o Sorting products before downloading to be consistent across runs.

o Storing the URL for each product.

2 files changed, 22 insertions, 9 deletions

diff --git a/src/ee/digikey/__init__.py b/src/ee/digikey/__init__.py
index 934c0d2..957aa40 100644
--- a/src/ee/digikey/__init__.py
+++ b/src/ee/digikey/__init__.py
@@ -71,9 +71,10 @@ class Digikey(object):
 
 @total_ordering
 class DigikeyProduct(object):
-    def __init__(self, part_number, mpn, attributes=None, categories=None):
+    def __init__(self, part_number, mpn, url, attributes=None, categories=None):
         self.part_number = _clean(part_number)
         self.mpn = _clean(mpn)
+        self.url = url
         self.attributes = attributes or []
         self.categories = categories or []
         self.quantity_available = None
@@ -95,9 +96,14 @@ class DigikeyProduct(object):
         return next((a for a in self.attributes if a.attribute_type.id == _id), None)
 
     def to_ini(self, c: configparser.ConfigParser):
+        def set(cfg, key, value):
+            if value:
+                cfg[key] = value
+
         c["overview"] = {};
         overview = c["overview"]
         overview["part_number"] = self.part_number
+        set(overview, "url", self.url)
         if self.mpn:
             overview["mpn"] = self.mpn
         c["attributes"] = {};
@@ -113,13 +119,19 @@ class DigikeyProduct(object):
 
     @staticmethod
     def from_ini(digikey, c):
+        def get(c, key):
+            try:
+              return c[key]
+            except KeyError:
+              return None
+
         overview = c["overview"]
         attributes = []
         for k, value in c.items("attributes"):
             (type_id, label) = DigikeyProduct.from_ini_r.match(k).groups()
             a_type = digikey.get_attribute_type(int(type_id), label)
             attributes.append(DigikeyAttributeValue(value, a_type))
-        return DigikeyProduct(overview["part_number"], overview["mpn"], attributes)
+        return DigikeyProduct(overview["part_number"], overview["mpn"], get(overview, "url"), attributes)
 
 
 class DigikeyAttributeType(object):
@@ -211,6 +223,8 @@ class DigikeyClient(object):
         attributes = []
         categories = []
 
+        url = _first((link.get("href") for link in tree.xpath("/html/head/link[@rel='canonical' and @href]")))
+
         part_number = mpn = None
         for n in tree.xpath("//*[@itemprop='productID' and @content]"):
             part_number = n.get("content")
@@ -241,7 +255,7 @@ class DigikeyClient(object):
                 attributes.append(DigikeyAttributeValue(value, a_type))
 
         if part_number and mpn:
-            p = DigikeyProduct(part_number, mpn, attributes, categories)
+            p = DigikeyProduct(part_number, mpn, url, attributes, categories)
             for n in tree.xpath("//*[@itemprop='description']"):
                 p.description = _to_string(n)
             return p
@@ -252,13 +266,14 @@ class DigikeyClient(object):
         products = tree.xpath("//*[@itemtype='http://schema.org/Product']")
 
         for product in products:
+            url = _first((a.get("href") for a in product.xpath(".//*[@class='tr-image']//a[@href]")))
             part_number = _first(product.xpath(".//*[@itemprop='productid' and @content]"))
             mpn = _first(product.xpath(".//*[@itemprop='name']"))
 
             if part_number is not None and mpn is not None:
                 res.append(DigikeyProduct(
                     part_number.get("content").strip().replace('sku:', ''),
-                    mpn.text))
+                    mpn.text, url))
 
         return len(products)
 
diff --git a/src/ee/tools/digikey_download_facts.py b/src/ee/tools/digikey_download_facts.py
index 9842c40..cc75f74 100644
--- a/src/ee/tools/digikey_download_facts.py
+++ b/src/ee/tools/digikey_download_facts.py
@@ -97,14 +97,12 @@ for q in queries:
 
         if len(viable_products):
             # Pick the first one, should be as good as any
-            part_number = viable_products[0].part_number
+            part_number = sorted(viable_products, key=lambda p: p.part_number)[0].part_number
 
             log.info("Got many hits for term '{}', will use {} for downloading attributes.".format(q.query, part_number))
             todos.append(part_number)
         else:
-            for k, g in hits:
-                log.info("Got many results with many parts: {}: {}".format(k, list(g)))
-                on_product(list(g)[0])
+            log.warn("Got many results: {}".format(", ".join([p.part_number for p in response.products])))
     elif response.response_type == SearchResponseTypes.TOO_MANY:
         log.warn("Too many results ({}), select a category first".format(response.count))
     elif response.response_type == SearchResponseTypes.NO_MATCHES:
@@ -114,7 +112,7 @@ for q in queries:
         response = client.search(part_number)
 
         if response.response_type == SearchResponseTypes.SINGLE:
-            p = response.products[0]
+            p = sorted(response.products, key=lambda p: p.part_number)[0]
             log.info("Downloaded {}".format(p.mpn))
             on_product(p)
         else: