Log spaCy, model and model version
Change-Id: I1809448c5b2d167943787d5f0c8ea2db6e42f0b1
diff --git a/systems/parse_spacy.py b/systems/parse_spacy.py
index e4259bf..d2bf1b1 100644
--- a/systems/parse_spacy.py
+++ b/systems/parse_spacy.py
@@ -138,6 +138,16 @@
spacy_de.tokenizer = WhitespaceTokenizer(spacy_de.vocab) # We won't re-tokenize to respect how the source CoNLL are tokenized!
write_out = open(args.output_file, "w")
lemmatizer = GermaLemma()
+
+ # Log version information
+ logger.info(f"spaCy version: {spacy.__version__}")
+ logger.info(f"spaCy model: {args.spacy_model}")
+ logger.info(f"spaCy model version: {spacy_de.meta.get('version', 'unknown')}")
+ try:
+ import germalemma
+ logger.info(f"GermaLemma version: {germalemma.__version__}")
+ except AttributeError:
+ logger.info("GermaLemma version: unknown (no __version__ attribute)")
if args.text_file: write_plain = open(args.text_file, "w")
if ".gz" == args.input_file[-3:]:
diff --git a/systems/parse_spacy3.py b/systems/parse_spacy3.py
index 9696a9d..1c71db6 100644
--- a/systems/parse_spacy3.py
+++ b/systems/parse_spacy3.py
@@ -125,6 +125,16 @@
write_out = open(args.output_file, "w")
lemmatizer = GermaLemma()
+ # Log version information
+ print(f"spaCy version: {spacy.__version__}")
+ print(f"spaCy model: {args.spacy_model}")
+ print(f"spaCy model version: {spacy_de.meta.get('version', 'unknown')}")
+ try:
+ import germalemma
+ print(f"GermaLemma version: {germalemma.__version__}")
+ except AttributeError:
+ print("GermaLemma version: unknown (no __version__ attribute)")
+
if ".gz" == args.input_file[-3:]:
in_file = fu.expand_file(args.input_file)
else:
diff --git a/systems/parse_spacy_pipe.py b/systems/parse_spacy_pipe.py
index 74b110b..a4f1693 100644
--- a/systems/parse_spacy_pipe.py
+++ b/systems/parse_spacy_pipe.py
@@ -172,6 +172,16 @@
spacy_de.tokenizer = WhitespaceTokenizer(spacy_de.vocab) # We won't re-tokenize to respect how the source CoNLL are tokenized!
lemmatizer = GermaLemma()
+ # Log version information
+ logger.info(f"spaCy version: {spacy.__version__}")
+ logger.info(f"spaCy model: {args.spacy_model}")
+ logger.info(f"spaCy model version: {spacy_de.meta.get('version', 'unknown')}")
+ try:
+ import germalemma
+ logger.info(f"GermaLemma version: {germalemma.__version__}")
+ except AttributeError:
+ logger.info("GermaLemma version: unknown (no __version__ attribute)")
+
start = time.time()
total_processed_sents = 0