From 733cfdc87036577b77ea83a7ee1c209c4bee44f7 Mon Sep 17 00:00:00 2001
From: Fuhu Xia <fxia@reisystems.com>
Date: Wed, 8 Nov 2023 13:08:11 -0500
Subject: [PATCH 1/2] add more date parser for IIS server

---
 ckanext/spatial/harvesters/waf.py | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/ckanext/spatial/harvesters/waf.py b/ckanext/spatial/harvesters/waf.py
index 5338d87e..31570a32 100644
--- a/ckanext/spatial/harvesters/waf.py
+++ b/ckanext/spatial/harvesters/waf.py
@@ -252,6 +252,15 @@ def fetch_stage(self, harvest_object):
            parse.Word(parse.alphas)
          , adjacent=False, joinString=' ').setResultsName('date')
          ) \
+         + parse.Optional(parse.Combine(
+            parse.Word(parse.alphas+',') +
+            parse.Word(parse.alphas) +
+            parse.Word(parse.nums+',') +
+            parse.Word(parse.nums) +
+            parse.Word(parse.nums+':') +
+            parse.Word(parse.alphas)
+        , adjacent=False, joinString=' ').setResultsName('date')
+        ) \
          + parse.Word(parse.nums).suppress() \
          + parse.Literal('<A HREF=').suppress() \
          + parse.quotedString.setParseAction(parse.removeQuotes).setResultsName('url')
@@ -328,6 +337,8 @@ def _extract_waf(content, base_url, scraper, results = None, depth=0):
             except Exception as e:
                 raise
                 date = None
+        if not date:
+            log.debug('failed to get date for %s', url)
         results.append((urljoin(base_url, record.url), date))
 
     return results

From 0beefa24a778804661483a0674fd777d0eefdff0 Mon Sep 17 00:00:00 2001
From: Fuhu Xia <fxia@reisystems.com>
Date: Thu, 9 Nov 2023 13:58:00 -0500
Subject: [PATCH 2/2] indentation

---
 ckanext/spatial/harvesters/waf.py | 24 ++++++++++++------------
 1 file changed, 12 insertions(+), 12 deletions(-)

diff --git a/ckanext/spatial/harvesters/waf.py b/ckanext/spatial/harvesters/waf.py
index 31570a32..c9299d5b 100644
--- a/ckanext/spatial/harvesters/waf.py
+++ b/ckanext/spatial/harvesters/waf.py
@@ -244,15 +244,15 @@ def fetch_stage(self, harvest_object):
         ,adjacent=False, joinString=' ').setResultsName('date')
         )
 
-iis =      parse.SkipTo("<br>").suppress() \
-         + parse.OneOrMore("<br>").suppress() \
-         + parse.Optional(parse.Combine(
-           parse.Word(parse.alphanums+'/') +
-           parse.Word(parse.alphanums+':') +
-           parse.Word(parse.alphas)
-         , adjacent=False, joinString=' ').setResultsName('date')
-         ) \
-         + parse.Optional(parse.Combine(
+iis     = parse.SkipTo("<br>").suppress() \
+        + parse.OneOrMore("<br>").suppress() \
+        + parse.Optional(parse.Combine(
+            parse.Word(parse.alphanums+'/') +
+            parse.Word(parse.alphanums+':') +
+            parse.Word(parse.alphas)
+        , adjacent=False, joinString=' ').setResultsName('date')
+        ) \
+        + parse.Optional(parse.Combine(
             parse.Word(parse.alphas+',') +
             parse.Word(parse.alphas) +
             parse.Word(parse.nums+',') +
@@ -261,9 +261,9 @@ def fetch_stage(self, harvest_object):
             parse.Word(parse.alphas)
         , adjacent=False, joinString=' ').setResultsName('date')
         ) \
-         + parse.Word(parse.nums).suppress() \
-         + parse.Literal('<A HREF=').suppress() \
-         + parse.quotedString.setParseAction(parse.removeQuotes).setResultsName('url')
+        + parse.Word(parse.nums).suppress() \
+        + parse.Literal('<A HREF=').suppress() \
+        + parse.quotedString.setParseAction(parse.removeQuotes).setResultsName('url')
 
 other = parse.SkipTo(parse.CaselessLiteral("<a href="), include=True).suppress() \
         + parse.quotedString.setParseAction(parse.removeQuotes).setResultsName('url')