Allow specifying an editor as well as/instead of an author
[books.alexwlchan.net] / scripts / render_html.py
index c9f11e3..048a7fc 100755 (executable)
@@ -10,6 +10,8 @@ import subprocess
 import sys
 
 import attr
+import bs4
+import cssmin
 import frontmatter
 from jinja2 import Environment, FileSystemLoader, select_autoescape
 import markdown
@@ -24,11 +26,55 @@ def rsync(dir1, dir2):
     subprocess.check_call(["rsync", "--recursive", "--delete", dir1, dir2])
 
 
+def git(*args):
+    return subprocess.check_output(["git"] + list(args)).strip().decode("utf8")
+
+
+def set_git_timestamps():
+    """
+    For everything in the covers/ directory, set the last modified timestamp to
+    the last time it was modified in Git.  This should make tint colour computations
+    stable across machines.
+    """
+    root = git("rev-parse", "--show-toplevel")
+
+    now = datetime.datetime.now().timestamp()
+
+    for f in os.listdir("src/covers"):
+        path = os.path.join("src/covers", f)
+
+        if not os.path.isfile(path):
+            continue
+
+        stat = os.stat(path)
+
+        # If the modified time is >7 days ago, skip setting the modified time.  This means
+        # the script stays pretty fast when doing a regular sync.
+        if now - stat.st_mtime > 7 * 24 * 60 * 60 and "--reset" not in sys.argv:
+            continue
+
+        revision = git("rev-list", "--max-count=1", "HEAD", path)
+
+        if not revision:
+            continue
+
+        timestamp, *_ = git("show", "--pretty=format:%ai", "--abbrev-commit", revision).splitlines()
+        modified_time = datetime.datetime.strptime(timestamp, "%Y-%m-%d %H:%M:%S %z").timestamp()
+
+        access_time = stat.st_atime
+
+        os.utime(path, times=(access_time, modified_time))
+
+
 @attr.s
 class Book:
+    slug = attr.ib()
     title = attr.ib()
-    author = attr.ib()
     publication_year = attr.ib()
+
+    author = attr.ib(default="")
+    editor = attr.ib(default="")
+
     cover_image = attr.ib(default="")
     cover_desc = attr.ib(default="")
 
@@ -40,6 +86,7 @@ class Book:
 class Review:
     date_read = attr.ib()
     text = attr.ib()
+    date_order = attr.ib(default=1)
     format = attr.ib(default=None)
     rating = attr.ib(default=None)
     did_not_finish = attr.ib(default=False)
@@ -66,6 +113,7 @@ def get_review_entry_from_path(path):
         except KeyError:
             pass
 
+    kwargs["slug"] = os.path.basename(os.path.splitext(path)[0])
     book = Book(**kwargs)
 
     review = Review(**post["review"], text=post.content)
@@ -88,7 +136,9 @@ class CurrentlyReadingEntry:
 def get_reading_entry_from_path(path):
     post = frontmatter.load(path)
 
-    book = Book(**post["book"])
+    slug = os.path.basename(os.path.splitext(path)[0])
+    book = Book(slug=slug, **post["book"])
+
     reading = CurrentlyReading(text=post.content)
 
     return CurrentlyReadingEntry(path=path, book=book, reading=reading)
@@ -117,7 +167,9 @@ class PlanEntry:
 def get_plan_entry_from_path(path):
     post = frontmatter.load(path)
 
-    book = Book(**post["book"])
+    slug = os.path.basename(os.path.splitext(path)[0])
+    book = Book(slug=slug, **post["book"])
+
     plan = Plan(date_added=post["plan"]["date_added"], text=post.content)
 
     return PlanEntry(path=path, book=book, plan=plan)
@@ -163,18 +215,22 @@ def render_date(date_value):
         return date_obj.strftime("%B %Y")
 
