[scripts] RNNLM script fix: to accept successive spaces in configs (e…

…tc.) kaldi-asr#2595 (kaldi-asr#2597)

egs/wsj/s5/steps/libs/nnet3/xconfig/basic_layers.py

@@ -748,7 +748,8 @@ def check_configs(self):
         if self.config['target-rms'] < 0.0:
             raise RuntimeError("target-rms has invalid value {0}"
                                .format(self.config['target-rms']))
-        if self.config['learning-rate-factor'] <= 0.0:
+        if (self.config['learning-rate-factor'] != '' and
+            self.config['learning-rate-factor'] <= 0.0):
             raise RuntimeError("learning-rate-factor has invalid value {0}"
                                .format(self.config['learning-rate-factor']))
 

scripts/rnnlm/choose_features.py

@@ -11,7 +11,7 @@
 sys.stdout = open(1, 'w', encoding='latin-1', closefd=False)
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="This script chooses the sparse feature representation of words. "
                                              "To be more specific, it chooses the set of features-- you compute "

scripts/rnnlm/get_special_symbol_opts.py

@@ -9,7 +9,7 @@
 import sys
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="This script checks whether the special symbols "
                                  "appear in words.txt with expected values, if not, it will "

scripts/rnnlm/get_unigram_probs.py

@@ -8,7 +8,7 @@
 import sys
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="This script gets the unigram probabilities of words.",
                                  epilog="E.g. " + sys.argv[0] + " --vocab-file=data/rnnlm/vocab/words.txt "

scripts/rnnlm/get_vocab.py

@@ -9,7 +9,7 @@
 sys.stdout = open(1, 'w', encoding='latin-1', closefd=False)
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="This script get a vocab from unigram counts "
                                  "of words produced by get_unigram_counts.sh",

scripts/rnnlm/get_word_features.py

@@ -10,7 +10,7 @@
 from collections import defaultdict
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="This script turns the words into the sparse feature representation, "
                                              "using features from rnnlm/choose_features.py.",

scripts/rnnlm/prepare_split_data.py

@@ -9,7 +9,7 @@
 import sys
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="This script prepares files containing integerized text, "
                                  "for consumption by nnet3-get-egs.",

scripts/rnnlm/show_word_features.py

@@ -9,7 +9,7 @@
 sys.stdout = open(1, 'w', encoding='latin-1', closefd=False)
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="This script turns the word features to a human readable format.",
                                  epilog="E.g. " + sys.argv[0] + "exp/rnnlm/word_feats.txt exp/rnnlm/features.txt "

scripts/rnnlm/validate_features.py

@@ -8,7 +8,7 @@
 import sys
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="Validates features file, produced by rnnlm/choose_features.py.",
                                  epilog="E.g. " + sys.argv[0] + " exp/rnnlm/features.txt",

scripts/rnnlm/validate_text_dir.py

@@ -8,7 +8,7 @@
 import sys
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="Validates data directory containing text "
                                  "files from one or more data sources, including dev.txt.",

scripts/rnnlm/validate_word_features.py

@@ -8,7 +8,7 @@
 import sys
 
 import re
-tab_or_space = re.compile('[ \t]')
+tab_or_space = re.compile('[ \t]+')
 
 parser = argparse.ArgumentParser(description="Validates word features file, produced by rnnlm/get_word_features.py.",
                                  epilog="E.g. " + sys.argv[0] + " --features-file=exp/rnnlm/features.txt "