Merge branch 'Fix_abstract' into 'develop'

Fix abstract

See merge request inf1603/irisaparser!35

irisaparser/abstract_lib.py

@@ -6,6 +6,8 @@
 def abstract_extractor(parsed):
     abstr = parsed["content"].upper().find('ABSTRACT')
     intro = parsed["content"].upper().find('INTRODUCTION')
+    if intro == -1:
+        intro = parsed["content"].upper().find('I N T R O D U C T I O N')
     end = len(parsed["content"])
 
     text = parsed["content"]
@@ -18,11 +20,19 @@ def abstract_extractor(parsed):
                 return x
         return "error"
 
+    if intro == -1:
+        splited_text = text[abstr:].split("\n\n")
+
+        for x in splited_text:
+            if len(x) > 200:
+                return x
+        return "error"
+
     else:
         cut = 0
         for x in range(intro, abstr, -1):
             if text[x] == "\n" and text[x-1] == "\n":
                 cut = x
                 break
-
-        return text[abstr+8:cut] if end-intro > 6000 else text[abstr+8:cut].split("\n\n")[0]
+    
+        return text[abstr+8:cut] if end-intro > 6000 else text[abstr+8:cut].split("\n\n")[0]