Merge pull request #29 from cu-mkp/issue28

Update paths in update.py and manuscript_helpers.py to account for new m-k-manuscript-data location

Author: gschare

manuscript_helpers.py

@@ -10,9 +10,10 @@
               'music', 'plant', 'place', 'personal_name',
               'profession', 'sensory', 'tool', 'time', 'weapon']
 
-cwd = os.getcwd()
-m_path = cwd if 'manuscript-object' not in cwd else f'{cwd}/../'
-m_k_data_to_thesaurus = f'{m_path}/manuscript-object/thesaurus'
+manuscript_data_path = os.path.dirname(os.getcwd()) + "/m-k-manuscript-data"
+thesaurus_path = os.getcwd() + "/thesaurus"
+assert(os.path.exists(manuscript_data_path)), ("Could not find manuscript data directory: " + manuscript_data_path)
+print("Using manuscript data directory:", manuscript_data_path)
 
 def use_thesaurus(entries: Dict[str, Recipe]) -> List[Recipe]:
   """
@@ -27,7 +28,7 @@ def use_thesaurus(entries: Dict[str, Recipe]) -> List[Recipe]:
   Output:
     entries: List[Recipe] -- same as above, but with the thesaurus corrections applied.
   """
-  if not os.path.exists(m_k_data_to_thesaurus):
+  if not os.path.exists(thesaurus_path):
     print('Thesaurus not found. Generating now.')
     os.system(f'python {cwd}/manuscript-object/thesaurus.py')
     print('Finished Generating Thesaurus')
@@ -36,7 +37,7 @@ def use_thesaurus(entries: Dict[str, Recipe]) -> List[Recipe]:
 
   for prop in properties:
     dct = {} # {verbatim_term: prefLabel_en}
-    df = pd.read_csv(f'{m_k_data_to_thesaurus}/{prop}.csv')
+    df = pd.read_csv(f'{thesaurus_path}/{prop}.csv')
 
     # manual_df = manual_corrections[manual_corrections['property'] == prop]
     # manual_dict = {} # verbatim_term, prefLabel_en pairs
@@ -108,7 +109,7 @@ def generate_complete_manuscript(apply_corrections=True) -> Dict[str, Recipe]:
   TODO: Instead of going version by version, consider going folio by folio. 
   """
   for version in versions: 
-    dir_path = f'{m_path}/ms-xml/{version}/'
+    dir_path = f'{manuscript_data_path}/ms-xml/{version}/'
     entry_dict = OrderedDict()
 
     for r, d, f in os.walk(dir_path):

update.py

@@ -1,12 +1,10 @@
-# Last Updated | 2020-08-21
+# Last Updated | 2020-08-24
 # Python Modules
 import os
 import sys
 import re
 from typing import List
 
-sys.path.insert(1, './manuscript-object/')
-
 # Third Party Modules
 import pandas as pd
 from datetime import datetime
@@ -25,7 +23,9 @@
               'profession': 'pro', 'sensory': 'sn', 'tool': 'tl', 'time': 'tmp', 'weapon': 'wp',
               'german': 'de', 'greek': 'el', 'italian': 'it', 'latin': 'la', 'occitan': 'oc', 'poitevin': 'po',}
 
-m_path = f'{os.getcwd()}'
+manuscript_data_path = os.path.dirname(os.getcwd()) + "/m-k-manuscript-data"
+assert(os.path.exists(manuscript_data_path)), ("Could not find manuscript data directory: " + manuscript_data_path)
+print("Using manuscript data directory:", manuscript_data_path)
 
 def update_metadata(manuscript: BnF) -> None:
   """
@@ -54,7 +54,7 @@ def update_metadata(manuscript: BnF) -> None:
   # remove entry column, since it only displays memory address
   df.drop(columns=['entry'], inplace=True)
 
-  df.to_csv(f'{m_path}/metadata/entry_metadata.csv', index=False)
+  df.to_csv(f'{manuscript_data_path}/metadata/entry_metadata.csv', index=False)
 
 def update_entries(manuscript: BnF) -> None:
   """
@@ -67,13 +67,13 @@ def update_entries(manuscript: BnF) -> None:
     None
   """
 
-  for path in [f'{m_path}/entries', f'{m_path}/entries/txt', f'{m_path}/entries/xml']:
+  for path in [f'{manuscript_data_path}/entries', f'{manuscript_data_path}/entries/txt', f'{manuscript_data_path}/entries/xml']:
     if not os.path.exists(path):
       os.mkdir(path)
 
   for version in versions:
-    txt_path = f'{m_path}/entries/txt/{version}'
-    xml_path = f'{m_path}/entries/xml/{version}'
+    txt_path = f'{manuscript_data_path}/entries/txt/{version}'
+    xml_path = f'{manuscript_data_path}/entries/xml/{version}'
 
     # If the entries/txt or xml directory does not exist, create it. Otherwise, clear the directory.
     for path in [txt_path, xml_path]:
@@ -121,7 +121,7 @@ def update_all_folios(manuscript: BnF) -> None:
         text = f'{text}\n\n{new_text}' if text else new_text
 
       # write file
-      f = open(f'{m_path}/allFolios/{folder}/all_{version}.{folder}', 'w')
+      f = open(f'{manuscript_data_path}/allFolios/{folder}/all_{version}.{folder}', 'w')
       f.write(text)
       f.close()
 
@@ -136,11 +136,11 @@ def update_ms(manuscript: BnF) -> None:
     None
   """
   for version in versions:
-    for r, d, f in os.walk(f'{m_path}/ms-xml/{version}'):
+    for r, d, f in os.walk(f'{manuscript_data_path}/ms-xml/{version}'):
       for filename in f: # iterate through /ms-xml/{version} folder
         # read xml file
         text = ''
-        filepath = f'{m_path}/ms-xml/{version}/{filename}'
+        filepath = f'{manuscript_data_path}/ms-xml/{version}/{filename}'
         with open(filepath, encoding="utf-8", errors="surrogateescape") as f:
           text = f.read()
 
@@ -188,7 +188,7 @@ def update():
   print('Updating allFolios')
   update_all_folios(manuscript)
 
-  update_time()
+  # update_time()
 
 if __name__ == "__main__":
   update()