sg-text-forge 0.3.2__tar.gz → 0.3.3__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (62) hide show
  1. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/PKG-INFO +1 -1
  2. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/pyproject.toml +1 -1
  3. sg_text_forge-0.3.3/text_forge/migrate_curation.py +334 -0
  4. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/wtd_books.py +123 -39
  5. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/wtd_books_build.py +6 -4
  6. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/uv.lock +1 -1
  7. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/.github/copilot-instructions.md +0 -0
  8. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/.github/skills/check-i18n/SKILL.md +0 -0
  9. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/.github/skills/check-i18n/references/example_translations.json +0 -0
  10. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/.github/skills/check-i18n/references/test_cases.md +0 -0
  11. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/.github/skills/check-i18n/scripts/check_i18n.py +0 -0
  12. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/.github/workflows/publish-pypi.yml +0 -0
  13. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/.gitignore +0 -0
  14. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/AGENTS.md +0 -0
  15. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/LICENSE +0 -0
  16. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/Makefile +0 -0
  17. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/README.md +0 -0
  18. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/action.yml +0 -0
  19. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/docs/hindsight-memory/SKILL.md +0 -0
  20. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/docs/hindsight-memory.md +0 -0
  21. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/epub/book_meta.yml +0 -0
  22. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/epub/epub.css +0 -0
  23. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/hooks/nobr_emoticons.py +0 -0
  24. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/assets/js/editor.js +0 -0
  25. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/assets/js/translations.json +0 -0
  26. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/assets/stylesheets/editor.css +0 -0
  27. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/assets/stylesheets/text-forge.css +0 -0
  28. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/main.html +0 -0
  29. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/partials/comments.html +0 -0
  30. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/partials/downloads.html +0 -0
  31. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/partials/editor.html +0 -0
  32. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/partials/header.html +0 -0
  33. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/mkdocs/overrides/partials/source-file.html +0 -0
  34. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/hindsight-ingest.py +0 -0
  35. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/hindsight-wipe-documents-by-tag.py +0 -0
  36. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/hindsight-wtd-ingest-wrapper.py +0 -0
  37. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/hindsight_common.py +0 -0
  38. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/mkdocs-combine.py +0 -0
  39. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/mkdocs_common.py +0 -0
  40. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/process-epub-meta.py +0 -0
  41. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/pymdown-pandoc.lua +0 -0
  42. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/wtd-inventory.py +0 -0
  43. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/scripts/wtd_sections_common.py +0 -0
  44. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/__init__.py +0 -0
  45. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/anchor_ids.py +0 -0
  46. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/build.py +0 -0
  47. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/cli.py +0 -0
  48. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/livelib.py +0 -0
  49. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/__init__.py +0 -0
  50. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/hotkeys.json +0 -0
  51. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/plugins/text-forge/main.js +0 -0
  52. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/plugins/text-forge/manifest.json +0 -0
  53. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/scripts/insert_block.js +0 -0
  54. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/scripts/insert_image.js +0 -0
  55. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/scripts/insert_link.js +0 -0
  56. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/templater.json +0 -0
  57. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/templates/insert_block.md +0 -0
  58. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/templates/insert_image.md +0 -0
  59. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/obsidian/templates/insert_link.md +0 -0
  60. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/plugin.py +0 -0
  61. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/wtd_books_p3.py +0 -0
  62. {sg_text_forge-0.3.2 → sg_text_forge-0.3.3}/text_forge/wtd_links.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.5
2
2
  Name: sg-text-forge
3
- Version: 0.3.2
3
+ Version: 0.3.3
4
4
  Summary: MkDocs plugin and build pipeline for Markdown text projects with EPUB generation
5
5
  Project-URL: Homepage, https://github.com/shared-goals/text-forge
6
6
  Project-URL: Repository, https://github.com/shared-goals/text-forge
@@ -1,6 +1,6 @@
1
1
  [project]
2
2
  name = "sg-text-forge"
3
- version = "0.3.2"
3
+ version = "0.3.3"
4
4
  description = "MkDocs plugin and build pipeline for Markdown text projects with EPUB generation"
5
5
  readme = "README.md"
6
6
  license = {text = "MIT"}
@@ -0,0 +1,334 @@
1
+ """Migrate the old JSONC curation file to the markdown sections file.
