Added chapter 14 and 16 code, formatting changes

Author: REMitchell

v2/Chapter04_CrawlingModels.ipynb

@@ -1022,7 +1022,7 @@
     "                return\n",
     "            title = self.safeGet(bs, site.titleTag)\n",
     "            body = self.safeGet(bs, site.bodyTag)\n",
-    "            if title != \"\" and body != \"\":\n",
+    "            if title != '' and body != '':\n",
     "                content = Content(topic, title, body, url)\n",
     "                content.print()\n",
     "\n",
@@ -1042,7 +1042,7 @@
     "    sites.append(Website(row[0], row[1], row[2],\n",
     "                         row[3], row[4], row[5], row[6], row[7]))\n",
     "\n",
-    "topics = [\"python\", \"data science\"]\n",
+    "topics = ['python', 'data science']\n",
     "for topic in topics:\n",
     "    print(\"GETTING INFO ABOUT: \" + topic)\n",
     "    for targetSite in sites:\n",
@@ -1246,24 +1246,24 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 1,
    "metadata": {},
    "outputs": [],
    "source": [
     "class Website:\n",
     "    \"\"\"Common base class for all articles/pages\"\"\"\n",
     "\n",
-    "    def __init__(self, type, name, url, searchUrl, resultListing, resultUrl, absoluteUrl, titleTag, bodyTag):\n",
+    "    def __init__(self, name, url, titleTag, bodyTag):\n",
     "        self.name = name\n",
     "        self.url = url\n",
     "        self.titleTag = titleTag\n",
     "        self.bodyTag = bodyTag\n",
-    "        self.pageType = pageType"
+    "        "
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": null,
+   "execution_count": 2,
    "metadata": {},
    "outputs": [],
    "source": [
@@ -1284,6 +1284,22 @@
     "        self.bodyTag = bodyTag\n",
     "        self.dateTag = dateTag"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "\n",
+    "\n",
+    "def parsePage(url):\n",
+    "    \n",
+    "    if '/ideas/' in url:\n",
+    "        \n",
+    "\n",
+    "oreilly = Website('O\\'Reilly', 'https://oreilly.com', 'h1' '')        "
+   ]
   }
  ],
  "metadata": {