Skip to content
Projects
Groups
Snippets
Help
Loading...
Help
Support
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Z
Zope
Project overview
Project overview
Details
Activity
Releases
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Issues
0
Issues
0
List
Boards
Labels
Milestones
Merge Requests
0
Merge Requests
0
Analytics
Analytics
Repository
Value Stream
Wiki
Wiki
Snippets
Snippets
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Create a new issue
Commits
Issue Boards
Open sidebar
Kirill Smelkov
Zope
Commits
5b7d9b5d
Commit
5b7d9b5d
authored
May 03, 2001
by
Andreas Jung
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
cleaner handling of punctuation, several fixes, enhanced
locale support
parent
7c1487b7
Changes
5
Hide whitespace changes
Inline
Side-by-side
Showing
5 changed files
with
18 additions
and
198 deletions
+18
-198
lib/python/StructuredText/ClassicDocumentClass.py
lib/python/StructuredText/ClassicDocumentClass.py
+10
-23
lib/python/StructuredText/DocumentClass.py
lib/python/StructuredText/DocumentClass.py
+4
-4
lib/python/StructuredText/HTMLClass.py
lib/python/StructuredText/HTMLClass.py
+1
-1
lib/python/StructuredText/MML.py
lib/python/StructuredText/MML.py
+0
-168
lib/python/StructuredText/STletters.py
lib/python/StructuredText/STletters.py
+3
-2
No files found.
lib/python/StructuredText/ClassicDocumentClass.py
View file @
5b7d9b5d
...
...
@@ -85,7 +85,7 @@
import
re
,
ST
,
STDOM
from
string
import
split
,
join
,
replace
,
expandtabs
,
strip
,
find
from
STletters
import
letters
from
STletters
import
letters
,
lettpunc
,
punctuations
StringType
=
type
(
''
)
ListType
=
type
([])
...
...
@@ -584,7 +584,7 @@ class DocumentClass:
def doc_emphasize(
self, s,
expr = re.compile('
\
s*
\
*([
\
n
%s0-9
.:/;,
\
'
\
"
\
?
\
=
\
-
\
>
\
<
\
(
\
)]+)
\
*(?!
\
*|-)
'
% letters
).search
expr = re.compile('
\
s*
\
*([
\
n
%s0-9
]+)
\
*(?!
\
*|-)' % lettpunc
).search
):
r=expr(s)
...
...
@@ -632,7 +632,7 @@ class DocumentClass:
def doc_underline(self,
s,
expr=re.compile("
\
s
+
\
_
([
%
s0
-
9
\
s
\
.,
\
?
\
/
]
+
)
\
_
" % letters
).search):
expr=re.compile("
\
s
+
\
_
([
0
-
9
%
s
]
+
)
\
_
" % lettpunc
).search):
result = expr(s)
if result:
...
...
@@ -644,7 +644,7 @@ class DocumentClass:
def doc_strong(self,
s,
expr = re.compile('
\
s*
\
*
\
*([
\
n%s0-9
.:/;
\
-,!
\
?
\
'
\
"
]+)
\
*
\
*' % letters
).search
expr = re.compile('
\
s*
\
*
\
*([
\
n%s0-9
]+)
\
*
\
*' % lettpunc
).search
):
r=expr(s)
...
...
@@ -657,19 +657,9 @@ class DocumentClass:
def doc_href(
self, s,
expr1 = re.compile("
(
\
"[ %s0-9
\
n
\
-
\
.
\
,
\
;
\
(
\
)
\
/
\
:
\
/
\
*
\
'
]+
\
"
)(:)([a-zA-Z0-9
\
:
\
/
\
.
\
~
\
-
]+)([,]*
\
s*)"
%
letters
).
search
,
expr2
=
re
.
compile
(
'(
\
"
[ %s0-9
\
n
\
-
\
.
\
:
\
;
\
(
\
)
\
/
\
*
\
'
]+
\
"
)([,]+
\
s+)([
a
-zA-Z0-9
\
@
\
.
