|
| 1 | +import re |
| 2 | + |
| 3 | +import requests |
| 4 | +from bs4 import BeautifulSoup |
| 5 | +from monster.models import * |
| 6 | + |
| 7 | +root_url = "https://dnd-5e.herokuapp.com/monsters/" |
| 8 | + |
| 9 | + |
| 10 | +def main(): |
| 11 | + html = requests.get(root_url).text |
| 12 | + parsed_html = BeautifulSoup(html, features="html.parser") |
| 13 | + list_entries_html = parsed_html.find("ul", attrs={"id": "monsters"}).find_all("a") |
| 14 | + list_entries = [entry["href"] for entry in list_entries_html] |
| 15 | + for entry in list_entries: |
| 16 | + monster_html = requests.get(root_url + entry).text |
| 17 | + monster_parsed_html = BeautifulSoup(monster_html, features="html.parser") |
| 18 | + monster_box = monster_parsed_html.find("div", attrs={"class": "monster-box"}) |
| 19 | + m_name = monster_box.find("h1").text |
| 20 | + if len(Monster.objects.filter(name=m_name)) == 0: |
| 21 | + m = Monster() |
| 22 | + m.name = monster_box.find("h1").text |
| 23 | + m.save() |
| 24 | + else: |
| 25 | + m = Monster.objects.get(name=m_name) |
| 26 | + print(m.name) |
| 27 | + p_list = monster_box.find_all("p") |
| 28 | + |
| 29 | + size_tmp = p_list[0].text.split(" ")[0].lower() |
| 30 | + m.size = Size.objects.get(size=size_tmp) |
| 31 | + |
| 32 | + race_temp = " ".join(p_list[0].text.split(" ")[1:]).split(",")[0] |
| 33 | + if "(" in race_temp: |
| 34 | + race_name = race_temp.split(" ")[0] |
| 35 | + race_comment = "".join(race_temp.split(" ")[-1][1:]) |
| 36 | + if race_comment.endswith(")"): |
| 37 | + race_comment = race_comment[:-1] |
| 38 | + if len(Race.objects.filter(race=race_name, comment=race_comment)) == 0: |
| 39 | + r = Race() |
| 40 | + r.race = race_name |
| 41 | + r.comment = race_comment |
| 42 | + r.save() |
| 43 | + m.race = Race.objects.get(race=race_name, comment=race_comment) |
| 44 | + else: |
| 45 | + if len(Race.objects.filter(race=race_temp, comment="")) == 0: |
| 46 | + r = Race() |
| 47 | + r.race = race_temp |
| 48 | + r.save() |
| 49 | + m.race = Race.objects.get(race=race_temp, comment="") |
| 50 | + |
| 51 | + al = p_list[0].text.split(",")[-1].lower().strip() |
| 52 | + if len(Alignment.objects.filter(alignment=al)) == 0: |
| 53 | + a = Alignment() |
| 54 | + a.alignment = al |
| 55 | + a.save() |
| 56 | + m.alignment = Alignment.objects.get(alignment=al) |
| 57 | + m.armor_class = int(p_list[1].text.split(" ")[2:][0]) |
| 58 | + m.armor_description = " ".join(p_list[1].text.split(" ")[3:]).strip("(").rstrip(")") |
| 59 | + m.hit_points = int(p_list[2].text.split(" ")[2:][0]) |
| 60 | + m.hit_points_alt = " ".join(p_list[2].text.split(" ")[3:]).split("(")[1].rstrip(")") |
| 61 | + m.speed_base = " ".join(p_list[3].text.split(" ")[1:]).split(",")[0].split(" ")[0] |
| 62 | + if len(" ".join(p_list[3].text.split(" ")[1:]).split(",")) > 2: |
| 63 | + m.speed_alt = ", ".join(" ".join(p_list[3].text.split(" ")[1:]).split(", ")[1:]) |
| 64 | + |
| 65 | + li_list = monster_box.find_all("li") |
| 66 | + m.strength = li_list[0].text.split(" ")[1] |
| 67 | +# m.strength_mod = int(li_list[0].text.split(" ")[2][1:-1]) |
| 68 | + m.dexterity = li_list[1].text.split(" ")[1] |
