move content handler specific selection copying into handlers

6 files changed, 295 insertions, 39 deletions

diff --git a/content/content_protected.h b/content/content_protected.h
index 9a3654beb..02bc8614d 100644
--- a/content/content_protected.h
+++ b/content/content_protected.h
@@ -47,6 +47,7 @@ struct browser_window_features;
 struct textsearch_context;
 struct box;
 struct selection;
+struct selection_string;
 
 typedef struct content_handler content_handler;
 
@@ -112,7 +113,10 @@ struct content_handler {
 	nserror (*textsearch_bounds)(struct content *c, unsigned start_idx, unsigned end_idx, struct box *start_ptr, struct box *end_ptr, struct rect *bounds_out);
 
 	/**
-	 * cause a region of the content to be marked invalid and hence redraw
+	 * redraw an area of selected text
+	 *
+	 * The defined text selection will cause an area of the
+	 *   content to be marked as invalid and hence redrawn.
 	 *
 	 * \param c The content being redrawn
 	 * \param start_idx The start index of the text region to be redrawn
@@ -122,6 +126,11 @@ struct content_handler {
 	nserror (*textselection_redraw)(struct content *c, unsigned start_idx, unsigned end_idx);
 
 	/**
+	 * copy selected text into selection string possibly with formatting
+	 */
+	nserror (*textselection_copy)(struct content *c, unsigned start_idx, unsigned end_idx, struct selection_string *selstr);
+
+	/**
 	 * create a selection object
 	 */
 	nserror (*create_selection)(struct content *c, struct selection **sel_out);
diff --git a/content/handlers/html/html.c b/content/handlers/html/html.c
index 2a3cb1774..a8ae9e452 100644
--- a/content/handlers/html/html.c
+++ b/content/handlers/html/html.c
@@ -2336,6 +2336,7 @@ static const content_handler html_content_handler = {
 	.textsearch_find = html_textsearch_find,
 	.textsearch_bounds = html_textsearch_bounds,
 	.textselection_redraw = html_textselection_redraw,
+	.textselection_copy = html_textselection_copy,
 	.create_selection = html_create_selection,
 	.no_share = true,
 };
diff --git a/content/handlers/html/textselection.c b/content/handlers/html/textselection.c
index bbd55732e..11e9deeb4 100644
--- a/content/handlers/html/textselection.c
+++ b/content/handlers/html/textselection.c
@@ -277,6 +277,7 @@ html_create_selection(struct content *c, struct selection **sel_out)
 	return NSERROR_OK;
 }
 
+/* exported interface documented in html/textselection.h */
 nserror
 html_textselection_redraw(struct content *c,
 			  unsigned start_idx,
@@ -301,3 +302,229 @@ html_textselection_redraw(struct content *c,
 
 	return NSERROR_OK;
 }
+
+/**
+ * Selection traversal routine for appending text to a string
+ *
+ * \param  text         pointer to text being added, or NULL for newline
+ * \param  length       length of text to be appended (bytes)
+ * \param  box          pointer to text box, or NULL if from textplain
+ * \param  len_ctx      Length conversion context
+ * \param  handle       selection string to append to
+ * \param  whitespace_text    whitespace to place before text for formatting
+ *                            may be NULL
+ * \param  whitespace_length  length of whitespace_text
+ * \return true iff successful and traversal should continue
+ */
+static bool
+selection_copy(const char *text,
+		       size_t length,
+		       struct box *box,
+		       const nscss_len_ctx *len_ctx,
+		       struct selection_string *handle,
+		       const char *whitespace_text,
+		       size_t whitespace_length)
+{
+	bool add_space = false;
+	plot_font_style_t style;
+	plot_font_style_t *pstyle = NULL;
+
+	/* add any whitespace which precedes the text from this box */
+	if (whitespace_text != NULL &&
+	    whitespace_length > 0) {
+		if (!selection_string_append(whitespace_text,
+					     whitespace_length,
+					     false,
+					     pstyle,
+					     handle)) {
+			return false;
+		}
+	}
+
+	if (box != NULL) {
+		/* HTML */
+		add_space = (box->space != 0);
+
+		if (box->style != NULL) {
+			/* Override default font style */
+			font_plot_style_from_css(len_ctx, box->style, &style);
+			pstyle = &style;
+		} else {
+			/* If there's no style, there must be no text */
+			assert(box->text == NULL);
+		}
+	}
+
+	/* add the text from this box */
+	if (!selection_string_append(text, length, add_space, pstyle, handle)) {
+		return false;
+	}
+
+	return true;
+}
+
+
+/**
+ * Traverse the given box subtree, calling selection copy for all
+ * boxes that lie (partially) within the given range
+ *
+ * \param box        box subtree
+ * \param len_ctx    Length conversion context.
+ * \param start_idx  start of range within textual representation (bytes)
+ * \param end_idx    end of range
+ * \param handler    handler function to call
+ * \param handle     handle to pass
+ * \param before     type of whitespace to place before next encountered text
+ * \param first      whether this is the first box with text
+ * \param do_marker  whether deal enter any marker box
+ * \return false iff traversal abandoned part-way through
+ */
+static bool
+traverse_tree(struct box *box,
+	      const nscss_len_ctx *len_ctx,
+	      unsigned start_idx,
+	      unsigned end_idx,
+	      struct selection_string *selstr,
+	      save_text_whitespace *before,
+	      bool *first,
+	      bool do_marker)
+{
+	struct box *child;
+	const char *whitespace_text = "";
+	size_t whitespace_length = 0;
+
+	assert(box);
+
+	/* If selection starts inside marker */
+	if (box->parent &&
+	    box->parent->list_marker == box &&
+	    !do_marker) {
+		/* set box to main list element */
+		box = box->parent;
+	}
+
+	/* If box has a list marker */
+	if (box->list_marker) {
+		/* do the marker box before continuing with the rest of the
+		 * list element */
+		if (!traverse_tree(box->list_marker,
+				   len_ctx,
+				   start_idx,
+				   end_idx,
+				   selstr,
+				   before,
+				   first,
+				   true)) {
+			return false;
+		}
+	}
+
+	/* we can prune this subtree, it's after the selection */
+	if (box->byte_offset >= end_idx) {
+		return true;
+	}
+
+	/* read before calling the handler in case it modifies the tree */
+	child = box->children;
+
+	/* If nicely formatted output of the selected text is required, work
+	 * out what whitespace should be placed before the next bit of text */
+	if (before) {
+		save_text_solve_whitespace(box,
+					   first,
+					   before,
+					   &whitespace_text,
+					   &whitespace_length);
+	} else {
+		whitespace_text = NULL;
+	}
+
+	if ((box->type != BOX_BR) &&
+	    !((box->type == BOX_FLOAT_LEFT ||
+	       box->type == BOX_FLOAT_RIGHT) &&
+	      !box->text)) {
+		unsigned start_offset;
+		unsigned end_offset;
+
+		if (selected_part(box,
+				  start_idx,
+				  end_idx,
+				  &start_offset,
+				  &end_offset)) {
+			if (!selection_copy(box->text + start_offset,
+					    min(box->length, end_offset) - start_offset,
+					    box,
+					    len_ctx,
+					    selstr,
+					    whitespace_text,
+					    whitespace_length)) {
+				return false;
+			}
+			if (before) {
+				*first = false;
+				*before = WHITESPACE_NONE;
+			}
+		}
+	}
+
+	/* find the first child that could lie partially within the selection;
+	 * this is important at the top-levels of the tree for pruning subtrees
+	 * that lie entirely before the selection */
+
+	if (child) {
+		struct box *next = child->next;
+
+		while (next && next->byte_offset < start_idx) {
+			child = next;
+			next = child->next;
+		}
+
+		while (child) {
+			/* read before calling the handler in case it modifies
+			 * the tree */
+			struct box *next = child->next;
+
+			if (!traverse_tree(child,
+					   len_ctx,
+					   start_idx,
+					   end_idx,
+					   selstr,
+					   before,
+					   first,
+					   false)) {
+				return false;
+			}
+
+			child = next;
+		}
+	}
+
+	return true;
+}
+
+/* exported interface documented in html/textselection.h */
+nserror
+html_textselection_copy(struct content *c,
+			unsigned start_idx,
+			unsigned end_idx,
+			struct selection_string *selstr)
+{
+	html_content *html = (html_content *)c;
+	save_text_whitespace before = WHITESPACE_NONE;
+	bool first = true;
+	bool res;
+
+	res = traverse_tree(html->layout,
+			    &html->len_ctx,
+			    start_idx,
+			    end_idx,
+			    selstr,
+			    &before,
+			    &first,
+			    false);
+
+	if (res == false) {
+		return NSERROR_NOMEM;
+	}
+	return NSERROR_OK;
+}
diff --git a/content/handlers/html/textselection.h b/content/handlers/html/textselection.h
index 0287f3c58..4a866b0b2 100644
--- a/content/handlers/html/textselection.h
+++ b/content/handlers/html/textselection.h
@@ -34,4 +34,6 @@ nserror html_create_selection(struct content *c, struct selection **sel_out);
 
 nserror html_textselection_redraw(struct content *c, unsigned start_idx, unsigned end_idx);
 
+nserror html_textselection_copy(struct content *c, unsigned start_idx, unsigned end_idx, struct selection_string *selstr);
+
 #endif
diff --git a/content/handlers/text/textplain.c b/content/handlers/text/textplain.c
index d8f61be89..275efda83 100644
--- a/content/handlers/text/textplain.c
+++ b/content/handlers/text/textplain.c
@@ -1514,6 +1514,43 @@ textplain_coords_from_range(struct content *c,
 
 
 /**
+ * Return a pointer to the raw UTF-8 data, as opposed to the reformatted
+ * text to fit the window width. Thus only hard newlines are preserved
+ * in the saved/copied text of a selection.
+ *
+ * \param[in] c     content of type CONTENT_TEXTPLAIN
+ * \param[in] start starting byte offset within UTF-8 text
+ * \param[in] end   ending byte offset
+ * \param[out] plen receives validated length
+ * \return pointer to text, or NULL if no text
+ */
+static char *
+textplain_get_raw_data(struct content *c,
+		       unsigned start,
+		       unsigned end,
+		       size_t *plen)
+{
+	textplain_content *text = (textplain_content *) c;
+	size_t utf8_size;
+
+	assert(c != NULL);
+
+	utf8_size = text->utf8_data_size;
+
+	/* any text at all? */
+	if (!utf8_size) return NULL;
+
+	/* clamp to valid offset range */
+	if (start >= utf8_size) start = utf8_size;
+	if (end >= utf8_size) end = utf8_size;
+
+	*plen = end - start;
+
+	return text->utf8_data + start;
+}
+
+
+/**
  * get bounds of a free text search match
  */
 static nserror
@@ -1570,6 +1607,23 @@ textplain_textselection_redraw(struct content *c,
 	return NSERROR_OK;
 }
 
+static nserror
+textplain_textselection_copy(struct content *c,
+			     unsigned start_idx,
+			     unsigned end_idx,
+			     struct selection_string *selstr)
+{
+	const char *text;
+	size_t length;
+	bool res;
+
+	text = textplain_get_raw_data(c, start_idx, end_idx, &length);
+	res = selection_string_append(text, length, false, NULL, selstr);
+	if (res == false) {
+		return NSERROR_NOMEM;
+	}
+	return NSERROR_OK;
+}
 
 /**
  * plain text content handler table
@@ -1593,6 +1647,7 @@ static const content_handler textplain_content_handler = {
 	.textsearch_find = textplain_textsearch_find,
 	.textsearch_bounds = textplain_textsearch_bounds,
 	.textselection_redraw = textplain_textselection_redraw,
+	.textselection_copy = textplain_textselection_copy,
 	.create_selection = textplain_create_selection,
 	.no_share = true,
 };
@@ -1643,28 +1698,4 @@ size_t textplain_size(struct content *c)
 
 
 
-/* exported interface documented in html/textplain.h */
-char *
-textplain_get_raw_data(struct content *c,
-		       unsigned start,
-		       unsigned end,
-		       size_t *plen)
-{
-	textplain_content *text = (textplain_content *) c;
-	size_t utf8_size;
 
-	assert(c != NULL);
-
-	utf8_size = text->utf8_data_size;
-
-	/* any text at all? */
-	if (!utf8_size) return NULL;
-
-	/* clamp to valid offset range */
-	if (start >= utf8_size) start = utf8_size;
-	if (end >= utf8_size) end = utf8_size;
-
-	*plen = end - start;
-
-	return text->utf8_data + start;
-}
diff --git a/content/handlers/text/textplain.h b/content/handlers/text/textplain.h
index 206c5831e..b94ee33d3 100644
--- a/content/handlers/text/textplain.h
+++ b/content/handlers/text/textplain.h
@@ -46,18 +46,4 @@ nserror textplain_init(void);
 size_t textplain_size(struct content *c);
 
 
-/**
- * Return a pointer to the raw UTF-8 data, as opposed to the reformatted
- * text to fit the window width. Thus only hard newlines are preserved
- * in the saved/copied text of a selection.
- *
- * \param[in] c     content of type CONTENT_TEXTPLAIN
- * \param[in] start starting byte offset within UTF-8 text
- * \param[in] end   ending byte offset
- * \param[out] plen receives validated length
- * \return pointer to text, or NULL if no text
- */
-char *textplain_get_raw_data(struct content *c, unsigned start, unsigned end, size_t *plen);
-
-
 #endif