dsh-docx-format 1.1.1 → 1.2.0

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.
package/lib/docxfmt.py CHANGED
@@ -73,7 +73,11 @@ DEFAULTS = {
73
73
  "spaceAfterPt": 0,
74
74
  "marginsCm": {"top": 3.6, "bottom": 3.0, "left": 2.7, "right": 2.7},
75
75
  "titleBlankBeforeLines": 2,
76
+ # 标题后的空行从“副标题”算起:主标题与副标题之间不留空行
76
77
  "titleBlankAfterLines": 1,
78
+ # 副标题:紧跟主标题(中间最多隔一个空段)的另一段短文本,仿宋三号居中
79
+ "subtitle": {"enabled": True, "mainMaxChars": 18, "maxChars": 40,
80
+ "gapTolerance": 1},
77
81
  "attachBlankBeforeLines": 1,
78
82
  "attachBlankAfterLines": 2,
79
83
  # 文件中的“附件1”标记:下与附件标题空出一行(参考要求未规定其上要空行)
@@ -86,8 +90,39 @@ DEFAULTS = {
86
90
  # 控制(未明确要求署名/日期时不生成落款文字)。
87
91
  "signature": {"enabled": True, "seal": False, "datePadSpaces": 2,
88
92
  "blankLinesBefore": 2},
89
- # 正文命中这些关键词时整篇跳过(白名单),如“实验报告”
90
- "skipContentKeywords": ["实验报告"],
93
+ # 字体档:standard 用常见简体字体,strict 用 GBK/GB2312 全字库(更接近
94
+ # GB/T 9704-2012 的“标准字体”,缺字更少)。
95
+ # strictFonts=true → 一律 strict
96
+ # 本条请求含 strictKeywords → strict
97
+ # 都不满足 → standard
98
+ "fontProfile": "standard",
99
+ "strictFonts": False,
100
+ "strictKeywords": ["严格公文格式", "严格公文"],
101
+ # strict 档在 standard 之上覆盖这些位置
102
+ "fontsStrict": {
103
+ "title": "方正小标宋_GBK",
104
+ "tableCaption": "方正小标宋_GBK",
105
+ "body": "仿宋_GB2312",
106
+ "h3": "仿宋_GB2312",
107
+ "tableBody": "仿宋_GB2312",
108
+ "h1": "方正黑体_GBK",
109
+ "tableHead": "方正黑体_GBK",
110
+ "h2": "楷体_GB2312",
111
+ "footer": "宋体",
112
+ },
113
+ # 字体检测与安装。检测只读注册表/字体目录,**从不联网**;
114
+ # 下载必须拿到 consent token(或 CLI --yes)。
115
+ # autoInstall: ask 缺字体时返回 consentRequired,由对话里问用户
116
+ # off 只报告,从不下载
117
+ "fontCheck": {
118
+ "enabled": True,
119
+ "onRun": True,
120
+ "aliases": {},
121
+ "presets": {},
122
+ "sources": [],
123
+ "autoInstall": "ask",
124
+ "installDir": "",
125
+ },
91
126
  "normalize": {
92
127
  "stripLeadingSpace": True,
93
128
  "fullwidthNumberDot": True,
@@ -96,6 +131,11 @@ DEFAULTS = {
96
131
  "boldOff": True,
97
132
  # 中文内容里英文前后不留空格(「使用AI工具」);纯英文语句的空格保持原样
98
133
  "cjkLatinSpace": True,
134
+ # 中文里的引号用“”‘’,不用「」『』
135
+ "quotePair": True,
136
+ # 「换行和分页」里的 6 个开关(孤行控制/与下段同页/段中不分页/段前分页/
137
+ # 取消行号/取消断字)一律不启用
138
+ "paginationOff": True,
99
139
  },
100
140
  }
101
141
 
@@ -106,6 +146,8 @@ RE_ATTACH_TAG = re.compile(r"^[ \u3000]*附件\s*\d{0,2}\s*$")
106
146
  RE_ATTACH_LINE = re.compile(r"^[ \u3000]*附件\s*[::]")
107
147
  RE_ATTACH_ITEM = re.compile(r"^[((]?\d{1,2}\s*[..、))]")
108
148
  RE_DATE = re.compile(r"\d{4}\s*年\s*\d{1,2}\s*月\s*\d{1,2}\s*日")
149
+ # 副标题常以破折号开头:「——在XX会议上的讲话」
150
+ RE_DASH_LEAD = re.compile(r"^[—–]{1,2}|^-{1,2}\s")
109
151
  RE_LEAD_SPACE = re.compile(r"^[\u3000 ]+")
110
152
  RE_FULLWIDTH_SPACE = "\u3000"
111
153
 
