Check lesson files and their contents.
"""
+from __future__ import print_function
import sys
import os
import glob
import re
from optparse import OptionParser
-from util import Reporter, read_markdown, load_yaml
+from util import Reporter, read_markdown, load_yaml, check_unwanted_files, require, IMAGE_FILE_SUFFIX
__version__ = '0.2'
# Pattern to match lines ending with whitespace.
P_TRAILING_WHITESPACE = re.compile(r'\s+$')
+# Pattern to match figure references in HTML.
+P_FIGURE_REFS = re.compile(r'<img[^>]+src="([^"]+)"[^>]*>')
+
+# Pattern to match internally-defined Markdown links.
+P_INTERNALLY_DEFINED_LINK = re.compile(r'\[[^\]]+\]\[[^\]]+\]')
+
# What kinds of blockquotes are allowed?
KNOWN_BLOCKQUOTES = {
'callout',
'source',
'bash',
'make',
+ 'matlab',
'python',
'r',
'sql'
check_config(args.reporter, args.source_dir)
docs = read_all_markdown(args.source_dir, args.parser)
check_fileset(args.source_dir, args.reporter, docs.keys())
+ check_unwanted_files(args.source_dir, args.reporter)
for filename in docs.keys():
checker = create_checker(args, filename, docs[filename])
checker.check()
+ check_figures(args.source_dir, args.reporter)
args.reporter.report()
config_file = os.path.join(source_dir, '_config.yml')
config = load_yaml(config_file)
reporter.check_field(config_file, 'configuration', config, 'kind', 'lesson')
+ reporter.check_field(config_file, 'configuration', config, 'carpentry', ('swc', 'dc'))
+ reporter.check_field(config_file, 'configuration', config, 'title')
+ reporter.check_field(config_file, 'configuration', config, 'email')
+
+ reporter.check({'values': {'root': '..'}} in config.get('defaults', []),
+ 'configuration',
+ '"root" not set to ".." in configuration')
def read_all_markdown(source_dir, parser):
seen)
+def check_figures(source_dir, reporter):
+ """Check that all figures are present and referenced."""
+
+ # Get references.
+ try:
+ all_figures_html = os.path.join(source_dir, '_includes', 'all_figures.html')
+ with open(all_figures_html, 'r') as reader:
+ text = reader.read()
+ figures = P_FIGURE_REFS.findall(text)
+ referenced = [os.path.split(f)[1] for f in figures if '/fig/' in f]
+ except FileNotFoundError as e:
+ reporter.add(all_figures_html,
+ 'File not found')
+ return
+
+ # Get actual image files (ignore non-image files).
+ fig_dir_path = os.path.join(source_dir, 'fig')
+ actual = [f for f in os.listdir(fig_dir_path) if os.path.splitext(f)[1] in IMAGE_FILE_SUFFIX]
+
+ # Report differences.
+ unexpected = set(actual) - set(referenced)
+ reporter.check(not unexpected,
+ None,
+ 'Unexpected image files: {0}',
+ ', '.join(sorted(unexpected)))
+ missing = set(referenced) - set(actual)
+ reporter.check(not missing,
+ None,
+ 'Missing image files: {0}',
+ ', '.join(sorted(missing)))
+
+
def create_checker(args, filename, info):
"""Create appropriate checker for file."""
return cls(args, filename, **info)
-def require(condition, message):
- """Fail if condition not met."""
-
- if not condition:
- print(message, file=sys.stderr)
- sys.exit(1)
-
-
class CheckBase(object):
"""Base class for checking Markdown files."""
self.check_trailing_whitespace()
self.check_blockquote_classes()
self.check_codeblock_classes()
+ self.check_defined_link_references()
def check_metadata(self):
cls)
+ def check_defined_link_references(self):
+ """Check that defined links resolve in the file.
+
+ Internally-defined links match the pattern [text][label]. If
+ the label contains '{{...}}', it is hopefully a references to
+ a configuration value - we should check that, but don't right
+ now.
+ """
+
+ result = set()
+ for node in self.find_all(self.doc, {'type' : 'text'}):
+ for match in P_INTERNALLY_DEFINED_LINK.findall(node['value']):
+ if '{{' not in match:
+ result.add(match)
+ self.reporter.check(not result,
+ self.filename,
+ 'Internally-defined links may be missing definitions: {0}',
+ ', '.join(sorted(result)))
+
+
def find_all(self, node, pattern, accum=None):
"""Find all matches for a pattern."""
super(CheckIndex, self).__init__(args, filename, metadata, metadata_len, text, lines, doc)
self.layout = 'lesson'
+ def check_metadata(self):
+ super(CheckIndex, self).check_metadata()
+ self.reporter.check(self.metadata.get('root', '') == '.',
+ self.filename,
+ 'Root not set to "."')
+
class CheckEpisode(CheckBase):
"""Check an episode page."""