Exemple #1
0
def create_entry2(request):
    json_data = simplejson.loads(request.raw_post_data)
    url = json_data["url"]
    name = json_data["name"].strip()
    fields = json_data["fields"]
    print fields
    url_obj = Url(url=url, name=name)
    url_obj.save()
    for field in fields:
        print "got here"
        match_text = field[1].split("\n")[0].replace(" ", "").strip()
        field_obj = Field(
            field_name=field[0].strip(),
            match_text=match_text,
            match_data=parser(url, match_text),
            ignore_breaks=field[2],
            url=url_obj,
            field_name_ns=field[0].strip().replace(" ", ""),
        )
        field_obj.save()
    return HttpResponse(url_obj.id)
Exemple #2
0
def create_entry(request):
    json_data = simplejson.loads(request.raw_post_data)
    url = json_data["url"]
    name = json_data["name"].strip()
    fields = json_data["fields"]
    print fields
    url_obj = Url(url=url, name=name, multiple_match=False)
    url_obj.save()
    write_data = "{"
    for field in fields:
        match_text = "".join(field[1].split("\n")[0].split()).strip()
        match_data = parser(url, match_text)
        field_obj = Field(
            field_name=field[0].strip(),
            match_text=match_text,
            match_data=match_data,
            url=url_obj,
            field_name_ns=field[0].strip().replace(" ", ""),
        )
        write_data += '"' + field[0].replace(" ", "") + '"' + ":" + match_data + ","
        print match_data
        field_obj.save()
    write_data = write_data[:-1] + "}"
    file_name = "scraper_" + name.replace(" ", "") + ".py"
    os.system("cp scraper_customized.py files/static/scripts/" + file_name)
    f = open("files/static/scripts/" + file_name, "r")
    lines = f.read().split("\n")
    f.close()
    f = open("files/static/scripts/" + file_name, "w")
    for line in lines:
        if "DATA = []" in line:
            f.write("DATA = '" + write_data + "'\n")
        else:
            f.write(line + "\n")
    f.close()
    return HttpResponse(url_obj.id)