Repository navigation
Support smart quotes (“Jack”), guillemets («Petit»), and CJK brackets as nickname delimiters #273
Description
The three built-in nickname delimiters (quoted_word, double_quotes, parenthesis) are ASCII-only, so nicknames wrapped in any typographic or non-Latin quote convention leak into the name pieces as garbage middle names (v1.3.0):
>>> HumanName('John "Jack" Kennedy').nickname # ASCII: works
'Jack'
>>> HumanName('John “Jack” Kennedy').middle # smart quotes: leaks
'“Jack”'
>>> HumanName('Jean «Petit» Dupont').middle # guillemets (French, Russian, ...)
'«Petit»'
>>> HumanName('Hans „Hansi“ Müller').middle # low-high (German, Polish, Czech, ...)
'„Hansi“'
>>> HumanName('山田「タロ」太郎').first # CJK corner brackets: unsplit blob
'山田「タロ」太郎'The smart-quote case is effectively English input: “…” (U+201C/201D) is what word processors and phone keyboards autocorrect plain quotes to, so real-world copy-paste text hits this constantly — same input-normalization class as #266.
Proposed additions to the built-in delimiter inventory, one ordered open/close pair each (the conventions can't be merged into a single character class — German's closing “ is English's opening quote, so each pair needs its own regex, exactly like the three existing ones):
| Pair | Convention |
|---|---|
“…” |
smart-quotes English; also Chinese |
«…» |
French, Russian, Italian, Greek — allow inner spaces (« Petit ») |
„…“ |
German, Polish, Czech, Hungarian |
»…« |
German alternative |
”…” |
Swedish, Finnish |
「…」, 『…』 |
Japanese corner brackets |
（…） |
fullwidth parentheses (CJK) |
All of these are safe as defaults: unlike the apostrophe, none can legitimately appear inside a name. Explicitly excluded: curly single quotes ‘…’ — U+2019 is also the preferred typographic apostrophe (O’Connor), so it would need the same whitespace-border rules as the ASCII quoted_word regex and deserves separate, careful treatment if ever.
The nickname_delimiters dict already accepts additional compiled patterns at runtime, so this is purely an expansion of the built-in inventory. Related: #266, #269.
Activity
[-]Nicknames in smart quotes (“Jack”), guillemets («Petit»), or CJK brackets are not extracted[/-][+]Support smart quotes (“Jack”), guillemets («Petit»), and CJK brackets as nickname delimiters[/+]feat: typographic nickname delimiters as defaults (closes #273)
derek73 commentedon Jul 18, 2026
Implemented on the 2.0 branch (PR #288, commit 7ee6e3a): all eight proposed pairs ship in
DEFAULT_NICKNAME_DELIMITERSand as named sentinels in the v1CONSTANTSAPI (smart_double_quotes,guillemets, ...), with curly single quotes excluded as proposed (U+2019 stays the apostrophe in "O'Connor"). The issue's five repro examples are pinned as test rows,« Petit »inner spacing is tolerated, and a delimiter character consumed by another pair's extraction (the shared-character overlap: German's closing“is English's opening quote) no longer emits a spuriousunbalanced-delimiterambiguity. Closes with the 2.0.0 merge.fix: extraction scans by position, not pair order (review critical)
test: regression guards from the review
A word-final apostrophe is not an unbalanced delimiter
Add a second differential corpus harvested from the issue tracker
Bound every rule by the corpus it claims, not by its category
Record what a rule claims in all three dimensions classify() uses
Protect the shapes the two entries claim to protect
Say which way an over-wide exclusion fails, and drop "for a year"
Stop feat(#273) pointing at a limit that no longer exists
Fix three claims this branch got wrong while fixing others
docs(rules): topic-review amendments from the July-arc session (prima…
Pin the delimited maiden marker and classify it against 1.4.0 (derek7…
Label the CJK honorific rows for the issues they depend on (derek73#324)
Let a [[never]] entry narrow by fields, not just by name
Fix what the review of derek73#597 found
docs(decisions): Standing blocks for fourteen small decisions files (#…