diff --git a/extract.py b/extract.py index 2c52ecc..06ed771 100644 --- a/extract.py +++ b/extract.py @@ -59,51 +59,6 @@ while content is not None: print(f"Gathering messages...({i}/{total_row_number})", end="\r") content = c.fetchone() print(f"Gathering messages...({total_row_number}/{total_row_number})", end="\r") -# Get media - -c.execute("""SELECT count() FROM message_media""") -total_row_number = c.fetchone()[0] -print(f"\nGathering media...(0/{total_row_number})", end="\r") -i = 0 -c.execute("""SELECT messages.key_remote_jid, message_row_id, file_path, message_url, mime_type, media_key FROM message_media INNER JOIN messages ON message_media.message_row_id = messages._id ORDER BY messages.key_remote_jid ASC""") -content = c.fetchone() -mime = MimeTypes() -while content is not None: - file_path = f"WhatsApp/{content[2]}" - if os.path.isfile(file_path): - with open(file_path, "rb") as f: - data[content[0]]["messages"][content[1]]["data"] = base64.b64encode(f.read()).decode("utf-8") - data[content[0]]["messages"][content[1]]["media"] = True - if content[4] is None: - guess = mime.guess_type(file_path)[0] - if guess is not None: - data[content[0]]["messages"][content[1]]["mime"] = guess - else: - data[content[0]]["messages"][content[1]]["mime"] = "image/jpeg" - else: - data[content[0]]["messages"][content[1]]["mime"] = content[4] - else: - if "https://mmg" in content[4]: - try: - r = requests.get(content[3]) - if r.status_code != 200: - raise RuntimeError() - except: - data[content[0]]["messages"][content[1]]["data"] = "{The media is missing}" - data[content[0]]["messages"][content[1]]["media"] = True - data[content[0]]["messages"][content[1]]["mime"] = "media" - else: - open('temp.file', 'wb').write(r.content) - open('temp.asdasda', "a").write(content[3]) - else: - data[content[0]]["messages"][content[1]]["data"] = "{The media is missing}" - data[content[0]]["messages"][content[1]]["media"] = True - data[content[0]]["messages"][content[1]]["mime"] = "media" - i += 1 - if i % 1000 == 0: - print(f"Gathering media...({i}/{total_row_number})", end="\r") - content = c.fetchone() -print(f"Gathering media...({total_row_number}/{total_row_number})", end="\r") templateLoader = jinja2.FileSystemLoader(searchpath="./") templateEnv = jinja2.Environment(loader=templateLoader) @@ -111,7 +66,7 @@ templateEnv.globals.update(determine_day=determine_day) TEMPLATE_FILE = "whatsapp.html" template = templateEnv.get_template(TEMPLATE_FILE) -total_row_number = len(data) +total_row_number_html = len(data) print(f"\nCreating HTML...(0/{total_row_number})", end="\r") if len(sys.argv) < 3: @@ -122,30 +77,74 @@ else: if not os.path.isdir(output_folder): os.mkdir(output_folder) -for current, i in enumerate(data): - if len(data[i]["messages"]) == 0: +list_of_contact = tuple(data.keys()) + +for current, contact in enumerate(list_of_contact): + if len(data[contact]["messages"]) == 0: continue - phone_number = i.split('@')[0] - if "-"in i: + # Get media + c.execute("""SELECT count() FROM message_media INNER JOIN messages ON message_media.message_row_id = messages._id WHERE messages.key_remote_jid=?""", [contact]) + total_row_number = c.fetchone()[0] + phone_number = contact.split('@')[0] + print(f"Creating HTML...({current}/{total_row_number_html})|Gathering media with {phone_number}...(0/{total_row_number})", end="\r") + j = 0 + c.execute("""SELECT messages.key_remote_jid, message_row_id, file_path, message_url, mime_type, media_key, file_hash FROM message_media INNER JOIN messages ON message_media.message_row_id = messages._id WHERE messages.key_remote_jid=? ORDER BY messages.key_remote_jid ASC""", [contact]) + contents = c.fetchall() + mime = MimeTypes() + for content in contents: + file_path = f"WhatsApp/{content[2]}" + data[content[0]]["messages"][content[1]]["media"] = True + if os.path.isfile(file_path): + with open(file_path, "rb") as f: + data[content[0]]["messages"][content[1]]["data"] = base64.b64encode(f.read()).decode("utf-8") + if content[4] is None: + guess = mime.guess_type(file_path)[0] + if guess is not None: + data[content[0]]["messages"][content[1]]["mime"] = guess + else: + data[content[0]]["messages"][content[1]]["mime"] = "image/jpeg" + else: + data[content[0]]["messages"][content[1]]["mime"] = content[4] + else: + # if "https://mmg" in content[2]: + # try: + # r = requests.get(content[2]) + # if r.status_code != 200: + # raise RuntimeError() + # except: + # data[contact]["messages"][content[0]]["data"] = "{The media is missing}" + # data[contact]["messages"][content[0]]["media"] = True + # data[contact]["messages"][content[0]]["mime"] = "media" + # else: + # open('temp.file', 'wb').write(r.content) + # open('temp.asdasda', "a").write(content[3]) + # else: + data[content[0]]["messages"][content[1]]["data"] = "{The media is missing}" + data[content[0]]["messages"][content[1]]["mime"] = "media" + j += 1 + if j % 100 == 0: + print(f"Creating HTML...({current}/{total_row_number_html})|Gathering media with {phone_number}...({j}/{total_row_number}) ", end="\r") + print(f"Creating HTML...({current}/{total_row_number_html})|Gathering media with {phone_number}...({total_row_number}/{total_row_number}) ", end="\r") + if "-" in contact: file_name = "" else: file_name = phone_number - if data[i]["name"] is not None: + if data[contact]["name"] is not None: if file_name != "": file_name += "-" - file_name += data[i]["name"].replace("/", "-") + file_name += data[contact]["name"].replace("/", "-") + if data[contact]["name"]: + name = data[contact]["name"] + else: + name = phone_number with open(f"{output_folder}/{file_name}.html", "w", encoding="utf-8") as f: - f.write(template.render(name=data[i]["name"] if data[i]["name"] is not None else phone_number, msgs=data[i]["messages"].values())) - if current % 10 == 0: - print(f"Creating HTML...({current}/{total_row_number})", end="\r") + f.write(template.render(name=name, msgs=data[contact]["messages"].values())) -print(f"Creating HTML...({total_row_number}/{total_row_number})", end="\r") - -with open("result.json", "w") as f: - data = json.dumps(data) - print(f"\nWriting JSON file...({int(len(data)/1024/1024)}MB)") - f.write(data) - -print("Everything is done!") + del data[contact] + #if current % 10 == 0: + #print(f"Creating HTML...({current}/{total_row_number})", end="\r") + +print(f"Creating HTML...({total_row_number_html}/{total_row_number_html}) ") +print("Everything is done! ")