| 69 | +# m.dexterity_mod = int(li_list[1].text.split(" ")[2][1:-1]) |
| 70 | + m.constitution = li_list[2].text.split(" ")[1] |
| 71 | +# m.constitution_mod = int(li_list[2].text.split(" ")[2][1:-1]) |
| 72 | + m.intelligence = li_list[3].text.split(" ")[1] |
| 73 | +# m.intelligence_mod = int(li_list[3].text.split(" ")[2][1:-1]) |
| 74 | + m.wisdom = li_list[4].text.split(" ")[1] |
| 75 | +# m.wisdom_mod = int(li_list[4].text.split(" ")[2][1:-1]) |
| 76 | + m.charisma = li_list[5].text.split(" ")[1] |
| 77 | +# m.charisma_mod = int(li_list[5].text.split(" ")[2][1:-1]) |
| 78 | + |
| 79 | + for p_element in p_list[4:]: |
| 80 | + if p_element.text.startswith("Saving Throws"): |
| 81 | + tmp = " ".join(p_element.text.split(" ")[2:]) |
| 82 | + if "," in tmp: |
| 83 | + tmp_list = tmp.split(", ") |
| 84 | + for x in [item.split(" ") for item in tmp_list]: |
| 85 | + if len(SavingThrow.objects.filter(attribute=x[0], value=int(x[1]))) == 0: |
| 86 | + st = SavingThrow() |
| 87 | + st.attribute = x[0] |
| 88 | + st.value = int(x[1]) |
| 89 | + st.save() |
| 90 | + m.saving_throws.add(SavingThrow.objects.get(attribute=x[0], value=int(x[1]))) |
| 91 | + else: |
| 92 | + if len(SavingThrow.objects.filter(attribute=tmp.split(" ")[0], value=int(tmp.split(" ")[1]))) == 0: |
| 93 | + st = SavingThrow() |
| 94 | + st.attribute = tmp.split(" ")[0] |
| 95 | + st.value = int(tmp.split(" ")[1]) |
| 96 | + st.save() |
| 97 | + m.saving_throws.add(SavingThrow.objects.get(attribute=tmp.split(" ")[0], value=int(tmp.split(" ")[1]))) |
| 98 | + |
| 99 | + elif p_element.text.startswith("Senses"): |
| 100 | + tmp = " ".join(p_element.text.split(" ")[1:]) |
| 101 | + if "," in tmp: |
| 102 | + for x in tmp.split(","): |
| 103 | + if len(Sense.objects.filter(sense=x)) == 0: |
| 104 | + s = Sense() |
| 105 | + s.sense = x |
| 106 | + s.save() |
| 107 | + m.senses.add(Sense.objects.get(sense=x)) |
| 108 | + else: |
| 109 | + if len(Sense.objects.filter(sense=tmp)) == 0: |
| 110 | + s = Sense() |
| 111 | + s.sense = tmp |
| 112 | + s.save() |
| 113 | + m.senses.add(Sense.objects.get(sense=tmp)) |
| 114 | + elif p_element.text.startswith("Languages"): |
| 115 | + tmp = " ".join(p_element.text.split(" ")[1:]) |
| 116 | + if "," in tmp: |
| 117 | + for language in tmp.split(","): |
| 118 | + if len(Language.objects.filter(language=language)) == 0: |
| 119 | + l = Language() |
| 120 | + l.language = language |
| 121 | + l.save() |
| 122 | + m.languages.add(l) |
| 123 | + else: |
| 124 | + m.languages.add(Language.objects.get(language=language)) |
| 125 | + else: |
| 126 | + if tmp != "-": |
| 127 | + if len(Language.objects.filter(language=tmp)) == 0: |
| 128 | + l = Language() |
| 129 | + l.language = tmp |
| 130 | + l.save() |
| 131 | + m.languages.add(Language.objects.get(language=tmp)) |
| 132 | + elif p_element.text.startswith("Challenge"): |
| 133 | + challenge_tmp = p_element.text.split(" ")[1:] |
| 134 | + if challenge_tmp[0] == "1/8": |
| 135 | + challenge_rating = 0.125 |
| 136 | + elif challenge_tmp[0] == "1/4": |
| 137 | + challenge_rating = 0.25 |
| 138 | + elif challenge_tmp[0] == "1/2": |
| 139 | + challenge_rating = 0.5 |
| 140 | + else: |
| 141 | + challenge_rating = challenge_tmp[0] |
| 142 | + m.challenge = float(challenge_rating) |
| 143 | + m.challenge_xp = int(challenge_tmp[1].strip("(").replace(",", "")) |
| 144 | + elif p_element.text.startswith("Skills"): |
| 145 | + tmp = " ".join(p_element.text.split(" ")[1:]) |
| 146 | + if "," in tmp: |
| 147 | + for x in [re.split(r"(?!\w)\s[+]", item) for item in tmp.split(", ")]: |
| 148 | + if len(Skill.objects.filter(skill=x[0], value=int(x[1]))) == 0: |
| 149 | + s = Skill() |
| 150 | + s.skill = x[0] |
| 151 | + s.value = int(x[1]) |
| 152 | + s.save() |
| 153 | + m.skills.add(Skill.objects.get(skill=x[0], value=int(x[1]))) |
| 154 | + else: |
| 155 | + if len(Skill.objects.filter(skill=tmp.split(" ")[0], value=int(tmp.split(" ")[1]))) == 0: |
| 156 | + s = Skill() |
| 157 | + s.skill = tmp.split(" ")[0] |
| 158 | + s.value = int(tmp.split(" ")[1]) |
| 159 | + s.save() |
| 160 | + m.skills.add(Skill.objects.get(skill=tmp.split(" ")[0], value=int(tmp.split(" ")[1]))) |
| 161 | + elif p_element.text.startswith("Damage Immunities"): |
| 162 | + tmp = " ".join(p_element.text.split(" ")[2:]).lower() |
| 163 | + if tmp.lower().startswith("one of the following: "): |
| 164 | + tmp = "".join(tmp[22:]) |
| 165 | + if ";" in tmp: |
| 166 | + x = tmp.split("; ") |
| 167 | + if len(DamageImmunity.objects.filter(damage_immunity=x[1])) == 0: |
| 168 | + di = DamageImmunity() |
| 169 | + di.damage_immunity = x[1].lower() |
| 170 | + di.save() |
| 171 | + m.damage_immunities.add(DamageImmunity.objects.get(damage_immunity=x[1])) |
| 172 | + for y in x[0].split(", "): |
| 173 | + if y.startswith("or "): |
| 174 | + y = "".join(y[3:]) |
| 175 | + if y.startswith("and "): |
| 176 | + y = "".join(y[4:]) |
| 177 | + if len(DamageImmunity.objects.filter(damage_immunity=y)) == 0: |
| 178 | + di = DamageImmunity() |
| 179 | + di.damage_immunity = y.lower() |
| 180 | + di.save() |
| 181 | + m.damage_immunities.add(DamageImmunity.objects.get(damage_immunity=y)) |
| 182 | + elif "," in tmp: |
| 183 | + for x in tmp.split(", "): |
| 184 | + if x.startswith("or "): |
| 185 | + x = "".join(x[3:]) |
| 186 | + if x.startswith("and "): |
| 187 | + x = "".join(x[4:]) |
| 188 | + if len(DamageImmunity.objects.filter(damage_immunity=x)) == 0: |
| 189 | + di = DamageImmunity() |
| 190 | + di.damage_immunity = x.lower() |
| 191 | + di.save() |
| 192 | + m.damage_immunities.add(DamageImmunity.objects.get(damage_immunity=x)) |
| 193 | + else: |
| 194 | + if len(DamageImmunity.objects.filter(damage_immunity=tmp)) == 0: |
| 195 | + di = DamageImmunity() |
| 196 | + di.damage_immunity = tmp.lower() |
| 197 | + di.save() |
| 198 | + m.damage_immunities.add(DamageImmunity.objects.get(damage_immunity=tmp)) |
| 199 | + elif p_element.text.startswith("Damage Resistances"): |
| 200 | + tmp = " ".join(p_element.text.split(" ")[2:]).lower() |
| 201 | + if tmp.lower().startswith("one of the following: "): |
| 202 | + tmp = "".join(tmp[22:]) |
| 203 | + if ";" in tmp: |
| 204 | + x = tmp.split("; ") |
| 205 | + if len(DamageResistance.objects.filter(damage_resistance=x[1])) == 0: |
| 206 | + di = DamageResistance() |
| 207 | + di.damage_resistance = x[1].lower() |
| 208 | + di.save() |
