webapp/app/helper.py

"""helper functions

"""
from bs4 import BeautifulSoup

from .dbinit import inverted_prince_bigram


def make_timeitem_from_filename(filename):
    """
    "anj_isa_i_1441_08_05a" -> "1441_08_05a"
    """
    trs_fname = filename.split('_')
    return "_".join(trs_fname[3:])

def plaintext_response(search, actecol, prince_bigram):
    """plain text search response 
    
    :search: the form's result
    :return: database query
    """
    query = [
        {
          "$match": {
            "xmlcontent": {
            "$regex": search,
            "$options" :'i' # case-insensitive
                          }
                    }
        },
        {
          '$group': {'_id': {'prince_name': '$prince_name', 
                            'house': '$house',
                            'date': '$date',
                            'prince_code': '$prince_code',
                            'filename': '$filename',
                            "place": "$place.name",
                            "diplo_state": "$diplo_state",
                            "diplo_type": "$diplo_type",
                            "xmlcontent": "$xmlcontent"
                            }
                    }
        }]
    
    results = list(actecol.aggregate(query))
    transformed_query = [pr['_id'] for pr in results]
    invert_prince_bigram = {val: key for key, val in prince_bigram.items()}

    for trs in transformed_query:
        trs['house'] = trs['house'].capitalize()
        trs['dateitem'] = make_timeitem_from_filename(trs['filename'])
        bigram, number = trs['prince_code'].split('_')
        long_prince_bigram = inverted_prince_bigram(bigram) + '_' + number
        trs['prince_url'] = prince_name=long_prince_bigram.capitalize()
        
        # find the word in the text
        cleantext = BeautifulSoup(trs["xmlcontent"], "lxml").text
        cleantext = cleantext.lower()
        searchtext = search.lower()
        begin = cleantext.find(searchtext)
        #end = cleantext.find(searchtext, begin, len(cleantext))
        strframe = 35
        begin_mark = begin - strframe
        if begin_mark < 0:
            begin_mark = 0
        end_mark = begin + len(searchtext)  + strframe
        trs['show_text'] = cleantext[begin_mark:end_mark]
        
    return transformed_query


#from flask import abort
#
#def find_one_or_404(collection, **kwargs):
#      """Find a single document or raise a 404.
#
#      This is like :meth:`~pymongo.collection.Collection.find_one`, but
#      rather than returning ``None``, cause a 404 Not Found HTTP status
#      on the request.
#
#      .. code-block:: python
#
#          usercollection = mydb['usercollection']
#
#          @app.route("/user/<username>")
#          def user_profile(username):
#              userfound = find_one_or_404(usercollection, {"_id": username})
#              return render_template("user.html",
#                  user=userfound)
#      """
#      found = collection.find_one(**kwargs)
#      if found is None:
#          abort(404)
#      return found