Issue #17150: pprint now uses line continuations to wrap long string literals.

2 files changed, 72 insertions, 5 deletions

diff --git a/Lib/pprint.py b/Lib/pprint.py
index ae96dde..1f98f5c 100644
--- a/Lib/pprint.py
+++ b/Lib/pprint.py
@@ -34,6 +34,7 @@ saferepr()
 
 """
 
+import re
 import sys as _sys
 from collections import OrderedDict as _OrderedDict
 from io import StringIO as _StringIO
@@ -158,13 +159,10 @@ class PrettyPrinter:
             return
         rep = self._repr(object, context, level - 1)
         typ = _type(object)
-        sepLines = _len(rep) > (self._width - 1 - indent - allowance)
+        max_width = self._width - 1 - indent - allowance
+        sepLines = _len(rep) > max_width
         write = stream.write
 
-        if self._depth and level > self._depth:
-            write(rep)
-            return
-
         if sepLines:
             r = getattr(typ, "__repr__", None)
             if issubclass(typ, dict):
@@ -242,6 +240,37 @@ class PrettyPrinter:
                 write(endchar)
                 return
 
+            if issubclass(typ, str) and len(object) > 0 and r is str.__repr__:
+                def _str_parts(s):
+                    """
+                    Return a list of string literals comprising the repr()
+                    of the given string using literal concatenation.
+                    """
+                    lines = s.splitlines(True)
+                    for i, line in enumerate(lines):
+                        rep = repr(line)
+                        if _len(rep) <= max_width:
+                            yield rep
+                        else:
+                            # A list of alternating (non-space, space) strings
+                            parts = re.split(r'(\s+)', line) + ['']
+                            current = ''
+                            for i in range(0, len(parts), 2):
+                                part = parts[i] + parts[i+1]
+                                candidate = current + part
+                                if len(repr(candidate)) > max_width:
+                                    if current:
+                                        yield repr(current)
+                                    current = part
+                                else:
+                                    current = candidate
+                            if current:
+                                yield repr(current)
+                for i, rep in enumerate(_str_parts(object)):
+                    if i > 0:
+                        write('\n' + ' '*indent)
+                    write(rep)
+                return
         write(rep)
 
     def _repr(self, object, context, level):
diff --git a/Lib/test/test_pprint.py b/Lib/test/test_pprint.py
index d492d75..bf136de 100644
--- a/Lib/test/test_pprint.py
+++ b/Lib/test/test_pprint.py
@@ -1,3 +1,5 @@
+# -*- coding: utf-8 -*-
+
 import pprint
 import test.support
 import unittest
@@ -475,6 +477,42 @@ class QueryTestCase(unittest.TestCase):
         self.assertEqual(pprint.pformat(dict.fromkeys(keys, 0)),
                          '{%r: 0, %r: 0}' % tuple(sorted(keys, key=id)))
 
+    def test_str_wrap(self):
+        # pprint tries to wrap strings intelligently
+        fox = 'the quick brown fox jumped over a lazy dog'
+        self.assertEqual(pprint.pformat(fox, width=20), """\
+'the quick brown '
+'fox jumped over '
+'a lazy dog'""")
+        self.assertEqual(pprint.pformat({'a': 1, 'b': fox, 'c': 2},
+                                        width=26), """\
+{'a': 1,
+ 'b': 'the quick brown '
+      'fox jumped over '
+      'a lazy dog',
+ 'c': 2}""")
+        # With some special characters
+        # - \n always triggers a new line in the pprint
+        # - \t and \n are escaped
+        # - non-ASCII is allowed
+        # - an apostrophe doesn't disrupt the pprint
+        special = "Portons dix bons \"whiskys\"\nà l'avocat goujat\t qui fumait au zoo"
+        self.assertEqual(pprint.pformat(special, width=20), """\
+'Portons dix bons '
+'"whiskys"\\n'
+"à l'avocat "
+'goujat\\t qui '
+'fumait au zoo'""")
+        # An unwrappable string is formatted as its repr
+        unwrappable = "x" * 100
+        self.assertEqual(pprint.pformat(unwrappable, width=80), repr(unwrappable))
+        self.assertEqual(pprint.pformat(''), "''")
+        # Check that the pprint is a usable repr
+        special *= 10
+        for width in range(3, 40):
+            formatted = pprint.pformat(special, width=width)
+            self.assertEqual(eval("(" + formatted + ")"), special)
+
 
 class DottedPrettyPrinter(pprint.PrettyPrinter):