@@ -504,6 +546,10 @@ def style_par(par, kind, report, label, cfg):
504
546
  if kind in ("title", "title_first", "title_mid", "title_last", "title_only"):
505
547
  want_align, want_font, want_size = CENTER, fonts["title"], sizes["title"]
506
548
  want_first_chars = None
549
+ elif kind == "subtitle":
550
+ # 副标题:仿宋三号、居中、无首行缩进,行距仍固定 29 磅
551
+ want_align, want_font, want_size = CENTER, fonts["body"], sizes["body"]
552
+ want_first_chars = None
507
553
  elif kind == "table_caption":
508
554
  want_align, want_font, want_size = CENTER, fonts["tableCaption"], sizes["title"]
509
555
  want_first_chars = None
@@ -807,6 +853,80 @@ def drop_char_ranges(runs, gaps):
807
853
  run.text = kept
808
854
 
809
855
 
856
+ # 「换行和分页」里那 6 个开关对应的 OOXML 属性,顺序即 pPr 的 schema 顺序
857
+ PAGINATION_ORDER = ("w:keepNext", "w:keepLines", "w:pageBreakBefore", "w:framePr",
858
+ "w:widowControl", "w:numPr", "w:suppressLineNumbers", "w:pBdr",
859
+ "w:shd", "w:tabs", "w:suppressAutoHyphens", "w:kinsoku",
860
+ "w:wordWrap", "w:overflowPunct", "w:topLinePunct",
861
+ "w:autoSpaceDE", "w:autoSpaceDN", "w:bidi", "w:adjustRightInd",
862
+ "w:snapToGrid", "w:spacing", "w:ind", "w:contextualSpacing",
863
+ "w:mirrorIndents", "w:suppressOverlap", "w:jc",
864
+ "w:textDirection", "w:textAlignment", "w:textboxTightWrap",
865
+ "w:outlineLvl", "w:divId", "w:cnfStyle", "w:rPr", "w:sectPr",
866
+ "w:pPrChange")
867
+ PAGINATION_TAGS = ("w:widowControl", "w:keepNext", "w:keepLines",
868
+ "w:pageBreakBefore", "w:suppressLineNumbers",
869
+ "w:suppressAutoHyphens")
870
+ PAGINATION_LABELS = {"w:widowControl": "孤行控制", "w:keepNext": "与下段同页",
871
+ "w:keepLines": "段中不分页", "w:pageBreakBefore": "段前分页",
872
+ "w:suppressLineNumbers": "取消行号",
873
+ "w:suppressAutoHyphens": "取消断字"}
874
+
875
+ # 中文引号:不用「」『』,用“”‘’
876
+ QUOTE_MAP = {"「": "“", "」": "”", "『": "‘", "』": "’"}
877
+
878
+
879
+ def _pagination_successors(tag):
880
+ return PAGINATION_ORDER[PAGINATION_ORDER.index(tag) + 1:]
881
+
882
+
883
+ def clear_pagination(par, report, label):
884
+ """段落「换行和分页」里的 6 个开关一律不启用(对话框里复选框全部留空)。
885
+
886
+ 这 6 项都不在 python-docx 的 API 里,直接写 pPr:
887
+ 孤行控制 / 与下段同页 / 段中不分页 / 段前分页 / 取消行号 / 取消断字。
888
+ widowControl 的 Word 默认值是“开”,所以必须显式写 0,不能靠删除元素。
889
+ """
890
+ pPr = par._p.get_or_add_pPr()
891
+ for tag in PAGINATION_TAGS:
892
+ el = pPr.find(qn(tag))
893
+ val = None if el is None else (el.get(qn("w:val")) or "1")
894
+ if el is not None and str(val).lower() in ("0", "false", "off"):
895
+ continue
896
+
897
+ def apply(t=tag, e=el, p=pPr):
898
+ node = e
899
+ if node is None:
900
+ node = OxmlElement(t)
901
+ try:
902
+ p.insert_element_before(node, *_pagination_successors(t))
903
+ except Exception:
904
+ p.append(node)
905
+ node.set(qn("w:val"), "0")
906
+
907
+ report.want("par.pagination." + tag.split(":")[1], label,
908
+ "%s:启用" % PAGINATION_LABELS[tag],
909
+ "%s:不启用" % PAGINATION_LABELS[tag], apply)
910
+
911
+
912
+ def quote_normalize(par, report, label):
913
+ """中文里的引号写成“”‘’,不写「」『』。"""
914
+ for idx, run in enumerate(par.runs):
915
+ t = run.text or ""
916
+ if not any(ch in QUOTE_MAP for ch in t):
917
+ continue
918
+ new = "".join(QUOTE_MAP.get(ch, ch) for ch in t)
919
+ report.want("text.quote", "%s#r%d" % (label, idx),
920
+ "使用「」『』", "使用“”‘’",
921
+ (lambda r=run, v=new: setattr(r, "text", v)))
922
+
923
+
924
+ def _iter_all_paragraphs(doc):
925
+ """文档里所有段落(正文 + 表格单元格 + 文本框),每个 XML 节点只回一次。"""
926
+ for node in doc.element.body.iter(qn("w:p")):
927
+ yield Paragraph(node, doc)
928
+
929
+
810
930
  def auto_space_off(par, report, label):
811
931
  """关闭 Word 的“自动调整中西文间距”(w:autoSpaceDE / w:autoSpaceDN)。
812
932
 