2
+
3
+ python -m text_forge.migrate_curation <in.jsonc> <out.md>
4
+
5
+ One-off, but repeatable: the same input always yields the same bytes, so a
6
+ re-run diffs empty and the migration stays auditable after the fact.
7
+
8
+ What it does:
9
+
10
+ * reads the old file with its own JSONC scanner — the loader's comment stripper
11
+ was deleted when the format changed, and this module is the only place that
12
+ syntax still has a right to exist;
13
+ * multiplies every `weight` by 10, which is why the migration exists at all:
14
+ 1..35 leaves no room to insert a book between two neighbours, 10..350 does;
15
+ * carries each entry's title across: `// title` above a key, or `// title`
16
+ trailing a one-line entry, becomes the heading's eye text;
17
+ * orders sections by weight ascending, then the weightless entries by id, so the
18
+ file reads as a ranked list with an unranked tail;
19
+ * drops `my_mark` (it reaches the frontmatter from LiveLib shelf data, never
20
+ from curation) and any section other than `books`.
21
+ """
22
+
23
+ import argparse
24
+ import json
25
+ import re
26
+ import sys
27
+ from pathlib import Path
28
+
29
+ #: old jsonc key -> new bullet key, in the order bullets are emitted
30
+ CURATION_MERGE_ORDER = ("weight", "author", "status", "merges")
31
+
32
+ #: the weight factor: the whole point of migrating, so it is a named constant
33
+ #: rather than a `* 10` someone has to look up
34
+ WEIGHT_FACTOR = 10
35
+
36
+ #: the header every migrated file carries: the format's only documentation,
37
+ #: written for a reader who does not program. It carries no bullets on purpose:
38
+ #: the loader reads a bullet only inside a `## <id>` section, so a bullet here
39
+ #: would be a hard error in the very file this module writes.
40
+ DOC_BLOCK = """# Curation
41
+
42
+ Одна секция — одна редакция книги.
43
+
44
+ Заголовок секции — это id издания с LiveLib: `## <id> — <название>`.
45
+ id слева от тире читает сборка; название справа — только для глаз, сборка его
46
+ не трогает. Строки ниже — поля секции.
47
+
48
+ Что можно писать под заголовком секции (любой набор, в любом порядке).
49
+ weight — вес книги, целое число. Вес умножен на 10 по сравнению со старым
50
+ файлом: между соседними книгами остаются дырки, чтобы новую книгу можно было
51
+ вставить между ними, не перекладывая половину списка (15, 25, 110).
52
+ merges — id других изданий той же работы, через запятую. Их цитаты и отзыв
53
+ достаются этой заметке. author — автор, если LiveLib его не указал или указал
54
+ неверно. status — read / reading / stop, если полка LiveLib врёт.
55
+
56
+ Поля `my_mark` здесь нет: оценку сборка берёт с полки LiveLib, ей и верить.
57
+ Пустая секция (без полей) тоже законна: id уже говорит «эта книга в списке».
58
+ """
59
+
60
+
61
+ def scan_jsonc_lines(text: str) -> list:
62
+ """Per line of the file: (json data, raw with trailing comment, whole-line
63
+ comment text).
64
+
65
+ `data` is the line with every comment removed, so the file's line numbers
66
+ survive and a json error still points at the real line. `raw` puts back a
67
+ `//` comment that trailed the data: that is the one comment a key can name
68
+ its own entry with (`"1": {"weight": 1} // Заголовок`). `whole` is the text
69
+ of a comment occupying the line on its own, which names the entry whose key
70
+ sits BELOW it.
71
+
72
+ A `//` inside a string is data, so this walks characters instead of regexing
73
+ the text, and `\\"` does not toggle string state: an escaped quote cannot leave
74
+ a following `//` looking like data. A `/* */` span names no entry at all —
75
+ it is prose — but the lines it covered keep their place in the file. (This
76
+ is the scanner the old loader stripped with; it moved here when the loader
77
+ stopped reading JSONC.)
78
+ """
79
+ rows = []
80
+ data_buf, comment_buf = [], []
81
+ # what the current line carries: "" (no comment), "//", or "*"
82
+ marker = [""]
83
+ i, n = 0, len(text)
84
+ in_string = False
85
+
86
+ def flush():
87
+ """Hand over the line built so far and start the next one empty."""
