| skipped 16 lines |
| 17 | 17 | | import getpass |
| 18 | 18 | | import click |
| 19 | 19 | | import re |
| 20 | | - | import textwrap |
| 21 | 20 | | import time |
| 22 | | - | import pprint |
| | 21 | + | |
| 23 | 22 | | |
| 24 | 23 | | from telepathy.utils import ( |
| 25 | 24 | | print_banner, |
| skipped 2 lines |
| 28 | 27 | | process_message, |
| 29 | 28 | | process_description, |
| 30 | 29 | | parse_tg_date, |
| 31 | | - | parse_html_page |
| | 30 | + | parse_html_page, |
| | 31 | + | print_shell, |
| | 32 | + | createPlaceholdeCls |
| 32 | 33 | | ) |
| 33 | 34 | | import telepathy.const as const |
| 34 | 35 | | |
| skipped 13 lines |
| 48 | 49 | | from telethon import TelegramClient, functions, types, utils |
| 49 | 50 | | from telethon.utils import get_display_name, get_message_id |
| 50 | 51 | | from alive_progress import alive_bar |
| 51 | | - | from bs4 import BeautifulSoup |
| 52 | | - | |
| 53 | 52 | | |
| 54 | 53 | | @click.command() |
| 55 | 54 | | @click.option( |
| skipped 152 lines |
| 208 | 207 | | exports = [] |
| 209 | 208 | | |
| 210 | 209 | | print("Exporting...") |
| 211 | | - | |
| 212 | 210 | | # progress bar |
| 213 | | - | |
| 214 | 211 | | for Dialog in await client.get_dialogs(): |
| 215 | 212 | | try: |
| 216 | 213 | | if Dialog.entity.username: |
| skipped 7 lines |
| 224 | 221 | | _desc = process_description( |
| 225 | 222 | | group_description, user_language |
| 226 | 223 | | ) |
| 227 | | - | |
| 228 | | - | original_language = _desc[ |
| 229 | | - | "original_language" |
| 230 | | - | ] |
| 231 | 224 | | translated_description = _desc["translated_text"] |
| 232 | 225 | | |
| 233 | 226 | | if Dialog.entity.broadcast is True: |
| skipped 65 lines |
| 299 | 292 | | |
| 300 | 293 | | except AttributeError: |
| 301 | 294 | | pass |
| 302 | | - | |
| 303 | 295 | | else: |
| 304 | | - | |
| 305 | 296 | | for t in target: |
| 306 | 297 | | target_clean = t |
| 307 | 298 | | alphanumeric = "" |
| 308 | | - | |
| 309 | 299 | | |
| 310 | 300 | | for character in target_clean: |
| 311 | 301 | | if character.isalnum(): |
| skipped 117 lines |
| 429 | 419 | | ] |
| 430 | 420 | | |
| 431 | 421 | | translated_description = _desc["translated_text"] |
| 432 | | - | |
| 433 | | - | preferredWidth = 70 |
| 434 | | - | descript = Fore.GREEN + "Description: " + Style.RESET_ALL |
| 435 | | - | prefix = descript |
| 436 | | - | wrapper_d = textwrap.TextWrapper( |
| 437 | | - | initial_indent=prefix, |
| 438 | | - | width=preferredWidth, |
| 439 | | - | subsequent_indent=" ", |
| 440 | | - | ) |
| 441 | | - | |
| 442 | | - | trans_descript = Fore.GREEN + "Translated: " + Style.RESET_ALL |
| 443 | | - | prefix = trans_descript |
| 444 | | - | wrapper_td = textwrap.TextWrapper( |
| 445 | | - | initial_indent=prefix, |
| 446 | | - | width=preferredWidth, |
| 447 | | - | subsequent_indent=" ", |
| 448 | | - | ) |
| 449 | | - | |
| 450 | 422 | | group_description = ('"' + group_description + '"') |
| 451 | 423 | | |
| 452 | 424 | | if entity.broadcast is True: |
| skipped 17 lines |
| 470 | 442 | | else: |
| 471 | 443 | | group_status = "None" |
| 472 | 444 | | |
| 473 | | - | restrict = Fore.GREEN + "Restrictions:" + Style.RESET_ALL |
| 474 | | - | prefix = restrict + " " |
| 475 | | - | preferredWidth = 70 |
| 476 | | - | wrapper_r = textwrap.TextWrapper( |
| 477 | | - | initial_indent=prefix, |
| 478 | | - | width=preferredWidth, |
| 479 | | - | subsequent_indent=" ", |
| 480 | | - | ) |
| 481 | | - | |
| | 445 | + | found_participants = 0 |
| | 446 | + | found_percentage = 0 |
| 482 | 447 | | if chat_type != "Channel": |
| 483 | 448 | | members = [] |
| 484 | | - | all_participants = [] |
| 485 | 449 | | all_participants = await client.get_participants(t, limit=5000) |
| 486 | | - | |
| 487 | 450 | | members_df = None |
| 488 | 451 | | for user in all_participants: |
| 489 | 452 | | members_df = pd.DataFrame( |
| skipped 39 lines |
| 529 | 492 | | else: |
| 530 | 493 | | pass |
| 531 | 494 | | |
| 532 | | - | color_print_green(" ┬ Chat details", "") |
| 533 | | - | color_print_green(" ├ Title: ", str(entity.title)) |
