parseESTERSyncV2_text.py 17.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11
#!/usr/bin/env python
# -*- coding: utf-8 -*-

from xml.etree import ElementTree as ET
from sys import argv
from num2words import num2words
from unidecode import unidecode
import re
import os.path

def transformation_text(text):
12 13 14 15 16
    # ESTER Problem "Mohamed v" ===> "Mohammed cinq"
    text=re.sub("mohammed vi","mohamed six",text)
    text=re.sub("mohammed v","mohamed cinq",text)
    # map all "mohamed" to "mohammed"
    text=re.sub("mohamed","mohammed",text)
17
    # character normalization:
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
18
    text=re.sub("&","et",text)
19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
    text=re.sub("æ","ae",text)
    text=re.sub("œ","oe",text)
    #if "###" in text or len(re.findall(r"\[.+\]", text)) > 0 or \
    #    len(re.findall(r"\p{L}+-[^\p{L}]|\p{L}+-$",text)) > 0 \
    #    or len(re.findall("[^\p{L}]-\p{L}+|^-\p{L}+", text)) > 0:
    #    bool=False
    #else:
    # ^^ remove
    text=re.sub(r"\^+","",text)
    text=re.sub(r"\_+","",text)
    # 4x4
    # Remove noise sound (BIP) over Name of places and person
    #text = re.sub(r"¤[^ ]+|[^ ]+¤|¤", "", text.strip())
    if len(re.findall(r"\dx\d",text))>0:
        text=re.sub(r"x","  ",text)
    if len(re.findall("\d+h\d+",text))>0:
        heures=re.findall("\d+h\d+",text)
        for h in heures:
            split_h=h.split('h')
            text_rep=split_h[0]+' heure '+split_h[1]
            text=text.replace(h, text_rep)
    text=re.sub(r',',' ',text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
41
    # remove silence character : OK
42
    #text=re.sub(r"(/.+/","remplacer par la 1er",text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
43
    # Liaison non standard remarquable
44
    text=re.sub(r'=','',text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
45
    # Comment Transcriber
46 47 48
    text=re.sub(r'\{.+\}','',text)
    text=re.sub(r'\(.+\}','',text)
    #print "detecter (///|/|<|>)"
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
49
    # Remove undecidable variant heared like on (n') en:
50 51 52 53 54 55 56
    text=re.sub(r"\(.+\)|\(\)","",text)
    #text = re.sub(r"(\+|[*]+|///|/|<|>)", "", text.strip())
    #text=re.sub(r"-|_|\."," ",text.strip())
    text=re.sub(r'(O.K.)','ok',text)
    text = re.sub(r'(O.K)', 'ok', text)
    # Replace . with ''
    text=re.sub(r'\.|,|;','',text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
57
    #text=re.sub(r"{[^{]+}"," ",text.strip())
58 59 60 61
    # Remove ? ! < > : OK
    #<[^\p{L}]|[^\p{L}]>|#+|<\p{L}+[ ]|<\p{L}+$
    text=re.sub(r":|\?|/|\!|#+","",text)
    text=re.sub(r"%","pour cent",text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
62
    # replace silence character with <sil> : OK
63 64 65 66 67
    #text=re.sub(r"(\+)", "<sil>", text)
    #text=re.sub(r"(\+)", "!SIL", text)
    #text=re.sub(r"(///)", "!SIL", text)
    #text=re.sub(r"(///)", "<long-sil>", text)
    #if len(re.findall(r"/.+/", text)) > 0:
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
68
    #print "AVANT***********"+text
69
    #    for unchoosen_text in re.findall(r"/.+/", text):
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
70
    # choose first undecideble word
71 72
    #        unchoosen_word=unchoosen_text.split(',')
    #        for choosen_word in unchoosen_word:
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
73
    # isn't incomprehensible word
74 75 76
    #            if len(re.findall(r"\*+|\d+", choosen_word))==0:
    #                choosen_word = choosen_word.replace('/', '')
    #                text = text.replace(unchoosen_text, choosen_word)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
77 78
    #print "Apres************"+text
    # Remove noise sound (BIP) over Name of places and person
79
    #text=re.sub(r"(¤.+¤)",'<NOISE>',text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
80
    # replace unkown syllable
81
    text=re.sub(r"\*+","",text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
82
    # cut of recording : OK
83
    #text=re.sub(r"\$+","",text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
84
    # remove " character: OK
85 86
    text = re.sub(r"\"+", "", text)
    # t 'avais
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
87
    text = re.sub(r"[ ]\'", "\'", text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
88
    text = re.sub(r"\'", "\' ", text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
89
    # for example : A43
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
90 91
    #num_list = re.findall("\w+?-?\d+", text)
    num_list = re.findall("[a-zA-Z|à|è|é|ù]+\'*[a-zA-Z|à|é|è|ù]*[-]?\d+""", text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
92 93
    if len(num_list) > 0:
        for s in num_list:
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
94
            split_between_char_int=re.findall(r'([a-zA-Z|à|é|è|ù]+\'*[a-zA-Z]*)-?(\d+)',s)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
95 96 97 98
            num_in_word = num2words(int(split_between_char_int[0][1]), lang='fr')
            #num_in_word=normalize('NFKD', num_in_word).encode('ascii', 'ignore')
            #text = re.sub(r"(^|[ ])"+str(s)+"([ ]|$)"," " + str(split_between_char_int[0][0]) +" "+ str(num_in_word) + " ",text)
            text = re.sub(r"(^|[ ])"+str(s)," " + str(split_between_char_int[0][0]) +" "+ str(num_in_word) + " ",text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
99 100
    #num_list = re.findall("\d+\w+", text)
    num_list = re.findall("\d+[a-zA-Z|à|é|è|ù]+\'*[a-zA-Z|à|é|è|ù]*", text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
101 102
    if len(num_list) > 0:
        for s in num_list:
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
103 104
            split_between_char_int=re.findall(r'(\d+)([a-zA-Z|à|é|è|ù]+\'*[a-zA-Z|à|é|è|ù]*)',s)
            #re.findall(r'\d+\w+',s)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
105 106 107 108 109
            num_in_word = num2words(int(split_between_char_int[0][0]), lang='fr')
            #num_in_word=normalize('NFKD', num_in_word).encode('ascii', 'ignore')
            #text = re.sub(r"(^|[ ])"+str(s)+"([ ]|$)"," " + str(split_between_char_int[0][0]) +" "+ str(num_in_word) + " ",text)
            text = re.sub(r"(^|[ ])"+str(s)," "+ str(num_in_word)+ " " + str(split_between_char_int[0][1]) + " ",text)
    # convert number if exist : OK
110 111 112 113 114 115 116 117
    num_list = re.findall("\d+", text)
    if len(num_list) > 0:
        #print text
        #print "********************************* NUM2WORD"
        for num in num_list:
            num_in_word = num2words(int(num), lang='fr')
            #num_in_word=normalize('NFKD', num_in_word).encode('ascii', 'ignore')
            text = re.sub(r"(^|[ ])"+str(num)+"([ ]|$)"," " + str(num_in_word) + " ",text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
118
            #print text
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
119
        # replace n succesive spaces with one space. : OK
120 121 122
    text=re.sub(r"\s{2,}"," ",text)
    text=re.sub(r" $","",text)
    text=re.sub("^ ", '', text)
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
123
    # change bounding | to < and > : OK
124 125
    #balise=set(re.findall(r"\|\w+_?\w+\|",text))
    #if len(balise)>0:
Abdelwahab HEBA's avatar
Abdelwahab HEBA committed
126
    #print(balise)
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
    #    for b in balise:
    #        new_balise='<'+b[1:len(b)-1]+'>'
    #        text=text.replace(b,new_balise)
    #print(text)
