bash: Add initial support for the content parsing of the entries

bash/src/bib2web.bash

@@ -39,6 +39,10 @@ main() {
 	processInputFile
 	checkResultAndAbortIfNeeded "$?"
 
+	# Process the entries
+	processEntries
+	checkResultAndAbortIfNeeded "$?"
+
 	# Clean up the temporary directory
 	cleanUp
 }

bash/src/error-codes.bash

@@ -2,20 +2,25 @@
 
 # Error codes for tool setup
 export BIB2WEB_TOOL_FINDER_NOT_FOUND="1"
-export BIB2WEB_GREP_NOT_FOUND="2"
-export BIB2WEB_AWK_NOT_FOUND="3"
-export BIB2WEB_RM_NOT_FOUND="4"
-export BIB2WEB_MKTEMP_NOT_FOUND="5"
-export BIB2WEB_TOUCH_NOT_FOUND="6"
-export BIB2WEB_CAT_NOT_FOUND="7"
+export BIB2WEB_BASENAME_NOT_FOUND="2"
+export BIB2WEB_GREP_NOT_FOUND="3"
+export BIB2WEB_AWK_NOT_FOUND="4"
+export BIB2WEB_RM_NOT_FOUND="5"
+export BIB2WEB_MKTEMP_NOT_FOUND="6"
+export BIB2WEB_TOUCH_NOT_FOUND="7"
+export BIB2WEB_CAT_NOT_FOUND="8"
+export BIB2WEB_FIND_NOT_FOUND="9"
 
 # Error codes for option parsing
-export BIB2WEB_MISSING_BIBTEX_FILE="8"
-export BIB2WEB_NOT_BIBTEX_FILE="9"
+export BIB2WEB_MISSING_BIBTEX_FILE="10"
+export BIB2WEB_NOT_BIBTEX_FILE="11"
 
 # Error codes for directory and file setup
-export BIB2WEB_CANNOT_CREATE_TMP_DIRECTORY="10"
-export BIB2WEB_CANNOT_CREATE_LOG_FILE="11"
+export BIB2WEB_CANNOT_CREATE_TMP_DIRECTORY="20"
+export BIB2WEB_CANNOT_CREATE_LOG_FILE="21"
 
 # Error codes for input processing
-export BIB2WEB_CANNOT_PROCESS_BIBTEX_FILE="12"
+export BIB2WEB_CANNOT_PROCESS_BIBTEX_FILE="30"
+
+# Error codes for output generation
+export BIB2WEB_CANNOT_FIND_ENTRIES="40"

