Jenni 和我审查了在链接线程中分享的 Paratext 6 Numbers.xml。 我们使用 AI 将其与 Paratext 9 中的当前版本进行了比较,结果发现遗漏了 72 处关于重量和度量的引用。 我们已在附上的更新版 Numbers.xml 文件中恢复了这些内容。
Numbers.xml
Numbers.xml (PT9) — 恢复报告
我做了什么
我将 PT6_Numbers.xml(105 个术语,基于 GUID 的 ID,UTF-16)与 PT9_Numbers.xml(78 个术语,语义化 ID 如 cardinal_012,UTF-8)进行了比较,通过实际的经文引用而非 ID 来匹配术语(这两个文件使用完全不同的 ID 方案,因此 ID 本身无法提供对应关系)。PT6 的经文引用带有额外的两位后缀(06600601200);通过去掉最后两位数字,这些引用可以规范化为 PT9 的 9 位 BBBCCCVVV 格式——由于所有 PT6 引用在该位置都使用 00,因此该规范化过程没有丢失任何信息。
关键发现:基数/序数/分数/倍数数字并未受损
PT6 中关于普通数字(基数、序数、分数、倍数)的 246 个唯一经文引用在 PT9 中全部存在——零丢失。PT9 还在此组中新增了 4 个引用。这就是你提到的“改进”:例如,PT6 曾将“第九/第十/第十一/第十二”合并为一个术语,并错误地指向启示录 21:20;PT9 正确地将这些拆分为独立且标签准确的术语。我未对 PT9 的这一部分做任何改动。
真正发生数据丢失的地方
PT6 有四个完整类别被 PT9 完全删除:金额(9 个术语——lepton, quadrans, assarion, drachma, didrachma, denarius, mina, argurion, talanton)、重量和度量(8 个术语——litra, talent, choinix, modios, saton, koros, batos, metretes)、距离(5 个术语——pekus/cubit, stadion, milion, a Sabbath day's journey, orguia/fathom)以及时间(24 个术语——daily, hour, month, week, "on the third day," "in those days," 等)。这意味着 46 个概念和大约 251 个唯一经文引用在 PT9 的 Numbers.xml 中不再被检查。
在这 46 个中:
5 个已在 PT9 的其他位置恢复,只是位置移动了:PT9 的 ordinal_001_day(次日)、ordinal_002_day(第二天)、ordinal_003_day(第三天)、ordinal_007_day(第七天)和 ordinal_008_day(第八天)是完全匹配的——相同的引用,相同的含义。我没有重复这些。
2 个被故意排除:PT6 的“Time thirty-eight”和“Time forty-six”术语将不相关的数字混合在一个错误标签下(例如,“Time thirty-eight”同时指向约翰福音 5:5 中 38 年的引用和使徒行传/希伯来书中几个不相关的 40 年引用)。PT9 已经正确修复了这一点——所有这些引用现在都位于标签准确的基数术语下(cardinal_038, cardinal_040, cardinal_046 等)。逐字恢复这些 PT6 术语会重新引入该错误,因此我将它们排除在外。
其余 39 个确实缺失,我已将它们添加回新文件中。
新 Numbers.xml 中包含的内容
总共 117 个术语:PT9 原有的 78 个术语完全未变,加上 39 个恢复的术语,涵盖四个丢失的类别,并采用 PT9 自身的风格编写:
延续 PT9 模式的语义化 snake_case ID:money_001_lepton … money_009_talanton, weight_001_litra … weight_008_metretes, distance_001_pekus … distance_005_orguia, time_001_daily … time_017_time_times_half。
每个术语都有一个 <Domain> 标签(PT9 仅对其时间相关的序数使用此标签,标记为 time;我将同样的概念扩展为 money, weightsAndMeasures 和 distance,用于其他三个类别——见下文注意事项)。
<Transliteration> 设置为与 ID 匹配,正如所有现有 PT9 条目所做的那样。
简短的 <Gloss>,保留 PT6 中的实质性定义(例如 batos (Hebrew bat, a liquid measure, ~39.5 litres)),而不是 PT6 冗长的“Weights and measures in the New Testament - 7. ...”框架。
每个恢复术语的完整原始 PT6 引用列表。
Jenni and I reviewed the Paratext 6 Numbers.xml that was shared in the linked thread. We used AI to compare it with the current version in Paratext 9 and it found 72 references for weights and measures which had been omitted. We have now restored them in an updated Numbers.xml file, attached.
