textareas: filter out images

2018-05-10 01:17:28 -07:00
parent 20aabb2e83
commit 45336c7c28
1 changed files with 2 additions and 1 deletions
@@ -714,7 +714,8 @@ class PageInfo:
        page = doc[self._pageno]
        text = page.getText('dict')
        for block in text['blocks']:
-            yield block['bbox']
+            if block['type'] == 0:  # 0=text, >0=image or something else
+                yield block['bbox']

    @property
    def xres(self):