change on ontologie level 2 more case

src/processing_mapping_results/chebi_id/recovery_of_associated_chebi_id.py

@@ -470,10 +470,18 @@ def ontologie_gap_kegg_chebi(csv_file, out_folder):
     ontologie_level_chebi = ["associated ontologie level of ChEBI ID"]
     ontologie_level_equiv_kegg  = ["positional difference betwen kegg et chebi ID"]
     chebi_on_kegg = ["ID chebi from KEGG"]
-    for i_ac_k, ac_kegg in enumerate(l_kegg):
-        accheb = l_cheb[i_ac_k].strip().upper()
+    for i_ac_k, ac_kegg in enumerate(l_kegg[1:]):
+        accheb = l_cheb[i_ac_k+1].strip().upper()
         ackegg = ac_kegg.strip().upper()
-        if accheb == "NA":
+        #print(accheb)
+        #print(ackegg)
+        if accheb == "NA" and ackegg == "NA":
+            chebi_on_kegg.append("NA")
+            name_chebi.append("NA")
+            ontologie_level_chebi.append("NA")
+            ontologie_level_equiv_kegg.append("NA")
+            name_kegg.append("NA")
+        elif accheb == "NA":
             url_kegg = "https://www.genome.jp/entry/"+ackegg
             print(url_kegg)
             soupk = BeautifulSoup(urllib.request.urlopen(url_kegg).read().decode("utf-8"), "html.parser")
@@ -519,19 +527,39 @@ def ontologie_gap_kegg_chebi(csv_file, out_folder):
             name_chebi.append("NA")
             ontologie_level_chebi.append("NA")
             ontologie_level_equiv_kegg.append("NA")
-    print(chebi_on_kegg)
-    #print(name_kegg)
-    out_df = pd.DataFrame([l_kegg, name_kegg, chebi_on_kegg, l_cheb, name_chebi,
-                           ontologie_level_chebi, ontologie_level_equiv_kegg]).transpose()
-    outf_name = out_folder + "KEGG_and_ChEBI_dif_ontologie_test.xlsx"
-    excel_file_writer(out_df, outf_name,  sheetname="resume_ontology")
-    """
+
         elif ackegg == "NA":
             name_kegg.append("NA")
-            name_chebi.append(accheb)
-            ontologie_level_chebi.append("NA")  # voir si on sort le niveau de l'ontologie
+            chebi_on_kegg.append("NA")
             ontologie_level_equiv_kegg.append("NA")
-
+            print(accheb)
+            url_chebi = "https://www.ebi.ac.uk/chebi/chebiOntology.do?chebiId="+accheb +"&treeView=true#vizualisation"
+            chebi_soup = BeautifulSoup(urllib.request.urlopen(url_chebi).read().decode("utf-8"), "html.parser")
+            text_chebi = chebi_soup.get_text()
+            #print(text_chebi)
+            i_beg_p_name = text_chebi.find("ChEBI Name")
+            inter_name = text_chebi[i_beg_p_name: i_beg_p_name+300]
+            i_end_p_name = inter_name.find("ChEBI ID")
+            name_cheb_treat= text_chebi[i_beg_p_name+10: i_beg_p_name + i_end_p_name]
+            name_cheb_treat = name_cheb_treat.strip()
+            print(name_cheb_treat)
+            name_chebi.append(name_cheb_treat)
+            onto_name = accheb + " " + name_cheb_treat
+            print(onto_name)
+            begin_tree = text_chebi.find("Tree View")
+            end_tree = text_chebi.find("Relationship Types")
+            tree = text_chebi[begin_tree: end_tree]
+            onto_pos =  tree.find(onto_name)
+            #print(tree)
+            print(onto_pos)
+            #print(tree[0:onto_pos])
+            ont_lev_che = 0
+            index = tree[0:onto_pos].find("CHEBI:")
+            while index != -1:
+                ont_lev_che += 1
+                index = tree[0:onto_pos].find("CHEBI:", index + 1)
+            ontologie_level_chebi.append(ont_lev_che)
+        """
         else:
             url = "https://www.ebi.ac.uk/chebi/searchId.do?chebiId="+ac_chebi
             soup = BeautifulSoup(urllib.request.urlopen(url).read().decode("utf-8"), "html.parser")
@@ -542,8 +570,12 @@ def ontologie_gap_kegg_chebi(csv_file, out_folder):
             name_chebi = text[i_beg_p_name+10: i_beg_p_name + i_end_p_name]
             name_chebi = name_chebi.strip()
     """
+    out_df = pd.DataFrame([l_kegg, name_kegg, chebi_on_kegg, l_cheb, name_chebi,
+                           ontologie_level_chebi, ontologie_level_equiv_kegg]).transpose()
+    outf_name = out_folder + "KEGG_and_ChEBI_dif_ontologie_test2.xlsx"
+    excel_file_writer(out_df, outf_name,  sheetname="resume_ontology")
 
 if __name__ == "__main__":
     LOCAL = "C:\\Users\\mumec\\Desktop\\Mini_codes\\ontologie_kegg_chebi\\"
-    infile = LOCAL  + "ontologie_gap_chebi_na.csv"
+    infile = LOCAL  + "ontologie_gap_2.csv"
     ontologie_gap_kegg_chebi(infile, LOCAL)
\ No newline at end of file