217 lines
8.4 KiB
Python
217 lines
8.4 KiB
Python
import fnmatch
|
|
import glob
|
|
import io
|
|
import itertools
|
|
import logging
|
|
import os
|
|
import subprocess
|
|
|
|
from sqlalchemy.orm.exc import NoResultFound
|
|
|
|
from aeromancer.db.models import Project, File, Line
|
|
from aeromancer import filehandler
|
|
from aeromancer import utils
|
|
|
|
|
|
LOG = logging.getLogger(__name__)
|
|
|
|
|
|
def discover(repo_root, organizations):
|
|
"""Discover project-like directories under the repository root"""
|
|
glob_patterns = ['%s/*' % o for o in organizations]
|
|
with utils.working_dir(repo_root):
|
|
return itertools.ifilter(
|
|
lambda x: os.path.isdir(os.path.join(repo_root, x)),
|
|
itertools.chain(*(glob.glob(g) for g in glob_patterns))
|
|
)
|
|
|
|
def _find_files_in_project(path):
|
|
"""Return a list of the files managed in the project and their sha hash.
|
|
|
|
Uses 'git ls-files -s'
|
|
"""
|
|
with utils.working_dir(path):
|
|
# Ask git to tell us the sha hash so we can tell if the file
|
|
# has changed since we looked at it last.
|
|
cmd = subprocess.Popen(['git', 'ls-files', '-z', '-s'],
|
|
stdout=subprocess.PIPE)
|
|
output = cmd.communicate()[0]
|
|
entries = output.split('\0')
|
|
for e in entries:
|
|
if not e:
|
|
continue
|
|
metadata, ignore, filename = e.partition('\t')
|
|
sha = metadata.split(' ')[1]
|
|
yield (filename, sha)
|
|
|
|
|
|
class ProjectManager(object):
|
|
|
|
_DO_NOT_READ = [
|
|
'*.doc', '*.docx', '*.graffle', '*.odp', '*.pptx', '*.vsd', '*.xsd',
|
|
'*.gif', '*.ico', '*.jpeg', '*.jpg', '*.png', '*.tiff', '*.JPG',
|
|
'*.gpg',
|
|
'*.jar', # Why do we check in jar files?!
|
|
'*.swf', '*.eot',
|
|
'*.ttf', '*.woff', # webfont; horizon
|
|
'*.xml',
|
|
'*.gz', '*.zip', '*.z',
|
|
'*.mo', '*.db',
|
|
]
|
|
|
|
def __init__(self, session):
|
|
self.file_handlers = filehandler.load_handlers()
|
|
self.session = session
|
|
|
|
def get_project(self, name):
|
|
"""Return an existing project, if there is one"""
|
|
query = self.session.query(Project).filter(Project.name == name)
|
|
try:
|
|
return query.one()
|
|
except NoResultFound:
|
|
return None
|
|
|
|
def add_or_update(self, name, path, force=False):
|
|
"""Create a new project definition or update an existing one"""
|
|
proj_obj = self.get_project(name)
|
|
if proj_obj:
|
|
proj_obj.path = path
|
|
LOG.info('updating project %s from %s', name, path)
|
|
else:
|
|
proj_obj = Project(name=name, path=path)
|
|
LOG.info('adding project %s from %s', name, path)
|
|
self.session.add(proj_obj)
|
|
self.session.flush()
|
|
assert proj_obj.id, 'No id for new project'
|
|
self.update(proj_obj, force=force)
|
|
return proj_obj
|
|
|
|
def update(self, proj_obj, force=False):
|
|
"""Update the settings for an existing project"""
|
|
self._update_project_files(proj_obj, force=force)
|
|
|
|
def remove(self, name):
|
|
"""Delete stored data for the named project"""
|
|
query = self.session.query(Project).filter(Project.name == name)
|
|
try:
|
|
proj_obj = query.one()
|
|
LOG.info('removing project %s', name)
|
|
except NoResultFound:
|
|
return
|
|
for file_obj in proj_obj.files:
|
|
self._remove_plugin_data_for_file(file_obj)
|
|
self.session.delete(proj_obj)
|
|
|
|
def _remove_plugin_data_for_file(self, file_obj):
|
|
# We have to explicitly have the handlers delete their data
|
|
# because the parent-child relationship of the tables is reversed
|
|
# because the plugins define the relationships.
|
|
for fh in self.file_handlers:
|
|
if fh.obj.supports_file(file_obj):
|
|
LOG.debug('removing %s plugin data for %s', fh.name, file_obj.name)
|
|
fh.obj.delete_data_for_file(self.session, file_obj)
|
|
|
|
def _remove_file_data(self, file_obj, reason='file has changed'):
|
|
"""Delete the data associated with the file, including plugin data and
|
|
file contents.
