add more stop words

CHANGELOG.md

@@ -4,8 +4,11 @@
 
 ## [v1.7.5] - 2023-09-12 Tue
 
+- Add: CSV and TSV files provide now verbatim authorship instead of normalized
+  one.
+- Add: a few more "termination words"
 - Fix [#254]: treat `fa` as forma.
-- Fix [#252]: process `dem` as an author word for `Von dem Bush` and like.
+- Fix [#253]: process `dem` as an author word for `Von dem Bush` and like.
 - Fix [#251]: do not process `y` as `and` for `Rafael Arango y Molina`.
 - Fix [#249]: allow `cf` at the end of the strings, cf for infraspecies.
 - Fix [#248]: do not escape double quotes for TSV output.

ent/internal/preparser/grammar.peg

@@ -15,7 +15,7 @@ TailPhrase <- TailLastWordJunk / TailPhrase4 / TailPhrase3 /
 
 TailLastWordJunk <- (("var" / "ined" / "ssp" / "subsp" / "subgen" ) '.'? /
  "sensu" / "new" / "non" / "nec" / "hybrid" / "von" / 'P.' _? 'P.' /
- "ms") '?'? &SpaceOrEnd
+ "ms" / 'CF') '?'? &SpaceOrEnd
 
 TailPhrase4 <- ("pro" _ "parte" / "nomen") &NotLetterOrEnd / 'p.' _? 'p.' /
   "nom." / "comb."
@@ -24,13 +24,13 @@ TailPhrase3 <- '('? 's' ('.' _? / _ ) ('s' '.'? &NotLetterOrEnd / 'l.' / 'str.'
    'lat.')
 
 TailStopWords <- ("environmental" / "enrichment" / "samples" /
-  ("species" _)? ("group" / "complex") / "clade" /
-  "author" / "nec" / "vide" / "fide" / "non" ) &NotLetterOrEnd
+  "species" / "group" / "complex" / "clade" /
+  "author" / "nec" / "vide" / "species" / "fide" / "non" / "not" ) &NotLetterOrEnd
 
 TailPhrase2 <- ("sero" ("var" / "type") / "sensu" / "auct" / "sec" / "near" /
   "str") '.'? &NotLetterOrEnd
 
-TailPhrase1 <- (('('? ('ht' / 'hort')) / 'spec' /
+TailPhrase1 <- (('('? ('ht' / 'hort')) / "S" 'pec' /
   'nov' '.'? _ 'spec') '.'? &NotLetterOrEnd
 
 SpaceOrEnd <- CommaSpace? END