| 209 | + m.damage_resistances.add(DamageResistance.objects.get(damage_resistance=x[1])) |
| 210 | + for y in x[0].split(", "): |
| 211 | + if y.startswith("or "): |
| 212 | + y = "".join(y[3:]) |
| 213 | + if y.startswith("and "): |
| 214 | + y = "".join(y[4:]) |
| 215 | + if len(DamageResistance.objects.filter(damage_resistance=y)) == 0: |
| 216 | + di = DamageResistance() |
| 217 | + di.damage_resistance = y.lower() |
| 218 | + di.save() |
| 219 | + m.damage_resistances.add(DamageResistance.objects.get(damage_resistance=y)) |
| 220 | + elif "," in tmp: |
| 221 | + for x in tmp.split(", "): |
| 222 | + if x.startswith("or "): |
| 223 | + x = "".join(x[3:]) |
| 224 | + if x.startswith("and "): |
| 225 | + x = "".join(x[4:]) |
| 226 | + if len(DamageResistance.objects.filter(damage_resistance=x)) == 0: |
| 227 | + di = DamageResistance() |
| 228 | + di.damage_resistance = x.lower() |
| 229 | + di.save() |
| 230 | + m.damage_resistances.add(DamageResistance.objects.get(damage_resistance=x)) |
| 231 | + else: |
| 232 | + if len(DamageResistance.objects.filter(damage_resistance=tmp)) == 0: |
| 233 | + di = DamageResistance() |
| 234 | + di.damage_resistance = tmp.lower() |
| 235 | + di.save() |
| 236 | + m.damage_resistances.add(DamageResistance.objects.get(damage_resistance=tmp)) |
| 237 | + elif p_element.text.startswith("Condition Immunities"): |
| 238 | + tmp = " ".join(p_element.text.split(" ")[2:]).lower() |
| 239 | + if tmp.lower().startswith("one of the following: "): |
| 240 | + tmp = "".join(tmp[22:]) |
| 241 | + if ";" in tmp: |
| 242 | + x = tmp.split("; ") |
| 243 | + if len(ConditionImmunity.objects.filter(condition_immunity=x[1])) == 0: |
| 244 | + di = ConditionImmunity() |
| 245 | + di.condition_immunity = x[1].lower() |
| 246 | + di.save() |
| 247 | + m.condition_immunities.add(ConditionImmunity.objects.get(condition_immunity=x[1])) |
| 248 | + for y in x[0].split(", "): |
| 249 | + if y.startswith("or "): |
| 250 | + y = "".join(y[3:]) |
| 251 | + if y.startswith("and "): |
| 252 | + y = "".join(y[4:]) |
| 253 | + if len(ConditionImmunity.objects.filter(condition_immunity=y)) == 0: |
| 254 | + di = ConditionImmunity() |
| 255 | + di.condition_immunity = y.lower() |
| 256 | + di.save() |
| 257 | + m.condition_immunities.add(ConditionImmunity.objects.get(condition_immunity=y)) |
| 258 | + elif "," in tmp: |
| 259 | + for x in tmp.split(", "): |
| 260 | + if x.startswith("or "): |
| 261 | + x = "".join(x[3:]) |
| 262 | + if x.startswith("and "): |
| 263 | + x = "".join(x[4:]) |
| 264 | + if len(ConditionImmunity.objects.filter(condition_immunity=x)) == 0: |
| 265 | + di = ConditionImmunity() |
| 266 | + di.condition_immunity = x.lower() |
| 267 | + di.save() |
| 268 | + m.condition_immunities.add(ConditionImmunity.objects.get(condition_immunity=x)) |
| 269 | + else: |
| 270 | + if len(ConditionImmunity.objects.filter(condition_immunity=tmp)) == 0: |
| 271 | + di = ConditionImmunity() |
| 272 | + di.condition_immunity = tmp.lower() |
| 273 | + di.save() |
| 274 | + m.condition_immunities.add(ConditionImmunity.objects.get(condition_immunity=tmp)) |
| 275 | + m.tag = Tag.objects.get(tag="monster") |
| 276 | + m.save() |
0 commit comments