    # c'est l'essaim ....
    text=text.lower()
    text=re.sub("[ ]-|-$","",text)
    return text
if __name__=="__main__":
    # Inputs
    file_trs=argv[1]
    #outdir=argv[2]
    basename=os.path.basename(file_trs.split('.')[0])
    # Read Trans File
    tree_trs = ET.parse(file_trs)
    trsdoc= tree_trs.getroot()
    #============================ Read MetaData =======================================
    #============================ Topic section (ID,DESC) =======================================
    for topic in trsdoc.iter('Topic'):
        topic_id=unidecode(topic.get('id'))
        topic_desc=unidecode(topic.get('desc'))
        #print(str(basename)+" "+topic_id+" "+topic_desc+"\n")
        #topic_file.write(str(basename)+" "+topic_id+" "+topic_desc+"\n")
    #============================ Speaker section (ID,GENDER) ===================================
    speaker_id=[]
    #namespk=[]
    speaker_gender=[]
    for spk in trsdoc.iter('Speaker'):
        id_spk=spk.get('id')
        #name_spk=unidecode(spk.get('name'))
        if spk.findall('type')==[]:
            gender="m"
        else:
            gender=unidecode(spk.get('type'))
            if gender =="female":
                gender="f"
            else:
                gender="m"
        #if isinstance(name_spk,str):
        #print(type(name_spk))
        #name_spk=normalize('NFKD', name_spk).encode('ascii', 'ignore')
        speaker_id.append(id_spk.replace(" ",""))
        speaker_gender.append([id_spk.replace(" ",""),gender.lower()])
        #namespk.append(name_spk.lower().replace(" ",""))
    #============================ Catch Transcription Segment and Topic Section ==================
    text=""
    Turn_count=0
    count=0
    has_attrib_speaker=False
    # set for uniq add
    Spk_that_contribute_to_meeting=set([])
    start_utt=0
    end_utt=0
    #Not used
    section_start_time=0
    section_end_time=0
    section_type=""
    section_topic=""
    nb_section=0
    spkr="spk1"
    for Element in trsdoc.iter():
        #OK validation
        #print("Print lekbirr "+str(Element.tail))
        #print("Print lekbirr "+str(Element.tail))
        if Element.tag=="Section":
            if nb_section>0:
                text = transformation_text(text)
                # File wav.scp
                # File utt2spk
                # File text
                # File speaker_gender
                if text!=""and has_attrib_speaker:
                    Spk_that_contribute_to_meeting.add(spkr)
                    #print("SAVED BY SECTION "+seg_id+" "+text)
                    seg_id = str(basename) + '_%s-%03d_Section%02d_Topic-%s_Turn-%03d_seg-%07d' % (
                    str(re.sub('\d+','',spkr)),int(re.sub('[a-zA-Z]','',spkr)),int(nb_section),str(section_topic), int(Turn_count), int(count))
                    spkr_id = str(basename)+'_%s-%03d' % (str(re.sub('\d+','',spkr)),int(re.sub('[a-zA-Z]','',spkr)))
                    #segments_file.write(seg_id+" "+basename+" "+str(start_utt)+" "+str(section_end_time)+"\n")
                    start_utt=section_end_time
                    #utt2spk_file.write(seg_id+" "+spkr_id+"\n")
                    #text_file.write(seg_id+" "+text+"\n")
                    print(text)
                    text=""
            # New section
            section_start_time=Element.get('startTime')
            section_end_time=Element.get('endTime')
            section_type=unidecode(Element.get('type'))
            #if Element.findall('topic')==[]:
            #    section_topic="None"
            #else:
            section_topic=unidecode(str(Element.get('topic')))
            if section_topic=="":
                section_topic="None"
            Turn_count=0
            count=0
            nb_section+=1
        elif Element.tag=="Turn":
            # if the turn is the spoken turn , not musical segment or noise
            #print(str(Element.tag))
            #print(Element.attrib)
            #print(Element.get("speaker"))
            if not "speaker" in Element.attrib:
                #print("pas de champ speaker")
                has_attrib_speaker=False
            else:
                if Element.get('speaker')=="":
                    #print("Cest vide: "+Element.get('speaker'))
                    has_attrib_speaker=False
                else:
                    #print(Element.get('speaker'))
                    # If the latest Utterance of previous Speaker is the latest one of his Turn speech
                    if Turn_count>0:
                        seg_id = str(basename) + '_%s-%03d_Section%02d_Topic-%s_Turn-%03d_seg-%07d' % (
                        str(re.sub('\d+','',spkr)),int(re.sub('[a-zA-Z]','',spkr)),int(nb_section),str(section_topic), int(Turn_count), int(count))
                        spkr_id = str(basename)+'_%s-%03d' % (str(re.sub('\d+','',spkr)),int(re.sub('[a-zA-Z]','',spkr)))
                        text = transformation_text(text)
                        # File wav.scp
                        # File utt2spk
                        # File text
                        # File speaker_gender
                        if bool and text!="":
                            #print("SAVED BY TURN "+seg_id+" "+text)
                            Spk_that_contribute_to_meeting.add(spkr)
                            #segments_file.write(seg_id+" "+basename+" "+str(start_utt)+" "+str(endTime)+"\n")
                            start_utt=endTime
                            #utt2spk_file.write(seg_id+" "+spkr_id+"\n")
                            #text_file.write(seg_id+" "+text+"\n")
                            print(text)
                            text=""
                        count = 0
                    # Get id_spkr
                    #print(Element.get('speaker'))
                    spkr=Element.get('speaker')
                    #print file_trs
                    has_attrib_speaker=True
                    spkr=spkr.split()[0]
                    #print spkr
                    # Get StartSegment
                    startTime = Element.get('startTime')
                    # Get EndSegment
                    endTime = Element.get('endTime')
                    # count sync for computing start and end utterance
                    Turn_count = Turn_count+1
        elif has_attrib_speaker:
            #print("Je rentre dans has_attrib_speaker et element.tail not null")
            #print(str(Element.tag))
            #print(str(Element.tail))
274
            if Element.tag=="Sync":
275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372
                #print("Je rentre Sync+Background"+ text +"| et le next c'est "+ Element.tail)
                #print(Element.tag+" "+Element.tail)
                Time_start_current_sync=Element.get('time')
                #if count>0:
                    #print("save after Turn")
                    #print(str(basename))
                    #print(str(section_topic))
                    #print(str(spkr))
                    #print(str(int(Turn_count)))
                    #print(str(int(count)))
                    #print text
                    ### Save Files For Kaldi ###
                seg_id = str(basename) + '_%s-%03d_Section%02d_Topic-%s_Turn-%03d_seg-%07d' % (
                str(re.sub('\d+','',spkr)),int(re.sub('[a-zA-Z]','',spkr)),int(nb_section),str(section_topic), int(Turn_count), int(count))
                spkr_id = str(basename)+'_%s-%03d' % (str(re.sub('\d+','',spkr)),int(re.sub('[a-zA-Z]','',spkr)))
                text = transformation_text(text)
                #print("Sync or Background: wizzz "+text)
                end_utt=Time_start_current_sync
                if text!="":
                    #print("SAVED BY SYNC or BACKGROUND "+seg_id+" "+text)
                    Spk_that_contribute_to_meeting.add(spkr)
                    #segments_file.write(seg_id+" "+basename+" "+str(start_utt)+" "+str(end_utt)+"\n")
                    #utt2spk_file.write(seg_id+" "+spkr_id+"\n")
                    #text_file.write(seg_id+" "+text+"\n")