| 534 | | - | color_print_green(" ├ ", wrapper_d.fill(group_description)) |
| 535 | | - | if translated_description != group_description: |
| 536 | | - | color_print_green(" ├ ", wrapper_td.fill(translated_description)) |
| 537 | | - | color_print_green( |
| 538 | | - | " ├ Total participants: ", str(total_participants) |
| 539 | | - | ) |
| 540 | | - | |
| | 495 | + | setattr(entity, "group_description", group_description) |
| | 496 | + | setattr(entity, "group_status", group_status) |
| | 497 | + | setattr(entity, "group_username", group_username) |
| | 498 | + | setattr(entity, "first_post", first_post) |
| | 499 | + | setattr(entity, "group_url", group_url) |
| | 500 | + | setattr(entity, "chat_type", chat_type) |
| | 501 | + | setattr(entity, "translated_description", translated_description) |
| | 502 | + | setattr(entity, "total_participants", total_participants) |
| 541 | 503 | | if chat_type != "Channel": |
| 542 | | - | color_print_green( |
| 543 | | - | " ├ Participants found: ", |
| 544 | | - | str(found_participants) |
| 545 | | - | + " (" |
| 546 | | - | + str(format(found_percentage, ".2f")) |
| 547 | | - | + "%)", |
| 548 | | - | ) |
| | 504 | + | setattr(entity, "found_participants", found_participants) |
| | 505 | + | setattr(entity, "found_percentage", found_percentage) |
| | 506 | + | setattr(entity, "memberlist_filename", memberlist_filename) |
| 549 | 507 | | else: |
| 550 | | - | found_participants = "N/A" |
| 551 | | - | |
| 552 | | - | color_print_green(" ├ Username: ", str(group_username)) |
| 553 | | - | color_print_green(" ├ URL: ", str(group_url)) |
| 554 | | - | color_print_green(" ├ Chat type: ", str(chat_type)) |
| 555 | | - | color_print_green(" ├ Chat id: ", str(entity.id)) |
| 556 | | - | color_print_green(" ├ Access hash: ", str(entity.access_hash)) |
| 557 | | - | |
| | 508 | + | setattr(entity, "found_participants", found_participants) |
| | 509 | + | print_flag = "group_recap" |
| 558 | 510 | | if chat_type == "Channel": |
| 559 | | - | scam_status = str(entity.scam) |
| 560 | | - | color_print_green(" ├ Scam: ", str(scam_status)) |
| 561 | | - | else: |
| 562 | | - | scam_status = "N/A" |
| | 511 | + | print_flag = "channel_recap" |
| 563 | 512 | | |
| 564 | | - | color_print_green(" ├ First post date: ", str(first_post)) |
| 565 | | - | |
| 566 | | - | if chat_type != "Channel": |
| 567 | | - | color_print_green( |
| 568 | | - | " ├ Memberlist saved to: ", memberlist_filename |
| 569 | | - | ) |
| 570 | | - | |
| 571 | | - | color_print_green( |
| 572 | | - | " â”” ", wrapper_r.fill(group_status) |
| 573 | | - | ) |
| 574 | | - | #print("\n") |
| | 513 | + | print_shell(print_flag, entity) |
| 575 | 514 | | |
| 576 | 515 | | log.append( |
| 577 | 516 | | [ |
| skipped 8 lines |
| 586 | 525 | | chat_type, |
| 587 | 526 | | entity.id, |
| 588 | 527 | | entity.access_hash, |
| 589 | | - | scam_status, |
| | 528 | + | str(entity.scam), |
| 590 | 529 | | date, |
| 591 | 530 | | mtime, |
| 592 | 531 | | group_status, |
| skipped 146 lines |
| 739 | 678 | | df01 = value_count.rename_axis("unique_values").reset_index( |
| 740 | 679 | | name="counts" |
| 741 | 680 | | ) |
| 742 | | - | |
| 743 | | - | top_forward_one = df01.iloc[0]["unique_values"] |
| 744 | | - | top_value_one = df01.iloc[0]["counts"] |
| 745 | | - | top_forward_two = df01.iloc[1]["unique_values"] |
| 746 | | - | top_value_two = df01.iloc[1]["counts"] |
| 747 | | - | top_forward_three = df01.iloc[2]["unique_values"] |
| 748 | | - | top_value_three = df01.iloc[2]["counts"] |
| 749 | | - | top_forward_four = df01.iloc[3]["unique_values"] |
| 750 | | - | top_value_four = df01.iloc[3]["counts"] |
| 751 | | - | top_forward_five = df01.iloc[4]["unique_values"] |
| 752 | | - | top_value_five = df01.iloc[4]["counts"] |
| 753 | | - | |
| 754 | | - | forward_one = ( |
| 755 | | - | str(top_forward_one) |
| | 681 | + | report_forward = createPlaceholdeCls() |
| | 682 | + | report_forward.forward_one = ( |
| | 683 | + | str(df01.iloc[0]["unique_values"]) |
| 756 | 684 | | + ", " |
| 757 | | - | + str(top_value_one) |
| | 685 | + | + str(df01.iloc[0]["counts"]) |
| 758 | 686 | | + " forwarded messages" |
| 759 | 687 | | ) |
| 760 | | - | forward_two = ( |
| 761 | | - | str(top_forward_two) |
| 762 | | - | + ", " |
| 763 | | - | + str(top_value_two) |
| 764 | | - | + " forwarded messages" |
| | 688 | + | report_forward.forward_two = ( |
| | 689 | + | str(df01.iloc[1]["unique_values"]) |
| | 690 | + | + ", " |
| | 691 | + | + str(df01.iloc[1]["counts"]) |
| | 692 | + | + " forwarded messages" |
| 765 | 693 | | ) |
| 766 | | - | forward_three = ( |
| 767 | | - | str(top_forward_three) |
| 768 | | - | + ", " |
| 769 | | - | + str(top_value_three) |
| 770 | | - | + " forwarded messages" |
| | 694 | + | report_forward.forward_three = ( |
| | 695 | + | str(df01.iloc[2]["unique_values"]) |
| | 696 | + | + ", " |
| | 697 | + | + str(df01.iloc[2]["counts"]) |
| | 698 | + | + " forwarded messages" |
| 771 | 699 | | ) |
| 772 | | - | forward_four = ( |
| 773 | | - | str(top_forward_four) |
| 774 | | - | + ", " |
| 775 | | - | + str(top_value_four) |
| 776 | | - | + " forwarded messages" |
| | 700 | + | report_forward.forward_four = ( |
| | 701 | + | str(df01.iloc[3]["unique_values"]) |
| | 702 | + | + ", " |
| | 703 | + | + str(df01.iloc[3]["counts"]) |
| | 704 | + | + " forwarded messages" |
| 777 | 705 | | ) |
| 778 | | - | forward_five = ( |
| 779 | | - | str(top_forward_five) |
| 780 | | - | + ", " |
| 781 | | - | + str(top_value_five) |
| 782 | | - | + " forwarded messages" |
| | 706 | + | report_forward.forward_five = ( |
| | 707 | + | str(df01.iloc[4]["unique_values"]) |
| | 708 | + | + ", " |
| | 709 | + | + str(df01.iloc[4]["counts"]) |
| | 710 | + | + " forwarded messages" |
| 783 | 711 | | ) |
| 784 | | - | |
| 785 | 712 | | df02 = forwards_df.Source.unique() |
| 786 | | - | unique_forwards = len(df02) |
| 787 | | - | |
| 788 | | - | #print("\n") |
| 789 | | - | color_print_green(" [+] Forward scrape complete", "") |
| 790 | | - | color_print_green(" ┬ Statistics", "") |
| 791 | | - | color_print_green( |
| 792 | | - | " ├ Forwarded messages found: ", str(forward_count) |
| 793 | | - | ) |
| 794 | | - | color_print_green( |
| 795 | | - | " ├ Forwards from active public chats: ", |
| 796 | | - | str(forwards_found), |
| 797 | | - | ) |
| 798 | | - | color_print_green( |
| 799 | | - | " ├ Unique forward sources: ", str(unique_forwards) |
| 800 | | - | ) |
| 801 | | - | color_print_green( |
| 802 | | - | " ├ Top forward source 1: ", str(forward_one) |
| 803 | | - | ) |
| 804 | | - | color_print_green( |
| 805 | | - | " ├ Top forward source 2: ", str(forward_two) |
| 806 | | - | ) |
| 807 | | - | color_print_green( |
| 808 | | - | " ├ Top forward source 3: ", str(forward_three) |
| 809 | | - | ) |
| 810 | | - | color_print_green( |
| 811 | | - | " ├ Top forward source 4: ", str(forward_four) |
| 812 | | - | ) |
| 813 | | - | color_print_green( |
| 814 | | - | " ├ Top forward source 5: ", str(forward_five) |
| 815 | | - | ) |
| 816 | | - | color_print_green(" â”” Edgelist saved to: ", edgelist_file) |
| 817 | | - | #print("\n") |
| | 713 | + | report_forward.unique_forwards = len(df02) |
| | 714 | + | report_forward.edgelist_file = edgelist_file |
| | 715 | + | print_shell("forwarder_stat",report_forward) |
| 818 | 716 | | |
| 819 | 717 | | else: |
| 820 | 718 | | print( |
| skipped 8 lines |
| 829 | 727 | | if comp_check is True: |
| 830 | 728 | | |
| 831 | 729 | | messages = client.iter_messages(t) |
| 832 | | - | |
| 833 | 730 | | message_list = [] |
| 834 | 731 | | forwards_list = [] |
| 835 | | - | |
| 836 | 732 | | user_reaction_list = [] |
| 837 | | - | |
| 838 | 733 | | replies_list = [] |
| 839 | 734 | | user_replier_list = [] |
| 840 | | - | |
| 841 | 735 | | timecount = [] |
| 842 | | - | |
| 843 | 736 | | forward_count = 0 |
| 844 | 737 | | private_count = 0 |
| 845 | 738 | | |
| skipped 104 lines |
| 950 | 843 | | "Message ID", |
| 951 | 844 | | "Reply ID", |
| 952 | 845 | | "Display_name", |
| 953 | | - | "ID", |
| | 846 | + | "User ID", |
| 954 | 847 | | "Message_text", |
