This repository contains a single command to manage a header section, e.g. copyright, for a source code tree.
Using UNIX glob patterns, addheader modifies an entire tree of source code at once. The program replaces existing headers with an updated version, and places the header after any shell magic at the top of the file.
As of version 0.3.0, Jupyter notebooks can also be handled. See Usage -> Adding headers to Jupyter Notebooks.
addheader is written in Python and can be simply installed from the PyPI package:
pip install addheader
If you want Jupyter Notebook support, add "jupyter" in square brackets after the name of the package (use the quotes unless you know your shell doesn't need them):
pip install 'addheader[jupyter]'
Use the command addheader
. Invokingaddheader -h
shows a detailed help message
for the command arguments and options. Below are some examples and comments on usage.
If you have the header file in "copyright.txt", and your source tree is a Python package located at "./mypackage", then you would invoke the program like this:
adddheader mypackage --text copyright.txt
By default, the header will not be added to "init.py" files.
If you want to see which files would be changed without modifying them, add
-n
or --dry-run
to the command line arguments.
If this argument is given, any arguments related to modifying or removing headers will be ignored.
If you want to remove existing headers instead of adding or updating them,
add -r
or --remove
to the command line arguments.
You can customize the files that
are modified with the -p
or --pattern
argument, which takes a UNIX glob-style pattern and can be
repeated as many times as you like. To help exclude files, if the '~' is the first letter of the pattern,
then the rest of the pattern is used to exclude (not include) files. So, for example, if you provide the
following source code tree:
mypackage
__init__.py
foo.py
bar.py
tests/
__init__.py
test_foo.py
test_bar.py
The following commands would match the following lists of files:
addheader mypackage -t header.txt -p *.py
mypackage/{init.py, foo.py, bar.py}, mypackage/tests/{init.py, test_foo.py, test_bar.py}addheader mypackage -t header.txt -p *.py -p ~__init__.py
mypackage/{foo.py, bar.py}, mypackage/tests/{test_foo.py, test_bar.py}addheader mypackage -t header.txt -p *.py -p ~__init__.py -p ~test_*.py
mypackage/{foo.py, bar.py}
The header itself is, by default, delimited by a line of 78 '#' characters. While detecting an existing header, the program will look for any separator of 10 or more '#' characters. For example, if you have a file that looks like this:
##########
my header with 10
hashes above and below
##########
hello
and a header text file containing simply "Hello, world!", then the modified header will be:
##############################################################################
# Hello, world!
##############################################################################
hello
The comment character and separator character, as well as the width of the separator, can be modified with command-line options. For example, to add a C/C++ style comment as a header, use these options:
addheader mypackage --comment "//" --sep "=" --sep-len 40 -t myheader.txt
This will insert a header that looks like this:
//========================================
// my text goes here
//========================================
Keep in mind that subsequent operations on files with this header, including
--remove
, will need the same --comment
and --sep
arguments so that the header can be properly identified. For example,
running addheader mypackage --remove
after the above command will not
remove anything, and addheader mypackage -t myheader.txt
will insert a
second header (using the default comment character and separator).
You can control whether the final line has a newline character appended with the --final-linesep
command-line option (or the final_linesep
configuration option). This is True by default for text files, but False for Jupyter notebooks. The logic is that Jupyter notebook headers are in their own cell -- and also, this avoids spurious modifications by the Black code reformatter.
To avoid passing command-line arguments every time, use the configuration file. See the "Configuration" section for more details.
Starting in version 0.3.0, you can add headers to Jupyter Notebooks as well.
To enable Jupyter notebooks, you must install the 'jupyter' optional dependencies, e.g.,
pip install addheader[jupyter]
.
To enable this, add a -j {suffix}
or --jupyter {suffix}
argument to the command-line, or
similarly add a jupyter: {suffix}
argument in the configuration file.
The {suffix}
indicates an alternate file suffix to use for identifying
whether a file is a Jupyter Notebook, where the default is ".ipynb".
In the configuration file, use jupyter: true
to use the default.
On the command-line, omit the value to use the default.
To set the Jupyter notebook format version, add --notebook-version {value}
to the command-line or, equivalently, notebook_version: {value}
to the configuration file.
Values can be from 1 to 4. The default value is 4.
The file pattern arguments (see Specifying file patterns, above) are still honored,
but if Jupyter notebooks are enabled, the pattern *{suffix}
will be automatically added
to the patterns to match. Thus, by default *.ipynb
will be added to the files to match.
If there is no existing header, the Jupyter notebook header will be inserted as the first 'cell', i.e. the first
item, in the notebook. An existing header will be found anywhere in the notebook (by its header
tag, see below).
Currently the header cell is of type "code", with every line of the cell commented (using a 'markdown' cell is another possibility, but the code cell is friendler to the Jupyterbook machinery, and also retains the header in exported versions of the notebook without markdown cells). The content of the header is the same as for text files. Two, optionally three, tags will be added to the cell metadata:
header
- Indicates this is the header cell, so it can be modified or removed later.hide-cell
- If you build documentation with Jupyterbook, this will hide the cell in the generated documentation behind a toggle button (see https://jupyterbook.org/interactive/hiding.html).
Just as for text files, Jupyter notebook headers can be updated or removed.
For reference, below is the form of the generated Jupyter notebook cell JSON (with the 'id' field):
{
"id": "1234567890abcdef1234567890abcdef",
"cell_type": "code",
"metadata": {
"tags": [
"header",
"hide-cell"
]
},
"source": [
"# Copyright info\n",
"# is placed here.\n"
],
"outputs": []
}
To avoid passing commandline arguments every time, you can create a configuration
file that simply lists them as key/value pairs (using the long-option name as
the key). By default, the program will look for a file addheader.cfg
in the
current directory, but this can also be specified on the command-line with
-c/--config
. For example:
addheader # looks for addheader.cfg, ok if not present
addheader -c myoptions.conf # uses myoptions.conf, fails if not present
The configuration file is in YAML format. For example:
text: myheader.txt
pattern:
- "*.py"
- "~__init__.py"
# C/Java style comment block
sep: "-"
comment: "//"
sep-len: 40
# Verbosity as a number instead of -vv
verbose: 2
Command-line arguments will override configuration arguments, even if the
configuration file is explicitly provided with -c/--config
.
The "action" arguments, -r/--remove
and -n/--dry-run
, will be
ignored in the configuration file.
To test your package for files missing headers, use the following formula:
import os
import mypackage
from addheader.add import FileFinder, detect_files
def test_headers():
root = os.path.dirname(mypackage.__file__)
# modify patterns to match the files that should have headers
ff = FileFinder(root, glob_pat=["*.py", "~__init__.py"])
has_header, missing_header = detect_files(ff)
assert len(missing_header) == 0
The addheader program was developed for use in the IDAES project and is maintained in the IDAES organization in Github at https://github.com/IDAES/addheader . The primary author and maintainer is Dan Gunter (dkgunter at lbl dot gov).
Please see the COPYRIGHT.md and LICENSE.md files in the repository for limitations on use and distribution of this software.