Merge pull request hildogjr#430 from daglem/improved_datasheet_url_va…

…lidation

Improved validation of datasheet URLs

kicost/spreadsheet.py

@@ -35,6 +35,7 @@
 import xlsxwriter # XLSX file interpreter.
 from xlsxwriter.utility import xl_rowcol_to_cell, xl_range, xl_range_abs
 from babel import numbers # For currency presentation.
+from validators import url as validate_url # URL validator.
 
 # KiCost libraries.
 from . import __version__ # Version control by @xesscorp and collaborator.
@@ -56,9 +57,6 @@
 
 WORKBOOK = None
 
-# Regular expression to the link for one datasheet.
-DATASHEET_LINK_REGEX = re.compile('^(http(s)?:\/\/)?(www.)?[0-9a-z\.]+\/[0-9a-z\.\/\%\-\_]+(.pdf)?$', re.IGNORECASE)
-
 # Extra information characteristics of the components gotten in the page that will be displayed as comment in the 'cat#' column.
 EXTRA_INFO_DISPLAY = ['value', 'tolerance', 'footprint', 'power', 'current', 'voltage', 'frequency', 'temp_coeff', 'manf', 'size']
 
@@ -528,7 +526,7 @@ def get_ref_key(part):
                     except:
                         cell_format = wrk_formats['part_format']
                         pass
-                    if link and re.match(DATASHEET_LINK_REGEX, link):
+                    if link and validate_url(link):
                         # Just put the link if is a valid format.
                         wks.write_url(row, start_col + columns['manf#']['col'],
                              link, string=string, cell_format=cell_format)

setup.py

@@ -87,6 +87,7 @@ def run(self):
     'requests >= 2.18.4', # Scrape, API and web modules.
     'CurrencyConverter >= 0.13', # Used to convert price to a not available currency in one distributor.
     'babel >= 2.6', # For currency format by the language in the spreadsheet.
+    'validators >= 0.18.2', # For validation of datasheet URLs in the spreadsheet.
 #    'wxPython >= 4.0', # Graphical package/library needed to user guide.
 ]