Core part of search improvement:

+ Split up browser window and content related elements of html & text search.
+ Associate search contexts with contents, rather than browser windows.
+ Remove highlighting of search terms dependency on current_redraw_browser.
+ Fixes issues with search contexts being applied to the wrong content.


svn path=/trunk/netsurf/; revision=12646

1 files changed, 35 insertions, 643 deletions

diff --git a/desktop/search.c b/desktop/search.c
index 014014e71..36027044b 100644
--- a/desktop/search.c
+++ b/desktop/search.c
@@ -34,6 +34,7 @@
 #include "desktop/selection.h"
 #include "render/box.h"
 #include "render/html.h"
+#include "render/search.h"
 #include "render/textplain.h"
 #include "utils/config.h"
 #include "utils/log.h"
@@ -42,102 +43,39 @@
 #include "utils/utils.h"
 
 
-#ifndef NOF_ELEMENTS
-#define NOF_ELEMENTS(array) (sizeof(array)/sizeof(*(array)))
-#endif
 
+bool browser_window_search_create_context(struct browser_window *bw, 
+		struct search_callbacks *callbacks, void *p)
+{
+	assert(bw != NULL);
 
-struct list_entry {
-	unsigned start_idx;	/* start position of match */
-	unsigned end_idx;	/* end of match */
-
-	struct box *start_box;	/* used only for html contents */
-	struct box *end_box;
+	if (bw->cur_search != NULL)
+		search_destroy_context(bw->cur_search);
+	bw->cur_search = NULL;
 
-	struct selection *sel;
+	if (!bw->current_content)
+		return false;
 
-	struct list_entry *prev;
-	struct list_entry *next;
-};
+	bw->cur_search = search_create_context(bw->current_content,
+			callbacks, p);
 
-struct search_context {
-	struct browser_window 		*bw;
-	char 				*string;
-	bool 				prev_case_sens;
-	bool 				newsearch;
-	bool 				insert;
-	void 				*p; /* front-specific data */
-	struct search_callbacks		*callbacks;
-	struct list_entry 		*found;
-	struct list_entry 		*current; /* first for select all */
-};
+	if (bw->cur_search == NULL)
+		return false;
 
-static void search_text(const char *string, int string_len,
-		struct search_context *context, search_flags_t flags);
-static const char *find_pattern(const char *string, int s_len,
-		const char *pattern, int p_len, bool case_sens, 
-		unsigned int *m_len);
-static bool find_occurrences_html(const char *pattern, int p_len,
-		struct box *cur, bool case_sens,
-		struct search_context *context);
-static bool find_occurrences_text(const char *pattern, int p_len,
-		struct content *c, bool case_sens,
-		struct search_context *context);
-static struct list_entry *add_entry(unsigned start_idx, unsigned end_idx,
-		struct search_context *context);
-static void free_matches(struct search_context *context);
+	return true;
+}
 
 
-/**
- * create a search_context
- * \param bw the browser_window the search_context is connected to
- * \param callbacks the callbacks to modify appearance according to results
- * \param p the pointer to send to the callbacks
- * \return true for success
- */
-bool search_create_context(struct browser_window *bw,
-		struct search_callbacks *callbacks, void *p)
+void browser_window_search_destroy_context(struct browser_window *bw)
 {
-	struct search_context *context;
-	struct list_entry *search_head;
-	
-	context = malloc(sizeof(struct search_context));
-	if (context == NULL) {
-		warn_user("NoMemory", 0);
-		return false;
-	}
-	
-	search_head = malloc(sizeof(struct list_entry));
-	if (search_head == NULL) {
-		warn_user("NoMemory", 0);
-		free(context);
-		return false;
-	}
-	
-	if (bw->search_context != NULL)
-		search_destroy_context(bw->search_context);
+	assert(bw != NULL);
 
-	search_head->start_idx = 0;
-	search_head->end_idx = 0;
-	search_head->start_box = NULL;
-	search_head->end_box = NULL;
-	search_head->sel = NULL;
-	search_head->prev = NULL;
-	search_head->next = NULL;
-	
-	context->found = search_head;
-	context->current = NULL;
-	context->string = NULL;
-	context->prev_case_sens = false;
-	context->newsearch = true;
-	context->insert = true;
-	context->bw = bw;
-	context->callbacks = callbacks;
-	context->p = p;
-	
-	bw->search_context = context;
-	return true;
+	if (bw->cur_search != NULL)
+		search_destroy_context(bw->cur_search);
+	bw->cur_search = NULL;
 }
+
+
 /**
  * to simplify calls to search_step(); checks that the browser_window is
  * non-NULL, creates a new search_context in case of a new search
@@ -146,579 +84,33 @@ bool search_create_context(struct browser_window *bw,
  * \param p a pointer returned to the callbacks
  * \return true for success
  */
-bool search_verify_new(struct browser_window *bw,
+bool browser_window_search_verify_new(struct browser_window *bw,
 		struct search_callbacks *callbacks, void *p)
 {
 	if (bw == NULL)
 		return false;
-	if (bw->search_context == NULL)
-		return search_create_context(bw, callbacks, p);
-	return true;
-}
-
-/**
- * Begins/continues the search process
- * Note that this may be called many times for a single search.
- *
- * \param bw the browser_window to search in
- * \param flags the flags forward/back etc
- * \param string the string to match
- */
-
-void search_step(struct search_context *context, search_flags_t flags,
-		const char *string)
-{
-	int string_len;
-	int i = 0;
-	
-	if ((context == NULL) || (context->callbacks == NULL)) {
-		warn_user("SearchError", 0);
-		return;
-	}
-	
-	
-	if (context->callbacks->add_recent != NULL)
-		context->callbacks->add_recent(string, context->p);
-
-	string_len = strlen(string);
-	for(i = 0; i < string_len; i++)
-		if (string[i] != '#' && string[i] != '*') break;
-	if (i >= string_len) {
-		free_matches(context);
-		if (context->callbacks->status != NULL)
-			context->callbacks->status(true, context->p);
-		if (context->callbacks->back_state != NULL)
-			context->callbacks->back_state(false, context->p);
-		if (context->callbacks->forward_state != NULL)
-			context->callbacks->forward_state(false, context->p);
-		browser_window_set_scroll(context->bw, 0, 0);
-		return;
-	}
-	search_text(string, string_len, context, flags);
-}
-
-/**
- * Release the memory used by the list of matches,
- * deleting selection objects too
- */
-
-void free_matches(struct search_context *context)
-{
-	struct list_entry *a;
-	struct list_entry *b;
-	
-	a = context->found->next;
-
-	/* empty the list before clearing and deleting the
-	   selections because the the clearing updates the
-	   screen immediately, causing nested accesses to the list */
-
-	context->found->prev = NULL;
-	context->found->next = NULL;
-
-	for (; a; a = b) {
-		b = a->next;
-		if (a->sel) {
-			selection_clear(a->sel, true);
-			selection_destroy(a->sel);
-		}
-		free(a);
-	}
-}
-
-/**
- * Search for a string in the box tree
- *
- * \param string the string to search for
- * \param string_len length of search string
- */
-void search_text(const char *string, int string_len,
-		struct search_context *context, search_flags_t flags)
-{
-	struct rect bounds;
-	hlcache_handle *h;
-	struct box *box = NULL;
-	bool case_sensitive, forwards, showall;
-	
-	case_sensitive = ((flags & SEARCH_FLAG_CASE_SENSITIVE) != 0) ?
-			true : false;
-	forwards = ((flags & SEARCH_FLAG_FORWARDS) != 0) ? true : false;
-	showall = ((flags & SEARCH_FLAG_SHOWALL) != 0) ? true : false;
-
-	if (context->bw == NULL)
-		return;
-	h = context->bw->current_content;
-
-	/* only handle html contents */
-	if ((!h) || (content_get_type(h) != CONTENT_HTML &&
-			content_get_type(h) != CONTENT_TEXTPLAIN))
-		return;
-        
-	if (content_get_type(h) == CONTENT_HTML) {
-		box = html_get_box_tree(h);
-
-		if (!box)
-			return;
-	}
-
-	/* LOG(("do_search '%s' - '%s' (%p, %p) %p (%d, %d) %d",
-		search_data.string, string, search_data.content, c, search_data.found->next,
-		search_data.prev_case_sens, case_sens, forwards)); */
-
-	/* check if we need to start a new search or continue an old one */
-	if (context->newsearch) {
-		bool res;
-
-		if (context->string != NULL)
-			free(context->string);
-		context->current = NULL;
-		free_matches(context);
-
-		context->string = malloc(string_len + 1);
-		if (context->string != NULL) {
-			memcpy(context->string, string, string_len);
-			context->string[string_len] = '\0';
-		}
-
-		if ((context->callbacks != NULL) && 
-				(context->callbacks->hourglass != NULL))
-			context->callbacks->hourglass(true, context->p);
-
-		if (content_get_type(h) == CONTENT_HTML)
-			res = find_occurrences_html(string, string_len,
-					box, case_sensitive, context);
-		else {
-			assert(content_get_type(h) == CONTENT_TEXTPLAIN);
-			res = find_occurrences_text(string, string_len,
-					hlcache_handle_get_content(h),
-					case_sensitive, context);
-		}
-
-		if (!res) {
-			free_matches(context);
-			if ((context->callbacks != NULL) && 
-					(context->callbacks->hourglass !=
-					NULL))
-				context->callbacks->hourglass(false,
-						context->p);
-			return;
-		}
-		if ((context->callbacks != NULL) && 
-				(context->callbacks->hourglass != NULL))
-			context->callbacks->hourglass(false, context->p);
-
-		context->prev_case_sens = case_sensitive;
-/* LOG(("%d %p %p (%p, %p)", new, search_data.found->next, search_data.current,
-		search_data.current->prev, search_data.current->next)); */
-		/* new search, beginning at the top of the page */
-		context->current = context->found->next;
-		context->newsearch = false;
-	}
-	else if (context->current != NULL) {
-		/* continued search in the direction specified */
-		if (forwards) {
-			if (context->current->next)
-				context->current = context->current->next;
