[PATCH] added line truncation to pdb_tidy for >80 chars

pdbtools/pdb_tidy.py

@@ -44,6 +44,8 @@
 
 import os
 import sys
+import re
+import textwrap
 
 __author__ = "Joao Rodrigues"
 __email__ = "j.p.g.l.m.rodrigues@gmail.com"
@@ -108,7 +110,7 @@ def check_input(args):
 
 def run(fhandle, strict=False):
     """
-    Add TER/END statements and pads all lines to 80 characters.
+    Add TER/END statements and truncates/pads all lines to 80 characters.
 
     This function is a generator.
 
@@ -161,8 +163,14 @@ def make_TER(prev_line):
         if line.startswith(ignored):  # to avoid matching END _and_ ENDMDL
             continue
 
-        # Check line length
-        line = "{:<80}\n".format(line)
+        # Check line length, wrapping and padding as necessary
+        # preserve the line prefix for wrapping
+        prefix = re.match(r"\S+\s*", line).group(0)
+        content = line[len(prefix):].lstrip()
+
+        line = "".join(
+            f"{prefix}{part:<{80 - len(prefix)}}\n"
+            for part in textwrap.wrap(content, width=80 - len(prefix)))
 
         yield line
 

tests/test_pdb_tidy.py

@@ -138,6 +138,26 @@ def test_tidy_removes_master(self):
         m_master = sum(1 for i in self.stdout if i.startswith('MASTER'))
         self.assertEqual(m_master, 0)
 
+    def test_tidy_rewraps_long_lines(self):
+        """Test pdb_tidy re-wraps long lines in the header maintaining padding"""
+        sys.argv = ['']
+        self.exec_module(
+            'TITLE     THIS IS A VERY LONG LINE WHICH SHOULD BE WRAPPED '
+            'CORRECTLY OVER MULTIPLE LINES WITH PADDING'
+        )
+
+        # Check no long lines in output
+        long_lines = [line for line in self.stdout if len(line) > 80]
+        self.assertEqual(len(long_lines), 0)
+
+        # Check correct wrapping and padding
+        self.assertEqual(self.stdout[0],
+                         'TITLE     THIS IS A VERY LONG LINE WHICH SHOULD BE '
+                         'WRAPPED CORRECTLY OVER       ')
+        self.assertEqual(self.stdout[1],
+                         'TITLE     MULTIPLE LINES WITH PADDING              '
+                         '                             ')
+
     def test_default_stdin(self):
         """$ cat data/dummy.pdb | pdb_tidy"""