scripts: Implemented recursive punescape nesting

The previous implementation of punescape nesting only supported one
layer, because I didn't want to rewrite the pure re.sub approach.

But regex is not a pushdown automaton!

I.e. it's impossible to match both of these correctly:

  %{a%}s%{b%}s
  %{a%{a%}s%}s

This rewrites punescape and psplit to properly parse the punescape
string, recursing when we encounter "%{" and terminating on "%}s".

As a plus this shows that the punescape grammar is sound. This was a bit
up in the air with the hacky regex globbing.

scripts/codemap.py

@@ -472,7 +472,7 @@ def si2(x):
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -481,7 +481,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -505,37 +506,62 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 # split %-escaped strings into chars
-def psplit(s):
+def psplit(s, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -544,10 +570,39 @@ def psplit(s):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
-    return [m.group() for m in re.finditer(pattern.pattern + '|.', s)]
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '{':
+                m__ = []
+                v = psplit(s, m.end(), None, m__)
+                if m__:
+                    s_.append(m.group() + ''.join(v))
+                    i = m__[0].end()
+                else:
+                    s_.append(m.group() + ''.join(v))
+                    i = end if end is not None else len(s)
+                continue
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+
+            s_.append(m.group())
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
+
+    return s_
 
 
 # a little ascii renderer

scripts/codemapsvg.py

@@ -334,7 +334,7 @@ def si2(x):
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -343,7 +343,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -367,34 +368,59 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 
 

scripts/csv.py

@@ -1710,7 +1710,7 @@ def si2(x):
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -1719,7 +1719,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -1743,34 +1744,59 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 class PunescapeGetattr:
     def __init__(self, r):

scripts/dbgbmap.py

@@ -4205,7 +4205,7 @@ def si2(x):
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -4214,7 +4214,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -4238,37 +4239,62 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 # split %-escaped strings into chars
-def psplit(s):
+def psplit(s, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -4277,10 +4303,39 @@ def psplit(s):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
-    return [m.group() for m in re.finditer(pattern.pattern + '|.', s)]
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '{':
+                m__ = []
+                v = psplit(s, m.end(), None, m__)
+                if m__:
+                    s_.append(m.group() + ''.join(v))
+                    i = m__[0].end()
+                else:
+                    s_.append(m.group() + ''.join(v))
+                    i = end if end is not None else len(s)
+                continue
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+
+            s_.append(m.group())
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
+
+    return s_
 
 
 # a little ascii renderer

scripts/dbgbmapsvg.py

@@ -4094,7 +4094,7 @@ def si2(x):
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -4103,7 +4103,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -4127,34 +4128,59 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 
 # naive space filling curve (the default)

scripts/dbgtrace.py

@@ -480,7 +480,7 @@ def si2(x):
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -489,7 +489,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -513,37 +514,62 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 # split %-escaped strings into chars
-def psplit(s):
+def psplit(s, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -552,10 +578,39 @@ def psplit(s):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
-    return [m.group() for m in re.finditer(pattern.pattern + '|.', s)]
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '{':
+                m__ = []
+                v = psplit(s, m.end(), None, m__)
+                if m__:
+                    s_.append(m.group() + ''.join(v))
+                    i = m__[0].end()
+                else:
+                    s_.append(m.group() + ''.join(v))
+                    i = end if end is not None else len(s)
+                continue
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+
+            s_.append(m.group())
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
+
+    return s_
 
 
 # a little ascii renderer

scripts/plot.py

@@ -638,7 +638,7 @@ class CsvAttr:
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -647,7 +647,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -671,37 +672,62 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 # split %-escaped strings into chars
-def psplit(s):
+def psplit(s, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -710,10 +736,39 @@ def psplit(s):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
-    return [m.group() for m in re.finditer(pattern.pattern + '|.', s)]
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '{':
+                m__ = []
+                v = psplit(s, m.end(), None, m__)
+                if m__:
+                    s_.append(m.group() + ''.join(v))
+                    i = m__[0].end()
+                else:
+                    s_.append(m.group() + ''.join(v))
+                    i = end if end is not None else len(s)
+                continue
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+
+            s_.append(m.group())
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
+
+    return s_
 