@@ -831,12 +951,32 @@ def auto_space_off(par, report, label):
831
951
  "开启(Word 自动在中英文之间加间距)", "关闭(autoSpaceDE/DN=0)", apply)
832
952
 
833
953
 
834
- def normalize_text(par, report, label, cfg):
954
+ def _normalize_cjk_latin(par, report, label, cfg):
955
+ """中文文段里英文词前后不留空格(「使用AI工具」);纯英文语句不动。"""
956
+ if not cfg["normalize"].get("cjkLatinSpace", True):
957
+ return
958
+ runs = par.runs
959
+ if not runs:
960
+ return
961
+ joined = "".join(r.text for r in runs)
962
+ gaps = cjk_latin_gaps(joined)
963
+ if gaps:
964
+ report.want("text.cjkLatinSpace", label,
965
+ "中文与英文之间夹着空白(共 %d 处)" % len(gaps),
966
+ "中文内英文前后不加空格(如「使用AI工具」)",
967
+ (lambda rs=runs, gs=gaps: drop_char_ranges(rs, gs)))
968
+
969
+
970
+ def normalize_text(par, report, label, cfg, light=False):
835
971
  p = par
836
972
  text = p.text
837
973
  if not text:
838
974
  return
839
975
  n = cfg["normalize"]
976
+ if light:
977
+ # 轻量模式:只做与版式无关的两件事(中英之间不留空格、引号用“”)
978
+ _normalize_cjk_latin(p, report, label, cfg)
979
+ return
840
980
  if n["stripLeadingSpace"] and RE_LEAD_SPACE.match(text):
841
981
  stripped = text.lstrip("\u3000 ")
842
982
  if stripped and p.runs:
@@ -865,14 +1005,7 @@ def normalize_text(par, report, label, cfg):
865
1005
  (lambda r=run, v=new: setattr(r, "text", v)))
866
1006
  break
867
1007
  if n.get("cjkLatinSpace", True):
868
- runs = p.runs
869
- joined = "".join(r.text for r in runs)
870
- gaps = cjk_latin_gaps(joined)
871
- if gaps:
872
- report.want("text.cjkLatinSpace", label,
873
- "中文与英文之间夹着空白(共 %d 处)" % len(gaps),
874
- "中文内英文前后不加空格(如「使用AI工具」)",
875
- (lambda rs=runs, gs=gaps: drop_char_ranges(rs, gs)))
1008
+ _normalize_cjk_latin(p, report, label, cfg)
876
1009
 
877
1010
 
878
1011
  # ---------------------------------------------------------------- 表格
@@ -1097,27 +1230,47 @@ def _looks_like_attach_wrap(par, cfg):
1097
1230
  return cur is not None and abs(cur - want) < 0.05
1098
1231
 
