openmedialibrary_platform_d.../lib/python3.7/idlelib/paragraph.py

195 lines
7 KiB
Python
Raw Normal View History

2018-12-31 23:25:26 +00:00
"""Format a paragraph, comment block, or selection to a max width.
2016-02-06 09:36:57 +00:00
Does basic, standard text formatting, and also understands Python
comment blocks. Thus, for editing Python source code, this
extension is really only suitable for reformatting these comment
blocks or triple-quoted strings.
Known problems with comment reformatting:
* If there is a selection marked, and the first line of the
selection is not complete, the block will probably not be detected
as comments, and will have the normal "text formatting" rules
applied.
* If a comment block has leading whitespace that mixes tabs and
spaces, they will not be considered part of the same block.
* Fancy comments, like this bulleted list, aren't handled :-)
"""
import re
2018-12-31 23:25:26 +00:00
from idlelib.config import idleConf
2016-02-06 09:36:57 +00:00
2018-12-31 23:25:26 +00:00
class FormatParagraph:
2016-02-06 09:36:57 +00:00
def __init__(self, editwin):
self.editwin = editwin
2018-12-31 23:25:26 +00:00
@classmethod
def reload(cls):
cls.max_width = idleConf.GetOption('extensions', 'FormatParagraph',
'max-width', type='int', default=72)
2016-02-06 09:36:57 +00:00
def close(self):
self.editwin = None
def format_paragraph_event(self, event, limit=None):
"""Formats paragraph to a max width specified in idleConf.
If text is selected, format_paragraph_event will start breaking lines
at the max width, starting from the beginning selection.
If no text is selected, format_paragraph_event uses the current
cursor location to determine the paragraph (lines of text surrounded
by blank lines) and formats it.
The length limit parameter is for testing with a known value.
"""
2018-12-31 23:25:26 +00:00
limit = self.max_width if limit is None else limit
2016-02-06 09:36:57 +00:00
text = self.editwin.text
first, last = self.editwin.get_selection_indices()
if first and last:
data = text.get(first, last)
comment_header = get_comment_header(data)
else:
first, last, comment_header, data = \
find_paragraph(text, text.index("insert"))
if comment_header:
newdata = reformat_comment(data, limit, comment_header)
else:
newdata = reformat_paragraph(data, limit)
text.tag_remove("sel", "1.0", "end")
if newdata != data:
text.mark_set("insert", first)
text.undo_block_start()
text.delete(first, last)
text.insert(first, newdata)
text.undo_block_stop()
else:
text.mark_set("insert", last)
text.see("insert")
return "break"
2018-12-31 23:25:26 +00:00
FormatParagraph.reload()
2016-02-06 09:36:57 +00:00
def find_paragraph(text, mark):
"""Returns the start/stop indices enclosing the paragraph that mark is in.
Also returns the comment format string, if any, and paragraph of text
between the start/stop indices.
"""
lineno, col = map(int, mark.split("."))
line = text.get("%d.0" % lineno, "%d.end" % lineno)
# Look for start of next paragraph if the index passed in is a blank line
while text.compare("%d.0" % lineno, "<", "end") and is_all_white(line):
lineno = lineno + 1
line = text.get("%d.0" % lineno, "%d.end" % lineno)
first_lineno = lineno
comment_header = get_comment_header(line)
comment_header_len = len(comment_header)
# Once start line found, search for end of paragraph (a blank line)
while get_comment_header(line)==comment_header and \
not is_all_white(line[comment_header_len:]):
lineno = lineno + 1
line = text.get("%d.0" % lineno, "%d.end" % lineno)
last = "%d.0" % lineno
# Search back to beginning of paragraph (first blank line before)
lineno = first_lineno - 1
line = text.get("%d.0" % lineno, "%d.end" % lineno)
while lineno > 0 and \
get_comment_header(line)==comment_header and \
not is_all_white(line[comment_header_len:]):
lineno = lineno - 1
line = text.get("%d.0" % lineno, "%d.end" % lineno)
first = "%d.0" % (lineno+1)
return first, last, comment_header, text.get(first, last)
# This should perhaps be replaced with textwrap.wrap
def reformat_paragraph(data, limit):
"""Return data reformatted to specified width (limit)."""
lines = data.split("\n")
i = 0
n = len(lines)
while i < n and is_all_white(lines[i]):
i = i+1
if i >= n:
return data
indent1 = get_indent(lines[i])
if i+1 < n and not is_all_white(lines[i+1]):
indent2 = get_indent(lines[i+1])
else:
indent2 = indent1
new = lines[:i]
partial = indent1
while i < n and not is_all_white(lines[i]):
# XXX Should take double space after period (etc.) into account
2018-12-31 23:25:26 +00:00
words = re.split(r"(\s+)", lines[i])
2016-02-06 09:36:57 +00:00
for j in range(0, len(words), 2):
word = words[j]
if not word:
continue # Can happen when line ends in whitespace
if len((partial + word).expandtabs()) > limit and \
partial != indent1:
new.append(partial.rstrip())
partial = indent2
partial = partial + word + " "
if j+1 < len(words) and words[j+1] != " ":
partial = partial + " "
i = i+1
new.append(partial.rstrip())
# XXX Should reformat remaining paragraphs as well
new.extend(lines[i:])
return "\n".join(new)
def reformat_comment(data, limit, comment_header):
"""Return data reformatted to specified width with comment header."""
# Remove header from the comment lines
lc = len(comment_header)
data = "\n".join(line[lc:] for line in data.split("\n"))
# Reformat to maxformatwidth chars or a 20 char width,
# whichever is greater.
format_width = max(limit - len(comment_header), 20)
newdata = reformat_paragraph(data, format_width)
# re-split and re-insert the comment header.
newdata = newdata.split("\n")
2018-12-31 23:25:26 +00:00
# If the block ends in a \n, we don't want the comment prefix
2016-02-06 09:36:57 +00:00
# inserted after it. (Im not sure it makes sense to reformat a
# comment block that is not made of complete lines, but whatever!)
# Can't think of a clean solution, so we hack away
block_suffix = ""
if not newdata[-1]:
block_suffix = "\n"
newdata = newdata[:-1]
return '\n'.join(comment_header+line for line in newdata) + block_suffix
def is_all_white(line):
"""Return True if line is empty or all whitespace."""
return re.match(r"^\s*$", line) is not None
def get_indent(line):
"""Return the initial space or tab indent of line."""
return re.match(r"^([ \t]*)", line).group()
def get_comment_header(line):
"""Return string with leading whitespace and '#' from line or ''.
A null return indicates that the line is not a comment line. A non-
null return, such as ' #', will be used to find the other lines of
a comment block with the same indent.
"""
m = re.match(r"^([ \t]*#*)", line)
if m is None: return ""
return m.group(1)
2018-12-31 23:25:26 +00:00
2016-02-06 09:36:57 +00:00
if __name__ == "__main__":
2018-12-31 23:25:26 +00:00
from unittest import main
main('idlelib.idle_test.test_paragraph', verbosity=2, exit=False)