 
 # a little ascii renderer

scripts/plotmpl.py

@@ -522,7 +522,7 @@ class CsvAttr:
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -531,7 +531,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -555,34 +556,59 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 
 # some classes for organizing subplots into a grid

scripts/treemap.py

@@ -559,7 +559,7 @@ def si2(x):
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -568,7 +568,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -592,37 +593,62 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)
 
 # split %-escaped strings into chars
-def psplit(s):
+def psplit(s, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -631,10 +657,39 @@ def psplit(s):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
-    return [m.group() for m in re.finditer(pattern.pattern + '|.', s)]
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '{':
+                m__ = []
+                v = psplit(s, m.end(), None, m__)
+                if m__:
+                    s_.append(m.group() + ''.join(v))
+                    i = m__[0].end()
+                else:
+                    s_.append(m.group() + ''.join(v))
+                    i = end if end is not None else len(s)
+                continue
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+
+            s_.append(m.group())
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
+
+    return s_
 
 
 # a little ascii renderer

scripts/treemapsvg.py

@@ -420,7 +420,7 @@ def si2(x):
 # parse %-escaped strings
 #
 # attrs can override __getitem__ for lazy attr generation
-def punescape(s, attrs=None):
+def punescape(s, attrs=None, start=None, end=None, submatch=None):
     pattern = re.compile(
         '%' '(?P<rep>[0-9]*)' '(?P<pun>'
             '[%ns]'
@@ -429,7 +429,8 @@ def punescape(s, attrs=None):
                 '|' 'U........'
                 '|' '\((?P<field>[^)]*)\)'
                     '(?P<format>[+\- #0-9\.]*[siIdboxXfFeEgG])'
-                '|' '\{(?P<substring>(?:[^%]|%[^}])*)%\}'
+                '|' '\{'
+                '|' '\}'
                     '(?P<subformat>[+\- #0-9\.]*[siIdboxXfFeEgG])' ')')
 
     def format(f, v):
@@ -453,34 +454,59 @@ def punescape(s, attrs=None):
         # note we need Python's new format syntax for binary
         return ('{:%s}' % f).format(v)
 
-    def unescape(m):
-        if m.group('pun')[0] == '%':    s = '%'
-        elif m.group('pun')[0] == 'n':  s = '\n'
-        elif m.group('pun')[0] == 's':  s = ' '
-        elif m.group('pun')[0] == 'x':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'u':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == 'U':  s = chr(int(m.group('pun')[1:], 16))
-        elif m.group('pun')[0] == '(':
+    s_ = []
+    i = start or 0
+    while i < (end if end is not None else len(s)):
+        m = pattern.match(s, i)
+        if m:
+            m_ = m.group('pun')[0]
+            i_ = m.end()
+            if m_ == '%':   v_ = '%'
+            elif m_ == 'n': v_ = '\n'
+            elif m_ == 's': v_ = ' '
+            elif m_ == 'x': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'u': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == 'U': v_ = chr(int(m.group('pun')[1:], 16))
+            elif m_ == '(':
                 if attrs is not None:
                     try:
                         v = attrs[m.group('field')]
                     except KeyError:
-                    return m.group()
+                        s_.append(m.group())
+                        i = i_
+                        continue
                 else:
-                return m.group()
-            s = format(m.group('format'), v)
-        elif m.group('pun')[0] == '{':
-            v = punescape(m.group('substring'), attrs)
-            s = format(m.group('subformat'), v)
+                    s_.append(m.group())
+                    i = i_
+                    continue
+                v_ = format(m.group('format'), v)
+            elif m_ == '{':
+                m__ = []
+                v = punescape(s, attrs, m.end(), None, m__)
+                if m__:
+                    v_ = format(m__[0].group('subformat'), v)
+                    i_ = m__[0].end()
                 else:
-            return m.group()
+                    v_ = str(v)
+                    i_ = end if end is not None else len(s)
+            elif m_ == '}':
+                if submatch is not None:
+                    submatch.append(m)
+                break
+            else:
+                s_.append(m.group())
+                i = i_
+                continue
 
             if m.group('rep'):
-            s = int(m.group('rep'), 10) * s
+                v_ = int(m.group('rep'), 10) * v_
+            s_.append(v_)
+            i = i_
+        else:
+            s_.append(s[i])
+            i += 1
 
-        return s
-
-    return re.sub(pattern, unescape, s)
+    return ''.join(s_)