fixed bugs; updated tests

.gitignore

@@ -161,3 +161,8 @@ cython_debug/
 
 # vscode
 .vscode/
+
+# some test files
+tests/tmp/
+tests/kinases/
+tests/panther/

pyuniprot/DB.py

@@ -30,7 +30,7 @@ class PANTHER:
 class SeqRange:
     """Store a sequence range record."""
 
-    seq_begein: int | str
+    seq_begin: int | str
     seq_end: int | str
 
 

pyuniprot/Uniprot.py

@@ -49,7 +49,7 @@ class EC:
 class DE:
     """Protein names"""
 
-    recommended_name: str
+    recommended_name: str | None
     ec_record: EC | None
     alt_names: list[str] | None
     contains: str | None
@@ -291,7 +291,7 @@ def _get_category_lines(self) -> None:
 
         line: str = u_file.readline()
         references: list = []
-        while line[0:2] != "//":
+        while line and line[0:2] != "//":
             if line.startswith("ID"):
                 items = line[2:-1].split()
                 entry_name = items[0]
@@ -321,9 +321,10 @@ def _get_category_lines(self) -> None:
                 continue
             elif line.startswith("DE"):
                 alt_names: list[str] = []
-                ec_line = "DE            EC="
+                ec_line: str = "DE            EC="
                 ec_record: EC | None = None
-                contains = ""
+                contains: str | None = None
+                recommended_name: str | None = None
                 while line.startswith("DE"):
                     if line.startswith("DE   RecName:"):
                         recommended_name = line[:-2].split("=")[-1]
@@ -480,8 +481,13 @@ def _get_category_lines(self) -> None:
                             seq = seq_covering.split("=")
                             chain_ids: list[str] = seq[0].split("/")
                             all_chain_ids.extend(chain_ids)
-                            uniprot_resid_start: int = int(seq[1].split("-")[0])
-                            uniprot_resid_end: int = int(seq[1].split("-")[1])
+                            if len(seq) > 1 and "-" != seq[1]:
+                                uniprot_resid_start: int | str = int(
+                                    seq[1].split("-")[0]
+                                )
+                                uniprot_resid_end: int | str = int(seq[1].split("-")[1])
+                            else:
+                                uniprot_resid_start, uniprot_resid_end = "", ""
                             all_seq_ranges.append(
                                 SeqRange(uniprot_resid_start, uniprot_resid_end)
                             )
@@ -625,4 +631,4 @@ def _get_category_lines(self) -> None:
         # print(line)
 
         self._category_lines = content_dict
-        # print(self.category_lines)
+        # print(self.category_lines["FT"].feature_tables)

P36952.txt → tests/P36952.txt

@@ -2,7 +2,7 @@ ID   SPB5_HUMAN              Reviewed;         375 AA.
 AC   P36952; B2R6Y4; Q6N0B4; Q8WW89;
 DT   01-JUN-1994, integrated into UniProtKB/Swiss-Prot.
 DT   05-MAY-2009, sequence version 2.
-DT   28-JUN-2023, entry version 197.
+DT   08-NOV-2023, entry version 199.
 DE   RecName: Full=Serpin B5;
 DE   AltName: Full=Maspin;
 DE   AltName: Full=Peptidase inhibitor 5;
@@ -194,13 +194,14 @@ DR   EPD; P36952; -.
 DR   jPOST; P36952; -.
 DR   MassIVE; P36952; -.
 DR   MaxQB; P36952; -.
-DR   PaxDb; P36952; -.
+DR   PaxDb; 9606-ENSP00000372221; -.
 DR   PeptideAtlas; P36952; -.
 DR   PRIDE; P36952; -.
 DR   ProteomicsDB; 55240; -. [P36952-1]
 DR   ProteomicsDB; 55241; -. [P36952-2]
+DR   Pumba; P36952; -.
 DR   TopDownProteomics; P36952-1; -. [P36952-1]
-DR   Antibodypedia; 4036; 620 antibodies from 43 providers.
+DR   Antibodypedia; 4036; 599 antibodies from 43 providers.
 DR   DNASU; 5268; -.
 DR   Ensembl; ENST00000382771.9; ENSP00000372221.4; ENSG00000206075.14. [P36952-1]
 DR   Ensembl; ENST00000489441.5; ENSP00000467158.1; ENSG00000206075.14. [P36952-2]
@@ -238,7 +239,7 @@ DR   GenomeRNAi; 5268; -.
 DR   Pharos; P36952; Tbio.
 DR   PRO; PR:P36952; -.
 DR   Proteomes; UP000005640; Chromosome 18.
-DR   RNAct; P36952; protein.
+DR   RNAct; P36952; Protein.
 DR   Bgee; ENSG00000206075; Expressed in skin of abdomen and 95 other tissues.
 DR   ExpressionAtlas; P36952; baseline and differential.
 DR   Genevisible; P36952; HS.
@@ -249,7 +250,6 @@ DR   GO; GO:0016528; C:sarcoplasm; IEA:Ensembl.
 DR   GO; GO:0004867; F:serine-type endopeptidase inhibitor activity; IBA:GO_Central.
 DR   GO; GO:0030198; P:extracellular matrix organization; IEA:Ensembl.
 DR   GO; GO:0002009; P:morphogenesis of an epithelium; IEA:Ensembl.
-DR   GO; GO:0010951; P:negative regulation of endopeptidase activity; IBA:GO_Central.
 DR   GO; GO:0060512; P:prostate gland morphogenesis; IEA:Ensembl.
 DR   GO; GO:0050678; P:regulation of epithelial cell proliferation; IEA:Ensembl.
 DR   CDD; cd02057; serpinB5_maspin; 1.