88
+ data = "".join(data_buf)
89
+ comment = "".join(comment_buf)
90
+ trails_data = marker[0] == "//" and bool(data.strip())
91
+ whole = comment[2:].strip() if (marker[0] == "//" and not data.strip()) else ""
92
+ rows.append((data, data + (comment if trails_data else ""), whole))
93
+ data_buf.clear()
94
+ comment_buf.clear()
95
+ marker[0] = ""
96
+
97
+ while i < n:
98
+ ch = text[i]
99
+ if ch == "\n":
100
+ flush()
101
+ i += 1
102
+ continue
103
+ if in_string:
104
+ if ch == "\\":
105
+ data_buf.append(text[i:i + 2])
106
+ i += 2
107
+ continue
108
+ data_buf.append(ch)
109
+ if ch == '"':
110
+ in_string = False
111
+ i += 1
112
+ continue
113
+ if ch == '"':
114
+ in_string = True
115
+ data_buf.append(ch)
116
+ i += 1
117
+ continue
118
+ if text[i:i + 2] == "//" and not marker[0]:
119
+ marker[0] = "//"
120
+ while i < n and text[i] != "\n":
121
+ comment_buf.append(text[i])
122
+ i += 1
123
+ continue
124
+ if text[i:i + 2] == "/*" and not marker[0]:
125
+ # a block comment names no entry: swallow it whole, but let the lines
126
+ # it covered keep their place in the file
127
+ marker[0] = "*"
128
+ end = text.find("*/", i + 2)
129
+ stop = n if end == -1 else end + 2
130
+ while i < stop:
131
+ if text[i] == "\n":
132
+ flush()
133
+ i += 1
134
+ continue
135
+ data_buf.append(ch)
136
+ i += 1
137
+ if data_buf or comment_buf:
138
+ flush()
139
+ return rows
140
+
141
+
142
+ def strip_jsonc_comments(text: str) -> str:
143
+ """The file with comments removed, line numbers intact — for json.loads.
144
+
145
+ One line per input line, so a syntax error points at the real line. The
146
+ walking lives in scan_jsonc_lines, the module's one implementation.
147
+ """
148
+ return "\n".join(data for data, _raw, _whole in scan_jsonc_lines(text)) + "\n"
149
+
150
+
151
+ def _title_of(block: list) -> str:
152
+ """The title among the comment lines of one entry's block.
153
+
154
+ The old file wrote the title first and its url as a footnote after it:
155
+
156
+ // Моменты счастья — Алекс Дубас
157
+ // @url:`https://www.livelib.ru/book/1003047920`
158
+
159
+ so the title is the FIRST line that is not an `@url:` footnote.
160
+ """
161
+ for body in block:
162
+ if not body.startswith("@url"):
163
+ return body
164
+ return ""
165
+
166
+
167
+ def _own_title(key: str, raw: str, trailing: str) -> str:
168
+ """The title a key takes from the comment trailing its own line, or ''.
169
+
170
+ `raw` still shows the marker, so the key's tail can be read: the comment
171
+ names the key only when that tail is the key's complete value — closed off
172
+ by the end of the line — and not a value still open (`"1": {`).
173
+ """
174
+ if not trailing:
175
+ return ""
176
+ tail = raw[raw.index(f'"{key}"'):]
177
+ tail = tail.split("//", 1)[0]
178
+ closed = (tail.count("{") + tail.count("[")) <= (tail.count("}") + tail.count("]"))
179
+ return _title_of([trailing]) if closed else ""
180
+
181
+
182
+ def _titles_from_comments(text: str) -> dict:
183
+ """Map an edition id to the title held by the comment naming it.
184
+
185
+ The old file put a title in either of two places, and both are carried over:
186
+
187
+ // Моменты счастья — Алекс Дубас <- whole-line block above a key
188
+ // @url:`https://www.livelib.ru/book/1`
189
+ "1003047920": {"weight": 1},
190
+ "1000699994": {"status": "read"} // Зона — Сергей Довлатов
191
+
192
+ so an entry takes its own trailing comment when it is written as one line,
193
+ and otherwise the block above its key. `/* */` asides name nothing, and real
194
+ data ends a block. The same look-up names each `merges` member sitting
195
+ inside the array.
196
+ """
197
+ views = scan_jsonc_lines(text)
198
+ n = len(views)
199
+
200
+ def block_above(index: int) -> list:
201
+ """The whole-line `//` comments naming the id at `index`, top first."""