| 955 | 848 | | "Original_language", |
| 956 | 849 | | "Translated_text", |
| skipped 103 lines |
| 1060 | 953 | | else: |
| 1061 | 954 | | has_media = 'FALSE' |
| 1062 | 955 | | |
| 1063 | | - | post_url = "https://t.me/s/" + t + "/" + message.id |
| | 956 | + | post_url = "https://t.me/s/" + t + "/" + str(message.id) |
| 1064 | 957 | | |
| 1065 | 958 | | message_list.append( |
| 1066 | 959 | | [ |
| skipped 191 lines |
| 1258 | 1151 | | pass |
| 1259 | 1152 | | |
| 1260 | 1153 | | messages_found = int(c_archive.To.count()) - 1 |
| 1261 | | - | message_frequency_count = {} |
| 1262 | | - | message_text = {} |
| 1263 | | - | word_count = {} |
| 1264 | | - | most_used_words = {} |
| 1265 | | - | most_used_words_filtered = {} |
| 1266 | | - | # message stats, top words |
| | 1154 | + | report_obj = createPlaceholdeCls() |
| | 1155 | + | report_obj.messages_found = messages_found |
| | 1156 | + | report_obj.file_archive = file_archive |
| 1267 | 1157 | | |
| 1268 | | - | if chat_type != "Channel": |
| 1269 | | - | pcount = c_archive.Display_name.count() |
| | 1158 | + | if chat_type == "Channel": |
| | 1159 | + | print_shell("channel_stat", report_obj) |
| | 1160 | + | else: |
| 1270 | 1161 | | pvalue_count = c_archive["Display_name"].value_counts() |
| 1271 | 1162 | | df03 = pvalue_count.rename_axis( |
| 1272 | 1163 | | "unique_values" |
| 1273 | 1164 | | ).reset_index(name="counts") |
| 1274 | 1165 | | |
| 1275 | | - | top_poster_one = str(df03.iloc[0]["unique_values"]) |
| 1276 | | - | top_pvalue_one = df03.iloc[0]["counts"] |
| 1277 | | - | top_poster_two = str(df03.iloc[1]["unique_values"]) |
| 1278 | | - | top_pvalue_two = df03.iloc[1]["counts"] |
| 1279 | | - | top_poster_three = str(df03.iloc[2]["unique_values"]) |
| 1280 | | - | top_pvalue_three = df03.iloc[2]["counts"] |
| 1281 | | - | top_poster_four = str(df03.iloc[3]["unique_values"]) |
| 1282 | | - | top_pvalue_four = df03.iloc[3]["counts"] |
| 1283 | | - | top_poster_five = str(df03.iloc[4]["unique_values"]) |
| 1284 | | - | top_pvalue_five = df03.iloc[4]["counts"] |
| | 1166 | + | ''' |
| | 1167 | + | message_frequency_count = {} |
| | 1168 | + | message_text = {} |
| | 1169 | + | word_count = {} |
| | 1170 | + | most_used_words = {} |
| | 1171 | + | most_used_words_filtered = {} |
| | 1172 | + | ''' |
| | 1173 | + | #message stats, top words |
| 1285 | 1174 | | |
| 1286 | | - | poster_one = ( |
| 1287 | | - | str(top_poster_one) |
| | 1175 | + | report_obj.poster_one = ( |
| | 1176 | + | str(df03.iloc[0]["unique_values"]) |
| 1288 | 1177 | | + ", " |
| 1289 | | - | + str(top_pvalue_one) |
| | 1178 | + | + str(df03.iloc[0]["counts"]) |
| 1290 | 1179 | | + " messages" |
| 1291 | 1180 | | ) |
| 1292 | | - | poster_two = ( |
| 1293 | | - | str(top_poster_two) |
| | 1181 | + | |
| | 1182 | + | report_obj.poster_two = ( |
| | 1183 | + | str(df03.iloc[1]["unique_values"]) |
| 1294 | 1184 | | + ", " |
| 1295 | | - | + str(top_pvalue_two) |
| | 1185 | + | + str(df03.iloc[2]["counts"]) |
| 1296 | 1186 | | + " messages" |
| 1297 | 1187 | | ) |
| 1298 | | - | poster_three = ( |
| 1299 | | - | str(top_poster_three) |
| | 1188 | + | |
| | 1189 | + | report_obj.poster_three = ( |
| | 1190 | + | str(df03.iloc[2]["unique_values"]) |
| 1300 | 1191 | | + ", " |
| 1301 | | - | + str(top_pvalue_three) |
| | 1192 | + | + str(df03.iloc[2]["counts"]) |
| 1302 | 1193 | | + " messages" |
| 1303 | 1194 | | ) |
| 1304 | | - | poster_four = ( |
| 1305 | | - | str(top_poster_four) |
| | 1195 | + | |
| | 1196 | + | report_obj.poster_four = ( |
| | 1197 | + | str(df03.iloc[3]["unique_values"]) |
| 1306 | 1198 | | + ", " |
| 1307 | | - | + str(top_pvalue_four) |
| | 1199 | + | + df03.iloc[3]["counts"] |
| 1308 | 1200 | | + " messages" |
| 1309 | 1201 | | ) |
| 1310 | | - | poster_five = ( |
| 1311 | | - | str(top_poster_five) |
| | 1202 | + | |
| | 1203 | + | report_obj.poster_five = ( |
| | 1204 | + | str(df03.iloc[4]["unique_values"]) |
| 1312 | 1205 | | + ", " |
| 1313 | | - | + str(top_pvalue_five) |
| | 1206 | + | + str(df03.iloc[4]["counts"]) |
| 1314 | 1207 | | + " messages" |
