diff --git a/tools/pdf2html.cgi b/tools/pdf2html.cgi index b696c21..a0389bd 100755 --- a/tools/pdf2html.cgi +++ b/tools/pdf2html.cgi @@ -15,10 +15,10 @@ # $ cp pdfminer/tools/pdf2html.cgi $CGIDIR # -import sys +import sys, os, os.path, re, time +import cgi, logging, traceback, random # comment out at this at runtime. #import cgitb; cgitb.enable() -import os, os.path, re, cgi, time, random, codecs, logging, traceback import pdfminer from pdfminer.pdfinterp import PDFResourceManager, process_pdf from pdfminer.converter import HTMLConverter, TextConverter @@ -71,26 +71,27 @@ def convert(outfp, infp, path, codec='utf-8', return -## PDF2HTMLApp +## WebApp ## -class PDF2HTMLApp(object): +class WebApp(object): - APPURL = '/convert' - MAXFILESIZE = 5000000 - MAXPAGES = 10 + TITLE = 'pdf2html demo' + APPPATH = '/' # absolute URL path to this application. + MAXFILESIZE = 5000000 # set to zero if unlimited. + MAXPAGES = 10 # set to zero if unlimited. - def __init__(self, outfp=sys.stdout, codec='utf-8'): + def __init__(self, infp=sys.stdin, outfp=sys.stdout, codec='utf-8'): self.outfp = outfp self.codec = codec self.remote_addr = os.environ.get('REMOTE_ADDR') self.path_info = os.environ.get('PATH_INFO') - self.method = os.environ.get('REQUEST_METHOD', 'GET') + self.method = os.environ.get('REQUEST_METHOD', 'GET').upper() self.server = os.environ.get('SERVER_SOFTWARE', '') self.logpath = os.environ.get('LOG_PATH', './var/log') self.tmpdir = os.environ.get('TEMP', './var/') - self.debug = os.environ.get('DEBUG') self.content_type = 'text/html; charset=%s' % codec self.cur_time = time.time() + self.form = cgi.FieldStorage(infp) return def put(self, *args): @@ -127,9 +128,9 @@ class PDF2HTMLApp(object): def coverpage(self): self.put( - '