Support xlsx files in initdb-aprendizations

Miguel Barão
1 parent 0c0035fd
Showing 2 changed files with 38 additions and 6 deletions Show diff stats
aprendizations/initdb.py
setup.py
@@ -5,8 +5,9 @@ Initializes or updates database
 '''
  
 # python standard libraries
-import csv
 import argparse
+import csv
+from pandas import read_excel
 import re
 from string import capwords
  
@@ -32,11 +33,11 @@ def parse_commandline_arguments():
                     'command line. If the database does not exist, a new one'
                     ' is created.')
  
-    argparser.add_argument('csvfile',
+    argparser.add_argument('files',
                            nargs='*',
                            type=str,
                            default='',
-                           help='CSV file to import (SIIUE)')
+                           help='CSV or Excel files to import (SIIUE)')
  
     argparser.add_argument('--db',
                            default='students.db',
@@ -72,6 +73,30 @@ def parse_commandline_arguments():
  
  
 # ===========================================================================
+def get_students_from_xlsx(filename):
+    excel_settings = {
+        'skiprows': 3,
+        'converters': { "Número" : str }
+    }
+    students = []
+    try:
+        file = read_excel(filename, **excel_settings)
+        names = file['Aluno']
+        nums = file['Número']
+        students = [{
+            'uid': num,
+            'name': capwords(re.sub(r'\(.*\)', '', name).strip())
+            } for num, name in zip(nums, names)]
+    except FileNotFoundError as e:
+        print(f'!!! File {filename} not found !!!')
+    except ValueError as e:
+        print(f'!!! File {filename} has wrong format !!!')
+    except Exception as e:
+        raise e
+
+    return students
+
+# ===========================================================================
 def get_students_from_csv(filename):
     '''Reads CSV file with enrolled students in SIIUE format.
     SIIUE names can have suffixes like "(TE)" and are sometimes capitalized.
@@ -138,8 +163,13 @@ def main():
     # --- make list of students to insert/update
     students = []
  
-    for csvfile in args.csvfile:
-        students += get_students_from_csv(csvfile)
+    for file in args.files:
+        if file.endswith('.csv'):
+            students += get_students_from_csv(file)
+        elif file.endswith('.xlsx'):
+            students += get_students_from_xlsx(file)
+        else:
+            print(f'!!! Ignoring file {file} !!!')
  
     if args.admin:
         students.append({'uid': '0', 'name': 'Admin'})
@@ -26,7 +26,9 @@ setup(
         'pygments>=2.14',
         'sqlalchemy>=2.0.0',
         'bcrypt>=4.0.1',
-        'networkx>=3.0'
+        'networkx>=3.0',
+        'pandas>=2.3',
+        'openpyxl'
         ],
     entry_points={
         'console_scripts': [
...	...	@@ -5,8 +5,9 @@ Initializes or updates database
5	5	'''
6	6
7	7	# python standard libraries
8		-import csv
9	8	import argparse
	9	+import csv
	10	+from pandas import read_excel
10	11	import re
11	12	from string import capwords
12	13
...	...	@@ -32,11 +33,11 @@ def parse_commandline_arguments():
32	33	'command line. If the database does not exist, a new one'
33	34	' is created.')
34	35
35		- argparser.add_argument('csvfile',
	36	+ argparser.add_argument('files',
36	37	nargs='*',
37	38	type=str,
38	39	default='',
39		- help='CSV file to import (SIIUE)')
	40	+ help='CSV or Excel files to import (SIIUE)')
40	41
41	42	argparser.add_argument('--db',
42	43	default='students.db',
...	...	@@ -72,6 +73,30 @@ def parse_commandline_arguments():
72	73
73	74
74	75	# ===========================================================================
	76	+def get_students_from_xlsx(filename):
	77	+ excel_settings = {
	78	+ 'skiprows': 3,
	79	+ 'converters': { "Número" : str }
	80	+ }
	81	+ students = []
	82	+ try:
	83	+ file = read_excel(filename, **excel_settings)
	84	+ names = file['Aluno']
	85	+ nums = file['Número']
	86	+ students = [{
	87	+ 'uid': num,
	88	+ 'name': capwords(re.sub(r'\(.*\)', '', name).strip())
	89	+ } for num, name in zip(nums, names)]
	90	+ except FileNotFoundError as e:
	91	+ print(f'!!! File {filename} not found !!!')
	92	+ except ValueError as e:
	93	+ print(f'!!! File {filename} has wrong format !!!')
	94	+ except Exception as e:
	95	+ raise e
	96	+
	97	+ return students
	98	+
	99	+# ===========================================================================
75	100	def get_students_from_csv(filename):
76	101	'''Reads CSV file with enrolled students in SIIUE format.
77	102	SIIUE names can have suffixes like "(TE)" and are sometimes capitalized.
...	...	@@ -138,8 +163,13 @@ def main():
138	163	# --- make list of students to insert/update
139	164	students = []
140	165
141		- for csvfile in args.csvfile:
142		- students += get_students_from_csv(csvfile)
	166	+ for file in args.files:
	167	+ if file.endswith('.csv'):
	168	+ students += get_students_from_csv(file)
	169	+ elif file.endswith('.xlsx'):
	170	+ students += get_students_from_xlsx(file)
	171	+ else:
	172	+ print(f'!!! Ignoring file {file} !!!')
143	173
144	174	if args.admin:
145	175	students.append({'uid': '0', 'name': 'Admin'})
...	...
...	...	@@ -26,7 +26,9 @@ setup(
26	26	'pygments>=2.14',
27	27	'sqlalchemy>=2.0.0',
28	28	'bcrypt>=4.0.1',
29		- 'networkx>=3.0'
	29	+ 'networkx>=3.0',
	30	+ 'pandas>=2.3',
	31	+ 'openpyxl'
30	32	],
31	33	entry_points={
32	34	'console_scripts': [
...	...