-		}
-		else {
-			if (context->current->prev)
-				context->current = context->current->prev;
-		}
-	}
-
-	if (context->callbacks == NULL)
-		return;
-	if (context->callbacks->status != NULL)
-		context->callbacks->status((context->current != NULL), 
-				context->p);
-	search_show_all(showall, context);
-
-	if (context->callbacks->back_state != NULL)
-		context->callbacks->back_state((context->current != NULL) &&
-				(context->current->prev != NULL),
-				context->p);
-	if (context->callbacks->forward_state != NULL)
-		context->callbacks->forward_state((context->current != NULL) &&
-				(context->current->next != NULL), context->p);
-
-	if (context->current == NULL)
-		return;
-
-	switch (content_get_type(h)) {
-		case CONTENT_HTML:
-			/* get box position and jump to it */
-			box_coords(context->current->start_box,
-					&bounds.x0, &bounds.y0);
-			/* \todo: move x0 in by correct idx */
-			box_coords(context->current->end_box,
-					&bounds.x1, &bounds.y1);
-			/* \todo: move x1 in by correct idx */
-			bounds.x1 += context->current->end_box->width;
-			bounds.y1 += context->current->end_box->height;
-			break;
-
-		default:
-			assert(content_get_type(h) == CONTENT_TEXTPLAIN);
-			textplain_coords_from_range(
-					hlcache_handle_get_content(h),
-					context->current->start_idx,
-					context->current->end_idx, &bounds);
-			break;
-	}
-
-	browser_window_scroll_visible(context->bw, &bounds);
-}
-
-/**
- * Find the first occurrence of 'match' in 'string' and return its index
- *
- * \param  string     the string to be searched (unterminated)
- * \param  s_len      length of the string to be searched
- * \param  pattern    the pattern for which we are searching (unterminated)
- * \param  p_len      length of pattern
- * \param  case_sens  true iff case sensitive match required
- * \param  m_len      accepts length of match in bytes
- * \return pointer to first match, NULL if none
- */
-
-const char *find_pattern(const char *string, int s_len, const char *pattern,
-		int p_len, bool case_sens, unsigned int *m_len)
-{
-	struct { const char *ss, *s, *p; bool first; } context[16];
-	const char *ep = pattern + p_len;
-	const char *es = string  + s_len;
-	const char *p = pattern - 1;  /* a virtual '*' before the pattern */
-	const char *ss = string;
-	const char *s = string;
-	bool first = true;
-	int top = 0;
-
-	while (p < ep) {
-		bool matches;
-		if (p < pattern || *p == '*') {
-			char ch;
-
-			/* skip any further asterisks; one is the same as many 
-			*/
-			do p++; while (p < ep && *p == '*');
-
-			/* if we're at the end of the pattern, yes, it matches 
-			*/
-			if (p >= ep) break;
-
-			/* anything matches a # so continue matching from
-			   here, and stack a context that will try to match
-			   the wildcard against the next character */
-
-			ch = *p;
-			if (ch != '#') {
-				/* scan forwards until we find a match for 
-				   this char */
-				if (!case_sens) ch = toupper(ch);
-				while (s < es) {
-					if (case_sens) {
-						if (*s == ch) break;
-					} else if (toupper(*s) == ch)
-						break;
-					s++;
-				}
-			}
-
-			if (s < es) {
-				/* remember where we are in case the match 
-				   fails; we may then resume */
-				if (top < (int)NOF_ELEMENTS(context)) {
-					context[top].ss = ss;
-					context[top].s  = s + 1;
-					context[top].p  = p - 1;
-					/* ptr to last asterisk */
-					context[top].first = first;
-					top++;
-				}
-
-				if (first) {
-					ss = s;
-					/* remember first non-'*' char */
-					first = false;
-				}
-
-				matches = true;
-			}
-			else
-				matches = false;
-		}
-		else if (s < es) {
-			char ch = *p;
-			if (ch == '#')
-				matches = true;
-			else {
-				if (case_sens)
-					matches = (*s == ch);
-				else
-					matches = (toupper(*s) == toupper(ch));
-			}
-			if (matches && first) {
-				ss = s;  /* remember first non-'*' char */
-				first = false;
-			}
-		}
-		else
-			matches = false;
-
-		if (matches) {
-			p++; s++;
-		}
-		else {
-			/* doesn't match, resume with stacked context if we have one */
-			if (--top < 0) return NULL;  /* no match, give up */
-
-			ss = context[top].ss;
-			s  = context[top].s;
-			p  = context[top].p;
-			first = context[top].first;
-		}
-	}
-
-	/* end of pattern reached */
-	*m_len = max(s - ss, 1);
-	return ss;
-}
-
-/**
- * Finds all occurrences of a given string in the html box tree
- *
- * \param pattern   the string pattern to search for
- * \param p_len     pattern length
- * \param cur       pointer to the current box
- * \param case_sens whether to perform a case sensitive search
- * \return true on success, false on memory allocation failure
- */
-bool find_occurrences_html(const char *pattern, int p_len, struct box *cur,
-		bool case_sens, struct search_context *context)
-{
-	struct box *a;
-
-	/* ignore this box, if there's no visible text */
-	if (!cur->object && cur->text) {
-		const char *text = cur->text;
-		unsigned length = cur->length;
-
-		while (length > 0) {
-			struct list_entry *entry;
-			unsigned match_length;
-			unsigned match_offset;
-			const char *new_text;
-			const char *pos = find_pattern(text, length,
-					pattern, p_len, case_sens,
-					&match_length);
-			if (!pos) break;
-
-			/* found string in box => add to list */
-			match_offset = pos - cur->text;
-
-			entry = add_entry(cur->byte_offset + match_offset,
-						cur->byte_offset +
-							match_offset +
-							match_length, context);
-			if (!entry)
-				return false;
-
-			entry->start_box = cur;
-			entry->end_box = cur;
-
-			new_text = pos + match_length;
-			length -= (new_text - text);
-			text = new_text;
-		}
-	}
-
-	/* and recurse */
-	for (a = cur->children; a; a = a->next) {
-		if (!find_occurrences_html(pattern, p_len, a, case_sens,
-				context))
-			return false;
-	}
+	if (bw->cur_search == NULL)
+		return browser_window_search_create_context(bw, callbacks, p);
 