Numbers.xml
Numbers.xml (PT9) — Restoration Report
What I did
I compared PT6_Numbers.xml (105 terms, GUID-based IDs, UTF-16) against PT9_Numbers.xml (78 terms, semantic IDs like cardinal_012, UTF-8), matching terms by their actual verse references rather than by ID (the two files use completely different ID schemes, so IDs themselves gave no correspondence). PT6's verse references carry an extra two-digit suffix (06600601200); these normalize to PT9's 9-digit BBBCCCVVV format by dropping the last two digits — every PT6 reference used 00 there, so nothing was lost in that normalization.
Key finding: the cardinal/ordinal/fraction/multiple numbers were not damaged
Every one of PT6's 246 unique verse references for plain numbers (cardinals, ordinals, fractions, multiples) is still present in PT9 — zero losses. PT9 also added 4 new references in this group. This is the "improvement" you mentioned: PT6 had, for example, one single term lumping "ninth/tenth/eleventh/twelfth" together under a wrong label pointing at Revelation 21:20; PT9 correctly split these into distinct, accurately-labeled terms. I left this whole part of PT9 untouched.
Where the real data loss happened
PT6 had four entire categories that PT9 dropped completely: Money amounts (9 terms — lepton, quadrans, assarion, drachma, didrachma, denarius, mina, argurion, talanton), Weights and measures (8 terms — litra, talent, choinix, modios, saton, koros, batos, metretes), Distances (5 terms — pekus/cubit, stadion, milion, a Sabbath day's journey, orguia/fathom), and Time (24 terms — daily, hour, month, week, "on the third day," "in those days," etc.). That's 46 concepts and roughly 251 unique verse references that simply aren't checked anymore in PT9's Numbers.xml.
Of those 46:
5 were already restored elsewhere in PT9, just relocated: PT9's ordinal_001_day (next day), ordinal_002_day (second day), ordinal_003_day (third day), ordinal_007_day (seventh day), and ordinal_008_day (eighth day) are exact matches — same references, same meaning. I did not duplicate these.
2 were deliberately left out: PT6's "Time thirty-eight" and "Time forty-six" terms each mixed unrelated numbers under one mislabeled gloss (e.g. "Time thirty-eight" pointed at both a 38-year reference in John 5:5 and several unrelated 40-year references in Acts/Hebrews). PT9 already fixed this correctly — every one of those references now sits under an accurately-labeled cardinal term (cardinal_038, cardinal_040, cardinal_046, etc.). Restoring these PT6 terms verbatim would have reintroduced the bug, so I left them out.
The remaining 39 were genuinely missing and are what I've added back to the new file.
What's in the new Numbers.xml
117 terms total: all 78 of PT9's original terms, completely unchanged, plus 39 restored terms covering the four lost categories, written in PT9's own style:
Semantic snake_case IDs continuing PT9's pattern: money_001_lepton … money_009_talanton, weight_001_litra … weight_008_metretes, distance_001_pekus … distance_005_orguia, time_001_daily … time_017_time_times_half.
A <Domain> tag on each (PT9 only uses this for its time-related ordinals, tagging them time; I extended the same idea with money, weightsAndMeasures, and distance for the other three categories — see caveat below).
<Transliteration> set to match the ID, exactly like every existing PT9 entry does.
A short <Gloss> that keeps the substantive definition from PT6 (e.g. batos (Hebrew bat, a liquid measure, ~39.5 litres)) instead of PT6's verbose "Weights and measures in the New Testament - 7. ..." framing.
The complete original PT6 reference list for each restored term.