From 4c402ab5021b4cbc51d3303c3290f8fa8f3ed545 Mon Sep 17 00:00:00 2001
From: Rob van Son <r.j.j.h.vanson@gmail.com>
Date: Fri, 29 May 2015 15:36:31 +0200
Subject: [PATCH] Test of pinyin to number conversion for input

---
 InitialiseSGC2.praat |  3 ++-
 MainPage.praat       | 32 ++++++++++++++++++++++++++------
 sgc2.praat           |  6 ++++--
 3 files changed, 32 insertions(+), 9 deletions(-)

diff --git a/InitialiseSGC2.praat b/InitialiseSGC2.praat
index d4abfab..ba24fdc 100644
--- a/InitialiseSGC2.praat
+++ b/InitialiseSGC2.praat
@@ -211,7 +211,8 @@ procedure init_window
 		sgc.pinyin$ = ""
 		select Table 'wordlist$'
 		if sgc.currentWord > 0 and sgc.currentWord <= sgc.numberOfWords
-			sgc.pinyin$ = Get value... 'sgc.currentWord' Pinyin
+			call readPinyin 'sgc.currentWord'
+			sgc.pinyin$ = readPinyin.pinyin$
 			.outputName$ = "'sgc.saveAudio$'/'sgc.pinyin$'.wav"
 			if fileReadable(.outputName$)
 				sgc.recordedSound = Read from file: .outputName$
diff --git a/MainPage.praat b/MainPage.praat
index 7c3f949..6777db7 100644
--- a/MainPage.praat
+++ b/MainPage.praat
@@ -319,7 +319,8 @@ procedure generate_example
 	if sgc.currentWord > 0 and sgc.currentWord <= sgc.numberOfWords
 		select Table 'wordlist$'
 		.sound$ = Get value... 'sgc.currentWord' Sound
-		sgc.pinyin$ = Get value... 'sgc.currentWord' Pinyin
+		call readPinyin 'sgc.currentWord'
+		sgc.pinyin$ = readPinyin.pinyin$
 		if .sound$ = "-" or .sound$ = ""
 			.sound$ = sgc.pinyin$+".wav"
 		endif
@@ -365,7 +366,8 @@ procedure draw_tone_contour
 	select Table 'wordlist$'
 	if sgc.currentWord > 0 and sgc.currentWord <= sgc.numberOfWords
 		.sound$ = Get value... 'sgc.currentWord' Sound
-		sgc.pinyin$ = Get value... 'sgc.currentWord' Pinyin
+		call readPinyin 'sgc.currentWord'
+		sgc.pinyin$ = readPinyin.pinyin$
 		call drawToneContour 'sgc.pinyin$' 'config.register'
 		call reset_viewport
 		
@@ -385,7 +387,8 @@ procedure recognizeTone
 		endif
 			
 		.sound$ = Get value... 'sgc.currentWord' Sound
-		sgc.pinyin$ = Get value... 'sgc.currentWord' Pinyin
+		call readPinyin 'sgc.currentWord'
+		sgc.pinyin$ = readPinyin.pinyin$
 		call align_recordedSound 'sgc.pinyin$'
         call sgc_ToneProt 'recordedSound$' 'sgc.pinyin$' 'config.register' '.strict' 'config.language$'
         # sgc_ToneProt manipulates the sound given. Reconnect
@@ -448,7 +451,8 @@ procedure display_text .color$
 	
