#! /bin/bash
#
# Configuration parameters for MPG (Max Planar Graph) parsing.
#
# This is an example config file; you might want to use one of the
# preconfigured files, e.g. `3-mpg-conf-fake.sh` or `3-mpg-conf-en.sh`
# --------------
#
# Directory where corpora files can be found. This can be anywhere;
# however, it is most convenient to make it the same as the output
# directory configured in the previous (pair-counting) stage. This is
# the setting of the `$COMPLETED_DIR` of the earlier stage.
export CORPORA_DIR=$TEXT_DIR/pair-counted
# Enable or disable sentence splitting.
# If the text corpora have one sentence per line, then splitting is not
# needed. If the corpora are arranged into paragraphs (as conventional
# for natural language), then the paragraphs must be split into distinct
# sentences.
export SENTENCE_SPLIT=false
# If splitting is enabled, then specify the splitting language. Choices
# include `en`, `fr`, `pl` and many more; see the splitter directory for more.
export SPLIT_LANG=en
# IPv4 hostname and port number of where the cogserver is running.
export HOSTNAME=localhost
export PORT=17001
export PROMPT="scheme@(mpg-parse)"
export COGSERVER_CONF=""
# Scheme function name for planar MST parsing. This is a scheme function
# that will be called to process each sentence. For example, if the corpus
# contains "Some sentence." then the cogserver will receive
# (observe-mpg "Some sentence.")
#
# Use `observe-mpg` to get planar MST parsing.
# Use `observe-mst` to get generic MST parsing.
# The trees created by the generic parser might not be planar.
#
export OBSERVE="observe-mpg"
# Location of the database where disjunct counts will be accumulated
export MST_DB=rocks://${ROCKS_DATA_DIR}/mpg_parse.rdb
export STORAGE_NODE="(RocksStorageNode \"rocks://${MST_DB}\")"
# Directories where in-process and completed files will be moved.
export IN_PROCESS_DIR=mpg-split
export COMPLETED_DIR=mpg-done
# Message printed for each processed file
export MSG="MPG-Processing"