Building new commands

grep is very useful, and you can call it from marcel, e.g.

grep "def render_compact" *.py

Running this in marcel's marcel/object directory yields:

error.py:    def render_compact(self):
file.py:    def render_compact(self):
historyrecord.py:    def render_compact(self):
process.py:    def render_compact(self):
renderable.py:    def render_compact(self):

This output is not structured, so it is difficult to process further. It would be nice if there were a marcel version of grep that produced structured output, e.g.

(error.py, '        return self.render_compact()')
(error.py, '    def render_compact(self):')
(error.py, '        out = self.render_compact()')
(file.py, '    def render_compact(self):')
(historyrecord.py, '    def render_compact(self):')
(process.py, '    def render_compact(self):')
(renderable.py, '        return self.render_compact()')
(renderable.py, '    def render_compact(self):')

This output consists of (File object, string from file) tuples.

A simple version of grep can be built in marcel itself, using parameterized pipelines:

grem = (| pattern, *files: read --label (files) | \
                          select (file, line: pattern in line) |)

This new (pseudo-) operator is named grem. It has two parameters: pattern, a string to search for; and files, one or more filenames (or globs). The read operator reads the contents of each file, and the --label flag includes a label with each line, the File containing that line. The select operator is used to keep only those (file, line) pairs in which the line contains the given pattern.

So now you can write grem "def set_env" *.py and get the structured result shown above.

If you want grem to be available in all marcel sessions, you can define it in your config file (~/.marcel.py), assigning it to the RUN_ON_STARTUP variable, e.g.

RUN_ON_STARTUP = '''
grem = (| pattern, *files: read -l (files) | select (f, l: pattern in l) |)
'''

(You can add more marcel commands to RUN_ON_STARTUP.)