senticr_piper_2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import os,json\n",
    "import sys\n",
    "import augur\n",
    "import sqlalchemy as s\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from scipy import stats\n",
    "import matplotlib.pyplot as plot\n",
    "import matplotlib.dates as mdates\n",
    "#!{sys.executable} -m pip install sklearn\n",
    "#!{sys.executable} -m pip install xlrd\n",
    "#!{sys.executable} -m pip install imblearn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/home/keanu/temp/augur_push/augur5/augur/SentiCR\n",
      "/home/keanu/temp/augur_push/augur5/augur/SentiCR/SentiCR\n",
      "Reading data from oracle..\n",
      "Training classifier model..\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/keanu/anaconda3/envs/augur/lib/python3.6/site-packages/sklearn/utils/deprecation.py:77: DeprecationWarning: Function _ratio_float is deprecated; Use a float for 'ratio' is deprecated from version 0.2. The support will be removed in 0.4. Use a dict, str, or a callable instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    }
   ],
   "source": [
    "if(\"notebooks\" in os.getcwd()):\n",
    "    os.chdir(\"../SentiCR\")\n",
    "    from SentiCR.SentiCR import SentiCR\n",
    "\n",
    "sentiment_analyzer = SentiCR()\n",
    "\n",
    "if(\"notebooks\" not in os.getcwd()):\n",
    "    os.chdir(\"../../notebooks\")\n",
    "\n",
    "augurApp = augur.Application('../augur.config.json')\n",
    "\n",
    "connect = augurApp.ghtorrentplus()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['github_issues', 'github_issues_2', 'github_issues_repo_jobs', 'github_issues_sentiment_scores', 'github_issues_sentiment_scores_senticr', 'github_pull_request_repo_jobs', 'github_pull_requests', 'github_pull_requests_2', 'github_pull_requests_sentiment_scores', 'github_pull_requests_sentiment_scores_2', 'github_pull_requests_sentiment_scores_senticr', 'issue_response_time', 'issue_response_time2', 'issue_response_time_3', 'mail_lists', 'mail_lists_sentiment_scores', 'mail_lists_sentiment_scores_senticr', 'mailing_list_jobs']\n",
      "         project\n",
      "0  gluster-users\n",
      "1       announce\n",
      "2  gluster-devel\n",
      "3  gluster-infra\n",
      "4     aalldp-dev\n"
     ]
    }
   ],
   "source": [
    "table_names = s.inspect(connect.db).get_table_names()\n",
    "print(table_names)\n",
    "\n",
    "if(\"mailing_list_jobs\" in table_names):\n",
    "    lists_createdSQL = s.sql.text(\"\"\"SELECT project FROM mailing_list_jobs\"\"\")\n",
    "    df1 = pd.read_sql(lists_createdSQL, connect.db)\n",
    "    print(df1)\n",
    "    val = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "gluster-users Mailing List\n",
      "Here!!!!\n",
      "\n",
      "\n",
      "\n",
      "announce Mailing List\n",
      "Here!!!!\n",
      "\n",
      "\n",
      "\n",
      "gluster-devel Mailing List\n",
      "Here!!!!\n",
      "\n",
      "\n",
      "\n",
      "gluster-infra Mailing List\n",
      "Here!!!!\n",
      "\n",
      "\n",
      "\n",
      "aalldp-dev Mailing List\n",
      "Here!!!!\n",
      "\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "col = 'score','sentiment'\n",
    "columns1 = 'augurmsgID', 'backend_name','project','mailing_list','category',\\\n",
    "           'message_part','message_parts_tot', 'subject','date',\\\n",
    "           'message_from','message_id','message_text'\n",
    "df3 = pd.DataFrame(columns = col)\n",
    "df_list = pd.DataFrame(columns = columns1)\n",
    "for i in df1['project'].values:\n",
    "    print(i,\"Mailing List\")\n",
    "    SQL = s.sql.text(\"\"\"SELECT * \n",
    "    FROM mail_lists WHERE mailing_list = \"\"\" +  \"'\" + i + \"'\")\n",
    "    df2 = pd.read_sql(SQL, connect.db)\n",
    "    df_list = df_list.append(df2)\n",
    "    #print(df2)\n",
    "    grouped = df2.groupby('message_id').groups\n",
    "    print(\"Here!!!!\")\n",
    "    #print(\"\\n\\n\",\"-\"*70,\"\\n\\n\")\n",
    "    for group in grouped:\n",
    "        parts = 0\n",
    "        numb = len(df2.loc[df2['message_id'] == group]['message_parts_tot'].tolist())\n",
    "        message = (df2.loc[df2['message_id'] == group]['message_text']).tolist()\n",
    "        message_text = ''.join(message)\n",
    "        #print(message_text)\n",
    "        score=sentiment_analyzer.get_sentiment_polarity(message_text)\n",
    "        score = int(score[0])\n",
    "        compound = parts = 0\n",
    "        sentiment = \"Positive\"\n",
    "        if(score == -1):\n",
    "            sentiment = \"Negative\"\n",
    "        #print(message)\n",
    "        for i in range(numb):\n",
    "            li = [ [score, sentiment] ]\n",
    "            df_temp = pd.DataFrame(li,columns = col)\n",
    "            #print(df_temp['score'])\n",
    "            df3 = df3.append(df_temp)\n",
    "            #print(df3)\n",
    "    print(\"\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  augurmsgID backend_name                                            project  \\\n",
      "0          1    Pipermail  https://lists.gluster.org/pipermail/gluster-us...   \n",
      "1          2    Pipermail  https://lists.gluster.org/pipermail/gluster-us...   \n",
      "2          3    Pipermail  https://lists.gluster.org/pipermail/gluster-us...   \n",
      "3          4    Pipermail  https://lists.gluster.org/pipermail/gluster-us...   \n",
      "4          5    Pipermail  https://lists.gluster.org/pipermail/gluster-us...   \n",
      "\n",
      "    mailing_list category message_part message_parts_tot  \\\n",
      "0  gluster-users  message            1                 2   \n",
      "1  gluster-users  message            2                 2   \n",
      "2  gluster-users  message            1                 2   \n",
      "3  gluster-users  message            2                 2   \n",
      "4  gluster-users  message            1                 3   \n",
      "\n",
      "                                             subject                date  \\\n",
      "0    [Gluster-users] glusterfs 3.1.0 disable-io-mode 2010-12-02 02:20:01   \n",
      "1    [Gluster-users] glusterfs 3.1.0 disable-io-mode 2010-12-02 02:20:01   \n",
      "2  [Gluster-users] Who's using Fedora in producti... 2010-12-02 12:34:11   \n",
      "3  [Gluster-users] Who's using Fedora in producti... 2010-12-02 12:34:11   \n",
      "4  [Gluster-users] RDMA Problems with GlusterFS 3... 2010-12-01 00:00:20   \n",
      "\n",
      "                                        message_from  \\\n",
      "0             kkakas at yahoo.gr (Kakaletris Kostas)   \n",
      "1             kkakas at yahoo.gr (Kakaletris Kostas)   \n",
      "2  mrogers at valuecommerce.co.jp (Mark \"Naoki\" R...   \n",
      "3  mrogers at valuecommerce.co.jp (Mark \"Naoki\" R...   \n",
      "4           stout.jeremy at gmail.com (Jeremy Stout)   \n",
      "\n",
      "                                          message_id  \\\n",
      "0                        <4CF6E631.3090805@yahoo.gr>   \n",
      "1                        <4CF6E631.3090805@yahoo.gr>   \n",
      "2               <4CF713B3.20904@valuecommerce.co.jp>   \n",
      "3               <4CF713B3.20904@valuecommerce.co.jp>   \n",
      "4  <AANLkTikd0VkCTSVWNDRBrRQ1_6HWbgFWJJ7ACGqzDFhT...   \n",
      "\n",
      "                                        message_text score sentiment  \n",
      "0   From what i remember i tried and the followin...    -1  Negative  \n",
      "1  sterfs -o direct-io-mode=off\\nmount -t gluster...    -1  Negative  \n",
      "2  Hi James,\\n\\nI'm using 3.1.1 on six bricks in ...    -1  Negative  \n",
      "3  n fedora12/13/14. I build the RPMs from source...    -1  Negative  \n",
      "4  As an update to my situation, I think I have G...    -1  Negative  \n",
      "(103973, 14)\n"
     ]
    }
   ],
   "source": [
    "df3 = df3.reset_index(drop=True)\n",
    "df_list = df_list.reset_index(drop=True)\n",
    "combine = (df_list.join(df3))\n",
    "\n",
    "print(combine.head())\n",
    "print(combine.shape)\n",
    "combine.to_sql(name='mail_lists_sentiment_scores_senticr',con=connect.db,\\\n",
    "               if_exists='replace',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10,)\n",
      "(10,)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(208,)\n",
      "(208,)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(45298,)\n",
      "(45298,)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7827,)\n",
      "(7827,)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(50630,)\n",
      "(50630,)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "grouped = combine.groupby('mailing_list').groups\n",
    "dates = []\n",
    "mail_list = None\n",
    "j = 0\n",
    "for group in grouped:\n",
    "    dates = (combine.loc[combine['mailing_list'] == group]['date']).tolist()\n",
    "    values = []\n",
    "    for val in dates:\n",
    "        #print(val)\n",
    "        values.append((combine.loc[combine['date'] == val]['score']).tolist()[0])\n",
    "    #values = (combine.loc[combine['mailing_list'] == group]['score']).unique().tolist()\n",
    "    x = np.array(dates)\n",
    "    #dates = np.unique(x)\n",
    "    #print(dates)\n",
    "    dates1 = [r.to_pydatetime() for r in x]\n",
    "    fin_dates = mdates.date2num(dates1)\n",
    "    fin_values = np.asarray(values)\n",
    "    print(fin_dates.shape)\n",
    "    print(fin_values.shape)\n",
    "    plot.plot_date(fin_dates, fin_values)\n",
    "    #plot.plot(fin_dates,fin_values)\n",
    "    plot.gcf().autofmt_xdate()\n",
    "    plot.title(\"Mailing lst: \"+ group)\n",
    "    plot.ylabel('Score')\n",
    "    plot.xlabel('Date')\n",
    "    #print(fin_dates)\n",
    "    slope, intercept, r_value, p_value, std_err = stats.linregress(fin_dates,fin_values)\n",
    "    line = slope*fin_dates+intercept\n",
    "    plot.plot(fin_dates,fin_values,'o', fin_dates, line)\n",
    "    #plot.savefig('foo' + str(x) + '.png')\n",
    "    #x+=1\n",
    "    plot.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (augur)",
   "language": "python",
   "name": "augur"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}