202
+ block = []
203
+ j = index - 1
204
+ while j >= 0:
205
+ data, _raw, whole = views[j]
206
+ if whole:
207
+ block.append(whole)
208
+ elif data.strip():
209
+ break # real data: anything above names some other entry
210
+ j -= 1
211
+ return list(reversed(block))
212
+
213
+ titles: dict = {}
214
+ for i, (data, raw, _whole) in enumerate(views):
215
+ keys = list(re.finditer(r'"(\d+)"\s*:', data))
216
+ for key in keys:
217
+ if key.group(1) in titles:
218
+ continue # a merges member names itself, see below
219
+ # a key takes its title from the comment trailing its own line (the
220
+ # one place a key can name itself), else from the whole-line block
221
+ # written above it
222
+ trailing = raw.split("//", 1)[1].strip() if "//" in raw else ""
223
+ own = _own_title(key.group(1), raw, trailing) if key is keys[-1] else ""
224
+ titles[key.group(1)] = own if own else _title_of(block_above(i))
225
+
226
+ # a merges array may break over lines with a comment above each member:
227
+ # walk the lines it spans and let a line holding an id and nothing else
228
+ # take the title of the block above it
229
+ if '"merges"' not in data:
230
+ continue
231
+ last = i
232
+ while data.count("[") > data.count("]") and last + 1 < n:
233
+ last += 1
234
+ for k in range(i + 1, last + 1):
235
+ member = re.fullmatch(r'\s*(\d+)\s*,?\s*', views[k][0])
236
+ if member:
237
+ titles.setdefault(member.group(1), _title_of(block_above(k)))
238
+ return titles
239
+
240
+
241
+ def _section_sort_key(item):
242
+ """Weight ascending, then the weightless entries by id, then id as the
243
+ tiebreaker — a total order, so the file is a function of the data."""
244
+ entry_id, entry = item
245
+ weight = entry.get("weight")
246
+ weight = int(weight) * WEIGHT_FACTOR if weight is not None else None
247
+ return (weight is None, weight if weight is not None else 0, entry_id)
248
+
249
+
250
+ def _eye(ident: str, title: str) -> str:
251
+ """An id plus whatever the eye should read beside it.
252
+
253
+ The loader reads only the part before the em dash, so a title (or a run of
254
+ titles, when the old file kept several for one id) goes after it and can
255
+ never steer the build.
256
+ """
257
+ return f"{ident} — {title}" if title else ident
258
+
259
+
260
+ def _section(entry_id: str, entry: dict, titles: dict) -> str:
261
+ """One `## <id> — <title>` heading plus its bullets.
262
+
263
+ A title only ever lands in the eye part of a line: the heading's text after
264
+ the em dash, or the ids' own annotation inside the `merges` bullet. That is
265
+ what makes the file safe to hand-edit — a wrong or missing title cannot steer
266
+ the build, because the loader reads ids and keys, never the eye text.
267
+ """
268
+ heading = "## " + _eye(entry_id, titles.get(entry_id, ""))
269
+ lines = [heading]
270
+
271
+ for key in CURATION_MERGE_ORDER:
272
+ value = entry.get(key)
273
+ if value is None:
274
+ continue
275
+ if key == "weight":
276
+ lines.append(f"- weight: {int(value) * WEIGHT_FACTOR}")
277
+ elif key == "merges":
278
+ # ids ONLY: the loader parses this value as the id list, so no title
279
+ # may be annotated into it — extra text would make the ids unreadable
280
+ lines.append("- merges: " + ", ".join(str(int(m)) for m in value))
281
+ else:
282
+ lines.append(f"- {key}: {value}")
283
+
284
+ return "\n".join(lines)
285
+
286
+
287
+ def migrate_text(jsonc_text: str) -> str:
288
+ """Convert old JSONC curation text to markdown sections text."""
289
+ data = json.loads(strip_jsonc_comments(jsonc_text))
290
+ if not isinstance(data, dict):
291
+ raise ValueError("curation: expected an object at the top level")
292
+
293
+ titles = _titles_from_comments(jsonc_text)
294
+ entries = data.get("books") or {}
295
+ if not isinstance(entries, dict):
296
+ raise ValueError("curation: `books` must be an object of edition id -> entry")
297
+
298
+ sections = [_section(entry_id, entry, titles)
299
+ for entry_id, entry in sorted(entries.items(), key=_section_sort_key)]
300
+ return DOC_BLOCK + "\n" + "\n\n".join(sections) + ("\n" if sections else "")
301
+
302
+
303
+ def migrate_file(src, out) -> Path:
304
+ """Read <src>.jsonc, write <out>.md, return the output path."""