bash/src/input-processor.bash

@@ -11,12 +11,12 @@ parseInputFile() {
 
 generateEntrySplitterFile() {
 	local splitterFile="$1"
-	"${BIB2WEB_CAT}" <<EOF > "${BIB2WEB_TMP_DIR}/${splitterFile}"
+	"${BIB2WEB_CAT}" <<EOF > "${splitterFile}"
 BEGIN {
 	data = ""
 	file = ""
 }
-/^@.*/ {
+/^@/ {
 	if (length(data) > 0) {
 		print data > file
 		data = ""
@@ -39,7 +39,7 @@ EOF
 splitEntries() {
 	local splitterFile="$1"
 	local output=
-	output=$("${BIB2WEB_AWK}" -f "${BIB2WEB_TMP_DIR}/${splitterFile}" "${BIB2WEB_BIBTEX_FILE}" 2>&1)
+	output=$("${BIB2WEB_AWK}" -f "${splitterFile}" "${BIB2WEB_BIBTEX_FILE}" 2>&1)
 	local result="$?"
 	vvverbose "Splitter result: ${result}"
 	if [ "${result}" -gt "0" ]; then
@@ -52,14 +52,127 @@ splitEntries() {
 processInputFile() {
 	verbose "Processing the input file..."
 	vverbose "Splitting the input BibTeX file into individual entries..."
-	local splitterFile="splitter.awk"
+	local splitterFile="${BIB2WEB_TMP_DIR}/splitter.awk"
 	generateEntrySplitterFile "${splitterFile}"
 	splitEntries "${splitterFile}"
 	local splitResult="$?"
 	if [ "${splitResult}" -gt "0" ]; then
 		error "Could not process the contents of ${BIB2WEB_BIBTEX_FILE}!"
 		return "${BIB2WEB_CANNOT_PROCESS_BIBTEX_FILE}"
 	fi
-	vverbose "Input processing completed!"
+}
+
+parseField() {
+	local field="$1"
+	local entry="$2"
+	local entryIntermediateFile="$3"
+	local parserFile="${BIB2WEB_TMP_DIR}/${field}-input.awk"
+	if [ ! -e "${parserFile}" ]; then
+		cat <<EOF > "${parserFile}"
+BEGIN {
+	FS = " = "
+}
+tolower(\$1) ~ /${field}$/ {
+	data = \$2
+	# Strip the trailing comma, if exists
+	sub(/,$/, "", data)
+	# Quoted data
+	sub(/^"/, "", data)
+	sub(/"$/, "", data)
+	# Data in curly braces
+	sub(/^{/, "", data)
+	sub(/}$/, "", data)
+	# Internal curly braces
+	gsub(/{/, "", data)
+	gsub(/}/, "", data)
+	# Replace tildes with a space
+	gsub(/~/, " ", data)
+	# Remove the special characters' leading slash
+	gsub(/\\\\/, "", data)
+	print data
+}
+EOF
+	fi
+	local fieldContent
+	fieldContent=$("${BIB2WEB_AWK}" -f "${parserFile}" "${entry}" 2>&1)
+	vvverbose "${field}: ${fieldContent}"
+	echo "${field}: ${fieldContent}" >> "${entryIntermediateFile}"
+}
+
+parseEntry() {
+	local entry="$1"
+	local entryFileName
+	entryFileName=$("${BIB2WEB_BASENAME}" "${entry}")
+	vverbose "Parsing entry ${entryFileName}..."
+	local entryIntermediateFile
+	entryIntermediateFile="${entry}.bib2web"
+	# Parse the entry type
+	local type
+	type=$("${BIB2WEB_AWK}" "/^@(.*){/ { split(\$0, keyrow, \"{\"); print tolower(keyrow[1])}" "${entry}")
+	vverbose "type: ${type}"
+	echo "type: ${type}" > "${entryIntermediateFile}"
+	# Parse the default fields
+	parseField "address" "${entry}" "${entryIntermediateFile}"
+	parseField "annote" "${entry}" "${entryIntermediateFile}"
+	parseField "author" "${entry}" "${entryIntermediateFile}"
+	parseField "booktitle" "${entry}" "${entryIntermediateFile}"
+	parseField "chapter" "${entry}" "${entryIntermediateFile}"
+	parseField "crossref" "${entry}" "${entryIntermediateFile}"
+	parseField "doi" "${entry}" "${entryIntermediateFile}"
+	parseField "edition" "${entry}" "${entryIntermediateFile}"
+	parseField "editor" "${entry}" "${entryIntermediateFile}"
+	parseField "email" "${entry}" "${entryIntermediateFile}"
+	parseField "howpublished" "${entry}" "${entryIntermediateFile}"
+	parseField "institution" "${entry}" "${entryIntermediateFile}"
+	parseField "journal" "${entry}" "${entryIntermediateFile}"
+	parseField "key" "${entry}" "${entryIntermediateFile}"
+	parseField "month" "${entry}" "${entryIntermediateFile}"
+	parseField "note" "${entry}" "${entryIntermediateFile}"
+	parseField "number" "${entry}" "${entryIntermediateFile}"
+	parseField "organization" "${entry}" "${entryIntermediateFile}"
+	parseField "pages" "${entry}" "${entryIntermediateFile}"
+	parseField "publisher" "${entry}" "${entryIntermediateFile}"
+	parseField "school" "${entry}" "${entryIntermediateFile}"
+	parseField "series" "${entry}" "${entryIntermediateFile}"
+	parseField "title" "${entry}" "${entryIntermediateFile}"
+	parseField "type" "${entry}" "${entryIntermediateFile}"
+	parseField "volume" "${entry}" "${entryIntermediateFile}"
+	parseField "year" "${entry}" "${entryIntermediateFile}"
+	return "0"
+}
+
+getEntries() {
+	"${BIB2WEB_FIND}" "${BIB2WEB_TMP_DIR}" -name "*.bib"
+}
+
+parseEntries() {
+	for entry in $(getEntries); do
+		parseEntry "${entry}"
+		local entryParsingResult="$?"
+		if [ "${entryParsingResult}" -gt "0" ]; then
+			error "Error when parsing an entry!"
+			return "${entryParsingResult}"
+			break
+		fi
+	done
+	return "0"
+}
+
+processEntries() {
+	verbose "Processing the entries..."
+	# Do a sanity check that the entries can be read
+	local entries
+	entries=$(getEntries)
+	local entryFindResult="$?"
+	if [ "${entryFindResult}" -gt "0" ]; then
+		error "Could not fetch the entries!"
+		return "${BIB2WEB_CANNOT_FIND_ENTRIES}"
+	fi
+	local entryCount
+	entryCount=$(echo "${entries}" | grep -c ".bib")
+	vverbose "Number of entries: ${entryCount}"
+	parseEntries
+	verbose "Input processing completed!"
 	verbose "${BIB2WEB_LOG_SEPARATOR}"
+	return "0"
 }

bash/src/logger.bash

@@ -53,8 +53,8 @@ verbose() {
 vverbose() {
 	if [ "${BIB2WEB_VERBOSE}" -gt "2" ]; then
 		local message="$*"
-		echo "- ${message}" >&1
-		echo "- ${message}" >> "${BIB2WEB_LOG_FILE}"
+		echo "* ${message}" >&1
+		echo "* ${message}" >> "${BIB2WEB_LOG_FILE}"
 	fi
 }
 

bash/src/options-parser.bash

@@ -10,7 +10,7 @@ source "${BIB2WEB_BASE_DIR}/error-codes.bash"
 # Generic version getter
 getVersion() {
 	local scriptName
-	scriptName=$(basename "$0")
+	scriptName=$("${BIB2WEB_BASENAME}" "$0")
 	printf "%s, version %s" "${scriptName}" "${BIB2WEB_VERSION}"
 }
 

bash/src/parameters.bash.in

@@ -4,12 +4,14 @@
 export BIB2WEB_VERSION="%BIB2WEB_VERSION%"
 
 # Shell tools used by the script
+export BIB2WEB_BASENAME
 export BIB2WEB_GREP
 export BIB2WEB_AWK
 export BIB2WEB_MKTEMP
 export BIB2WEB_RM
 export BIB2WEB_TOUCH
 export BIB2WEB_CAT
+export BIB2WEB_FIND
 
 # Temporary directory and files
 export BIB2WEB_TMP_DIR

bash/src/preprocessor.bash

@@ -43,6 +43,13 @@ setUpTools() {
 		printError "No tool finder tool available!"
 		return "${toolFinderResult}"
 	fi
+	# basename
+	BIB2WEB_BASENAME=$("${toolFinder}" "basename")
+	local basenameResult="$?"
+	if [ "${basenameResult}" -gt 0 ]; then
+		printError "basename not found!"
+		return "${BIB2WEB_BASENAME_NOT_FOUND}"
+	fi
 	# grep
 	BIB2WEB_GREP=$("${toolFinder}" "grep")
 	local grepResult="$?"
@@ -85,6 +92,13 @@ setUpTools() {
 		printError "cat not found!"
 		return "${BIB2WEB_CAT_NOT_FOUND}"
 	fi
+	# find
+	BIB2WEB_FIND=$("${toolFinder}" "find")
+	local findResult="$?"
+	if [ "${findResult}" -gt 0 ]; then
+		printError "find not found!"
+		return "${BIB2WEB_FIND_NOT_FOUND}"
+	fi
 	return 0
 }
 

bash/test/input-processor-tests.bash

@@ -30,7 +30,7 @@ testSplittingInputFileWithOnlyOneEntryResultsInASingleFile() {
 	assertEquals "1" "${numberOfSplitEntries}"
 }
 
-testSplittingInputFileWithOnlyMoreThanOneEntryResultsInMultipleFile() {
+testSplittingInputFileWithOnlyMoreThanOneEntryResultsInMultipleFiles() {
 	BIB2WEB_BIBTEX_FILE="${SOURCE_DIRECTORY}/../../testdata/combined.bib"
 	processInputFile
 	local numberOfSplitEntries
@@ -39,4 +39,32 @@ testSplittingInputFileWithOnlyMoreThanOneEntryResultsInMultipleFile() {
 	expectedNumberOfEntries=$(grep -c "^@" "${BIB2WEB_BIBTEX_FILE}")
 	assertNotEquals "1" "${numberOfSplitEntries}"
 	assertEquals "${expectedNumberOfEntries}" "${numberOfSplitEntries}"
+}
+
+testParseEntry() {
+	BIB2WEB_BIBTEX_FILE="${SOURCE_DIRECTORY}/../../testdata/article/minimum.bib"
+	processInputFile
+	local numberOfSplitEntries
+	numberOfSplitEntries=$(find "${BIB2WEB_TMP_DIR}" -name "*.bib" | grep -c ".bib")
+
+	local entryFile
+	entryFile=$(find "${BIB2WEB_TMP_DIR}" -name "*.bib")
+	parseEntry "${entryFile}"
+
+	local numberOfParsedEntries
+	numberOfParsedEntries=$(find "${BIB2WEB_TMP_DIR}" -name "*.bib2web" | grep -c ".bib2web")
+	assertEquals "${numberOfSplitEntries}" "${numberOfParsedEntries}"
+}
+
+testParseSeveralEntries() {
+	BIB2WEB_BIBTEX_FILE="${SOURCE_DIRECTORY}/../../testdata/combined.bib"
+	processInputFile
+	local numberOfSplitEntries
+	numberOfSplitEntries=$(find "${BIB2WEB_TMP_DIR}" -name "*.bib" | grep -c ".bib")
+
+	parseEntries
+
+	local numberOfParsedEntries
+	numberOfParsedEntries=$(find "${BIB2WEB_TMP_DIR}" -name "*.bib2web" | grep -c ".bib2web")
+	assertEquals "${numberOfSplitEntries}" "${numberOfParsedEntries}"
 }

bash/test/preprocessor-tests.bash

@@ -15,6 +15,17 @@ tearDown() {
 	fi
 }
 
+testToolsSetupSuccess() {
+	assertNotNull "${BIB2WEB_BASENAME}"
+	assertNotNull "${BIB2WEB_GREP}"
+	assertNotNull "${BIB2WEB_AWK}"
+	assertNotNull "${BIB2WEB_MKTEMP}"
+	assertNotNull "${BIB2WEB_RM}"
+	assertNotNull "${BIB2WEB_TOUCH}"
+	assertNotNull "${BIB2WEB_CAT}"
+	assertNotNull "${BIB2WEB_FIND}"
+}
+
 testSetUpFilesAndCleanUp() {
 	BIB2WEB_LOG_FILE="foo.log"
 

testdata/article/minimum.bib

@@ -1,6 +1,6 @@
 @article{ARTICLE-MINIMUM,
-    author = "Article Author",
-    title = "Article Title",
+    autHOr = "Article Author",
+    title = {Article Title},
     journal = "Article Journal",
     year = "Article Year",
     volume = "Article Volume"

testdata/combined.bib

@@ -1,7 +1,7 @@
 @article{ARTICLE-MINIMUM,
     author = "Article Author",
-    title = "Article Title",
-    journal = "Article Journal",
+    title = {Article {Title~Foo}},
+    journal = "Article\ Journal",
     year = "Article Year",
     volume = "Article Volume"
  }