fitz/: Fix pymupdf#2238 - use 'overlap' rather than 'contains' when e…

…xtracting text.

We now include chars that overlap with the clipbox, instead of only those that
are entirely contained within the clipbox.

Note that new fn JM_rects_overlap() still returns true if one of the rects is
empty. This allows things to work with ligatures, where component glyphs can
have zero width.

fitz/fitz.i

@@ -11528,7 +11528,7 @@ struct TextPage {
                             fz_rect linerect = fz_empty_rect;
                             for (ch = line->first_char; ch; ch = ch->next) {
                                 fz_rect cbbox = JM_char_bbox(gctx, line, ch);
-                                if (!fz_contains_rect(tp_rect, cbbox) &&
+                                if (!JM_rects_overlap(tp_rect, cbbox) &&
                                     !fz_is_infinite_rect(tp_rect)) {
                                     continue;
                                 }
@@ -11542,7 +11542,7 @@ struct TextPage {
                             blockrect = fz_union_rect(blockrect, linerect);
                         }
                         text = JM_EscapeStrFromBuffer(gctx, res);
-                    } else if (fz_contains_rect(tp_rect, block->bbox) || fz_is_infinite_rect(tp_rect)) {
+                    } else if (JM_rects_overlap(tp_rect, block->bbox) || fz_is_infinite_rect(tp_rect)) {
                         fz_image *img = block->u.i.image;
                         fz_colorspace *cs = img->colorspace;
                         text = PyUnicode_FromFormat("<image: %s, width: %d, height: %d, bpc: %d>", fz_colorspace_name(gctx, cs), img->w, img->h, img->bpc);
@@ -11610,7 +11610,7 @@ struct TextPage {
                         buflen = 0;                       // reset char counter
                         for (ch = line->first_char; ch; ch = ch->next) {
                             fz_rect cbbox = JM_char_bbox(gctx, line, ch);
-                            if (!fz_contains_rect(tp_rect, cbbox) &&
+                            if (!JM_rects_overlap(tp_rect, cbbox) &&
                                 !fz_is_infinite_rect(tp_rect)) {
                                 continue;
                             }

fitz/helper-other.i

@@ -1295,7 +1295,17 @@ fz_archive *JM_archive_from_py(fz_context *ctx, fz_archive *arch, PyObject *path
 }
 
 
-
+int JM_rects_overlap(const fz_rect a, const fz_rect b)
+{
+    if (0
+            || a.x0 >= b.x1
+            || a.y0 >= b.y1
+            || a.x1 <= b.x0
+            || a.y1 <= b.y0
+            )
+        return 0;
+    return 1;
+}
 
 //-----------------------------------------------------------------------------
 // dummy structure for various tools and utilities

fitz/helper-stext.i

@@ -209,7 +209,7 @@ JM_new_buffer_from_stext_page(fz_context *ctx, fz_stext_page *page)
             if (block->type == FZ_STEXT_BLOCK_TEXT) {
                 for (line = block->u.t.first_line; line; line = line->next) {
                     for (ch = line->first_char; ch; ch = ch->next) {
-                        if (!fz_contains_rect(rect, JM_char_bbox(ctx, line, ch)) &&
+                        if (!JM_rects_overlap(rect, JM_char_bbox(ctx, line, ch)) &&
                             !fz_is_infinite_rect(rect)) {
                             continue;
                         }
@@ -375,7 +375,7 @@ JM_search_stext_page(fz_context *ctx, fz_stext_page *page, const char *needle)
             for (line = block->u.t.first_line; line; line = line->next) {
                 for (ch = line->first_char; ch; ch = ch->next) {
                     if (!fz_is_infinite_rect(rect) &&
-                        !fz_contains_rect(rect, JM_char_bbox(ctx, line, ch))) {
+                        !JM_rects_overlap(rect, JM_char_bbox(ctx, line, ch))) {
                             goto next_char;
                         }
 try_new_match:
@@ -436,7 +436,7 @@ JM_print_stext_page_as_text(fz_context *ctx, fz_output *out, fz_stext_page *page
                 for (ch = line->first_char; ch; ch = ch->next) {
                     chbbox = JM_char_bbox(ctx, line, ch);
                     if (fz_is_infinite_rect(rect) ||
-                        fz_contains_rect(rect, chbbox)) {
+                        JM_rects_overlap(rect, chbbox)) {
                         last_char = ch->c;
                         n = fz_runetochar(utf, ch->c);
                         for (i = 0; i < n; i++) {
@@ -525,7 +525,7 @@ JM_make_spanlist(fz_context *ctx, PyObject *line_dict,
 
     for (ch = line->first_char; ch; ch = ch->next) {
         fz_rect r = JM_char_bbox(ctx, line, ch);
-        if (!fz_contains_rect(tp_rect, r) &&
+        if (!JM_rects_overlap(tp_rect, r) &&
             !fz_is_infinite_rect(tp_rect)) {
             continue;
         }
@@ -771,7 +771,7 @@ JM_copy_rectangle(fz_context *ctx, fz_stext_page *page, fz_rect area)
 				int line_had_text = 0;
 				for (ch = line->first_char; ch; ch = ch->next) {
 					fz_rect r = JM_char_bbox(ctx, line, ch);
-					if (fz_contains_rect(area, r)) {
+					if (JM_rects_overlap(area, r)) {
 						line_had_text = 1;
 						if (need_new_line) {
 							fz_append_string(ctx, buffer, "\n");