                    print(text)
                    text=""
                    count+=1
                start_utt=Time_start_current_sync
                text=Element.tail.replace('\n', '')
                #print(count)
                #count+=1
            elif Element.tag=="Comment" or Element.tag=="Background":
                text=text+" "+Element.tail.replace('\n', '')
            elif Element.tag=="Event":
            #    if Element.get('type')=='noise':
                    # ===== Respiration
                if Element.get('desc')=='r' or Element.get('desc')=='i' or Element.get('desc')=='e' or Element.get('desc')=='n':
                    text=text+" "+Element.tail.replace('\n', '')
                elif Element.get('desc')=='pf':
                    text=text+" "+Element.tail.replace('\n', '')
                # ===== Bruits bouches
                elif Element.get('desc')=='tx':
                    text=text+" "+Element.tail.replace('\n', '')
                elif Element.get('desc')=='bg':
                    text=text+" "+Element.tail.replace('\n', '')
                elif Element.get('desc')=='bb':
                    text=text+" "+Element.tail.replace('\n', '')
                elif Element.get('desc')=='rire':
                    text=text+" "+Element.tail.replace('\n', '')
                elif Element.get('desc')=='sif':
                    text=text+" "+Element.tail.replace('\n', '')
                elif Element.get('desc')=='ch' or Element.get('desc')=='ch-':
                    text=text+" "+Element.tail.replace('\n', '')
                # ====== Bruit exterieus a l'acte de parole
                elif Element.get('desc')=='b' or Element.get('desc')=='pap' or Element.get('desc')=='mic' or Element.get('desc')=='conv':
                    text=text+" "+Element.tail.replace('\n', '')
                elif Element.get('desc')=='top':
                    text=text+" "+Element.tail.replace('\n', '')
                # "pi" intellegible "pif" inaudible voir doc transcriber
            #elif Element.get('type')=='pronounce':
            #    text=text+" "+Element.tail.replace('\n', '')
            # desc="EN"
            #if Element.get('type')=='language':
            #    text=text+" "+Element.tail.replace('\n', '')
            #if Element.tag=="Who":
                else:
                    text=text+" "+Element.tail.replace('\n', '')
    if count > 0 and has_attrib_speaker and not Element.tail is None:
        #print text
        ### Save Files For Kaldi ###
        seg_id = str(basename) + '_%s-%03d_Section%02d_Topic-%s_Turn-%03d_seg-%07d' % (
        str(re.sub('\d+','',spkr)),int(re.sub('[a-zA-Z]','',spkr)),int(nb_section),str(section_topic), int(Turn_count), int(count))
        #seg_id = str(basename) + '_spk-%03d_Turn-%03d_seg-%07d' % (
        #int(spkr.split('spk')[1]), int(Turn_count), int(count))
        spkr_id = str(basename)+'_%s-%03d' % (str(re.sub('\d+','',spkr)),int(re.sub('[a-zA-Z]','',spkr)))
        text = transformation_text(text)
        if bool and text != "":
            #print("Last SAVE"+seg_id+" "+text)
            #segments_file.write(seg_id+" "+basename+" "+str(start_utt)+" "+str(endTime)+"\n")
            #utt2spk_file.write(seg_id+" "+spkr_id+"\n")
            #text_file.write(seg_id+" "+text+"\n")
            print(text)
    # Gender file edition
    #print(Spk_that_contribute_to_meeting)
    #print(len(Spk_that_contribute_to_meeting))
    #print(speaker_gender)
    #print(len(speaker_gender))
    #for spk in speaker_gender:
    #    if spk[0] in Spk_that_contribute_to_meeting:
    #        spk_id = str(basename)+'_%s-%03d' % (str(re.sub('\d+','',spk[0])),int(re.sub('[a-zA-Z]','',spk[0])))
    #        spk2gender.write(spk_id+" "+spk[1]+"\n")
    #wav_scp.write(basename+" sox "+os.path.dirname(file_trs) + '/' + basename + '.wav'+" -t wav -r 16000 -c 1 - |\n")
    #segments_file.close()
    #utt2spk_file.close()
    #text_file.close()
    #wav_scp.close()
    #topic_file.close()
    #spk2gender.close()