vphill

def processRecord( RecordClass, row ):

    language_dict = {
        "en": "eng",
        "fr": "fre", 
        "eng": "eng"
    }

    record = RecordClass("mphillips")

vphill

{
  "001": "Anderson County", 
  "003": "Andrews County", 
  "005": "Angelina County", 
  "007": "Aransas County", 
  "009": "Archer County", 
  "011": "Armstrong County", 
  "013": "Atascosa County", 
  "015": "Austin County", 
  "017": "Bailey County", 

vphill

for dirname in *
do 
    #echo moving into $dirname
    cd "$dirname"
    for pdf in *.pdf
    do 
        #echo extract pages in $pdf 
        pagesfrompdf=`pdfinfo "$pdf" | grep -a "Pages:" | awk '{print $2}'`
	echo $pagesfrompdf
	pages=$(($(tr -d '\r' <<< $pagesfrompdf) - 1))

vphill

for i in *
  do
       echo "processing $i ======="
       echo "creating pdfs for"
       find "$i" -maxdepth 1 -name "*.tif" | xargs -i -n 1 -P 6 convert -density 200x200 -compress jpeg -quality 80 "{}" "{}".pdf
       echo "compressing pdfs"
       find "$i" -maxdepth 1 -name "*.tif.pdf" | xargs -i -n 1 -P 5 gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/screen -dNOPAUSE -dQUIET -dBATCH -sOutputFile={}_out.pdf {}
       cd "$i"
       echo "concatinating pdfs"
       pdftk *_out.pdf cat output temp.pdf

vphill

   for i in *
   do
       echo "creating pdfs"
       find "$i" -maxdepth 1 -name "*.jpg" | xargs -i -n 1 -P 6 convert -density 200x200 -compress jpeg -quality 80 "{}" "{}".pdf
       echo "compressing pdfs"
       find "$i" -maxdepth 1 -name "*.jpg.pdf" | xargs -i -n 1 -P 5 gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 -dPDFSETTINGS=/screen -dNOPAUSE -dQUIET -dBATCH -sOutputFile={}_out.pdf {}
       cd "$i"
       echo "concatinating pdfs"
       pdftk *_out.pdf cat output temp.pdf
       echo "linerizing pdf"

vphill

# coding=UTF-8

import sys
import re
import json
from dateutil import parser
from bs4 import BeautifulSoup

if len(sys.argv) != 2:
    print "usage: parse_bill_html.py <bill_html>"

vphill

<form method="GET" action="http://texashistory.unt.edu/explore/partners/TSU/browse/">
    <fieldset style="display: inline-block">
        <legend>Tarleton State University - Portal to Texas History</legend>
        <input type="text" name="q" size="80"/>
        <input type="hidden" name="t" value="fulltext" />
        <input type="submit" value="submit" />
    </fieldset>
</form>

vphill

 gs -dUseCIEColor -dFAPIDEBUG -sDEVICE=jpeg -o out-%d.jpg -sFONTPATH=/Users/mphillips/Desktop/auditors/msfonts-master/fonts/ -sFontMap="/CenturyGothic,Bold (/Users/mphillips/Desktop/auditors/msfonts-master/fonts/gothicb.ttf)  ;" -r400x400 13-704.pdf

vphill

{
    "placenames": [
        "Afghanistan",
        "Africa",
        "Albania",
        "Algeria",
        "Algeria - Alger Department - Algiers",
        "America",
        "American Samoa",
        "Angola",

vphill

{
    "placenames": {
      "Afghanistan": 196,
      "Africa": 379,
      "Albania": 4,
      "Algeria": 16,
      "Algeria - Alger Department - Algiers": 1,
      "American Samoa": 8,
      "Angola": 6,
      "Antarctica": 29,