From fd0f9794f1f7c5a0cd79c2af07a267847a9a06e1 Mon Sep 17 00:00:00 2001
From: voje <kristjan.voje@gmail.com>
Date: Thu, 28 Feb 2019 10:15:14 +0100
Subject: [PATCH] added logger

---
 tools/gen_json.py  | 19 ++++++++-----------
 tools/parse_all.py | 14 +++++++-------
 tools/tools.cfg    |  1 +
 3 files changed, 16 insertions(+), 18 deletions(-)

diff --git a/tools/gen_json.py b/tools/gen_json.py
index 4c489a3..5475f67 100644
--- a/tools/gen_json.py
+++ b/tools/gen_json.py
@@ -3,12 +3,7 @@ from parser.parser import Parser
 import configparser
 import json
 import sys
-
-# defaults
-ORIGPATH = Path("../data/kres_example")  # we need the IDs
-INPATH = Path("../data/kres_example_srl")
-OUTPATH = Path("../data/kres_example_json")
-DEBUG = False
+import logging
 
 # parse config
 config = configparser.ConfigParser()
@@ -16,8 +11,11 @@ config.read("tools.cfg")
 ORIGPATH = Path(config["tools"]["kres_orig"])
 INPATH = Path(config["tools"]["kres_srl"])
 OUTPATH = Path(config["tools"]["kres_json"])
+LOGFILE = Path(config["tools"]["logfile"])
 DEBUG = config["tools"]["debug"] == "True"
 
+logging.basicConfig(filename=LOGFILE, level=logging.INFO)
+
 def get_origfile(filename):
 	for origfile in ORIGPATH.iterdir():
 		if filename.name.split('.')[0] == origfile.name.split('.')[0]:
@@ -46,8 +44,7 @@ def match_sentence_id(sentence, orig_dict):
 	raise KeyError
 
 def get_dep_rel(token):
-	if DEBUG:
-		print(token)
+	logging.debug(token)
 	for i, field in enumerate(token[14:]):
 		if field != "_":
 			return {
@@ -61,7 +58,7 @@ def get_dep_rel(token):
 par = Parser()
 OUTPATH.mkdir(exist_ok=True)
 
-print("Start generating .josn files.")
+logging.info("Start generating .josn files.")
 for infile in [x for x in INPATH.iterdir() if x.is_file()]:
 	origfile = get_origfile(infile)
 	orig_dict = par.parse_tei(origfile)
@@ -99,5 +96,5 @@ for infile in [x for x in INPATH.iterdir() if x.is_file()]:
 	outfile = (OUTPATH / infile.name).with_suffix(".json")
 	with outfile.open("w") as fp:
 		json.dump(outdata, fp)
-		print("SRL relations written to: ", outfile)
-print("Finished generating .json files.")
\ No newline at end of file
+		logging.info("SRL relations written to: ", outfile)
+logging.info("Finished generating .json files.")
diff --git a/tools/parse_all.py b/tools/parse_all.py
index 6b76782..1218b67 100644
--- a/tools/parse_all.py
+++ b/tools/parse_all.py
@@ -6,10 +6,7 @@ import re
 import sys
 import cProfile
 import configparser
-
-# some defaults
-INDIR = Path("../data/kres_example")
-OUTDIR = Path("../data/kres_example_tsv")
+import logging
 
 SSJ500K_2_1 = 27829  # number of sentences
 par = Parser()
@@ -19,6 +16,9 @@ config = configparser.ConfigParser()
 config.read("tools.cfg")
 INDIR = Path(config["tools"]["kres_orig"])
 OUTDIR = Path(config["tools"]["kres_tsv"])
+LOGFILE = Path(config["tools"]["logfile"])
+
+logging.basicConfig(filename=LOGFILE, level=logging.INFO)
 
 """
 print("parsing ssj")
@@ -28,14 +28,14 @@ ssj_dict = par.parse_tei(ssj_file)
 print("end parsing ssj")
 """
 
-print("parsing kres")
+logging.info("parsing kres")
 # kres_file = "../data/kres_example/F0019343.xml.parsed.xml"
 OUTDIR.mkdir(exist_ok=True)
 
 infiles = [x for x in INDIR.iterdir() if x.is_file()]
 for i, kres_file in enumerate(infiles):
 
-    print("Processing file ({}/{}): {}".format(i+1, len(infiles), str(kres_file)))
+    logging.info("Processing file ({}/{}): {}".format(i+1, len(infiles), str(kres_file)))
     res_dict = par.parse_tei(kres_file)
     kres_out_str = ""
 
@@ -44,4 +44,4 @@ for i, kres_file in enumerate(infiles):
 
     with (OUTDIR / kres_file.name).with_suffix(".tsv").open("wb+") as fp:
         fp.write(kres_out_str.encode("utf-8"))
-print("end parsing kres")
+logging.info("end parsing kres")
diff --git a/tools/tools.cfg b/tools/tools.cfg
index b32cb78..d684723 100644
--- a/tools/tools.cfg
+++ b/tools/tools.cfg
@@ -3,4 +3,5 @@ kres_orig = /kres_mount/kres_parsed/tei
 kres_tsv = ../data/kres_out/1_tsv
 kres_srl = ../data/kres_out/_srl
 kres_json = ../data/kres_out/final_json
+logfile = "../progress.log"
 debug = False