Refactor dependency writer to make it easier to reason about indices

1 files changed, 67 insertions, 21 deletions

diff --git a/repo/www/make-deps.py b/repo/www/make-deps.py
index 4ca4e6e..26a4037 100755
--- a/repo/www/make-deps.py
+++ b/repo/www/make-deps.py
@@ -1,8 +1,34 @@
 #!/usr/bin/env python3
 
+"""Write dependencies for all website pages in makefile syntax.
+
+We want to compute:
+
+- a list of leaf pages,
+- a list of auto-generated indices,
+
+- dependencies for leaf pages:
+  OUTPUT/foo/bar.html: foo/bar.txt | OUTPUT/foo
+
+    - special case for READMEs:
+      OUTPUT/foo/index.html: foo/README.txt foo | OUTPUT/foo
+
+- dependencies for auto-generated indices:
+  OUTPUT/foo/index.html: foo | OUTPUT/foo
+"""
+
+from collections import defaultdict
+from dataclasses import dataclass, field
 from os import path
 from subprocess import run
 from sys import argv, exit
+from typing import List, Set
+
+
+@dataclass
+class Directory:
+    files: List[str] = field(default_factory=list)
+    subfolders: Set[str] = field(default_factory=set)
 
 
 def parse_arguments(args):
@@ -14,6 +40,7 @@ def parse_arguments(args):
 
 def find_sources(top_dir):
     p = run(
+        # TODO: use git ls-files
         ('find', top_dir, '-name', '*.md', '-o', '-name', '*.org'),
         capture_output=True, check=True, text=True
     )
@@ -21,41 +48,60 @@ def find_sources(top_dir):
     return p.stdout.splitlines()
 
 
-def html_path(source_path, top_dir, out_dir):
-    fname = path.basename(source_path)
-    dname = path.dirname(source_path)
+def compute_directories(files, top_dir):
+    directories = defaultdict(Directory)
 
-    name, _ = path.splitext(fname)
+    for f in files:
+        fpath, fname = path.split(f)
+        fdir = (
+            path.relpath(fpath, top_dir)
+            if fpath != top_dir
+            else ''
+        )
 
-    if name == 'README':
-        name = 'index'
+        directories[fdir].files.append(fname)
 
-    return path.join(
-        out_dir, path.relpath(dname, top_dir), f'{name}.html'
-    )
+        if fdir:
+            parent, child = path.split(fdir)
+            directories[parent].subfolders.add(child)
+
+    return directories
 
 
-def write_dependencies(output, sources, top_dir, out_dir):
-    pages = []
+def write_dependencies(deps, directories, top_dir, out_dir):
+    pages = list()
 
-    for src in sources:
-        html = html_path(src, top_dir, out_dir)
-        html_dir = path.dirname(html)
+    for dpath, d in directories.items():
 
-        print(f'{html}: {src} | {html_dir}', file=output)
+        for f in d.files:
+            name, _ = path.splitext(f)
+            if name == 'README':
+                name = 'index'
 
-        pages.append(html)
+            html_dir = (
+                path.join(out_dir, dpath)
+                if dpath
+                else out_dir
+            )
 
-    print(file=output)
-    print(f'pages = {" ".join(pages)}', file=output)
+            html_path = path.join(html_dir, name+'.html')
+            src_path = path.join(top_dir, dpath, f)
 
+            print(f'{html_path}: {src_path} | {html_dir}', file=deps)
+            pages.append(html_path)
 
-def main(argv):
-    top_dir, out_dir = parse_arguments(argv)
+    print(file=deps)
+    print(f'pages = {" ".join(pages)}', file=deps)
+
+
+def main(arguments):
+    top_dir, out_dir = parse_arguments(arguments)
     source_files = find_sources(top_dir)
 
+    directories = compute_directories(source_files, top_dir)
+
     with open('deps.mk', 'w') as deps:
-        write_dependencies(deps, source_files, top_dir, out_dir)
+        write_dependencies(deps, directories, top_dir, out_dir)
 
 
 if __name__ == '__main__':