| 1315 | 1208 | | ) |
| 1316 | 1209 | | |
| 1317 | 1210 | | df04 = c_archive.Display_name.unique() |
| 1318 | | - | plength = len(df03) |
| 1319 | 1211 | | unique_active = len(df04) |
| | 1212 | + | report_obj.unique_active = unique_active |
| | 1213 | + | print_shell("group_stat", report_obj) |
| 1320 | 1214 | | |
| 1321 | | - | elif reply_analysis is True: |
| | 1215 | + | if reply_analysis is True: |
| | 1216 | + | |
| 1322 | 1217 | | if len(replies_list) > 0: |
| 1323 | | - | replier_count = c_repliers["User id"].count() |
| 1324 | | - | replier_value_count = c_repliers["User id"].value_counts() |
| | 1218 | + | replier_value_count = c_repliers["User ID"].value_counts() |
| 1325 | 1219 | | replier_df = replier_value_count.rename_axis( |
| 1326 | 1220 | | "unique_values" |
| 1327 | 1221 | | ).reset_index(name="counts") |
| 1328 | 1222 | | |
| 1329 | | - | top_replier_one = str(replier_df.iloc[0]["unique_values"]) |
| 1330 | | - | top_replier_value_one = replier_df.iloc[0]["counts"] |
| 1331 | | - | top_replier_two = str(replier_df.iloc[1]["unique_values"]) |
| 1332 | | - | top_replier_value_two = replier_df.iloc[1]["counts"] |
| 1333 | | - | top_replier_three = str(replier_df.iloc[2]["unique_values"]) |
| 1334 | | - | top_replier_value_three = replier_df.iloc[2]["counts"] |
| 1335 | | - | top_replier_four = str(replier_df.iloc[3]["unique_values"]) |
| 1336 | | - | top_replier_value_four = replier_df.iloc[3]["counts"] |
| 1337 | | - | top_replier_five = str(replier_df.iloc[4]["unique_values"]) |
| 1338 | | - | top_replier_value_five = replier_df.iloc[4]["counts"] |
| 1339 | | - | |
| 1340 | | - | replier_one = ( |
| 1341 | | - | str(top_replier_one) |
| | 1223 | + | repliers = createPlaceholdeCls() |
| | 1224 | + | repliers.replier_one = ( |
| | 1225 | + | str(replier_df.iloc[0]["unique_values"]) |
| 1342 | 1226 | | + ", " |
| 1343 | | - | + str(top_replier_value_one) |
| | 1227 | + | + str(replier_df.iloc[0]["counts"]) |
| 1344 | 1228 | | + " replies" |
| 1345 | 1229 | | ) |
| 1346 | | - | replier_two = ( |
| 1347 | | - | str(top_replier_two) |
| | 1230 | + | |
| | 1231 | + | repliers.replier_two = ( |
| | 1232 | + | str(replier_df.iloc[1]["unique_values"]) |
| 1348 | 1233 | | + ", " |
| 1349 | | - | + str(top_replier_value_two) |
| | 1234 | + | + str(replier_df.iloc[1]["counts"]) |
| 1350 | 1235 | | + " replies" |
| 1351 | 1236 | | ) |
| 1352 | | - | replier_three = ( |
| 1353 | | - | str(top_replier_three) |
| | 1237 | + | |
| | 1238 | + | repliers.replier_three = ( |
| | 1239 | + | str(replier_df.iloc[2]["unique_values"]) |
| 1354 | 1240 | | + ", " |
| 1355 | | - | + str(top_replier_value_three) |
| | 1241 | + | + str(replier_df.iloc[2]["counts"]) |
| 1356 | 1242 | | + " replies" |
| 1357 | 1243 | | ) |
| 1358 | | - | replier_four = ( |
| 1359 | | - | str(top_replier_four) |
| | 1244 | + | |
| | 1245 | + | repliers.replier_four = ( |
| | 1246 | + | str(replier_df.iloc[3]["unique_values"]) |
| 1360 | 1247 | | + ", " |
| 1361 | | - | + str(top_replier_value_four) |
| | 1248 | + | + str(replier_df.iloc[3]["counts"]) |
| 1362 | 1249 | | + " replies" |
| 1363 | 1250 | | ) |
| 1364 | | - | replier_five = ( |
| 1365 | | - | str(top_replier_five) |
| | 1251 | + | |
| | 1252 | + | repliers.replier_five = ( |
| | 1253 | + | str(replier_df.iloc[3]["counts"]) |
| 1366 | 1254 | | + ", " |
| 1367 | | - | + str(top_replier_value_five) |
| | 1255 | + | + str(replier_df.iloc[4]["counts"]) |
| 1368 | 1256 | | + " replies" |
| 1369 | 1257 | | ) |
| 1370 | 1258 | | |
| 1371 | | - | replier_count_df = c_repliers["User id"].unique() |
| 1372 | | - | replier_length = len(replier_df) |
| | 1259 | + | replier_count_df = c_repliers["User ID"].unique() |
| 1373 | 1260 | | replier_unique = len(replier_count_df) |
| 1374 | 1261 | | |
| 1375 | | - | else: |
| 1376 | | - | pass |
| 1377 | | - | |
| 1378 | | - | #print("\n") |
| 1379 | | - | color_print_green(" [+] Chat archive saved", "") |
| 1380 | | - | color_print_green(" ┬ Chat statistics", "") |
| 1381 | | - | color_print_green( |
| 1382 | | - | " ├ Number of messages found: ", str(messages_found) |
