julienne
very ALPHA, use at your own risk, interface may change!
When writing code for teaching, you often need multiple versions of your code, showing progress to your students as you introduce new concepts. Keeping several versions is painful though, especially when you find a bug that is common to each copy.
Enter: julienne. It slices, it dices, well... it actually only slices. This
library comes with the juli
script which reads code and interprets special
directives in the comments, generating multiple versions of the code. The
directives allow you to limit which versions a block of code exists in.
The goal for this toolset once complete is to allow you to maintain a single
version of your project in its completed state. Running juli
on your
project will generate a separate copy of each version of your code.
Juli Comment Markers
When using juli
, you have one copy of your code in its final state. You
mark sections of your code with comments to indicate that a line or block only
participates in certain versions. Each version is called a chapter. When you
run the juli
command it will create a directory for each chapter found in
your code.
# This is a sample file
a = "In all chapters" # inline comment
b = "In chapters 1-3" #@= 1-3 comment on conditional
c = "In chapters 1-2" #@= -2
d = "In chapters 2 on" #@= 2-
#@@ 1-2 x = "In chapters 1-2"
#@+ 3-4
#@- e = "In chapters 3 to 4" # inline comment
#@- f = " as a block"
for x in range(10):
#@+ 1-2 block header with comment
#@- g = "In chapters 1 and 2"
h = "In all chapters"
#@[ 3- uncommented conditional block
def foo():
print("Blah de blah")
#@]
Juli can process Python style files (anything that uses #
as a comment, or
XML style files (anything that uses <!-- -->
as a comment block). The
markers for the two files are similar, with a small variation for blocks of
content.
Python-style Markers
Python-style juli
comment markers start with #@
followed by the
julienne type which determines how the marker behaves. The types are as
follows:
-
#@=
-- A single line conditional to a range of chapters, comment after the code -
#@@
-- Single line conditional to a range where code is commented out -
#@+
-- Start a conditional block that is commented out, applies to a range of chapters -
#@-
-- Part of a conditional block that is commented out. Must appear after a#@+
-
#@[
-- Start a conditional block that is not commented out, applies to a range of chapters -
#@]
-- End a conditional block that is not commented. Must appear after a#@[
-
#@*
-- A julienne comment, one not included in the output code
The #@=
, #@@`, ``#@+
, and #@[
markers expect a range that
indicates what chapters a line or block participates within. Ranges can
indicate a single chapter, a range of chapters, up-to-and-including a chapter,
and including-and-after a chapter. A space is expected between the julienne
type and the beginning of the range specifier. Example ranges:
-
#@= 3
-- this line only shows up in chapter 3 -
#@@ 2
-- the code in the comment is uncommented in chapter 2 -
#@+ 2-4
-- the following commented block is uncommented in chapters 2, 3, and 4 -
#@= 2-
-- this line is in chapters 2 and above -
#@[ -4
-- the following uncommented block starts appearing in chapter 4
All markers except #@@
support trailing comments. Generated code will
insert a comment without the juli
marker containing whatever comes after
your marker. Markers without trailing comments will not be included in the
results. Any indentation before a marker is respected if the marked line
results in output.
The sample code above will generate four chapters. Chapter one would contain:
# This is a sample file
a = "In all chapters" # inline comment
b = "In chapters 1-3" # comment on conditional
c = "In chapters 1-2"
x = "In chapters 1-2"
for x in range(10):
# block header with comment
g= "In chapters 1 and 2"
h = "In all chapters"
Chapter four would contain:
# This is a sample file
a = "In all chapters" # inline comment
d = "In chapters 2 on"
e = "In chapters 3 to 4" # inline comment
f = " as a block"
for x in range(10):
h = "In all chapters"
# uncommented conditional block
def foo():
print("Blah de blah")
Note that files that contain only conditional lines will not be included if they aren't in chapter range.
XML-Style Markers
XML-style markers are also comments. The markers begin with <!--@
, note
there must not be any white space between the comment marker and the @
. As
with the Python-style, a marker type follows the opening. The types are as
follows:
-
<!--@= 1-3 comment -->
-- Inline marker, anything appearing before this on the line is included in the range. -
<!--@+ 1-3 comment
-- Opening for a block. Subsequent lines between this and the closing marker are conditional. -
@+-->
-- Closing for a block, must be paired with an opening -
<!--@[ 1-3 comment -->
-- opening for a block that is not commented out, all content until the matching closing marker is conditional -
<!--@] -->
-- closing maker for a block -
<!--@* -->
-- a julienne comment, one not included in the output code, only use on a single line!
The same kinds of range specifiers are supported as Python-style (3, 1-3, 1-, and -3). Any additional text found in a comment marker is added as a comment in the result. If there is no additional comment in the marker, there is no corresponding line in the result.
Configuring Your Project
The juli
uses a TOML file for configuration. The file
must contain two key/value pairs that indicate the source and output
directories for the parser.
output_dir = 'last_output'
src_dir = 'code'
The above will cause juli
to look for a directory named code
relative
to the configuration file. The source found in that directory will be parsed.
The generated chapters will be put in a directory named last_output
. If
your source specified two chapters, running juli
will result in the
creation of two directories: last_output/ch1/code
and
last_output/ch2/code
.
Both the output_dir
and src_dir
values can be absolute paths or
relative to the TOML configuration file.
Additional, optional configuration values are:
-
black
-- if true (TOML uses lower case), runs the black formatting processor on your output code directories. Defaults to false. -
chapter_prefix
-- Specify what the prefix part of a chapter directory is named. If not specified, defaults to "ch" -
delete_output
-- if true (TOML uses lower case), removes any existing output directory before generating a new one. Defaults to false. -
pound_globs
-- A glob pattern that indicates which Python-style files participate in the parsing. Defaults to['**/*.py', ]
, meaning all files ending in ".py" -
xml_globs
-- A glob pattern that indicates which XML-style files participate in the parsing. Defaults to['**/*.xml', '**/*.htm', '**/*.html']
, meaning all files ending in ".xml", ".htm", or ".html" -
skip_dirs
-- A list of sub-directories that should not be processed. -
skip_patterns
-- A list of strings that if they show up in the path the path is ignored. Useful for things like __pycache__ -
[chapter_map]
-- Chapter numbers are integers, but you may not always want that in your output structure. This map allows you to change the suffix part of a chapter directory name. Keys in the map are the chapter numbers while values are what should be used in the chapter suffix. -
[ranged_files.XYZ]
-- Files or directories can be marked as conditional using this TOML map. This map must specifyrange
andfiles
attributes. Therange
attribute indicates what chapters this directory participates in, andfiles
is listing of file or directory names. In the case of files they will only participate in parsing if the match the range value. If a file contains a marker outside the range it will be ignored. TheXYZ
portion of the TOML nested map is ignored, it is there so you can have multiple conditional directories.
Here is a full example of a configuration file:
output_dir = 'last_output'
src_dir = 'code'
skip_dirs = ['bad_dir', ]
skip_patterns = ['__pycache__', ]
chapter_prefix = "chap"
[chapter_map]
4 = 'Four'
5 = '5.0'
[ranged_files.foo]
range = '2-4'
files = ['code/between24', 'only24.py']
[ranged_files.bar]
range = '4-'
files = ['code/after4', ]
If your code directory contained:
code/script.py
code/only24.py
code/readme.txt
code/between24/two_to_four.py
code/after4/later_on.txt
code/bad_dir/something.py
Then running juli example.toml
, the sample configuration would result
in the following:
last_output/chap1/code/script.py
last_output/chap1/code/readme.txt
last_output/chap2/code/script.py
last_output/chap2/code/only24.py
last_output/chap2/code/readme.txt
last_output/chap2/code/between24/two_to_four.py
last_output/chap3/code/script.py
last_output/chap3/code/only24.py
last_output/chap3/code/readme.txt
last_output/chap3/code/between24/two_to_four.py
last_output/chapFour/code/script.py
last_output/chapFour/code/only24.py
last_output/chapFour/code/readme.txt
last_output/chapFour/code/between24/two_to_four.py
last_output/chapFour/code/after4/later_on.txt
last_output/chap5.0/code/script.py
last_output/chap5.0/code/readme.txt
last_output/chap5.0/code/after4/later_on.txt
The script.py
, two_to_four.py
, and only24.py
files will be
processed for conditional content. The readme.txt
and later_on.txt
files will be straight copies as they aren't covered by the active glob.
Command Line Arguments
The juli
has one required argument, the name of the TOML
configuration
file. It also supports the following optional arguments:
-
--help
,-h
: show help info -
--verbose
,-v
: print information while processing -
--info
,-i
: only print the info don't do the processing -
--chapter CHAPTER
,-c CHAPTER
: process only the given chapter number (CHAPTER)
Uh, Oh
Warning
There is a known bug in Python where the shutil.copy2 method does not copy metadata on MacOS or Windows even though it is supposed to. This means group ownership flags and execution bits will get lost on those operating systems. See: