char *	leafname,
		const uint8_t *	data,
		size_t	data_len,
		lwc_string *	mime_type
	)