Add more word lists and scripts

pom.xml

@@ -148,10 +148,15 @@
                         <prurl>${project.url}</prurl>
                     </properties>
                     <mapping>
+                        <R>SCRIPT_STYLE</R>
+                        <g4>DOUBLESLASH_STYLE</g4>
                         <template>JAVADOC_STYLE</template>
                     </mapping>
                     <includes>
                         <include>src/*/java/**/*.java</include>
+                        <include>src/*/antlr4/**/*.g4</include>
+                        <include>src/*/shell/**/*.sh</include>
+                        <include>src/*/R/**/*.R</include>
                         <include>**/pom.xml</include>
                     </includes>
                 </configuration>

src/main/R/sample-words.R

@@ -1,4 +1,13 @@
 #!/usr/bin/Rscript
+#
+# This file is part of cryptator, https://github.com/arnaud-m/cryptator
+#
+# Copyright (c) 2022, Université Côte d'Azur. All rights reserved.
+#
+# Licensed under the BSD 3-clause license.
+# See LICENSE file in the project root for full license information.
+#
+
 
 
 ScanWords <- function(file) scan(file = file, what = character(), quiet = TRUE)

src/main/antlr4/Cryptator.g4

@@ -1,3 +1,12 @@
+//
+// This file is part of cryptator, https://github.com/arnaud-m/cryptator
+//
+// Copyright (c) 2022, Université Côte d'Azur. All rights reserved.
+//
+// Licensed under the BSD 3-clause license.
+// See LICENSE file in the project root for full license information.
+//
+
 grammar Cryptator;
 
 @header{

src/main/shell/demo-generate.sh

@@ -1,4 +1,13 @@
 #!/bin/sh
+#
+# This file is part of cryptator, https://github.com/arnaud-m/cryptator
+#
+# Copyright (c) 2022, Université Côte d'Azur. All rights reserved.
+#
+# Licensed under the BSD 3-clause license.
+# See LICENSE file in the project root for full license information.
+#
+
 ## sh demo-generate.sh > demo-generate-output.md
 JAR="../../../target/cryptator-*-with-dependencies.jar"
 DIR="../words"
@@ -12,7 +21,7 @@ function solve() {
 echo "# Search cryptarithms with a UNIQUE solution"
 
 ######
-echo -e "\n## Generate from a word list\n"
+echo -e "\n## Generate from a words list\n"
 
 echo "- Planets"
 solve $DIR/planets.txt

src/main/shell/demo-play.sh

@@ -1,4 +1,13 @@
 #!/bin/sh
+#
+# This file is part of cryptator, https://github.com/arnaud-m/cryptator
+#
+# Copyright (c) 2022, Université Côte d'Azur. All rights reserved.
+#
+# Licensed under the BSD 3-clause license.
+# See LICENSE file in the project root for full license information.
+#
+
 JAR="../../../target/cryptator-*-with-dependencies.jar"
 
 java -cp $JAR cryptator.Cryptamancer 'send+more=money'

src/main/shell/demo-solve.sh

@@ -1,4 +1,13 @@
 #!/bin/sh
+#
+# This file is part of cryptator, https://github.com/arnaud-m/cryptator
+#
+# Copyright (c) 2022, Université Côte d'Azur. All rights reserved.
+#
+# Licensed under the BSD 3-clause license.
+# See LICENSE file in the project root for full license information.
+#
+
 JAR="../../../target/cryptator-*-with-dependencies.jar"
 CMD="java -cp $JAR cryptator.Cryptator -c TRUE -g TRUE"
 

src/main/shell/format-word-list.sh

@@ -0,0 +1,121 @@
+#!/bin/sh
+#
+# This file is part of cryptator, https://github.com/arnaud-m/cryptator
+#
+# Copyright (c) 2022, Université Côte d'Azur. All rights reserved.
+#
+# Licensed under the BSD 3-clause license.
+# See LICENSE file in the project root for full license information.
+#
+
+#--------------------------------------------------------------------
+# Setup Global Variables
+#--------------------------------------------------------------------
+
+PROG=`basename $0`
+OVERWRITE=0
+#--------------------------------------------------------------------
+# Version and help messages
+#--------------------------------------------------------------------
+
+
+version() {
+cat <<EOF
+$PROG 0.1
+
+This file is part of cryptator, https://github.com/arnaud-m/cryptator
+
+Copyright (c) 2022, Université Côte d'Azur. All rights reserved.
+
+Licensed under the BSD 3-clause license.
+See LICENSE file in the project root for full license information.
+EOF
+}
+
+help() {
+cat <<EOF
+$PROG formats a word list.
+
+$PROG formats a word list and output its results into a file.
+It removes diacritics, non-alphanumeric and whitespace characters, and then sort the word list while eliminating duplicates.
+By default, the output file is renamed by adding the suffix '.clean', but an option can force overwrite..
+
+Usage: $PROG [OPTION] FILES...
+
+Options:
+ -f       force overwriting the original file instead of adding it a suffix.
+
+ -h        display this help and exit
+ -v        output version information and exit
+
+Example:
+$PROG planets.txt
+
+Report bugs to <arnaud (dot) malapert (at) univ-cotedazur (dot) fr>."
+EOF
+}
+
+#--------------------------------------------------------------------
+# Test for prerequisites
+#--------------------------------------------------------------------
+
+while getopts ":hvf" opt; do
+    case $opt in
+        f)
+            OVERWRITE=1
+            ;;
+        h)
+            help
+            exit 0
+            ;;
+        v)
+            version;
+            exit 0
+            ;;
+        \?)
+            echo "Invalid option: -$OPTARG" >&2
+            exit 1
+            ;;
+        :)
+            echo "Option -$OPTARG requires an argument." >&2
+            exit 1
+            ;;
+    esac
+done
+
+shift $((OPTIND-1))
+#--------------------------------------------------------------------
+# Do something
+#--------------------------------------------------------------------
+
+## Create temporary filenames
+TMP1=`mktemp`
+TMP2=`mktemp`
+
+
+
+for FILE in $* ; do
+    if [ -f $FILE ] && [ -r $FILE ] ; then
+        echo "Process $FILE"
+        ## Set output filename
+        if [ $OVERWRITE -ne 0 ] ; then
+            echo "Overwrite $FILE"
+            OUTFILE=$FILE
+        else
+            OUTFILE=$FILE.clean
+        fi
+        ## Remove all diacritics
+        ## https://stackoverflow.com/questions/10207354/how-to-remove-all-of-the-diacritics-from-a-file
+        cat $FILE | iconv -f utf8 -t ascii//TRANSLIT//IGNORE > $TMP1
+        ## Convert to lowercase
+        cat $TMP1 | tr '[:upper:]' '[:lower:]' > $TMP2
+        ## Remove non alphanumerics and blank lines
+        sed -e 's/[^a-z0-9]//g' -e '/^[[:space:]]*$/d'  $TMP2 > $TMP1
+        ## Remove duplicates
+        sort -u $TMP1 > $OUTFILE
+    else
+        echo "Ignore $FILE"
+    fi
+
+
+done

