single file conversion works

array2xls/gui.py

@@ -1,8 +1,11 @@
+import os
 import tkinter as tk
 import tkinter.ttk as ttk
 from tkinter import filedialog
 
+
 import validators
+import inout
 
 APP_STATE_1 = 'no valid files selected'
 APP_STATE_2 = 'no valid fields selected'
@@ -176,15 +179,10 @@ class Application(tk.Frame):
         state = self._state
         if state not in (APP_STATE_3, APP_STATE_4):
             return self.reset()
-        i = len(self.validated_files)
         if state == APP_STATE_3:
-            text = 'Converting %d data files to %d excel files ...' % (i, i)
+            text = self._convert_to_single_files()
-            self.status_panel.set_text(text)
-            self._convert_to_single_files()
         else:
-            text = 'Combining %d data files into one excel file ...' % i
+            text = self._combine_data_files()
-            self.status_panel.set_text(text)
-            self._combine_data_files()
         self.status_panel.set_text(text + ' Done')
 
     @property
@@ -200,7 +198,17 @@ class Application(tk.Frame):
             return APP_STATE_2
 
     def _convert_to_single_files(self):
-        pass
+        total = len(self.validated_files)
+        text = 'Converting data file %d of %d to single excel file ...'
+        for i, data_file in enumerate(self.validated_files):
+            self.status_panel.set_text(text % (i+1, total))
+            all_fields = data_file.validator.id_fields + data_file.validator.data_fields
+            data_frame = inout.read_data_file(data_file, all_fields)
+            data_frame = data_frame.set_index(data_file.validator.id_fields)
+            name, ext = os.path.splitext(data_file.path)
+            path = name + ext.replace('.', '_') + '.xlsx'
+            inout.write_excel_file(path, data_frame)
+        return text % (i+1, total)
 
     def _combine_data_files(self):
         pass

array2xls/inout.py

@@ -0,0 +1,14 @@
+import os
+import pandas
+
+import validators
+
+def read_data_file(data_file, usecols):
+    with open(data_file.path, 'r', encoding='UTF-8') as file_handle:
+        return pandas.read_csv(
+            file_handle, sep='\t', decimal=data_file.separator, header=0,
+            index_col=False, skiprows=data_file.skip, usecols=usecols)
+
+
+def write_excel_file(path, data_frame):
+    data_frame.to_excel(path)

array2xls/validators.py

@@ -6,43 +6,44 @@ class ValidationError(ValueError):
     pass
 
 Validator = namedtuple('Validator', 'type extension validate id_fields data_fields defaults')
-DataFile = namedtuple('DataFile', 'path separator')
+DataFile = namedtuple('DataFile', 'path separator skip validator')
 
 def validate_stx(lines):
-    iterator = iter(lines)
+    iterator = enumerate(lines)
-    line = next(iterator)
+    _, line = next(iterator)
     if not line.startswith('Report_Format:\t2'):
         raise ValidationError('1 Unsupported File;' + line)
-    for line in iterator:
+    for i, line in iterator:
         if line.startswith('Probe_Name\t'):
             break
     else:
         raise ValidationError('1 Unsupported File')
     try:
-        line = next(iterator)
+        _, line = next(iterator)
         _, numeric_data = line.split('\t', 1)
     except (ValueError, StopIteration):
         raise ValidationError('No Data Present')
     separator = ',' if numeric_data.count(',') > numeric_data.count('.') else '.'
-    return separator
+    return separator, i
 
 
 def validate_dat(lines):
-    iterator = iter(lines)
+    iterator = enumerate(lines)
-    if not next(iterator).startswith('Report_Format:\t2'):
+    _, line = next(iterator)
+    if not line.startswith('Report_Format:\t2'):
         raise ValidationError('Unsupported File')
-    for line in iterator:
+    for i, line in iterator:
         if line.startswith('Dot_Number\t'):
             break
     else:
         raise ValidationError('Unsupported File')
     try:
-        line = next(iterator)
+        _, line = next(iterator)
         _, _, _, numeric_data = line.split('\t', 3)
     except (ValueError, StopIteration):
         raise ValidationError('No Data Present')
     separator = ',' if numeric_data.count(',') > numeric_data.count('.') else '.'
-    return separator
+    return separator, i
 
 
 def validate_csv(lines):
@@ -55,7 +56,7 @@ def validate_csv(lines):
     except (ValueError, StopIteration):
         raise ValidationError('No Data Present')
     separator = ',' if numeric_data.count(',') > numeric_data.count('.') else '.'
-    return separator
+    return separator, 0
 
 
 stx_validator = Validator(
@@ -96,7 +97,7 @@ def validate_files(unvalidated, selected_validator):
     for file_path in unvalidated:
         try:
             with open(file_path, mode='r', encoding='utf-8') as file_handle:
-                separator = selected_validator.validate(file_handle)
+                separator, skip = selected_validator.validate(file_handle)
-                yield DataFile(file_path, separator)
+                yield DataFile(file_path, separator, skip, selected_validator)
         except (IOError, UnicodeError, ValidationError) as e:
             print(e)