305
+ text = Path(src).read_text(encoding="utf-8")
306
+ target = Path(out)
307
+ target.write_text(migrate_text(text), encoding="utf-8")
308
+ return target
309
+
310
+
311
+ def main(argv=None) -> int:
312
+ parser = argparse.ArgumentParser(
313
+ prog="python -m text_forge.migrate_curation",
314
+ description="Convert the old curation.jsonc into Curation.md "
315
+ "(markdown sections, weights ×10).")
316
+ parser.add_argument("input", help="path to the old curation.jsonc")
317
+ parser.add_argument("output", help="path to write the new Curation.md")
318
+ args = parser.parse_args(argv)
319
+
320
+ try:
321
+ target = migrate_file(args.input, args.output)
322
+ except (OSError, ValueError) as exc:
323
+ # nothing partial is left behind: a half-written Curation.md would drop
324
+ # curation decisions from the next build without saying so
325
+ print(f"curation migration failed: {exc}", file=sys.stderr)
326
+ return 1
327
+
328
+ entries = target.read_text(encoding="utf-8").count("\n## ")
329
+ print(f"curation: wrote {target} ({entries} sections, weights ×{WEIGHT_FACTOR})")
330
+ return 0
331
+
332
+
333
+ if __name__ == "__main__":
334
+ sys.exit(main())
@@ -100,46 +100,129 @@ def _edition_id(url: str) -> str:
100
100
  return m.group(1) if m else ""
101
101
 
102
102
 
103
+ #: the curation keys a section may carry, and how each value reads
104
+ CURATION_INT_KEYS = ("weight",)
105
+ CURATION_TEXT_KEYS = ("author", "status")
106
+ CURATION_ID_LIST_KEYS = ("merges",)
107
+ CURATION_KEYS = CURATION_INT_KEYS + CURATION_TEXT_KEYS + CURATION_ID_LIST_KEYS
108
+
109
+ # a `## ` heading: a hash pair on its own or followed by the section's text —
110
+ # but NOT a deeper `###`/`####` heading, and NOT a closing Setext underline
111
+ # (`#####` under a title, which is prose). Whatever precedes the em dash is the
112
+ # id; whatever follows it is for the eye only.
113
+ CURATION_HEADING_RE = re.compile(r"^##(?=[\s\u2014-]|$)(.*?)\s*$")
114
+ # a bullet: a list marker at (or within) the section's indentation
115
+ CURATION_BULLET_RE = re.compile(r"^\s*[-*+]\s+(\S.*)$")
116
+
117
+
118
+ class CurationError(ValueError):
119
+ """A curation file the build refuses to guess at.
120
+
121
+ Every message names the edition id it tripped on, plus the line it sits on:
122
+ a silently dropped entry would still build a plausible-looking corpus, and
123
+ that is the failure this format is not allowed to have.
124
+ """
125
+
126
+
127
+ def _curation_error(lineno: int, entry_id: str, why: str, line: str) -> CurationError:
128
+ where = f"line {lineno}"
129
+ # "- " is the shape every accepted bullet has, so a rejected one reads back
130
+ # in the same shape: what the eye wrote is what the message shows
131
+ who = f" (entry {entry_id})" if entry_id else " (no `## <id>` section above)"
132
+ return CurationError(
133
+ f"curation: {where}{who}: {why}\n {line}\n"
134
+ "A curation section is `## <edition id> — <title for the eye>` followed "
135
+ "by `- key: value` bullets, keys: " + ", ".join(CURATION_KEYS) + "."
136
+ )
137
+
138
+
139
+ def _curation_int(lineno: int, entry_id: str, key: str, raw: str, line: str) -> int:
140
+ if not raw.isdigit():
141
+ raise _curation_error(lineno, entry_id, f"{key} must be a whole number, got {raw!r}", line)
142
+ return int(raw)
143
+
144
+
145
+ def _curation_ids(lineno: int, entry_id: str, key: str, raw: str, line: str) -> list[int]:
146
+ ids = []
147
+ for part in raw.split(","):
148
+ part = part.strip()
149
+ if not part.isdigit():
150
+ raise _curation_error(
151
+ lineno, entry_id, f"{key} must be comma-separated edition ids, got {part!r}", line)
152
+ ids.append(int(part))
153
+ if not ids:
154
+ raise _curation_error(lineno, entry_id, f"{key} names no id", line)
155
+ return ids
156
+
157
+
103
158
  def load_curation(path) -> dict:
104
- """Parse a .jsonc curation file: JSON with // line and /* */ block comments.
