setup.py: Adding install_requires.

digikey: Updating tests. Making sure the directory exist before writing facts.
Stop recursing into new searches when a search returns multiple hits. Let the frontends do that.

2 files changed, 42 insertions, 24 deletions

diff --git a/src/ee/digikey/__init__.py b/src/ee/digikey/__init__.py
index bd9a86b..64d8943 100644
--- a/src/ee/digikey/__init__.py
+++ b/src/ee/digikey/__init__.py
@@ -66,11 +66,11 @@ class Digikey(object):
 
 @total_ordering
 class DigikeyProduct(object):
-    def __init__(self, part_number, mpn, attributes, categories):
+    def __init__(self, part_number, mpn, attributes=None, categories=None):
         self.part_number = _clean(part_number)
         self.mpn = _clean(mpn)
-        self.attributes = attributes
-        self.categories = categories
+        self.attributes = attributes or []
+        self.categories = categories or []
         self.quantity_available = None
         self.description = None
 
@@ -152,10 +152,10 @@ class DigikeySearchResponse(object):
         self.count = count
         self.response_type = response_type
 
-        self.products = set()
+        self.products = list()
 
     def append(self, product):
-        self.products.add(product)
+        self.products.append(product)
 
 
 class DigikeyClient(object):
@@ -173,7 +173,7 @@ class DigikeyClient(object):
         # self.sess.mount('http://', adapter)
         # self.sess.mount('https://', adapter)
 
-    def req(self, url, params=None):
+    def _req(self, url, params=None):
         if not url.startswith("http"):
             url = "https://www.digikey.com" + url
         s = "" if not params else "?" + urllib.parse.urlencode(params)
@@ -221,24 +221,24 @@ class DigikeyClient(object):
         return None
 
     def _search_process_multiple_results(self, tree: html, res: DigikeySearchResponse):
-        product_ids = [e.get("content").strip().replace('sku:', '') for e in
-                       tree.xpath("//*[@itemprop='productid' and @content]")]
+        products = tree.xpath("//*[@itemtype='http://schema.org/Product']")
 
-        for product_id in product_ids:
-            tmp = self.search(product_id)
-            if isinstance(tmp, DigikeyProduct):
-                res.append(tmp)
-            else:
-                [res.append(p) for p in tmp.products]
+        for product in products:
+            part_number = _first(product.xpath("//*[@itemprop='productid' and @content]"))
+            mpn = _first(product.xpath("//*[@itemprop='name']"))
 
-        return len(product_ids)
+            if part_number is not None and mpn is not None:
+                res.append(DigikeyProduct(
+                    part_number.get("content").strip().replace('sku:', ''),
+                    mpn.text))
 
-    def search(self, query: str) -> DigikeySearchResponse:
-        page_size = 10
+        return len(products)
 
+    def search(self, query: str, page_size=10) -> DigikeySearchResponse:
         # http://www.digikey.com/products/en?x=0&y=0&lang=en&site=us&keywords=553-2320-1-ND
+        params = {'lang': 'en', 'site': 'us', 'keywords': query, 'pageSize': str(page_size), 'x': 0, 'y': 0}
         params = {'lang': 'en', 'site': 'us', 'keywords': query, 'pageSize': str(page_size)}
-        page = self.req("https://www.digikey.com/products/en", params=params)
+        page = self._req("https://www.digikey.com/products/en", params=params)
         # print("page: ")
         # print(page.content)
 
diff --git a/src/ee/tools/digikey_download_facts.py b/src/ee/tools/digikey_download_facts.py
index a1e242b..08a8029 100644
--- a/src/ee/tools/digikey_download_facts.py
+++ b/src/ee/tools/digikey_download_facts.py
@@ -37,9 +37,16 @@ def mpn_to_path(mpn):
     return "{}/{}.yaml".format(args.out, mpn)
 
 
-def on_product(p: DigikeyProduct):
-    y = p.to_yaml()
-    with open(mpn_to_path(p.mpn), "w") as f:
+def on_product(product: DigikeyProduct):
+    y = product.to_yaml()
+
+    filename = mpn_to_path(product.mpn)
+    dirname = os.path.dirname(filename)
+
+    if not os.path.isdir(dirname):
+        os.mkdir(dirname)
+
+    with open(filename, "w") as f:
         yaml.dump(y, f, encoding="utf-8", allow_unicode=True)
 
 
@@ -54,11 +61,22 @@ for p in args.parts:
 
     if response.response_type == SearchResponseTypes.SINGLE:
         p = response.products[0]
-        print(color("Found {}".format(p.mpn), "white"))
+        print(color("Direct match {}".format(p.mpn), "white"))
         on_product(p)
     elif response.response_type == SearchResponseTypes.MANY:
-        for k, g in groupby(sorted(response.products), lambda p: p.mpn):
-            print(color("Found {}".format(k), "white"))
+        hits = list(groupby(sorted(response.products), lambda p: p.mpn))
+
+        if len(hits) == 1:
+            (mpn, products) = hits[0]
+            products = list(products)
+
+            if len(products) == 1:
+                print(color("Got many results, but they all point to the same part: {}".format(mpn), "white"))
+                on_product(products[0])
+                continue
+
+        for k, g in hits:
+            print(color("Got many results with many parts: {}: {}".format(k, list(g)), "white"))
             on_product(list(g)[0])
     elif response.response_type == SearchResponseTypes.TOO_MANY:
         print(color("Too many results ({}), select a category first".format(response.count), 'red'))