\
,
\
?
\
!
\
/
\
:
\
;
\
-
\
#]+)(
\
s*)
'
% letters).search):
expr1 = re.compile("
(
\
"[ %s0-9
\
n
\
-
\
.
\
,
\
;
\
(
\
)
\
/
\
:
\
/
\
*
\
'
]+
\
"
)(:)([a-zA-Z0-9
\
@
\
.
\
,
\
?
\
!
\
/
\
:
\
;
\
-
\
#
\
~
]+)([,]*
\
s*)"
%
letters
).
search
,
expr2
=
re
.
compile
(
'(
\
"
[ %s0-9
\
n
\
-
\
.
\
:
\
;
\
(
\
)
\
/
\
*
\
'
]+
\
"
)([,]+
\
s+)([
a
-zA-Z0-9
\
@
\
.
\
,
\
?
\
!
\
/
\
:
\
;
\
-
\
#
\
~
]+)(
\
s*)'
%
letters
).
search
):
#expr1=re.compile('
\
"([ a-zA-Z0-9.:/;,
\
n
\
~
\
(
\
)
\
-]+)
\
"
'
# ':'
# '([a-zA-Z0-9.:/;,
\
n
\
~]+)(?=(
\
s+|
\
.|
\
!|
\
?))
'
# ).search,
#expr2=re.compile('
\
"
([ a-zA-Z0-9./:]+)
\
"
'
# ',
\
s+
'
# '([ a-zA-Z0-9@.:/;]+)(?=(
\
s+|
\
.|
\
!|
\
?))'
# ).search,
punctuation = re.compile("
[
\
,
\
.
\
?
\
!
\
;]
+
").match
r
=
expr1
(
s
)
or
expr2
(
s
)
if
r
:
...
...
@@ -679,19 +669,16 @@ class DocumentClass:
start
,
e
=
r
.
span
(
1
)
name
=
s
[
start
:
e
]
name
=
replace
(
name
,
'"'
,
''
,
2
)
#start = start + 1
st
,
end
=
r
.
span
(
3
)
if punctuation(s[end-1:end]):
end = end -1
link = s[st:end]
#end = end - 1
if
s
[
end
-
1
:
end
]
in
punctuations
:
end
-=
1
link
=
s
[
st
:
end
]
# name is the href title, link is the target
# of the href
return
(
StructuredTextLink
(
name
,
href
=
link
),
start
,
end
)
#return (StructuredTextLink(s[start:end], href=s[start:end]),
# start, end)
else
:
return
None
lib/python/StructuredText/DocumentClass.py
View file @
5b7d9b5d
...
...
@@ -85,7 +85,7 @@
import
re
,
ST
,
STDOM
from
string
import
split
,
join
,
replace
,
expandtabs
,
strip
,
find
,
rstrip
from
STletters
import
letters
from
STletters
import
*
StringType
=
type
(
''
)
...
...
@@ -867,7 +867,7 @@ class DocumentClass:
def doc_emphasize(
self, s,
expr = re.compile(r'\
s*
\*([ \n%s0-9
.:/;,\'\"\
?
\-\
_
\/\
=
\-\
>
\<\
(
\)]+)\
*(?!
\*|-)' % letters
).search
expr = re.compile(r'\
s*
\*([ \n%s0-9
]+)\
*(?!
\*|-)' % lettpunc
).search
):
r=expr(s)
...
...
@@ -914,7 +914,7 @@ class DocumentClass:
def doc_underline(self,
s,
expr=re.compile(r"
\
s
+
\
_
([
%
s0
-
9
\
s
\
.,
\
?
]
+
)
\
_
" % letters
).search):
expr=re.compile(r"
\
s
+
\
_
([
%
s0
-
9
\
s
]
+
)
\
_
" % lettpunc
).search):
result = expr(s)
if result:
...
...
@@ -926,7 +926,7 @@ class DocumentClass:
def doc_strong(self,
s,
expr = re.compile(r'
\
s*
\
*
\
*([
\
n%s
Z0-9.:/;
\
-,!
\
?
\
'
\
"
]+)
\
*
\
*' % letters
).search
expr = re.compile(r'
\
s*
\
*
\
*([
\
n%s
0-9]+)
\
*
\
*' % lettpunc
).search
):
r=expr(s)
...
...
lib/python/StructuredText/HTMLClass.py
View file @
5b7d9b5d
...
...
@@ -215,7 +215,7 @@ class HTMLClass:
else
:
getattr
(
self
,
self
.
element_types
[
c
.
getNodeName
()])(
c
,
level
,
output
)
output
(
'</p>'
)
output
(
'</p>
\
n
'
)
def
link
(
self
,
doc
,
level
,
output
):
output
(
'<a href="%s">'
%
doc
.
href
)
...
...
lib/python/StructuredText/MML.py
deleted
100644 → 0
View file @
7c1487b7
##############################################################################
#
# Zope Public License (ZPL) Version 1.0
# -------------------------------------
#
# Copyright (c) Digital Creations. All rights reserved.
#
# This license has been certified as Open Source(tm).
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met:
#
# 1. Redistributions in source code must retain the above copyright
# notice, this list of conditions, and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright
# notice, this list of conditions, and the following disclaimer in
# the documentation and/or other materials provided with the
# distribution.
#
# 3. Digital Creations requests that attribution be given to Zope
# in any manner possible. Zope includes a "Powered by Zope"
# button that is installed by default. While it is not a license
# violation to remove this button, it is requested that the
# attribution remain. A significant investment has been put
# into Zope, and this effort will continue if the Zope community
# continues to grow. This is one way to assure that growth.
#
# 4. All advertising materials and documentation mentioning
# features derived from or use of this software must display
# the following acknowledgement:
#
# "This product includes software developed by Digital Creations
# for use in the Z Object Publishing Environment
# (http://www.zope.org/)."
#
# In the event that the product being advertised includes an
# intact Zope distribution (with copyright and license included)
# then this clause is waived.
#
# 5. Names associated with Zope or Digital Creations must not be used to
# endorse or promote products derived from this software without
# prior written permission from Digital Creations.
#
# 6. Modified redistributions of any form whatsoever must retain
# the following acknowledgment:
#
# "This product includes software developed by Digital Creations
# for use in the Z Object Publishing Environment
# (http://www.zope.org/)."
#
# Intact (re-)distributions of any official Zope release do not
# require an external acknowledgement.
#
# 7. Modifications are encouraged but must be packaged separately as
# patches to official Zope releases. Distributions that do not
# clearly separate the patches from the original work must be clearly
# labeled as unofficial distributions. Modifications which do not
# carry the name Zope may be packaged in any form, as long as they
# conform to all of the clauses above.
#
#
# Disclaimer
#
# THIS SOFTWARE IS PROVIDED BY DIGITAL CREATIONS ``AS IS'' AND ANY
# EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL DIGITAL CREATIONS OR ITS
# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
# USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
# ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
# OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
# SUCH DAMAGE.
#
#
# This software consists of contributions made by Digital Creations and
# many individuals on behalf of Digital Creations. Specific
# attributions are listed in the accompanying credits file.
#
##############################################################################
from
StructuredText
import
*
# :-)
def
ctag
(
s
):
# Blech, wish we could use character tags
if
s
is
None
:
s
=
''
s
=
gsub
(
strong
,
'
\
\
1<bold>
\
\
2<plain>
\
\
3'
,
s
)
s
=
gsub
(
code
,
'
\
\
1<family Courier>
\
\
2<family Times>
\
\
3'
,
s
)
s
=
gsub
(
em
,
'
\
\
1<italic>
\
\
2<plain>
\
\
3'
,
s
)
return
join
(
map
(
strip
,
split
(
s
,
'
\
n
'
)),
'
\
n
'
)
class
MML
(
StructuredText
):
'''
\
An MML structured text formatter.
'''
\
def
__str__
(
self
,
):
'''
\
Return an HTML string representation of the structured text data.
'''
s
=
self
.
_str
(
self
.
structure
,
self
.
level
)
return
s
def
ul
(
self
,
before
,
p
,
after
):
return
(
"%s
\
n
\
n
<Bulleted>
\
n
%s%s"
%
(
before
,
ctag
(
p
),
after
))
def
ol
(
self
,
before
,
p
,
after
):
return
(
"%s
\
n
\
n
<Numbered>
\
n
%s%s"
%
(
before
,
ctag
(
p
),
after
))
def
dl
(
self
,
before
,
t
,
d
,
after
):
return
(
"%s
\
n
\
n
<Term>
\
n
%s
\
n
\
n
<Definition>
\
n
%s%s"
%
(
before
,
ctag
(
t
),
ctag
(
d
),
after
))
def
head
(
self
,
before
,
t
,
level
,
d
):
return
(
"%s
\
n
\
n
<Heading%d>
\
n
%s%s"
%
(
before
,
level
,
ctag
(
t
),
d
))
def
normal
(
self
,
before
,
p
,
after
):
return
"%s
\
n
\
n
<Body>
\
n
%s%s"
%
(
before
,
ctag
(
p
),
after
)
def
pre
(
self
,
structure
,
r
=
None
):
if
r
is
None
:
r
=
[
''
]
for
s
in
structure
:
for
line
in
split
(
s
[
0
],
'
\
n
'
):
r
.
append
(
'
\
n
<PRE>'
)
r
.
append
(
line
)
self
.
pre
(
s
[
1
],
r
)
return
join
(
r
,
'
\
n
'
)
def
_str
(
self
,
structure
,
level
):
r
=
''
for
s
in
structure
:
# print s[0],'\n', len(s[1]), '\n\n'
if
bullet
.
match
(
s
[
0
])
>=
0
:
p
=
bullet
.
group
(
1
)
r
=
self
.
ul
(
r
,
p
,
self
.
_str
(
s
[
1
],
level
))
elif
ol
.
match
(
s
[
0
])
>=
0
:
p
=
ol
.
group
(
3
)
r
=
self
.
ol
(
r
,
p
,
self
.
_str
(
s
[
1
],
level
))
elif
olp
.
match
(
s
[
0
])
>=
0
:
p
=
olp
.
group
(
1
)
r
=
self
.
ol
(
r
,
p
,
self
.
_str
(
s
[
1
],
level
))
elif
dl
.
match
(
s
[
0
])
>=
0
:
t
,
d
=
dl
.
group
(
1
,
2
)
r
=
self
.
dl
(
r
,
t
,
d
,
self
.
_str
(
s
[
1
],
level
))
elif
example
.
search
(
s
[
0
])
>=
0
and
s
[
1
]:
# Introduce an example, using pre tags:
r
=
self
.
normal
(
r
,
s
[
0
],
self
.
pre
(
s
[
1
]))
elif
s
[
0
][
-
2
:]
==
'::'
and
s
[
1
]:
# Introduce an example, using pre tags:
r
=
self
.
normal
(
r
,
s
[
0
][:
-
1
],
self
.
pre
(
s
[
1
]))
elif
nl
.
search
(
s
[
0
])
<
0
and
s
[
1
]
and
s
[
0
][
-
1
:]
!=
':'
:
# Treat as a heading
t
=
s
[
0
]
r
=
self
.
head
(
r
,
t
,
level
,
self
.
_str
(
s
[
1
],
level
and
level
+
1
))
else
:
r
=
self
.
normal
(
r
,
s
[
0
],
self
.
_str
(
s
[
1
],
level
))
return
r
lib/python/StructuredText/STletters.py
View file @
5b7d9b5d
...
...
@@ -7,8 +7,9 @@ try:
except
:
pass
import
string
letters
=
string
.
letters
letters
=
string
.
letters
punctuations
=
string
.
punctuation
lettpunc
=
letters
+
punctuations
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment