Fix engrafo version

* add dummy config
* add docker-py as dependency
* fix busy box timeout syntax
* fix engrafo version

Author: Ubuntu

clean_html.sh

@@ -4,4 +4,4 @@ SOURCE=$(realpath "$1")
 mkdir -p $(dirname "$2")
 OUTPUT=$(realpath "$2")
 
-docker run --rm -v "$SOURCE":/files/index.html:ro --entrypoint '' zenika/alpine-chrome:73 timeout -t 20 -s KILL chromium-browser --headless --disable-gpu --disable-software-rasterizer --no-sandbox --timeout=30000 --dump-dom /files/index.html > "$OUTPUT"
+docker run --rm -v "$SOURCE":/files/index.html:ro --entrypoint '' zenika/alpine-chrome:73 timeout -s KILL 20 chromium-browser --headless --disable-gpu --disable-software-rasterizer --no-sandbox --timeout=30000 --dump-dom /files/index.html > "$OUTPUT"

docker-latex2html.sh

@@ -6,4 +6,4 @@ OUTPUT=$(realpath "$2")		#~/arxiv/htmls/1701/1701.xyz.html
 OUTPUT_DIR=$(dirname "$OUTPUT")	#~/arxiv/htmls/1701
 FILENAME=$(basename "$OUTPUT") #1701.xyz.html
 
-docker run --rm -v $PWD/latex2html.sh:/files/latex2html.sh:ro -v $PWD/guess_main.py:/files/guess_main.py:ro -v $PWD/patches:/files/patches:ro -v "$SOURCE_DIR":/files/ro-source:ro -v "$OUTPUT_DIR":/files/htmls arxivvanity/engrafo /files/latex2html.sh "$FILENAME"
+docker run --rm -v $PWD/latex2html.sh:/files/latex2html.sh:ro -v $PWD/guess_main.py:/files/guess_main.py:ro -v $PWD/patches:/files/patches:ro -v "$SOURCE_DIR":/files/ro-source:ro -v "$OUTPUT_DIR":/files/htmls arxivvanity/engrafo:b3db888fefa118eacf4f13566204b68ce100b3a6 /files/latex2html.sh "$FILENAME"

environment.yml

@@ -17,3 +17,4 @@ dependencies:
 - ipython=7.5.0
 - joblib=0.13.2
 - python-magic=0.4.15
+- docker-py=4.1.0

sota_extractor2/config.py

@@ -1,10 +0,0 @@
-import logging
-from .. import config # to get logging init
-
-logger = logging.getLogger(__name__)
-
-try:
-    from db import *
-except:
-    logger.info("Unable to intialise django falling back to json data")
-    from json import *