__filename__ = "speaker.py" __author__ = "Bob Mottram" __license__ = "AGPL3+" __version__ = "1.2.0" __maintainer__ = "Bob Mottram" __email__ = "bob@freedombone.net" __status__ = "Production" import os import random from auth import createBasicAuthHeader from session import getJson from utils import loadJson from utils import getFullDomain speakerRemoveChars = ('.\n', '. ', ',', ';', '?', '!') def getSpeakerPitch(displayName: str, screenreader: str, gender) -> int: """Returns the speech synthesis pitch for the given name """ random.seed(displayName) rangeMin = 1 rangeMax = 100 if 'She' in gender: rangeMin = 50 elif 'Him' in gender: rangeMax = 50 if screenreader == 'picospeaker': rangeMin = -8 rangeMax = 3 if 'She' in gender: rangeMin = -1 elif 'Him' in gender: rangeMax = -1 return random.randint(rangeMin, rangeMax) def getSpeakerRate(displayName: str, screenreader: str) -> int: """Returns the speech synthesis rate for the given name """ random.seed(displayName) if screenreader == 'picospeaker': return random.randint(0, 10) return random.randint(50, 120) def getSpeakerRange(displayName: str) -> int: """Returns the speech synthesis range for the given name """ random.seed(displayName) return random.randint(300, 800) def speakerPronounce(baseDir: str, sayText: str, translate: {}) -> str: """Screen readers may not always pronounce correctly, so you can have a file which specifies conversions. File should contain line items such as: Epicyon -> Epi-cyon """ pronounceFilename = baseDir + '/accounts/speaker_pronounce.txt' convertDict = { "Epicyon": "Epi-cyon", "espeak": "e-speak", "emoji": "emowji", "clearnet": "clear-net", "https": "H-T-T-P-S", "HTTPS": "H-T-T-P-S", "Tor": "Toor", "🤔": ". " + translate["thinking emoji"], "RT @": "Re-Tweet ", "#": translate["hashtag"], ":D": '. ' + translate["laughing"], ":-D": '. ' + translate["laughing"], ":)": '. ' + translate["smile"], ";)": '. ' + translate["wink"], ":(": '. ' + translate["sad face"], ":-)": '. ' + translate["smile"], ":-(": '. ' + translate["sad face"], ";-)": '. ' + translate["wink"], "*": "" } if os.path.isfile(pronounceFilename): with open(pronounceFilename, 'r') as fp: pronounceList = fp.readlines() for conversion in pronounceList: separator = None if '->' in conversion: separator = '->' elif ';' in conversion: separator = ';' elif ':' in conversion: separator = ':' elif ',' in conversion: separator = ',' if not separator: continue text = conversion.split(separator)[0].strip() converted = conversion.split(separator)[1].strip() convertDict[text] = converted for text, converted in convertDict.items(): if text in sayText: sayText = sayText.replace(text, converted) return sayText def speakerReplaceLinks(sayText: str, translate: {}, detectedLinks: []) -> str: """Replaces any links in the given text with "link to [domain]". Instead of reading out potentially very long and meaningless links """ text = sayText for ch in speakerRemoveChars: text = text.replace(ch, ' ') replacements = {} wordsList = text.split(' ') linkedStr = translate['Linked'] prevWord = '' for word in wordsList: if word.startswith(':'): if word.endswith(':'): replacements[word] = ', emowji ' + word.replace(':', '') + ',' continue # replace mentions, but not re-tweets if word.startswith('@') and not prevWord.endswith('RT'): replacements[word] = \ translate['mentioning'] + ' ' + word[1:] + ',' prevWord = word domain = None domainFull = None if 'https://' in word: domain = word.split('https://')[1] domainFull = 'https://' + domain elif 'http://' in word: domain = word.split('http://')[1] domainFull = 'http://' + domain if not domain: continue if '/' in domain: domain = domain.split('/')[0] if domain.startswith('www.'): domain = domain.replace('www.', '') replacements[domainFull] = '. ' + linkedStr + ' ' + domain + '.' detectedLinks.append(domainFull) for replaceStr, newStr in replacements.items(): sayText = sayText.replace(replaceStr, newStr) return sayText.replace('..', '.') def _addSSMLemphasis(sayText: str) -> str: """Adds emphasis to *emphasised* text """ if '*' not in sayText: return sayText text = sayText for ch in speakerRemoveChars: text = text.replace(ch, ' ') wordsList = text.split(' ') replacements = {} for word in wordsList: if word.startswith('*'): if word.endswith('*'): replacements[word] = \ '' + \ word.replace('*', '') + \ '' for replaceStr, newStr in replacements.items(): sayText = sayText.replace(replaceStr, newStr) return sayText def getSpeakerFromServer(baseDir: str, session, nickname: str, password: str, domain: str, port: int, httpPrefix: str, debug: bool, projectVersion: str) -> {}: """Returns some json which contains the latest inbox entry in a minimal format suitable for a text-to-speech reader """ if not session: print('WARN: No session for getSpeakerFromServer') return 6 domainFull = getFullDomain(domain, port) authHeader = createBasicAuthHeader(nickname, password) headers = { 'host': domain, 'Content-type': 'application/json', 'Authorization': authHeader } url = \ httpPrefix + '://' + \ domainFull + '/users/' + nickname + '/speaker' speakerJson = \ getJson(session, url, headers, None, __version__, httpPrefix, domain) return speakerJson def speakerEndpointJson(displayName: str, summary: str, content: str, imageDescription: str, links: [], gender: str) -> {}: """Returns a json endpoint for the TTS speaker """ speakerJson = { "name": displayName, "summary": summary, "say": content, "imageDescription": imageDescription, "detectedLinks": links } if gender: speakerJson['gender'] = gender return speakerJson def _speakerEndpointSSML(displayName: str, summary: str, content: str, imageDescription: str, links: [], language: str, instanceTitle: str, gender: str) -> str: """Returns an SSML endpoint for the TTS speaker https://en.wikipedia.org/wiki/Speech_Synthesis_Markup_Language https://www.w3.org/TR/speech-synthesis/ """ langShort = 'en' if language: langShort = language[:2] if not gender: gender = 'neutral' else: if langShort == 'en': gender = gender.lower() if 'he/him' in gender: gender = 'male' elif 'she/her' in gender: gender = 'female' else: gender = 'neutral' content = _addSSMLemphasis(content) voiceParams = 'name="' + displayName + '" gender="' + gender + '"' return '\n' + \ '\n' + \ ' \n' + \ ' ' + \ instanceTitle + ' inbox\n' + \ ' \n' + \ '

\n' + \ ' \n' + \ ' \n' + \ ' ' + content + '\n' + \ ' \n' + \ ' \n' + \ '

\n' + \ '\n' def getSSMLbox(baseDir: str, path: str, domain: str, systemLanguage: str, instanceTitle: str, boxName: str) -> str: """Returns SSML for the given timeline """ nickname = path.split('/users/')[1] if '/' in nickname: nickname = nickname.split('/')[0] speakerFilename = \ baseDir + '/accounts/' + nickname + '@' + domain + '/speaker.json' if not os.path.isfile(speakerFilename): return None speakerJson = loadJson(speakerFilename) if not speakerJson: return None gender = None if speakerJson.get('gender'): gender = speakerJson['gender'] return _speakerEndpointSSML(speakerJson['name'], speakerJson['summary'], speakerJson['say'], speakerJson['imageDescription'], speakerJson['detectedLinks'], systemLanguage, instanceTitle, gender)