 	return true;
 }
 
-/**
- * Finds all occurrences of a given string in a textplain content
- *
- * \param pattern   the string pattern to search for
- * \param p_len     pattern length
- * \param c         the content to be searched
- * \param case_sens wheteher to perform a case sensitive search
- * \return true on success, false on memory allocation failure
- */
 
-bool find_occurrences_text(const char *pattern, int p_len,
-		struct content *c, bool case_sens,
-		struct search_context *context)
+void browser_window_search_step(struct browser_window *bw,
+		search_flags_t flags, const char *string)
 {
-	int nlines = textplain_line_count(c);
-	int line;
-
-	for(line = 0; line < nlines; line++) {
-		size_t offset, length;
-		const char *text = textplain_get_line(c, line,
-				&offset, &length);
-		if (text) {
-			while (length > 0) {
-				struct list_entry *entry;
-				unsigned match_length;
-				size_t start_idx;
-				const char *new_text;
-				const char *pos = find_pattern(text, length,
-						pattern, p_len, case_sens,
-						&match_length);
-				if (!pos) break;
-
-				/* found string in line => add to list */
-				start_idx = offset + (pos - text);
-				entry = add_entry(start_idx, start_idx +
-						match_length, context);
-				if (!entry)
-					return false;
-
-				new_text = pos + match_length;
-				offset += (new_text - text);
-				length -= (new_text - text);
-				text = new_text;
-			}
-		}
-	}
+	assert(bw != NULL);
 