 	if sgc.currentWord > 0 and sgc.currentWord <= sgc.numberOfWords
 		.displayText$ = ""
-		.displayPinyin$ = Get value... 'sgc.currentWord' Pinyin
+		call readPinyin 'sgc.currentWord'
+		.displayPinyin$ = readPinyin.pinyin$
 		.displayChar$ = Get value... 'sgc.currentWord' Character
 		.displayTrans$ = Get value... 'sgc.currentWord' Translation
 		if .displayPinyin$ <> "-" and (config.displayPinyin or sgc.writeAll)
@@ -553,7 +557,7 @@ endproc
 # NEEDS WORK AND TESTING!!
 # Convert unicode Pinyin into tone numbers
 procedure pinyin2numbers .pinyin$
-	.intermediatePinyin$ = .pinyin$
+	.intermediatePinyin$ = replace_regex$(.pinyin$, "[A-Z]", "\L&", 0)
 	# Convert all special characters to numbers
 	# Tone 1
 	.intermediatePinyin$ = replace_regex$(.intermediatePinyin$, "ā([iaeouü]*)", "a\11", 0)
@@ -600,11 +604,27 @@ procedure pinyin2numbers .pinyin$
 	.intermediatePinyin$ = replace_regex$(.intermediatePinyin$, "([aeuiov]+)([^0-9aeuiov]|$)", "\10\2", 0)
 
 	# Move numbers to the end of the syllable
-	.intermediatePinyin$ = replace_regex$(.intermediatePinyin$, "([aeuiov])([0-9])((ng?)([^aeuiov]|$))?", "\1\3\2\4", 0)
+	# Syllables ending in (ng?)
+	.intermediatePinyin$ = replace_regex$(.intermediatePinyin$, "([aeuiov])([0-9])(ng?)([^aeuiov]|$)", "\1\3\2\4", 0)
+	# Syllables ending in r
+	.intermediatePinyin$ = replace_regex$(.intermediatePinyin$, "([aeuiov])([0-9])(r)([^aeuiov]|$)", "\1\3\2\4", 0)
+	# Other syllables
+	.intermediatePinyin$ = replace_regex$(.intermediatePinyin$, "([aeuiov])([0-9])(n)", "\1\3\2", 0)
+	# Remove quotes etc
+	.intermediatePinyin$ = replace_regex$(.intermediatePinyin$, "[\\']", "", 0)
 	
 	.numberstext$ = .intermediatePinyin$
 endproc
 
+procedure readPinyin .currentWord
+	select Table 'wordlist$'
+	.pinyin$ = Get value... '.currentWord' Pinyin
+	if index_regex(.pinyin$, "[āēūīōǖáéúíóǘǎěǔǐǒǚàèùìòǜåe̊ůi̊o̊ü̊̊]")
+		call pinyin2numbers '.pinyin$'
+		.pinyin$ = pinyin2numbers.numberstext$
+	endif 
+endproc
+
 # Includes
 include ToneProt/SGC_ToneProt.praat
 include ToneProt/DrawToneContour.praat
diff --git a/sgc2.praat b/sgc2.praat
index 463157b..299db30 100644
--- a/sgc2.praat
+++ b/sgc2.praat
@@ -714,7 +714,8 @@ procedure count_syllables
 	select Table 'wordlist$'
 	if sgc.currentWord > 0 and sgc.currentWord <= sgc.numberOfWords
 		.sound$ = Get value... 'sgc.currentWord' Sound
-		.pinyin$ = Get value... 'sgc.currentWord' Pinyin
+		call readPinyin 'sgc.currentWord'
+		.pinyin$ = readPinyin.pinyin$
 	endif
 	call add_missing_neutral_tones '.pinyin$'
 	.pinyin$ = add_missing_neutral_tones.pinyin$
@@ -843,7 +844,8 @@ procedure record_sound .recordingTime
 			.pinyin$ = ""
 			select Table 'wordlist$'
 			if sgc.currentWord > 0 and sgc.currentWord <= sgc.numberOfWords
-				.pinyin$ = Get value... 'sgc.currentWord' Pinyin
+				call readPinyin 'sgc.currentWord'
+				.pinyin$ = readPinyin.pinyin$
 				.outputName$ = "'sgc.saveAudio$'/'.pinyin$'.wav"
 				select sgc.recordedSound
 				Save as WAV file: .outputName$
-- 
2.11.4.GIT