fixes paper id and authors list handling

WolfgangFahl · Mar 22, 2023 · 0a85e11 · 0a85e11
1 parent b3dd56f
commit 0a85e11
Show file tree

Hide file tree

Showing 3 changed files with 9 additions and 1 deletion.
diff --git a/ceurws/ceur_ws.py b/ceurws/ceur_ws.py
@@ -488,6 +488,8 @@ def __init__(self):
                                             primaryKey="id",
                                             entityPluralName="papers",
                                             config=CEURWS.CONFIG,
+                                            handleInvalidListTypes=True,
+                                            listSeparator=",",
                                             name=self.__class__.__name__)
 
 

diff --git a/ceurws/papertocparser.py b/ceurws/papertocparser.py
@@ -39,14 +39,20 @@ def parsePapers(self):
         paper_records=[]
         toc=self.soup.find(attrs={"class": "CEURTOC"})
         if toc:
+            index=0
+            paper_ids=[]
             for paper_li in toc.findAll('li'):
+                index+=1
                 paper_record = self.scrape.parseWithScrapeDescription(paper_li, self.scrapeDescr)
                 paper_record["vol_number"]=self.number
                 href=paper_li.find('a', href=True)
                 if href:
                     paper_record["pdf_name"]=href.attrs["href"]
                 if "id" in paper_li.attrs:
                     paper_id=paper_li.attrs["id"]
+                    if paper_id in paper_ids:
+                        paper_id=f"{paper_id}-duplicate-{index}"
+                    paper_ids.append(paper_id)
                     paper_record["id"]=f"Vol-{self.number}/{paper_id}"
                 paper_records.append(paper_record)
                 pass

diff --git a/tests/test_papertocparser.py b/tests/test_papertocparser.py
@@ -66,7 +66,7 @@ def test_volExamples(self):
         """
         tests parsing of volume examples
         """
-        vol_examples = [(83,12),(1,15),(3264,10),(3343,7)]
+        vol_examples = [(2376,35),(2379,8),(1,15),(83,12),(3264,10),(3343,7)]
         counter=Counter()
         debug=self.debug
         debug=True