Refactor textbook generator to check redirects

scripts/generate_book.py

@@ -8,6 +8,7 @@
 from tqdm import tqdm
 import numpy as np
 from glob import glob
+from uuid import uuid4
 import argparse
 
 DESCRIPTION = ("Convert a collection of Jupyter Notebooks into Jekyll "
@@ -26,6 +27,8 @@
 parser.add_argument("--path-toc", default=None, help="Path to the Table of Contents YAML file")
 parser.add_argument("--overwrite", action='store_true', help="Overwrite md files if they already exist.")
 parser.add_argument("--execute", action='store_true', help="Execute notebooks before converting to MD.")
+parser.add_argument("--local-build", action='store_true',
+                    help="Specify you are building site locally for later upload.")
 parser.set_defaults(overwrite=False, execute=False)
 
 # Defaults
@@ -75,6 +78,28 @@ def _copy_non_content_files():
         sh.copy2(ifile, new_path)
 
 
+def _case_sensitive_fs(path):
+    """True when filesystem at `path` is case sensitive, False otherwise.
+
+    Checks this by attempting to write two files, one w/ upper case, one
+    with lower. If after this only one file exists, the system is case-insensitive.
+
+    Makes directory `path` if it does not exist.
+    """
+    if not op.exists(path):
+        os.makedirs(path)
+    root = op.join(path, uuid4().hex)
+    fnames = [root + suffix for suffix in 'aA']
+    try:
+        for fname in fnames:
+            with open(fname, 'wt') as fobj:
+                fobj.write('text')
+        written = glob(root + '*')
+    finally:
+        for fname in written:
+            os.unlink(fname)
+    return len(written) == 2
+
 
 if __name__ == '__main__':
     ###############################################################################
@@ -118,6 +143,7 @@ def _copy_non_content_files():
 
     n_skipped_files = 0
     n_built_files = 0
+    case_check = _case_sensitive_fs(BUILD_FOLDER) and args.local_build
     print("Convert and copy notebook/md files...")
     for ix_file, page in enumerate(tqdm(list(toc))):
         url_page = page.get('url', None)
@@ -233,10 +259,19 @@ def _copy_non_content_files():
         # Front-matter YAML
         yaml_fm = []
         yaml_fm += ['---']
-        yaml_fm += ['redirect_from:']
-        yaml_fm += ['  - "{}"'.format(_prepare_url(url_page).replace('_', '-').lower())]
-        if ix_file == 0:
-            yaml_fm += ['  - "/"']
+        # In case pre-existing links are sanitized
+        sanitized = url_page.lower().replace('_', '-')
+        if sanitized != url_page:
+            if case_check and url_page.lower() == sanitized:
+                raise RuntimeError(
+                    'Redirect {} clashes with page {} for local build on '
+                    'case-insensitive FS\n'.format(sanitized, url_page) +
+                    'Rename source page to lower case or build on a case '
+                    'sensitive FS, e.g. case-sensitive disk image on Mac')
+            yaml_fm += ['redirect_from:']
+            yaml_fm += ['  - "{}"'.format(sanitized)]
+            if ix_file == 0:
+                yaml_fm += ['  - "/"']
         if path_url_page.endswith('.ipynb'):
             interact_path = 'content/' + path_url_page.split('content/')[-1]
             yaml_fm += ['interact_link: {}'.format(interact_path)]

scripts/tests/test_build.py

@@ -31,9 +31,10 @@ def is_not_in(lines, check):
 if op.isdir(op.join(curdir, 'site', '_build')):
     sh.rmtree(op.join(curdir, 'site', '_build'))
 
+print("Building site for test suite...")
 cmd = ["python", op.join(curdir, "..", "generate_book.py"),
        "--site-root", op.join(curdir, "site"), "--path-template", op.join(curdir, "..", "templates", "jekyllmd.tpl")]
-out = subprocess.call(cmd)
+out = subprocess.check_call(cmd)
 
 ####################################################
 # Check outputs

scripts/tests/test_license.py

@@ -10,6 +10,9 @@
 
 def test_license():
     # Not yes/no answers should error
+    if op.exists(op.join(curdir, "site", "content", "LICENSE.md")):
+        os.remove(op.join(curdir, "site", "content", "LICENSE.md"))
+
     cmd_error = ["python", op.join(curdir, "..", "license.py"),
                 "--path", op.join(curdir, "site", "content"), "--use-license", "blah"]