Fix IndexError when abbrv is longer than original; Close #12 (#14)

* Fix IndexError when abbrv is longer than original

In some cases, there is a mismatch between abbreviation and original,
where a dot is added to an unabbreviated word, e.g., "Control".
If this occurs, the dot is removed and the abbreviation is reduced to
the length of the original word.

* Fix missed detection of single letter part names

Fix the wrong classification of single letter part names, if the single
letter is also a stopword.

pyiso4/lexer.py

@@ -132,7 +132,7 @@ def yield_hyphenated(word: str, base_pos: int) -> Iterable[Token]:
                     yield Token(TokenType.PART, word, self.start_word)
                     was_part = self.count
                 # check if ordinal (preceded by PART)
-                elif IS_ORDINAL.match(word) and self.count == was_part + 1:
+                elif IS_ORDINAL.fullmatch(word) and self.count == was_part + 1:
                     yield Token(TokenType.ORDINAL, word, self.start_word)
                 # check if article (after ordinal, so "a" is detected as ordinal if preceded by PART)
                 elif lower_word in ARTICLES:
@@ -155,6 +155,7 @@ def yield_hyphenated(word: str, base_pos: int) -> Iterable[Token]:
             # yield the remaining symbols, if any
             if len(end_symbols) > 0:
                 yield Token(TokenType.SYMBOLS, end_symbols, self.pos - len(end_symbols))
+                was_part = self.count
 
             self.next()
 

pyiso4/ltwa.py

@@ -177,6 +177,9 @@ def match_capitalization_and_diacritic(abbrv: str, original: str) -> str:
         """Matches the capitalization and diacritics of the `original` word, as long as they are similar
         """
 
+        if len(abbrv) > len(original):
+            abbrv = abbrv[:len(original)]
+
         normalized_abbrv = list(normalize(abbrv, Level.SOFT))
         for i, c in enumerate(normalized_abbrv):
             unided = unidecode(original[i])

tests/tests.tsv

@@ -40,4 +40,12 @@ Zeitschrift des Deutschen Palästina-Vereins	Z. Dtsch. Paläst.-Ver.
 International Journal of e-Collaboration	Int. J. e-Collab.
 Proceedings of A. Razmadze Mathematical Institute	Proc. A. Razmadze Math. Inst.
 Norsk Militært Tidsskrift	Nor. Mil. Tidsskr.
-Proceedings of the 2024 Conference on Science	Proc. 2024 Conf. Sci.
+Proceedings of the 2024 Conference on Science	Proc. 2024 Conf. Sci.
+IEEE Power and Energy Magazine	IEEE Power Energy Mag.
+IEEE Transactions on Automatic Control	IEEE Trans. Autom. Control
+E.S.A. bulletin	E.S.A. bull.
+Acta Universitatis Carolinae. Iuridica	Acta Univ. Carol., Iurid.
+Physical Review. A	Phys. Rev., A
+Physical Review. D	Phys. Rev., D
+Physical Review. E	Phys. Rev., E
+Physical Review. I	Phys. Rev., I