[INFRA] build pdf from mkdocs.yml (#1343)

* build pdf from mkdocs.yml

* fix for project fname

* Apply suggestions from code review

Co-authored-by: Chris Markiewicz <[email protected]>

* FIX: dedent

* Update pdf_build_src/pandoc_script.py

Co-authored-by: Chris Markiewicz <[email protected]>

* add root back

* FIX: Bad loop

* RF: Abstract finding needed files

Co-authored-by: Chris Markiewicz <[email protected]>

Author: sappelhoff

pdf_build_src/pandoc_script.py

@@ -3,8 +3,19 @@
 This is done once the duplicate src directory is processed.
 """
 import os
-import pathlib
 import subprocess
+import yaml
+from pathlib import Path
+
+HERE = Path(__file__).absolute()
+
+
+def _find(path, filename):
+    return next(
+        parent / filename
+        for parent in path.parents
+        if Path.is_file(parent / filename)
+    )
 
 
 def build_pdf(filename="bids-spec.pdf", logfile="bids-spec_pandoc_log.json"):
@@ -17,19 +28,26 @@ def build_pdf(filename="bids-spec.pdf", logfile="bids-spec_pandoc_log.json"):
     logfile : str
         Name of the log file. Defaults to "bids-spec_pandoc_log.json".
     """
-    # Files that are not supposed to be built into the PDF
-    EXCLUDE = ["./index.md", "./schema/README.md", "./pregh-changes.md"]
-
-    # Get all input files
-    markdown_list = []
-    for root, dirs, files in os.walk("."):
-        for file in files:
-            fpath = os.path.join(root, file)
-            if fpath.endswith(".md") and fpath not in EXCLUDE:
-                markdown_list.append(fpath)
-            elif fpath.endswith("index.md"):
-                # Special role for index.md
-                index_page = fpath
+
+    def _flatten_values(lst):
+        """Flatten a list of dicts of lists to a list of values."""
+        for obj in lst:
+            for val in obj.values():
+                if isinstance(val, str):
+                    yield val
+                else:
+                    yield from _flatten_values(val)
+
+    fname_mkdocs_yml = _find(HERE, "mkdocs.yml")
+
+    with open(fname_mkdocs_yml, "r") as stream:
+        mkdocs_yml = yaml.safe_load(stream)
+
+    sections = mkdocs_yml["nav"][0]["The BIDS Specification"]
+
+    # special files
+    index_page = "./index.md"
+    pandoc_metadata = _find(HERE, "metadata.yml")
 
     # Prepare the command options
     cmd = [
@@ -43,26 +61,23 @@ def build_pdf(filename="bids-spec.pdf", logfile="bids-spec_pandoc_log.json"):
         f"--output={filename}",
     ]
 
-    # location of this file: This is also the working directory when
-    # the pdf is being built using `cd build_pdf_src` and then
-    # `bash build_pdf.sh`
-    root = pathlib.Path(__file__).parent.absolute()
-
     # Resources are searched relative to the working directory, but
     # we can add additional search paths using <path>:<another path>, ...
     # When in one of the appendices/ files there is a reference to
     # "../04-modality-specific-files/images/...", then we need to use
     # appendices/ as a resource-path so that the relative files can
     # be found.
-    cmd += [f'--resource-path=.:{str(root / "appendices")}']
+    build_root = HERE.parent
+    cmd += [f'--resource-path=.:{build_root / "appendices"}']
 
     # Add input files to command
     # The filenames in `markdown_list` will ensure correct order when sorted
-    cmd += [str(root / index_page)]
-    cmd += [str(root / i) for i in ["../../metadata.yml"] + sorted(markdown_list)]
+    cmd += [str(build_root / index_page)]
+    cmd += [str(pandoc_metadata)]
+    cmd += [str(build_root / md) for md in _flatten_values(sections)]
 
     # print and run
-    print("running: \n\n" + "\n".join(cmd))
+    print("pandoc command being run: \n\n" + "\n".join(cmd))
     subprocess.run(cmd)