-	return true;
-}
-
-/**
- * Determines whether any portion of the given text box should be
- * selected because it matches the current search string.
- *
- * \param  bw            browser window
- * \param  start_offset  byte offset within text of string to be checked
- * \param  end_offset    byte offset within text
- * \param  start_idx     byte offset within string of highlight start
- * \param  end_idx       byte offset of highlight end
- * \return true iff part of the box should be highlighted
- */
-
-bool search_term_highlighted(struct browser_window *bw,
-		unsigned start_offset, unsigned end_offset,
-		unsigned *start_idx, unsigned *end_idx,
-		struct search_context *context)
-{
-	if (bw == context->bw) {
-		struct list_entry *a;
-		for(a = context->found->next; a; a = a->next)
-			if (a->sel && selection_defined(a->sel) &&
-				selection_highlighted(a->sel,
-					start_offset, end_offset,
-					start_idx, end_idx))
-				return true;
-	}
-
-	return false;
+	if (bw->cur_search != NULL)
+		search_step(bw->cur_search, flags, string);
 }
 
-/**
- * Specifies whether all matches or just the current match should
- * be highlighted in the search text.
- */
 
-void search_show_all(bool all, struct search_context *context)
+void browser_window_search_show_all(bool all, struct browser_window *bw)
 {
-	struct list_entry *a;
-
-	for (a = context->found->next; a; a = a->next) {
-		bool add = true;
-		if (!all && a != context->current) {
-			add = false;
-			if (a->sel) {
-				selection_clear(a->sel, true);
-				selection_destroy(a->sel);
-				a->sel = NULL;
-			}
-		}
-		if (add && !a->sel) {
-			hlcache_handle *h = context->bw->current_content;
-			struct content *c = hlcache_handle_get_content(h);
+	assert(bw != NULL);
 
-			switch (content_get_type(h)) {
-			case CONTENT_HTML:
-				a->sel = selection_create(c, true);
-				if (!a->sel)
-					continue;
-
-				selection_init(a->sel, html_get_box_tree(h));
-				break;
-			default:
-				assert(content_get_type(h) ==
-						CONTENT_TEXTPLAIN);
-				a->sel = selection_create(c, false);
-				if (!a->sel)
-					continue;
-
-				selection_init(a->sel, NULL);
-				break;
-			}
-
-			selection_set_start(a->sel, a->start_idx);
-			selection_set_end(a->sel, a->end_idx);
-		}
-	}
-}
-
-/**
- * Add a new entry to the list of matches
- *
- * \param  start_idx  offset of match start within textual representation
- * \param  end_idx    offset of match end
- * \return pointer to added entry, NULL iff failed
- */
-
-struct list_entry *add_entry(unsigned start_idx, unsigned end_idx,
-		struct search_context *context)
-{
-	struct list_entry *entry;
-
-	/* found string in box => add to list */
-	entry = calloc(1, sizeof(*entry));
-	if (!entry) {
-		warn_user("NoMemory", 0);
-		return NULL;
-	}
-
-	entry->start_idx = start_idx;
-	entry->end_idx = end_idx;
-	entry->sel = NULL;
-
-	entry->next = 0;
-	entry->prev = context->found->prev;
-	if (context->found->prev == NULL)
-		context->found->next = entry;
-	else
-		context->found->prev->next = entry;
-	context->found->prev = entry;
-
-	return entry;
-}
-
-/**
- * Ends the search process, invalidating all state
- * freeing the list of found boxes
- */
-void search_destroy_context(struct search_context *context)
-{
-	if (context->bw != NULL)
-		context->bw->search_context = NULL;
-	if ((context->string != NULL) && (context->callbacks != NULL) &&
-			(context->callbacks->add_recent != NULL)) {
-		context->callbacks->add_recent(context->string, context->p);
-		free(context->string);
-	}
-	free_matches(context);
-	free(context);
+	if (bw->cur_search != NULL)
+		search_show_all(all, bw->cur_search);
 }