src/main/shell/generate-doubly-true-output.md

@@ -0,0 +1,123 @@
+# Doubly true cryptarithms between 1 and 100
+- en
+  - one + two + five + nine + eleven + twelve + fifty = ninety 
+  - one + five + ten + eleven + nineteen + fortyfive = ninetyone 
+  - one + nine + twenty + fifty = eighty 
+  - five + seven + eleven + twelve + fifteen + twenty = seventy 
+  - three + seven + ten + twenty + thirty = seventy 
+  - one + seven + nine + fifteen + sixteen + seventeen = sixtyfive 
+  - two + fourteen + fifteen + twenty = fiftyone 
+  - three + nine + ten + fourteen + fifteen = fiftyone 
+  - six + nine + eleven + sixteen + nineteen = sixtyone 
+  - six + seven + nine + twelve + sixteen + twenty = seventy 
+  - five + six + sixteen + nineteen + twenty = sixtysix 
+  - eight + ten + sixteen + seventeen + eighteen = sixtynine 
+- hi
+  - सात + सैंतीस + चालीस = चौरासी 
+  - तीन + सात + बीस + उनतीस + तीस = नवासी 
+  - सात + बीस + पच्चीस + बत्तीस = चौरासी 
+  - एक + बीस + उनतीस + तीस = अस्सी 
+  - चार + बीस + चौबीस + छत्तीस = चौरासी 
+  - चार + चौबीस + छब्बीस + तीस = चौरासी 
+  - चार + सात + चौबीस + उनचास = चौरासी 
+  - सात + बारह + बीस + तैंतीस = बहत्तर 
+  - सात + बारह + छब्बीस + बत्तीस = सतहत्तर 
+  - सात + बारह + सत्रह + छत्तीस = बहत्तर 
+  - तीन + छह + सात + उन्नीस + बीस + बत्तीस = सत्तासी 
+  - तीन + छह + सात + सत्रह + छत्तीस = उनहत्तर 
+  - छह + बीस + चौबीस + तीस = अस्सी 
+  - छह + तेरह + छब्बीस + बत्तीस = सतहत्तर 
+  - तीन + तेरह + बीस + छत्तीस = बहत्तर 
+  - छह + तेरह + बत्तीस + छत्तीस = सत्तासी 
+  - पाँच + सात + पच्चीस = सैंतीस 
+  - सात + आठ + सड़सठ = बयासी 
+  - चार + चालीस + बयालीस = छियासी 
+  - बीस + बाईस + बयालीस = चौरासी 
+  - तीन + चार + तेईस + पचपन = पचासी 
+  - छह + तेईस + छब्बीस + बत्तीस = सत्तासी 
+  - तीन + उन्नीस + उनतीस + बत्तीस = तिरासी 
+  - छह + सात + बारह + छब्बीस + बत्तीस = तिरासी 
+  - चार + सात + पैंतीस + सैंतीस = तिरासी 
+  - तीन + चार + सात + उनतीस + चालीस = तिरासी 
+  - चार + सत्रह + तीस + बत्तीस = तिरासी 
+  - चार + सात + बत्तीस + चालीस = तिरासी 
+  - सत्रह + तीस + छत्तीस = तिरासी 
+  - चार + सात + सत्रह + पच्चीस + तीस = तिरासी 
+  - तीन + नौ + तेरह + सत्रह + तीस = बहत्तर 
+  - नौ + सत्रह + बीस + छब्बीस = बहत्तर 
+  - तीन + नौ + तीस + चालीस = बयासी 
+  - तीन + इक्कीस + इकतीस + बत्तीस = सत्तासी 
+  - चार + बीस + तीस + इकतीस = पचासी 
+  - बीस + इक्कीस + बत्तीस = तिहत्तर 
+  - सात + दस + बीस + चौबीस + छब्बीस = सत्तासी 
+  - तीन + छह + दस + सत्रह + छत्तीस = बहत्तर 
+  - दस + छब्बीस + छत्तीस = बहत्तर 
+  - दस + इक्कीस + पच्चीस + इकतीस = सत्तासी 
+  - सात + नौ + बत्तीस + उनतालीस = सत्तासी 
+- es
+  - seis + catorce + setenta = noventa 
+  - tres + seis + siete + catorce + sesenta = noventa 
+  - tres + seis + siete + once + trece + veinte + treinta = noventa 
+  - dos + cinco + siete + doce + catorce + treinta = setenta 
+  - dos + once + trece + catorce + treinta = setenta 
+  - uno + dos + tres + cuatro + once + doce + trece + catorce = sesenta 
+  - uno + cuatro + cinco + seis + once + trece + treinta = setenta 
+  - cuatro + cinco + siete + catorce + treinta = sesenta 
+  - tres + cinco + seis + ocho + once + trece + catorce = sesenta 
+  - cero + seis + siete = trece 
+  - cero + uno + tres + cuatro + cinco + trece + catorce + treinta = setenta 
+  - cero + tres + seis + siete + once + trece + veinte = sesenta 
+  - cero + seis + siete + trece + catorce + veinte = sesenta 
+  - cero + seis + siete + trece + catorce + veinte + treinta = noventa 
+  - cero + uno + tres + veinticinco = veintinueve 
+- fr
+  - zero + un + trois + onze + quinze = trente 
+- ar
+- ru
+- pt
+  - cinco + seis + sete + oito + catorze + trinta = setenta 
+  - sete + oito + onze + catorze + trinta = setenta 
+  - seis + sete + oito + onze + dezoito + trinta = oitenta 
+  - cinco + oito + treze + catorze + vinte + trinta = noventa 
+  - onze + doze + treze + catorze + trinta = oitenta 
+  - zero + doze + dezoito + vinte + trinta = oitenta 
+  - nove + onze + doze + dezoito + trinta = oitenta 
+  - cinco + oito + dez + doze + treze + catorze + dezoito = oitenta 
+- id
+- de
+  - eins + zwei + sieben + sechzig = siebzig 
+- tr
+  - bes + on + oniki + elliuc = seksen 
+  - yedi + dokuz + onyedi + ondokuz = elliiki 
+  - iki + uc + oniki + otuz + otuzuc = seksen 
+  - sıfır + iki + bes + on + onbir + oniki + kırk = seksen 
+  - onbir + kırkbir = elliiki 
+  - oniki + ondokuz = otuzbir 
+  - sıfır + iki + oniki = ondort 
+  - iki + bes + sekiz + oniki + onsekiz = kırkbes 
+  - bir + iki + sekiz + oniki + onsekiz = kırkbir 
+  - dokuz + otuziki = kırkbir 
+  - sıfır + bir + iki + on + oniki + onbes + kırk = seksen 
+  - sıfır + iki + onbir + oniki + onbes + kırk = seksen 
+  - bes + sekiz + on + oniki + onbes + otuz = seksen 
+  - iki + on + onbes + elliuc = seksen 
+  - iki + bes + onbir + oniki + elli = seksen 
+  - bir + iki + oniki + onbes + elli = seksen 
+  - uc + oniki + onbes + elli = seksen 
+- it
+  - zero + sei + sette + otto + nove = trenta 
+  - due + sette + dieci + undici = trenta 
+  - tre + sei + dieci + undici = trenta 
+  - otto + nove + tredici + venti + trenta = ottanta 
+  - uno + due + tre + otto + undici + dodici + tredici + trenta = ottanta 
+- uk
+- pl
+- ro
+- el
+  - ενα + επτα + εννεα + εντεκα + δωδεκα = σαραντα 
+  - ενα + οκτω + εννεα + δωδεκα = τριαντα 
+  - μηδεν + ενα + εννεα + δεκα = εικοσι 
+  - ενα + τρια + πεντε + εντεκα + εικοσι = σαραντα 
+  - τρια + τεσσερα + δεκα + εντεκα + δωδεκα = σαραντα 
+  - τρια + τεσσερα + πεντε + επτα + εντεκα = τριαντα 
+  - ενα + τεσσερα + πεντε + εικοσι = τριαντα 

src/main/shell/generate-doubly-true.sh

@@ -0,0 +1,30 @@
+#!/bin/sh
+#
+# This file is part of cryptator, https://github.com/arnaud-m/cryptator
+#
+# Copyright (c) 2022, Université Côte d'Azur. All rights reserved.
+#
+# Licensed under the BSD 3-clause license.
+# See LICENSE file in the project root for full license information.
+#
+
+
+## sh generate-doubly-true.sh  > generate-doubly-true-output.md
+JAR="../../../target/cryptator-*-with-dependencies.jar"
+LANGUAGES=language-codes.csv
+MIN=1
+MAX=100
+
+## Execute the command.
+## Filter the output: print only the cryptarithm.
+function solve() {
+    java -cp $JAR cryptator.Cryptagen -c TRUE $* | sed -n 's/\(.*+.*=.*\)/  - \1/p'
+}
+
+echo "# Doubly true cryptarithms between $MIN and $MAX"
+# Read the csf files with a list of country codes, and languages codes
+while IFS="," read ctry lang ; do
+    echo "- $lang"
+    # Search doubly true cryptarithms in the given language
+    solve -ctry $ctry -lang $lang $MIN $MAX
+done < <(grep -v "^#\|^$" $LANGUAGES)