added terpenes directory tree

demo.py

@@ -8,7 +8,14 @@
 
 
 def create_phrases_file(phrases_dir, phrases_file, dictionary_dir=ETHICS_DICTIONARY_DIR):
-    global terms_xml_path
+    """
+    Creates file for phrases based on the dictionary dir and dictionary name
+
+    creates directories if they do not exist
+    :param phrases_dir: directory holding the phrases file
+    :param phrases_file: file holding phrases file
+
+    """
     terms_xml_dir = Path(dictionary_dir, phrases_dir)
     if not terms_xml_dir.exists():
         terms_xml_dir.mkdir()
@@ -68,7 +75,7 @@ def create_and_write_list_for_fields(dict_for_entities, field, out_filename):
     run_analysis(
         get_or_create_corpus_dir(TERPENES),
         create_phrases_file("terpenes_key_phrases", "terpenes_key_phrases.xml", dictionary_dir="terpenes_dictionary"),
-        query=TERPENES,
+        query="terpenes",
         hits = 20,
     )
 

docanalysis/extract_entities.py

@@ -58,7 +58,8 @@ def extract_entities_from_papers(self, corpus_path, terms_xml_path, query=None,
             self.create_project_files(query, hits, corpus_path)
         if install_ami:
             logging.info(f"installing ami3 (check whether this is a good idea)")
-            self.install_ami()
+            logging.info(f"please check independently that ami is installed")
+            # self.install_ami()
 
         logging.info(f"dict with parsed xml in {corpus_path}")
         dict_with_parsed_xml = self.make_dict_with_parsed_xml(corpus_path)
@@ -88,10 +89,15 @@ def create_project_files(self, QUERY, HITS, OUTPUT):
         os.system(f'pygetpapers -q "{QUERY}" -k {HITS} -o {OUTPUT} -x')
         os.system(f"ami -p {OUTPUT} section")
 
-    def install_ami(self):
-        os.system("git clone https://github.com/petermr/ami3.git")
-        os.system("cd ami3")
-        os.system("mvn install -Dmaven.test.skip=true")
+    """
+    removed as too complex. 
+    TODO maybe have a check that `ami` is installed, but install elsewhere
+    """
+    # def install_ami(self):
+    #     os.system("git clone https://github.com/petermr/ami3.git")
+    #     os.system("cd ami3")
+    #     os.system("mvn install -Dmaven.test.skip=true")
+
 
     def make_dict_with_parsed_xml(self, output):
 

terpenes_dictionary/terpenes_key_phrases/terpenes_key_phrases.xml

@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<dictionary title="terpenes_key_phrases">
+ <entry term="menthol" name="menthol" />
+</dictionary>