add some data files

data_clean.py

+import sys, os, time, re
+
+list_dir = "./"
+brand_dir = "brands/"
+
+def getKey(name, name_en):
+    if name != "":
+        return name
+    else:
+        return name_en
+
+def findBrand(argv):
+    listf = open(list_dir + "all_products.csv", "r")
+    brandf = open(brand_dir + "brands.csv", "r")
+
+    outputbrand = "brands_clean.csv"
+    outputproduct = "products_clean.csv"
+    obf = open(outputbrand, "w")
+    opf = open(outputproduct, "w")
+
+    brands = []
+    for line in brandf:
+        temp = line.split("; ")
+        brand = getKey(temp[0], temp[1])
+        if brand in brands:
+            continue
+        newbls = temp[:2] + [temp[2].split(" ")[0]] + temp[3:]
+        if len(newbls) != 4:
+            print(newbls)
+            return
+        obf.write("; ".join(newbls))
+        # print(brand)
+        brands.append(brand)
+    obf.close()
+
+    for line in listf:
+        fields = line.split("; ")
+        key = getKey(fields[0], fields[1])
+        if key is None:
+            print("No product key !!!!!!!!")
+            return
+        # print("%s-------->%s" % (key, brand))
+        brand = fields[-1].strip()
+        if brand not in brands:
+            print("There is not brand of %s in product %s" % (brand, key))
+            continue
+        newline = fields[:2]
+
+        try:
+            score = float(fields[2]) / 2
+        except:
+            score = 0
+        newline.append(str(score))
+
+        comment_num = ''
+        for ch in fields[3]:
+            if ch == "人":
+                break
+            try:
+                temp = int(ch)
+                comment_num += ch
+            except:
+                continue
+        newline.append(str(comment_num))
+
+        if fields[4] != "":
+            temp = fields[4].split("·")
+            if len(temp) == 2:
+                newline.append(re.sub(r'\s+','', temp[0]))
+                price = re.sub(r'\s+','', temp[1])
+                price = price.split('/')[0]
+                if price[0] != "¥":
+                    print(price)
+                    return
+                newline.append(price[1:])
+            else:
+                newline.append(re.sub(r'\s+','', temp[0]))
+                newline.append("")
+        else:
+            newline += ["", ""]
+
+        newline += fields[5:]
+        if len(newline) != 9:
+            print(newline)
+            return
+
+        opf.write("; ".join(newline))
+
+    # print("===================totally %s product not found=========================" % len(not_found))
+    # for nf in not_found:
+    #     print(nf)
+    opf.close()
+
+if __name__=='__main__':
+    findBrand(sys.argv)
\ No newline at end of file