X-Git-Url: https://git.arvados.org/rnaseq-cwl-training.git/blobdiff_plain/9a5ab181b0ee63b72319678370de7b77c6dd895d..7f8e13d704e49b625b25b7b655d4c21db70c7ccf:/bin/lesson_check.py

diff --git a/bin/lesson_check.py b/bin/lesson_check.py
index 3a43ba3..311687e 100755
--- a/bin/lesson_check.py
+++ b/bin/lesson_check.py
@@ -4,6 +4,7 @@
 Check lesson files and their contents.
 """
 
+from __future__ import print_function
 import sys
 import os
 import glob
@@ -11,7 +12,7 @@ import json
 import re
 from optparse import OptionParser
 
-from util import Reporter, read_markdown, load_yaml, check_unwanted_files, require
+from util import Reporter, read_markdown, load_yaml, check_unwanted_files, require, IMAGE_FILE_SUFFIX
 
 __version__ = '0.2'
 
@@ -46,6 +47,9 @@ 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',
@@ -67,6 +71,7 @@ KNOWN_CODEBLOCKS = {
     'source',
     'bash',
     'make',
+    'matlab',
     'python',
     'r',
     'sql'
@@ -149,13 +154,10 @@ def check_config(reporter, source_dir):
     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_field(config_file, 'configuration', config, 'repo')
-    reporter.check_field(config_file, 'configuration', config, 'root')
-    if ('repo' in config) and ('root' in config):
-        reporter.check(config['repo'].endswith(config['root']),
-                       config_file,
-                       'Repository name "{0}" not consistent with root "{1}"',
-                       config['repo'], config['root'])
+
+    reporter.check({'values': {'root': '..'}} in config.get('defaults', []),
+                   'configuration',
+                   '"root" not set to ".." in configuration')
 
 
 def read_all_markdown(source_dir, parser):
@@ -227,9 +229,9 @@ def check_figures(source_dir, reporter):
                      'File not found')
         return
 
-    # Get actual files.
+    # 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 not f.startswith('.')]
+    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)
@@ -279,6 +281,7 @@ class CheckBase(object):
         self.check_trailing_whitespace()
         self.check_blockquote_classes()
         self.check_codeblock_classes()
+        self.check_defined_link_references()
 
 
     def check_metadata(self):
@@ -336,6 +339,26 @@ class CheckBase(object):
                                 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."""
 
@@ -405,6 +428,12 @@ class CheckIndex(CheckBase):
         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."""