1099
1232
 
1233
+ def _gap_all_blank(items, a, b):
1234
+ """items[a] 与 items[b] 之间(不含两端)是否全是空段落。"""
1235
+ for k in range(a + 1, b):
1236
+ t, obj = items[k]
1237
+ if t != "p" or (obj.text or "").strip():
1238
+ return False
1239
+ return True
1240
+
1241
+
1100
1242
  def classify(items, cfg):
1101
1243
  """返回 [(kind, obj, text)],kinds: title/table_caption/h1/h2/h3/attach_tag/attach_line/
1102
1244
  sign_org/sign_date/image/body/blank/tbl"""
1103
1245
  out = []
1104
1246
  paras = [(i, obj) for i, (t, obj) in enumerate(items) if t == "p"]
1105
1247
  nonempty = [(i, p) for i, p in paras if p.text.strip()]
1106
- title_idx = set()
1248
+ title_map = {}
1107
1249
  if nonempty:
1108
1250
  i0 = nonempty[0][0]
1109
- title_idx.add(i0)
1110
- # 标题续行:紧跟其后、短、非标题模式、非编号
1111
- prev_len = len(nonempty[0][1].text.strip())
1112
- for i, p in nonempty[1:3]:
1113
- txt = p.text.strip()
1114
- if (i == i0 + 1 and prev_len <= 18 and len(txt) <= 40
1115
- and not txt.endswith(("。", ":", ",", ";", "!", "?", "、", ":"))
1116
- and not RE_H1.match(txt) and not RE_H2.match(txt)
1117
- and not RE_ATTACH_LINE.match(txt) and not RE_ATTACH_TAG.match(txt)):
1118
- title_idx.add(i)
1119
- else:
1120
- break
1251
+ title_map[i0] = "title"
1252
+ # 副标题:紧跟主标题(中间最多隔 gapTolerance 个空段)的另一段短文本
1253
+ sub = cfg.get("subtitle") or {}
1254
+ if sub.get("enabled", True):
1255
+ tol = max(0, int(sub.get("gapTolerance", 1)))
1256
+ main_max = int(sub.get("mainMaxChars", 18))
1257
+ sub_max = int(sub.get("maxChars", 40))
1258
+ prev_len = len(nonempty[0][1].text.strip())
1259
+ for i, p in nonempty[1:3]:
1260
+ txt = p.text.strip()
1261
+ gap_ok = (i - i0 - 1) <= tol and _gap_all_blank(items, i0, i)
1262
+ # 破折号开头基本就是副标题;否则要求主标题够短,避免把正文首段吃进来
1263
+ dash_lead = bool(RE_DASH_LEAD.match(txt))
1264
+ if (gap_ok and len(txt) <= sub_max
1265
+ and (dash_lead or prev_len <= main_max)
1266
+ and not txt.endswith(("。", ":", ",", ";", "!", "?", "、", ":"))
1267
+ and not RE_H1.match(txt) and not RE_H2.match(txt)
1268
+ and not RE_ATTACH_LINE.match(txt) and not RE_ATTACH_TAG.match(txt)
1269
+ and not RE_DATE.search(txt)):
1270
+ title_map[i] = "subtitle"
1271
+ prev_len = len(txt)
1272
+ else:
1273
+ break
1121
1274
 
1122
1275
  # 落款:末尾日期行
1123
1276
  sign = {}
@@ -1138,8 +1291,8 @@ def classify(items, cfg):
1138
1291
  continue
1139
1292
  p = obj
1140
1293
  txt = p.text.strip()
1141
- if i in title_idx:
1142
- out.append(("title", p, txt))
1294
+ if i in title_map:
1295
+ out.append((title_map[i], p, txt))
1143
1296
  elif i in sign:
1144
1297
  out.append((sign[i], p, txt))
1145
1298
  elif not txt:
@@ -1205,7 +1358,7 @@ def classify(items, cfg):
1205
1358
  out[k] = ("caption_spacer", out[k][1], "")
1206
1359
  # 空行也带格式:标题上空、多行标题之间、标题下固定那 1 行 → 沿用标题格式;
1207
1360
  # 其余空行沿用正文格式——和“手工回车”继承出来的结果一致。
1208
- title_positions = [i for i, (t, _o, _x) in enumerate(out) if t == "title"]
1361
+ title_positions = [i for i, (t, _o, _x) in enumerate(out) if t in ("title", "subtitle")]
1209
1362
  if title_positions:
1210
1363
  first_t, last_t = title_positions[0], title_positions[-1]
1211
1364
  title_side = set(range(0, first_t)) | set(range(first_t + 1, last_t))
@@ -1352,7 +1505,7 @@ def ensure_blank_lines(kinds, report, cfg):
1352
1505
  if old is None or old[1] < want:
1353
1506
  plan[key] = (par, want, label, blank_kind)
1354
1507
 
1355
- titles = [obj for k, obj, _t in kinds if k == "title"]
1508
+ titles = [obj for k, obj, _t in kinds if k in ("title", "subtitle")]
1356
1509
  if titles:
1357
1510
  bump(titles[0], cfg["titleBlankBeforeLines"], "标题前", "blank_title")
1358
1511
  bump(_next_anchor_par(titles[-1]), cfg["titleBlankAfterLines"], "标题后", "blank_title")
@@ -1629,12 +1782,127 @@ def split_attach_paragraphs(doc, report, cfg):
1629
1782
  return changed
1630
1783
 
