From b0ce7c63dd27257d9c979fde9013243b8ae38c98 Mon Sep 17 00:00:00 2001
From: "James R. Barlow" <jim@purplerock.ca>
Date: Thu, 10 May 2018 16:09:43 -0700
Subject: [PATCH] Since PyMuPDF 1.13.3 corrupts text, pin 1.12.5 and work
 around it

---
 setup.py                | 2 +-
 src/ocrmypdf/pdfinfo.py | 7 ++++++-
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/setup.py b/setup.py
index ab634951..ccb85522 100644
--- a/setup.py
+++ b/setup.py
@@ -251,7 +251,7 @@ setup(
         'ruffus == 2.6.3',        # pinned - ocrmypdf implements a 2.6.3 workaround
     ],
     extras_require={
-        'fitz': ['PyMuPDF >= 1.13.3'],
+        'fitz': ['PyMuPDF == 1.12.5'],
     },
     tests_require=tests_require,
     entry_points={
diff --git a/src/ocrmypdf/pdfinfo.py b/src/ocrmypdf/pdfinfo.py
index 9aa32618..80bd8135 100644
--- a/src/ocrmypdf/pdfinfo.py
+++ b/src/ocrmypdf/pdfinfo.py
@@ -586,7 +586,12 @@ def _page_has_text(infile, pageno):
     "Smarter text detection that ignores text in margins"
     
     doc = fitz.Document(infile)
-    text = doc[pageno].getText('dict')
+    if fitz.version[0] >= '1.13.0':
+        text = doc[pageno].getText('dict')
+    else:
+        import json
+        textjson = doc[pageno].getText('json')
+        text = json.loads(textjson)
     if not text:
         return