| 1383 | | - | ) |
| 1384 | | - | |
| 1385 | | - | if chat_type != "Channel": |
| 1386 | | - | color_print_green( |
| 1387 | | - | " ├ Top poster 1: ", str(poster_one) |
| 1388 | | - | ) |
| 1389 | | - | color_print_green( |
| 1390 | | - | " ├ Top poster 2: ", str(poster_two) |
| 1391 | | - | ) |
| 1392 | | - | color_print_green( |
| 1393 | | - | " ├ Top poster 3: ", str(poster_three) |
| 1394 | | - | ) |
| 1395 | | - | color_print_green( |
| 1396 | | - | " ├ Top poster 4: ", str(poster_four) |
| 1397 | | - | ) |
| 1398 | | - | color_print_green( |
| 1399 | | - | " ├ Top poster 5: ", str(poster_five) |
| 1400 | | - | ) |
| 1401 | | - | color_print_green( |
| 1402 | | - | " ├ Total unique posters: ", str(unique_active) |
| 1403 | | - | ) |
| 1404 | | - | |
| 1405 | | - | else: |
| 1406 | | - | pass |
| 1407 | | - | # timestamp analysis |
| 1408 | | - | # print(Fore.GREEN |
| 1409 | | - | # + ' ├ Number of messages: ' |
| 1410 | | - | # + Style.RESET_ALL |
| 1411 | | - | # + str(message_count)) |
| 1412 | | - | |
| 1413 | | - | color_print_green( |
| 1414 | | - | " â”” Archive saved to: ", str(file_archive) |
| 1415 | | - | ) |
| 1416 | | - | |
| 1417 | | - | if reply_analysis is True: |
| 1418 | | - | if len(replies_list) > 0: |
| 1419 | | - | middle_char = "├" |
| 1420 | | - | if user_replier_list == 0: |
| 1421 | | - | middle_char = "â””" |
| 1422 | | - | |
| 1423 | | - | #print("\n") |
| 1424 | | - | color_print_green(" [+] Replies analysis ", "") |
| 1425 | | - | color_print_green(" ┬ Chat statistics", "") |
| 1426 | | - | color_print_green( |
| 1427 | | - | f" {middle_char} Archive of replies saved to: ", |
| 1428 | | - | str(reply_file_archive), |
| 1429 | | - | ) |
| 1430 | | - | if len(user_replier_list) > 0: |
| 1431 | | - | color_print_green( |
| 1432 | | - | " â”” Active members list who replied to messages, saved to: ", |
| 1433 | | - | str(reply_memberlist_filename), |
| 1434 | | - | ) |
| 1435 | | - | |
| 1436 | | - | color_print_green( |
| 1437 | | - | " ├ Top replier 1: ", str(replier_one) |
| 1438 | | - | ) |
| 1439 | | - | color_print_green( |
| 1440 | | - | " ├ Top replier 2: ", str(replier_two) |
| 1441 | | - | ) |
| 1442 | | - | color_print_green( |
| 1443 | | - | " ├ Top replier 3: ", str(replier_three) |
| 1444 | | - | ) |
| 1445 | | - | color_print_green( |
| 1446 | | - | " ├ Top replier 4: ", str(replier_four) |
| 1447 | | - | ) |
| 1448 | | - | color_print_green( |
| 1449 | | - | " ├ Top replier 5: ", str(replier_five) |
| 1450 | | - | ) |
| 1451 | | - | color_print_green( |
| 1452 | | - | " ├ Total unique repliers: ", str(replier_unique) |
| 1453 | | - | ) |
| | 1262 | + | repliers.user_replier_list_len = len(user_replier_list) |
| | 1263 | + | repliers.reply_file_archive = str(reply_file_archive) |
| | 1264 | + | repliers.reply_memberlist_filename = str(reply_memberlist_filename) |
| | 1265 | + | repliers.replier_unique = str(replier_unique) |
| | 1266 | + | print_shell("reply_stat", repliers) |
| 1454 | 1267 | | |
| 1455 | 1268 | | if forwards_check is True: |
| 1456 | 1269 | | if forward_count >= 15: |
| skipped 2 lines |
| 1459 | 1272 | | c_f_stats = value_count.rename_axis( |
| 1460 | 1273 | | "unique_values" |
| 1461 | 1274 | | ).reset_index(name="counts") |
| 1462 | | - | |
| 1463 | | - | top_forward_one = c_f_stats.iloc[0]["unique_values"] |
| 1464 | | - | top_value_one = c_f_stats.iloc[0]["counts"] |
| 1465 | | - | top_forward_two = c_f_stats.iloc[1]["unique_values"] |
| 1466 | | - | top_value_two = c_f_stats.iloc[1]["counts"] |
| 1467 | | - | top_forward_three = c_f_stats.iloc[2][ |
| 1468 | | - | "unique_values" |
| 1469 | | - | ] |
| 1470 | | - | top_value_three = c_f_stats.iloc[2]["counts"] |
| 1471 | | - | top_forward_four = c_f_stats.iloc[3][ |
| 1472 | | - | "unique_values" |
| 1473 | | - | ] |
| 1474 | | - | top_value_four = c_f_stats.iloc[3]["counts"] |
| 1475 | | - | top_forward_five = c_f_stats.iloc[4][ |
| 1476 | | - | "unique_values" |
| 1477 | | - | ] |
| 1478 | | - | top_value_five = c_f_stats.iloc[4]["counts"] |
| 1479 | | - | |
| 1480 | | - | forward_one = ( |
| 1481 | | - | str(top_forward_one) |
| 1482 | | - | + ", " |
| 1483 | | - | + str(top_value_one) |
| 1484 | | - | + " forwarded messages" |
| | 1275 | + | report_forward = createPlaceholdeCls() |
| | 1276 | + | report_forward.forward_one = ( |
| | 1277 | + | str(c_f_stats.iloc[0]["unique_values"]) |
| | 1278 | + | + ", " |
| | 1279 | + | + str(c_f_stats.iloc[0]["counts"]) |
| | 1280 | + | + " forwarded messages" |
| 1485 | 1281 | | ) |
| 1486 | | - | forward_two = ( |
| 1487 | | - | str(top_forward_two) |
| 1488 | | - | + ", " |
| 1489 | | - | + str(top_value_two) |
| 1490 | | - | + " forwarded messages" |
| | 1282 | + | report_forward.forward_two = ( |
| | 1283 | + | str(c_f_stats.iloc[1]["unique_values"]) |
| | 1284 | + | + ", " |
| | 1285 | + | + str(c_f_stats.iloc[1]["counts"]) |
| | 1286 | + | + " forwarded messages" |
| 1491 | 1287 | | ) |
| 1492 | | - | forward_three = ( |
| 1493 | | - | str(top_forward_three) |
| 1494 | | - | + ", " |
| 1495 | | - | + str(top_value_three) |
| 1496 | | - | + " forwarded messages" |
| | 1288 | + | report_forward.forward_three = ( |
| | 1289 | + | str(c_f_stats.iloc[2]["unique_values"]) |
| | 1290 | + | + ", " |
| | 1291 | + | + str(c_f_stats.iloc[2]["counts"]) |
| | 1292 | + | + " forwarded messages" |
| 1497 | 1293 | | ) |
| 1498 | | - | forward_four = ( |
| 1499 | | - | str(top_forward_four) |
| 1500 | | - | + ", " |
| 1501 | | - | + str(top_value_four) |
| 1502 | | - | + " forwarded messages" |
| | 1294 | + | report_forward.forward_four = ( |
| | 1295 | + | str(c_f_stats.iloc[3]["unique_values"]) |
| | 1296 | + | + ", " |
| | 1297 | + | + str(c_f_stats.iloc[3]["counts"]) |
| | 1298 | + | + " forwarded messages" |
| 1503 | 1299 | | ) |
| 1504 | | - | forward_five = ( |
| 1505 | | - | str(top_forward_five) |
| 1506 | | - | + ", " |
| 1507 | | - | + str(top_value_five) |
| 1508 | | - | + " forwarded messages" |
| | 1300 | + | report_forward.forward_five = ( |
| | 1301 | + | str(c_f_stats.iloc[4]["unique_values"]) |
| | 1302 | + | + ", " |
| | 1303 | + | + str(c_f_stats.iloc[4]["counts"]) |
| | 1304 | + | + " forwarded messages" |
| 1509 | 1305 | | ) |
| 1510 | 1306 | | |
| 1511 | 1307 | | c_f_unique = c_forwards.Source.unique() |
| 1512 | | - | unique_forwards = len(c_f_unique) |
| 1513 | | - | |
| 1514 | | - | #print("\n") |
| 1515 | | - | color_print_green(" [+] Edgelist saved", "") |
| 1516 | | - | color_print_green( |
| 1517 | | - | " ┬ Forwarded message statistics", "" |
| 1518 | | - | ) |
| 1519 | | - | color_print_green( |
| 1520 | | - | " ├ Forwarded messages found: ", |
| 1521 | | - | str(forward_count), |
| 1522 | | - | ) |
| 1523 | | - | color_print_green( |
| 1524 | | - | " ├ Forwards from active public chats: ", |
| 1525 | | - | str(forwards_found), |
| 1526 | | - | ) |
| 1527 | | - | color_print_green( |
| 1528 | | - | " ├ Forwards from private (or now private) chats: ", |
| 1529 | | - | str(private_count), |
| 1530 | | - | ) |
| 1531 | | - | color_print_green( |
| 1532 | | - | " ├ Unique forward sources: ", |
| 1533 | | - | str(unique_forwards), |
| 1534 | | - | ) |
| 1535 | | - | color_print_green( |
| 1536 | | - | " ├ Top forward source 1: ", str(forward_one) |
| 1537 | | - | ) |
| 1538 | | - | color_print_green( |
| 1539 | | - | " ├ Top forward source 2: ", str(forward_two) |
| 1540 | | - | ) |
| 1541 | | - | color_print_green( |
| 1542 | | - | " ├ Top forward source 3: ", |
| 1543 | | - | str(forward_three), |
| 1544 | | - | ) |
| 1545 | | - | color_print_green( |
| 1546 | | - | " ├ Top forward source 4: ", str(forward_four) |
| 1547 | | - | ) |
| 1548 | | - | color_print_green( |
| 1549 | | - | " ├ Top forward source 5: ", str(forward_five) |
| 1550 | | - | ) |
| 1551 | | - | color_print_green( |
| 1552 | | - | " â”” Edgelist saved to: ", edgelist_file |
| 1553 | | - | ) |
| 1554 | | - | #print("\n") |
| 1555 | | - | |
| | 1308 | + | report_forward.unique_forwards = len(c_f_unique) |
| | 1309 | + | report_forward.edgelist_file = edgelist_file |
| | 1310 | + | report_forward.private_count = private_count |
