Move verse selector to htmx

utils/compile_bible.py

@@ -45,6 +45,7 @@ from __future__ import annotations
 
 import argparse
 import html
+import json
 import os
 import re
 import sys
@@ -777,6 +778,91 @@ def render_index(books, translation, out_dir):
     return path
 
 
+# ==========================================================================
+# WRITING -- the reference map
+# ==========================================================================
+
+# bible/verses.json: every book's chapters, and every chapter's verse labels.
+#
+# The navbar's reference picker is built from this, by the service worker. It
+# used to be built in the browser by reading the book already on the page --
+# the chapters were the sections and a chapter's verses were the anchors inside
+# it -- which was free but only worked where there was a document to read. A
+# worker has no DOM, and the picker now lives in a shell the worker composes
+# before any document exists, so what the picker needs has to be a fact on
+# disk rather than something recovered from markup.
+#
+# Written here, from the same tree the fragments are rendered from, so the two
+# cannot disagree: every option this file produces names an anchor render_book
+# wrote, because both come from the same verse nodes.
+#
+# The encoding is compact on purpose. A chapter whose verses run 1..n is
+# written as the number n; only the eight in the corpus with a gap or a range
+# -- Sirach 26 ends "19-27", 28, 29 -- spell their labels out:
+#
+#     {"GEN": [31, 25, 24, ...], "SIR": [..., ["1", ..., "19-27", "28", "29"]]}
+#
+# That is 6 KB for 38,000 verses against 180 KB of spelled-out labels, which is
+# the difference between a file the worker can parse every time it wakes and
+# one it would have to find somewhere to keep.
+#
+# Chapter numbers are not in the file at all: they are the array index plus
+# one. render_verses checks that, rather than assuming it.
+
+VERSES_FILE = "verses.json"
+
+
+def chapter_verses(chapter):
+    """The verse labels of one chapter, in document order.
+
+    Blocks only. Footnotes hang inside the chapter section too, but a note's
+    content never carries a verse node -- and if one ever did, it would be a
+    citation rather than a place in the text.
+    """
+    labels = []
+
+    def walk(nodes):
+        for node in nodes:
+            if node["t"] == "verse":
+                labels.append(node["label"])
+            elif "c" in node:
+                walk(node["c"])
+
+    for block in chapter["blocks"]:
+        walk(block.get("content", []))
+    return labels
+
+
+def render_verses(books, out_dir):
+    """Write the reference map. Returns its path and the verses in it.
+
+    See the note above for the encoding.
+    """
+    out = {}
+    total = 0
+    for book in books:
+        numbers = [c["number"] for c in book["chapters"]]
+        if numbers != list(range(1, len(numbers) + 1)):
+            # The map addresses chapters by position, so a book that skips or
+            # repeats one would silently mislabel every option after the gap.
+            # Better to stop the build than to ship a picker that lies.
+            raise RuntimeError("%s: chapters are not 1..%d (%s)"
+                               % (book["code"], len(numbers), numbers))
+        chapters = []
+        for chapter in book["chapters"]:
+            labels = chapter_verses(chapter)
+            plain = [str(i + 1) for i in range(len(labels))]
+            chapters.append(len(labels) if labels == plain else labels)
+            total += len(labels)
+        out[book["code"]] = chapters
+
+    path = os.path.join(out_dir, VERSES_FILE)
+    with open(path, "w", encoding="utf-8") as fh:
+        json.dump(out, fh, separators=(",", ":"), ensure_ascii=False)
+        fh.write("\n")
+    return path, total
+
+
 # ==========================================================================
 # Entry point
 # ==========================================================================
@@ -797,6 +883,7 @@ def main(argv=None):
     for book in data["books"]:
         total += os.path.getsize(render_book(book, data["translation"], args.out))
     render_index(data["books"], data["translation"], args.out)
+    verses, count = render_verses(data["books"], args.out)
 
     chapters = sum(len(b["chapters"]) for b in data["books"])
     notes = sum(len(c["notes"]) for b in data["books"] for c in b["chapters"])
@@ -807,6 +894,9 @@ def main(argv=None):
     print("skipped  %d (contents pages)" % len(skipped), file=sys.stderr)
     print("html     %.1f MB total, %.0f KB average"
           % (total / 1e6, total / len(data["books"]) / 1e3), file=sys.stderr)
+    print("verses   %d in %s (%.1f KB)"
+          % (count, os.path.basename(verses), os.path.getsize(verses) / 1e3),
+          file=sys.stderr)
     print("wrote    %s/" % args.out.rstrip("/"), file=sys.stderr)
     return 0