python-tldp/tldp/sources.py

85 lines
3.1 KiB
Python
Raw Normal View History

2016-02-11 03:22:23 +00:00
#! /usr/bin/python
from __future__ import absolute_import, division, print_function
2016-02-11 03:22:23 +00:00
import os
import errno
2016-02-11 03:22:23 +00:00
from .utils import logger
from .typeguesser import guess, knownextensions
2016-02-11 03:22:23 +00:00
2016-02-16 04:58:44 +00:00
class SourceDirs(object):
2016-02-11 03:22:23 +00:00
def __repr__(self):
return '<%s:(%s docs)>' % \
(self.__class__.__name__, len(self.docs))
2016-02-11 03:22:23 +00:00
def __init__(self, args):
self.sourcedirs = [os.path.abspath(x) for x in args]
2016-02-11 03:22:23 +00:00
self.docs = list()
self.validateDirs()
2016-02-11 03:22:23 +00:00
self.enumerateDocuments()
2016-02-12 20:53:30 +00:00
self.docs.sort(key=lambda x: x.stem.lower())
2016-02-11 03:22:23 +00:00
def validateDirs(self):
results = [os.path.exists(x) for x in self.sourcedirs]
if not all(results):
2016-02-13 07:49:04 +00:00
for result, sdir in zip(results, self.sourcedirs):
logger.critical("Directory does not exist: " + sdir)
raise IOError(errno.ENOENT, os.strerror(errno.ENOENT), sdir)
2016-02-11 03:22:23 +00:00
def enumerateDocuments(self):
for sdir in self.sourcedirs:
docs = list()
for fname in os.listdir(sdir):
possible = os.path.join(sdir, fname)
if os.path.isfile(possible):
docs.append(SourceDocument(possible))
elif os.path.isdir(fname):
stem = os.path.basename(fname)
for ext in knownextensions:
possible = os.path.join(sdir, fname, stem + ext)
if os.path.isfile(possible):
docs.append(SourceDocument(possible))
logger.debug("Discovered %s documents in %s", len(docs), sdir)
self.docs.extend(docs)
logger.info("Discovered %s documents total", len(self.docs))
2016-02-11 03:22:23 +00:00
class SourceDocument(object):
def __repr__(self):
return '<%s:%s (%s)>' % \
(self.__class__.__name__, self.filename, self.doctype)
2016-02-11 03:22:23 +00:00
def __init__(self, filename):
# -- canonicalize the pathname we are given.
self.filename = os.path.abspath(filename)
if not os.path.exists(self.filename):
logger.critical("Missing source document: %s", self.filename)
raise IOError(errno.ENOENT, os.strerror(errno.ENOENT), self.filename)
2016-02-13 07:59:13 +00:00
if not os.path.isfile(self.filename):
logger.critical("Source document is not a plain file: %s", self.filename)
raise TypeError("Wrong type, not a plain file: " + self.filename)
2016-02-11 03:22:23 +00:00
logger.debug("Found existing %s", self.filename)
self.dirname, self.basename = os.path.split(self.filename)
2016-02-11 03:22:23 +00:00
self.stem, self.ext = os.path.splitext(self.basename)
self.stat = os.stat(self.filename)
self.resources = False # -- assume no ./images/, ./resources/
self.singlefile = True # -- assume only one file
parentdir = os.path.basename(self.dirname)
2016-02-11 03:22:23 +00:00
if parentdir == self.stem:
self.singlefile = False
for rdir in ('resources', 'images'):
if os.path.exists(os.path.join(self.dirname, rdir)):
2016-02-11 03:22:23 +00:00
self.resources = True
@property
2016-02-11 03:22:23 +00:00
def doctype(self):
return guess(self.filename)
2016-02-11 03:22:23 +00:00
# -- end of file