| | 1311 | + | print_shell("forwarder_stat", report_forward) |
| 1556 | 1312 | | else: |
| 1557 | 1313 | | #print("\n") |
| 1558 | 1314 | | color_print_green( |
| skipped 37 lines |
| 1596 | 1352 | | user_restrictions = str(ios_restriction) |
| 1597 | 1353 | | else: |
| 1598 | 1354 | | user_restrictions = "None" |
| 1599 | | - | |
| 1600 | | - | color_print_green(" [+] ", "User details for " + t) |
| 1601 | | - | color_print_green(" ├ Username: ", str(my_user.username)) |
| 1602 | | - | color_print_green(" ├ Name: ", str(user_full_name)) |
| 1603 | | - | color_print_green(" ├ Verification: ", str(my_user.verified)) |
| 1604 | | - | color_print_green(" ├ Photo ID: ", str(user_photo)) |
| 1605 | | - | color_print_green(" ├ Phone number: ", str(my_user.phone)) |
| 1606 | | - | color_print_green( |
| 1607 | | - | " ├ Access hash: ", str(my_user.access_hash) |
| 1608 | | - | ) |
| 1609 | | - | color_print_green(" ├ Language: ", str(my_user.lang_code)) |
| 1610 | | - | color_print_green(" ├ Bot: ", str(my_user.bot)) |
| 1611 | | - | color_print_green(" ├ Scam: ", str(my_user.scam)) |
| 1612 | | - | color_print_green(" â”” Restrictions: ", str(user_restrictions)) |
| | 1355 | + | setattr(my_user, "user_restrictions", str(user_restrictions)) |
| | 1356 | + | setattr(my_user, "user_full_name", str(user_full_name)) |
| | 1357 | + | setattr(my_user, "user_photo", str(user_photo)) |
| | 1358 | + | setattr(my_user, "target", t) |
| | 1359 | + | print_shell("user", my_user) |
| 1613 | 1360 | | |
| 1614 | 1361 | | except ValueError: |
| 1615 | 1362 | | pass |
| skipped 70 lines |
| 1686 | 1433 | | except: |
| 1687 | 1434 | | pass |
| 1688 | 1435 | | |
| 1689 | | - | d_500 = 0 |
| 1690 | | - | d_1000 = 0 |
| 1691 | | - | d_2000 = 0 |
| 1692 | | - | d_3000 = 0 |
| | 1436 | + | distance_obj = createPlaceholdeCls() |
| | 1437 | + | distance_obj.d500 = 0 |
| | 1438 | + | distance_obj.d1000 = 0 |
| | 1439 | + | distance_obj.d2000 = 0 |
| | 1440 | + | distance_obj.d3000 = 0 |
| 1693 | 1441 | | |
| 1694 | 1442 | | for account, distance in user_df.itertuples(index=False): |
| 1695 | 1443 | | account = int(account) |
| skipped 3 lines |
| 1699 | 1447 | | distance = int(distance) |
| 1700 | 1448 | | |
| 1701 | 1449 | | if distance == 500: |
| 1702 | | - | d_500 += 1 |
| | 1450 | + | distance_obj.d500 += 1 |
| 1703 | 1451 | | elif distance == 1000: |
| 1704 | | - | d_1000 += 1 |
| | 1452 | + | distance_obj.d1000 += 1 |
| 1705 | 1453 | | elif distance == 2000: |
| 1706 | | - | d_2000 += 1 |
| | 1454 | + | distance_obj.d2000 += 1 |
| 1707 | 1455 | | elif distance == 3000: |
| 1708 | | - | d_3000 += 1 |
| | 1456 | + | distance_obj.d3000 += 1 |
| 1709 | 1457 | | |
| 1710 | 1458 | | with open( |
| 1711 | 1459 | | save_file, "w+", encoding="utf-8" |
| 1712 | | - | ) as f: |
| | 1460 | + | ) as f: # could one day append, including access time to differentiate |
| 1713 | 1461 | | user_df.to_csv(f, sep=";", index=False) |
| 1714 | 1462 | | |
| 1715 | 1463 | | total = len(locations_list) |
| 1716 | | - | |
| 1717 | | - | color_print_green(" [+] Users located", "") |
| 1718 | | - | color_print_green(" ├ Users within 500m: ", str(d_500)) |
| 1719 | | - | color_print_green(" ├ Users within 1000m: ", str(d_1000)) |
| 1720 | | - | color_print_green(" ├ Users within 2000m: ", str(d_2000)) |
| 1721 | | - | color_print_green(" ├ Users within 3000m: ", str(d_3000)) |
| 1722 | | - | color_print_green(" ├ Total users found: ", str(total)) |
| 1723 | | - | color_print_green(" â”” Location list saved to: ", save_file) |
| 1724 | | - | |
| 1725 | | - | user_df.iloc[0:0] |
| | 1464 | + | distance_obj.save_file = save_file |
| | 1465 | + | distance_obj.total = total |
| | 1466 | + | print_shell("location_report",distance_obj) |
| | 1467 | + | # can also do the same for channels with similar output file to users |
| | 1468 | + | # may one day add trilateration to find users closest to exact point |
| 1726 | 1469 | | |
| 1727 | 1470 | | with client: |
| 1728 | 1471 | | client.loop.run_until_complete(main()) |
| skipped 4 lines |
Jordan Wildon
committed
with
GitHub
2 years ago