|
|
|
|
"""
|
|
LOG.debug('removing cached contents of %s: %s', file_obj.name, reason)
|
|
self._remove_plugin_data_for_file(file_obj)
|
|
self.session.query(Line).filter(Line.file_id == file_obj.id).delete()
|
|
self.session.delete(file_obj)
|
|
|
|
def _update_project_files(self, proj_obj, force):
|
|
"""Update the files stored for each project"""
|
|
LOG.debug('reading file contents in %s', proj_obj.name)
|
|
|
|
# Collect the known files in a project so we can test their
|
|
# SHAs quickly.
|
|
known = {f.name: f for f in proj_obj.files}
|
|
|
|
# Track the files we've seen so we can delete any files that
|
|
# are no longer present.
|
|
seen = set()
|
|
|
|
# Now load the files currently being managed by git.
|
|
for filename, sha in _find_files_in_project(proj_obj.path):
|
|
# Remember that we have seen the file in the project.
|
|
seen.add(filename)
|
|
# Skip things that are not files (usually symlinks).
|
|
fullname = os.path.join(proj_obj.path, filename)
|
|
if not os.path.isfile(fullname):
|
|
continue
|
|
try:
|
|
existing_file = known[filename]
|
|
if existing_file.sha == sha and not force:
|
|
# File has not changed, we can use the content we
|
|
# already have.
|
|
LOG.debug('using cached version of %s', filename)
|
|
continue
|
|
self._remove_file_data(existing_file)
|
|
except KeyError:
|
|
pass
|
|
new_file = File(project=proj_obj, name=filename, path=fullname, sha=sha)
|
|
self.session.add(new_file)
|
|
self.session.flush() # make sure new_file gets an id
|
|
assert new_file.id, 'No id for new file'
|
|
if any(fnmatch.fnmatch(filename, dnr) for dnr in self._DO_NOT_READ):
|
|
LOG.debug('ignoring contents of %s', fullname)
|
|
else:
|
|
LOG.debug('reading %s', fullname)
|
|
with io.open(fullname, mode='r', encoding='utf-8') as f:
|
|
try:
|
|
body = f.read()
|
|
except UnicodeDecodeError:
|
|
# FIXME(dhellmann): Be smarter about trying other
|
|
# encodings?
|
|
LOG.warn('Could not read %s as a UTF-8 encoded file, ignoring',
|
|
fullname)
|
|
continue
|
|
lines = body.splitlines()
|
|
# Use SQLalchemy's core mode to bulk insert the lines.
|
|
if lines:
|
|
self.session.execute(
|
|
Line.__table__.insert(),
|
|
[{'file_id': new_file.id,
|
|
'number': num,
|
|
'content': content}
|
|
for num, content in enumerate(lines, 1)]
|
|
)
|
|
LOG.debug('%s/%s has %s lines', proj_obj.name, filename, len(lines))
|
|
|
|
# Invoke plugins for processing files in special ways
|
|
for fh in self.file_handlers:
|
|
if fh.obj.supports_file(new_file):
|
|
fh.obj.process_file(self.session, new_file)
|
|
|
|
self.session.flush()
|
|
|
|
# Remove files that we have in the database but that were no
|
|
# longer seen in the git repository.
|
|
for name, obj in known.items():
|
|
if name not in seen:
|
|
self._remove_file_data(obj, reason='file no longer exists')
|
|
self.session.flush()
|
|
|
|
def grep(self, pattern, prj_filter):
|
|
"""Given a pattern, search for lines in files in all projects that match it.
|
|
|
|
Returns results of the query, including the four columns line
|
|
number, line content, filename, and project name.
|
|
|
|
"""
|
|
# TODO: Would it be more efficient to register the regexp
|
|
# function on the db session here instead of when we connect?
|
|
# We could pre-compile the regex and not pass it to each
|
|
# invocation of the function.
|
|
query = self.session.query(
|
|
Line.number, Line.content, File.name, Project.name,
|
|
).join(File, Project).filter(
|
|
Line.content.op('regexp')(pattern)
|
|
)
|
|
if prj_filter:
|
|
query = prj_filter.update_query(query)
|
|
query = query.order_by(Project.name, File.name, Line.number)
|
|
return query.yield_per(20).all()
|