1631
1784
 
1632
- def process(path, mode, cfg):
1785
+ def _save(doc, path):
1786
+ tmp = path + ".tmp.docx"
1787
+ doc.save(tmp)
1788
+ os.replace(tmp, path)
1789
+
1790
+
1791
+ def light_pass(doc, report, cfg):
1792
+ """与版式无关的全局规范化:不论是否套用公文版式都执行。
1793
+
1794
+ - 「换行和分页」里 6 个开关一律不启用
1795
+ - 中文里不写「」『』,改“”‘’
1796
+ - 中文文段里的英文前后不留空格(并关闭 Word 自动中西文间距)
1797
+ """
1798
+ n = cfg["normalize"]
1799
+ pagination = n.get("paginationOff", True)
1800
+ quote = n.get("quotePair", True)
1801
+ for i, par in enumerate(_iter_all_paragraphs(doc)):
1802
+ label = "l%d" % i
1803
+ if pagination:
1804
+ clear_pagination(par, report, label)
1805
+ if quote:
1806
+ quote_normalize(par, report, label)
1807
+ _normalize_cjk_latin(par, report, label, cfg)
1808
+ auto_space_off(par, report, label)
1809
+ return report
1810
+
1811
+
1812
+ def drop_blanks_between_titles(items, kinds):
1813
+ """主标题与副标题之间不留空行(空行从副标题之后开始算)。"""
1814
+ idxs = [i for i, (k, _o, _t) in enumerate(kinds) if k in ("title", "subtitle")]
1815
+ if len(idxs) < 2:
1816
+ return 0
1817
+ removed = 0
1818
+ for a, b in zip(idxs, idxs[1:]):
1819
+ for k in range(b - 1, a, -1):
1820
+ if kinds[k][0] in BLANK_KINDS:
1821
+ el = kinds[k][1]._p
1822
+ if el.getparent() is not None:
1823
+ el.getparent().remove(el)
1824
+ removed += 1
1825
+ return removed
1826
+
1827
+
1828
+ # ---------------------------------------------------------------- 字体档
1829
+ def apply_font_profile(cfg, profile=None):
1830
+ """按档位覆盖 cfg["fonts"],返回实际生效的档位名。
1831
+
1832
+ standard 用常见简体字体;strict 用 GBK/GB2312 全字库。
1833
+ """
1834
+ p = str(profile or cfg.get("fontProfile") or "standard").strip().lower()
1835
+ if p not in ("standard", "strict"):
1836
+ p = "standard"
1837
+ if p == "strict":
1838
+ fonts = cfg.setdefault("fonts", {})
1839
+ for k, v in (cfg.get("fontsStrict") or {}).items():
1840
+ if v:
1841
+ fonts[k] = v
1842
+ return p
1843
+
1844
+
1845
+ def check_fonts(report, cfg):
1846
+ """检测缺哪些字体。
1847
+
1848
+ **不联网**:只读注册表与字体目录。check 模式把缺失记成偏差;
1849
+ apply 模式只写 warnings(不影响排版,也不阻断写回)。
1850
+ """
1851
+ fc = cfg.get("fontCheck") or {}
1852
+ if not fc.get("enabled", True):
1853
+ return None
1854
+ try:
1855
+ here = os.path.dirname(os.path.abspath(__file__))
1856
+ if here not in sys.path:
1857
+ sys.path.insert(0, here)
1858
+ import fontcheck
1859
+ except Exception: # noqa: BLE001
1860
+ return None
1861
+ try:
1862
+ info = fontcheck.check(cfg)
1863
+ except Exception: # noqa: BLE001
1864
+ return None
1865
+ if not info.get("missing"):
1866
+ return info
1867
+ cr = info.get("consentRequired") or {}
1868
+ ask = cr.get("ask") or ("缺少字体:%s" % "、".join(info["missing"]))
1869
+ if report.mode == "check":
1870
+ for name in info["missing"]:
1871
+ report.want("font.missing:" + name, "字体", "未安装", "已安装")
1872
+ else:
1873
+ report.warnings.append(ask)
1874
+ if cr.get("command"):
1875
+ report.warnings.append("修复:" + cr["command"])
1876
+ report.stats["missingFonts"] = len(info["missing"])
1877
+ return info
1878
+
1879
+
1880
+ def process(path, mode, cfg, full=True, profile=None):
1633
1881
  t0 = time.time()
1634
1882
  fix = mode == "apply"
1635
1883
  doc = Document(path)
1636
1884
  report = Report(os.path.abspath(path), mode, fix)
1637
1885
 
1886
+ profile = apply_font_profile(cfg, profile)
1887
+ fontinfo = None
1888
+ if (cfg.get("fontCheck") or {}).get("onRun", True):
1889
+ fontinfo = check_fonts(report, cfg)
1890
+
1891
+ light_pass(doc, report, cfg)
1892
+
1893
+ if not full:
1894
+ changed = report.total > 0
1895
+ if fix and changed:
1896
+ _save(doc, path)
1897
+ out = report.as_dict(changed)
1898
+ out["light"] = True
1899
+ out["writeBack"] = bool(fix and changed)
1900
+ out["elapsedMs"] = int((time.time() - t0) * 1000)
1901
+ out["fontProfile"] = profile
1902
+ if fontinfo:
1903
+ out["fonts"] = fontinfo
1904
+ return out
1905
+
1638
1906
  for section in doc.sections:
1639
1907
  m = cfg["marginsCm"]
1640
1908
  for attr, key in (("top_margin", "top"), ("bottom_margin", "bottom"),
@@ -1658,6 +1926,12 @@ def process(path, mode, cfg):
1658
1926
  break
1659
1927
  items = _collect_body(doc)
1660
1928
  kinds = classify(items, cfg)
1929
+ removed = drop_blanks_between_titles(items, kinds)
1930
+ if removed:
1931
+ report.want("title.subtitleGap", "doc.标题",
1932
+ "%d 个空行" % removed, "0 个空行(主标题与副标题紧邻)")
1933
+ items = _collect_body(doc)
1934
+ kinds = classify(items, cfg)
1661
1935
  ensure_blank_lines(kinds, report, cfg)
1662
1936
  format_table_caption_spacer(items, [k for k, _o, _t in kinds], report, cfg)
1663
1937
  items = _collect_body(doc)
@@ -1697,14 +1971,18 @@ def process(path, mode, cfg):
1697
1971
  # check 模式:只统计表标题与表格之间是否缺空行,内存里插入不写回
1698
1972
  format_table_caption_spacer(items, [k for k, _o, _t in kinds], report, cfg)
1699
1973
 
1974
+ # 前面插过空段、拆过附件段,新段落还没被全局规范化覆盖:收尾再扫一遍
1975
+ light_pass(doc, report, cfg)
1976
+
1700
1977
  changed = report.total > 0
1701
1978
  if fix and changed:
1702
- tmp = path + ".tmp.docx"
1703
- doc.save(tmp)
1704
- os.replace(tmp, path)
1979
+ _save(doc, path)
1705
1980
  out = report.as_dict(changed)
1706
1981
  out["writeBack"] = bool(fix and changed)
1707
1982
  out["elapsedMs"] = int((time.time() - t0) * 1000)
1983
+ out["fontProfile"] = profile
1984
+ if fontinfo:
1985
+ out["fonts"] = fontinfo
1708
1986
  return out
1709
1987
 
1710
1988
 
@@ -1753,7 +2031,7 @@ def doc_text(path):
1753
2031
 
1754
2032
 
1755
2033
  def skip_reason(path, cfg):
1756
- """白名单:正文命中 skipContentKeywords(默认「实验报告」)时整篇跳过。"""
2034
+ """保留的扩展点:默认不按正文内容跳过任何文档。"""
1757
2035
  kws = cfg.get("skipContentKeywords") or []
1758
2036
  if not kws:
1759
2037
  return None
@@ -1786,13 +2064,145 @@ def _collect(paths, recursive):
1786
2064
  return out
1787
2065
 
1788
2066
 
2067
+ def _fonts_cli(args, cfg):
2068
+ """--fonts / --install-fonts 的处理入口。"""
2069
+ import tempfile
2070
+ import shutil
2071
+
2072
+ here = os.path.dirname(os.path.abspath(__file__))
2073
+ if here not in sys.path:
2074
+ sys.path.insert(0, here)
2075
+ import fontcheck
2076
+
2077
+ want_install = (args.install_fonts is not None or args.preset
2078
+ or args.from_zip)
2079
+
2080
+ if args.confirm:
2081
+ # 用户授权令牌:满足它等同于 --yes,且令牌只能用一次
2082
+ rec = fontcheck.consume_consent(args.confirm)
2083
+ if not rec:
2084
+ msg = "授权令牌无效或已过期(请重新运行 --fonts 取新的 token)"
2085
+ if args.json:
2086
+ print(json.dumps({"ok": False, "error": msg,
2087
+ "reason": "bad-consent"}, ensure_ascii=False))
2088
+ else:
2089
+ print(msg)
2090
+ return 3
2091
+ args.yes = True
2092
+ if not want_install:
2093
+ # 用户只回传了 token:按当初登记的来源装
2094
+ src = str(rec.get("source") or "")
2095
+ if src.lower().endswith(".zip") or src.startswith(("http://", "https://")):
2096
+ args.from_zip = src
2097
+ elif src in fontcheck.PRESETS:
2098
+ args.preset = src
2099
+
2100
+ if not want_install and not args.yes:
2101
+ info = fontcheck.check(cfg)
2102
+ if args.json:
2103
+ print(json.dumps(info, ensure_ascii=False, indent=2))
2104
+ else:
2105
+ print("平台:%s 安装目录:%s"
2106
+ % (info["platform"], info["installDir"]))
2107
+ for name in info["required"]:
2108
+ hit = info["found"].get(name)
2109
+ print(" %-4s %s" % ("OK" if hit else "缺失", name))
2110
+ if info["missing"]:
2111
+ cr = info.get("consentRequired") or {}
2112
+ print("")
2113
+ print(info["hint"])
2114
+ if cr:
2115
+ print(cr["ask"])
2116
+ print("命令:%s" % cr["command"])
2117
+ return 0 if info["ok"] else 2
2118
+
2119
+ if not args.yes:
2120
+ args.dry_run = True
2121
+ print("需要授权:加 --yes 才会下载安装(当前只做 dry-run)")
2122
+
2123
+ tmpdir = tempfile.mkdtemp(prefix="docxfmt-fonts-")
2124
+ results = []
2125
+ try:
2126
+ if args.from_zip:
2127
+ results += fontcheck.extract_zip(args.from_zip, tmpdir)
2128
+ if args.preset:
2129
+ results += fontcheck.download_preset(args.preset, tmpdir)
2130
+ for src in (args.install_fonts or []):
2131
+ if os.path.isdir(src):
2132
+ for fn in sorted(os.listdir(src)):
2133
+ if fn.lower().endswith(fontcheck.FONT_EXTS):
2134
+ results.append({"file": os.path.join(src, fn),
2135
+ "ok": True})
2136
+ elif os.path.isfile(src):
2137
+ results.append({"file": src, "ok": True})
2138
+ else:
2139
+ results.append({"file": src, "ok": False,
2140
+ "reason": "路径不存在"})
2141
+ good = [r["file"] for r in results if r.get("ok")]
2142
+ if not good:
2143
+ print("没有可安装的字体文件")
2144
+ if args.json:
2145
+ print(json.dumps({"results": results}, ensure_ascii=False,
2146
+ indent=2))
2147
+ return 1
2148
+ inst = fontcheck.install(good, dry_run=args.dry_run)
2149
+ if args.json:
2150
+ print(json.dumps({"downloaded": results, "installed": inst,
2151
+ "dryRun": args.dry_run},
2152
+ ensure_ascii=False, indent=2))
2153
+ else:
2154
+ for it in results:
2155
+ if not it.get("ok"):
2156
+ print(" [跳过] %s %s" % (it.get("file"), it.get("reason")))
2157
+ for it in inst:
2158
+ if not it.get("ok"):
2159
+ print(" [失败] %s %s" % (it.get("file"), it.get("reason")))
2160
+ continue
2161
+ verb = {"installed": "已安装", "exists": "已存在",
2162
+ "copy": "将安装"}.get(it["action"], it["action"])
2163
+ print(" [%s] %s → %s" % (verb, it.get("family"),
2164
+ it.get("dest")))
2165
+ if not args.dry_run:
2166
+ fams = [i["family"] for i in inst if i.get("ok")]
2167
+ if fams:
2168
+ fontcheck.save_consent(
2169
+ fams,
2170
+ args.from_zip or args.preset or "local",
2171
+ [i["dest"] for i in inst if i.get("ok")])
2172
+ print("\n重启 Word 后生效;以管理员身份运行的 Word "
2173
+ "看不到用户级字体。")
2174
+ return 0 if all(i.get("ok") for i in inst) else 1
2175
+ finally:
2176
+ shutil.rmtree(tmpdir, ignore_errors=True)
2177
+
2178
+
1789
2179
  def main(argv=None):
1790
2180
  ap = argparse.ArgumentParser(description="docx 公文格式规范化")
1791
- ap.add_argument("path", nargs="+")
2181
+ ap.add_argument("path", nargs="*")
1792
2182
  ap.add_argument("--mode", choices=["apply", "check"], default="apply")
2183
+ ap.add_argument("--light", action="store_true",
2184
+ help="只做全局规范化(分页开关/中文引号/中英之间不留空格),不套公文版式")
1793
2185
  ap.add_argument("--recursive", action="store_true")
1794
2186
  ap.add_argument("--json", action="store_true")
1795
2187
  ap.add_argument("--config", default=None, help="JSON 配置字符串或 .json 文件路径")
2188
+ ap.add_argument("--fonts", action="store_true",
2189
+ help="只报告字体安装情况(不处理文档)")
2190
+ ap.add_argument("--install-fonts", nargs="*", default=None,
2191
+ help="安装字体:本地文件/目录(可留空配合 --preset)")
2192
+ ap.add_argument("--preset", default=None,
2193
+ help="预置清单:gongwen(普通档)/ gongwen-strict(严格档)")
2194
+ ap.add_argument("--from-zip", default=None,
2195
+ help="从 zip(URL 或本地路径)提取字体后安装")
2196
+ ap.add_argument("--yes", action="store_true",
2197
+ help="确认下载安装;缺省时只做 dry-run")
2198
+ ap.add_argument("--confirm", default=None,
2199
+ help="授权令牌(由 --fonts --json 的 consentRequired.token 给出,"
2200
+ "用户同意后回传;一次性、10 分钟内有效)")
2201
+ ap.add_argument("--dry-run", action="store_true", help="只报告不落盘")
2202
+ ap.add_argument("--font-profile", choices=["standard", "strict"], default=None,
2203
+ help="字体档:standard(默认)/ strict(GBK/GB2312)")
2204
+ ap.add_argument("--no-font-check", action="store_true",
2205
+ help="跳过字体检测")
1796
2206
  args = ap.parse_args(argv)
1797
2207
 
1798
2208
  cfg = DEFAULTS
@@ -1803,6 +2213,21 @@ def main(argv=None):
1803
2213
  raw = fh.read()
1804
2214
  cfg = merge_config(json.loads(raw))
1805
2215
 
2216
+ if args.font_profile:
2217
+ cfg["fontProfile"] = args.font_profile
2218
+ if args.no_font_check:
2219
+ cfg.setdefault("fontCheck", {})["enabled"] = False
2220
+ if args.config and args.font_profile == "strict":
2221
+ cfg.setdefault("fontCheck", {})
2222
+ apply_font_profile(cfg, cfg.get("fontProfile"))
2223
+
2224
+ if args.fonts or args.install_fonts is not None or args.preset \
2225
+ or args.from_zip:
2226
+ return _fonts_cli(args, cfg)
2227
+
2228
+ if not args.path:
2229
+ ap.error("缺少要处理的 docx 路径(或用 --fonts 查看字体)")
2230
+
1806
2231
  results = []
1807
2232
  failed = False
1808
2233
  violations = 0
@@ -1818,7 +2243,7 @@ def main(argv=None):
1818
2243
  })
1819
2244
  continue
1820
2245
  try:
1821
- res = process(path, args.mode, cfg)
2246
+ res = process(path, args.mode, cfg, full=not args.light)
1822
2247
  except Exception as exc:
1823
2248
  import traceback
1824
2249
  res = {"ok": False, "file": os.path.abspath(path), "mode": args.mode,
package/locale/en.json CHANGED
@@ -8,11 +8,12 @@
8
8
  "applied": "normalized",
9
9
  "clean": "already compliant",
10
10
  "violations": "deviations",
11
- "skipped": "skipped (no trigger keyword / excluded)",
12
- "skipReason": "official-document formatting not applied (no trigger keyword or excluded)",
11
+ "skipped": "skipped (no trigger keyword, global normalization off)",
12
+ "skipReason": "official-document formatting not applied (no trigger keyword)",
13
+ "lightApplied": "global normalization only (no 公文 in the request: pagination switches, Chinese quotes, no space around Latin words)",
13
14
  "config": {
14
- "triggerKeywords": "trigger words; the request text must contain one of them, default ['公文']",
15
- "skipRequestKeywords": "exclusion words; a hit disables the whole plugin, default ['实验报告']",
15
+ "triggerKeywords": "trigger words; the request text must contain one of them for the full layout, default ['公文']",
16
+ "globalNormalize": "still apply global normalization when no trigger word matched (pagination off, Chinese quotes, no CJK-Latin spacing)",
16
17
  "autoFormat": "queue any .docx path seen in tool results",
17
18
  "formatExisting": "also normalize files created before the plugin loaded"
18
19
  }