 
-def render_individual_review(env, *, review_entry, **kwargs):
-    template = env.get_template("review.html")
-    html = template.render(
-        review_entry=review_entry,
-        title=f"My review of {review_entry.book.title}",
-        **kwargs
-    )
-
-    out_name = review_entry.out_path() / "index.html"
-    out_path = pathlib.Path("_html") / out_name
+def save_html(template, out_name="", **kwargs):
+    html = template.render(**kwargs)
+    out_path = pathlib.Path("_html") / out_name / "index.html"
     out_path.parent.mkdir(exist_ok=True, parents=True)
-    out_path.write_text(html)
+
+    soup = bs4.BeautifulSoup(html, "html.parser")
+
+    # Minify the CSS in all inline <style> tags.
+    for style_tag in soup.find_all("style"):
+        style_tag.string = cssmin.cssmin(style_tag.string)
+
+    # Remove any comments
+    for comment in soup(text=lambda text: isinstance(text, bs4.Comment)):
+        comment.extract()
+
+    out_path.write_text(str(soup))
 
 
 def _create_new_thumbnail(src_path, dst_path):
@@ -231,6 +287,8 @@ def css_hash(_):
 
 
 def main():
+    set_git_timestamps()
+
     env = Environment(
         loader=FileSystemLoader("templates"),
         autoescape=select_autoescape(["html", "xml"]),
@@ -256,18 +314,21 @@ def main():
         get_entries(dirpath="src/reviews", constructor=get_review_entry_from_path)
     )
     all_reviews = sorted(
-        all_reviews, key=lambda rev: str(rev.review.date_read), reverse=True
+        all_reviews, key=lambda rev: f"{rev.review.date_read}/{rev.review.date_order}", reverse=True
     )
 
     for review_entry in all_reviews:
-        render_individual_review(
-            env,
+        save_html(
+            template=env.get_template("review.html"),
+            out_name=review_entry.out_path(),
             review_entry=review_entry,
+            title=f"My review of {review_entry.book.title}",
             tint_colors=tint_colors
         )
 
-    template = env.get_template("list_reviews.html")
-    html = template.render(
+    save_html(
+        template=env.get_template("list_reviews.html"),
+        out_name="reviews",
         all_reviews=[
             (year, list(reviews))
             for (year, reviews) in itertools.groupby(
@@ -276,12 +337,9 @@ def main():
         ],
         title="books i’ve read",
         this_year=str(datetime.datetime.now().year),
-        tint_colors=tint_colors,
+        tint_colors=tint_colors
     )
 
-    out_path = pathlib.Path("_html") / "reviews/index.html"
-    out_path.write_text(html)
-
     # Render the "currently reading" page
 
     all_reading = list(
@@ -290,17 +348,14 @@ def main():
         )
     )
 
-    template = env.get_template("list_reading.html")
-    html = template.render(
+    save_html(
+        template=env.get_template("list_reading.html"),
+        out_name="reading",
         all_reading=all_reading,
         title="books i’m currently reading",
         tint_colors=tint_colors
     )
 
-    out_path = pathlib.Path("_html") / "reading/index.html"
-    out_path.parent.mkdir(exist_ok=True, parents=True)
-    out_path.write_text(html)
-
     # Render the "want to read" page
 
     all_plans = list(
@@ -309,17 +364,14 @@ def main():
 
     all_plans = sorted(all_plans, key=lambda plan: plan.plan.date_added, reverse=True)
 
-    template = env.get_template("list_plans.html")
-    html = template.render(
+    save_html(
+        template=env.get_template("list_plans.html"),
+        out_name="to-read",
         all_plans=all_plans,
         title="books i want to read",
         tint_colors=tint_colors,
     )
 
-    out_path = pathlib.Path("_html") / "to-read/index.html"
-    out_path.parent.mkdir(exist_ok=True, parents=True)
-    out_path.write_text(html)
-
     # Render the "never going to read this page"
 
     all_retired = list(
@@ -330,29 +382,23 @@ def main():
         all_retired, key=lambda plan: plan.plan.date_added, reverse=True
     )
 
-    template = env.get_template("list_will_never_read.html")
-    html = template.render(
+    save_html(
+        template=env.get_template("list_will_never_read.html"),
+        out_name="will-never-read",
         all_retired=all_retired,
         title="books i&rsquo;m never going to read",
         tint_colors=tint_colors
     )
 
-    out_path = pathlib.Path("_html") / "will-never-read/index.html"
-    out_path.parent.mkdir(exist_ok=True, parents=True)
-    out_path.write_text(html)
-
     # Render the front page
 
-    index_template = env.get_template("index.html")
-    html = index_template.render(
+    save_html(
+        template=env.get_template("index.html"),
         text=open("src/index.md").read(),
         reviews=all_reviews[:5],
         tint_colors=tint_colors
     )
 
-    index_path = pathlib.Path("_html") / "index.html"
-    index_path.write_text(html)
-
     print("✨ Rendered HTML files to _html ✨")