159
+ """Parse a Curation.md file: one `## <id>` section per livelib edition.
160
+
161
+ The heading's digits ARE the entry key; everything after the em dash (and
162
+ any prose line) is for the eye only, the same property the old jsonc `//`
163
+ comments had. A section carries `- key: value` bullets, any subset of
164
+ `weight`/`author`/`status`/`merges`, and reads back the mapping build_books
165
+ consumed before: {"books": {id: {weight: int, merges: [int], ...}}}.
105
166
 
106
- Comment markers inside string literals are data, not comments: a title like
107
- "Письма о Сезанне https://..." must survive intact. Scanning character by
108
- character (rather than regexing the text) is what makes that true, and it
109
- also keeps `\"` from pretending to close a string.
167
+ Malformed input raises CurationError naming the id and the line; a line that
168
+ is neither a heading, a bullet, nor blank is the old JSONC shape and raises
169
+ too — the previous format is gone, and it must not be read by accident.
110
170
  """
111
171
  text = pathlib.Path(path).read_text(encoding="utf-8")
112
- out = []
113
- i, n = 0, len(text)
114
- in_string = False
115
- while i < n:
116
- ch = text[i]
117
- if in_string:
118
- if ch == "\\":
119
- out.append(text[i:i + 2])
120
- i += 2
121
- continue
122
- if ch == '"':
123
- in_string = False
124
- out.append(ch)
125
- i += 1
126
- continue
127
- if ch == '"':
128
- in_string = True
129
- out.append(ch)
130
- i += 1
131
- continue
132
- if ch == "/" and i + 1 < n and text[i + 1] == "/":
133
- while i < n and text[i] != "\n":
134
- i += 1
172
+
173
+ # tombstone for the previous format: braces mean JSON/JSONC, and that file
174
+ # needs migrating (text_forge.migrate_curation), not guessing. Without this
175
+ # check such a file would read as ZERO entries and drop every curation
176
+ # decision while still building a plausible corpus.
177
+ if re.search(r"^\s*[{}\[\]]\s*$", text, re.MULTILINE):
178
+ raise CurationError(
179
+ f"curation: {path} is the old JSONC format, not markdown sections — "
180
+ "run `python -m text_forge.migrate_curation <in.jsonc> <out.md>`"
181
+ )
182
+
183
+ entries: dict = {}
184
+ entry_id = ""
185
+ for lineno, line in enumerate(text.splitlines(), 1):
186
+ heading = CURATION_HEADING_RE.match(line)
187
+ if heading:
188
+ # the heading's id keys the entry; the text after the em dash is
189
+ # eye-only, so a wrong title can never steer the build
190
+ entry_id = heading.group(1).lstrip("#").split("\u2014")[0].strip()
191
+ if not entry_id.isdigit():
192
+ raise _curation_error(lineno, "", f"section id must be digits, got {entry_id!r}", line)
193
+ if entry_id in entries:
194
+ raise _curation_error(lineno, entry_id, "duplicate section for this edition id", line)
195
+ entries[entry_id] = {}
135
196
  continue
136
- if ch == "/" and i + 1 < n and text[i + 1] == "*":
137
- end = text.find("*/", i + 2)
138
- i = n if end == -1 else end + 2
197
+
198
+ bullet = CURATION_BULLET_RE.match(line)
199
+ if not bullet:
200
+ # nothing else in the file carries data: blank lines, the doc block's
201
+ # prose, and any other heading are for the eye
139
202
  continue
140
- out.append(ch)
141
- i += 1
142
- return json.loads("".join(out))
203
+
204
+ key, sep, raw = bullet.group(1).partition(":")
205
+ if not sep:
206
+ raise _curation_error(lineno, entry_id, "bullet needs `key: value`", line)
207
+ key = key.strip()
208
+ raw = raw.strip()
209
+ if key not in CURATION_KEYS:
210
+ raise _curation_error(lineno, entry_id, f"unknown key {key!r}", line)
211
+ if not entry_id:
212
+ raise _curation_error(lineno, "", "bullet outside a `## <id>` section", line)
213
+ if key in entries[entry_id]:
214
+ raise _curation_error(lineno, entry_id, f"key {key!r} given twice", line)
215
+ if key in CURATION_INT_KEYS:
216
+ value = _curation_int(lineno, entry_id, key, raw, line)
217
+ elif key in CURATION_ID_LIST_KEYS:
218
+ value = _curation_ids(lineno, entry_id, key, raw, line)
219
+ else:
220
+ if not raw:
221
+ raise _curation_error(lineno, entry_id, f"{key} is empty", line)
222
+ value = raw
223
+ entries[entry_id][key] = value
224
+
225
+ return {"books": entries}
143
226
 
144
227
 
145
228
 
@@ -845,9 +928,10 @@ def build_books(books_path, quotes_path, reviews_path, out_dir, *, curation_path
845
928
  work_id_by_edition=None, fetch=None) -> dict:
846
929
  """Build books/<slug>.md notes from LiveLib export JSON files.
847
930
 
848
- curation_path — ONE curation.jsonc in normal form: one entry per livelib
849
- edition id under `books`, each carrying weight?/status?/my_mark?/author?/
850
- merges?. Anything livelib can tell us is not stored.
931
+ curation_path — the ONE curation file (Curation.md by default, or a mapping
932
+ already built): sections keyed by livelib edition id, read back as {"books":
933
+ {id: ...}}, each entry carrying weight?/status?/author?/merges?. Anything
934
+ livelib can tell us is not stored.
851
935
 
852
936
  Reference books need no section of their own: pass `text_dir` and the build
853
937
  scans the texts for livelib edition ids no note would declare, resolving
@@ -1159,8 +1243,8 @@ def _note_index(books_dir) -> dict:
1159
1243
 
1160
1244
  def _curation_section(path_or_map, section):
1161
1245
  """Read one section from a curation source given as a path to a consolidated
1162
- .jsonc or as the mapping itself (the form build_books emits, e.g. its
1163
- `link_map`)."""
1246
+ file (Curation.md, or a json side-car the build writes) or as the mapping
1247
+ itself (the form build_books emits, e.g. its `link_map`)."""
1164
1248
  if path_or_map is None:
1165
1249
  return None
1166
1250
  if isinstance(path_or_map, (dict, list)):
@@ -7,8 +7,8 @@ Single entry point for the `make books` target (whattodo/Makefile):
7
7
  Steps:
8
8
  1. Export fresh data (shelves incl. statuses, quotes, reviews) into
9
9
  <root>/livelib-export/ — anonymous API + SSR, no credentials needed.
10
- 2. build_books() with the ONE curation file
11
- <root>/text/books/_curation/curation.jsonc, and the texts themselves passed
10
+ 2. build_books() with the ONE curation file <root>/Curation.md — markdown
11
+ sections, one per livelib edition id — and the texts themselves passed
12
12
  as `text_dir` so a work a chapter cites but no shelf holds still gets its
13
13
  note (title/author from the edition card JSON).
14
14
  3. Replace <root>/text/books/*.md with the generated set (_curation/ untouched).
@@ -121,8 +121,10 @@ def main(root_str: str, user_id: int = 345497) -> int:
121
121
  print("export:", counts)
122
122
 
123
123
  tmp = Path(tempfile.mkdtemp(prefix="wtd-books-"))
124
- # one curation file, normal form: one entry per livelib edition id
125
- curation = cur / "curation.jsonc"
124
+ # ONE curation file, at the corpus root: markdown sections, one per livelib
125
+ # edition id. (It used to sit beside the notes as
126
+ # text/books/_curation/curation.jsonc; the link map below still does.)
127
+ curation = root / "Curation.md"
126
128
  if not curation.exists():
127
129
  print(f"FAIL: curation file not found: {curation}")
128
130
  return 1
@@ -1320,7 +1320,7 @@ wheels = [
1320
1320
 
1321
1321
  [[package]]
1322
1322
  name = "sg-text-forge"
1323
- version = "0.3.2"
1323
+ version = "0.3.3"
1324
1324
  source = { editable = "." }
1325
1325
  dependencies = [
1326
1326
  { name = "click" },
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes