Add JSON output format for programmatic access

Implements structured JSON output with hierarchical schema including book metadata, chapters, verses, and footnotes. All existing flags (-F, -g, -L) are respected in JSON mode.
2025-12-16 18:49:27 +01:00
parent 6432d37df3
commit 7ef782dda2
3 changed files with 228 additions and 0 deletions
--- a/allioli.awk
+++ b/allioli.awk
@@ -329,6 +329,52 @@ function printfootnote(footnote_num, footnote,    word_count, characters_printed
 }

 function processline() {
+	# JSON mode: collect data instead of printing
+	if (ENVIRON["ALLIOLI_JSON_OUTPUT"] != "" && ENVIRON["ALLIOLI_JSON_OUTPUT"] != "0") {
+		# Store book info (will be used in END block)
+		if (json_book_name == "") {
+			json_book_name = $1
+			json_book_abbr = $2
+			json_book_num = $3
+		}
+
+		# Check if this is a footnote
+		if ($6 == "" && $7 ~ /^[0-9]+$/ && NF >= 8) {
+			json_footnotes[$4, $5, $7] = $8
+			json_footnote_nums[$4, $5, ++json_footnote_count[$4, $5]] = $7
+		}
+		# Check if this is an introduction (chapter 0)
+		else if ($4 == 0 && $6 == "") {
+			if (json_intro == "") {
+				json_intro = $7
+			} else {
+				json_intro = json_intro " " $7
+			}
+		}
+		# Verse with content
+		else if ($6 != "" || ($7 != "" && $7 !~ /^[0-9]+$/)) {
+			# Store verse data
+			json_latin[$4, $5] = $6
+			json_german[$4, $5] = $7
+
+			# Track unique verses per chapter
+			if (!json_verse_seen[$4, $5]) {
+				json_verse_seen[$4, $5] = 1
+				json_verses[$4, ++json_verse_count[$4]] = $5
+			}
+
+			# Track chapters
+			if (!json_chapter_seen[$4]) {
+				json_chapter_seen[$4] = 1
+				json_chapters[++json_chapter_total] = $4
+			}
+		}
+
+		outputted_records++
+		return
+	}
+
+	# Normal text output mode
 	if (printed_intrudction && $4 != 0){
 		printf("\n\n")
 		printed_intrudction=0
@@ -401,6 +447,133 @@ cmd == "ref" && mode == "search" && (p["book"] == "" || bookmatches($1, $2, p["b
 }

 END {
+	# JSON output mode
+	if (cmd == "ref" && ENVIRON["ALLIOLI_JSON_OUTPUT"] != "" && ENVIRON["ALLIOLI_JSON_OUTPUT"] != "0") {
+		if (outputted_records == 0) {
+			print "Unknown reference: " ref
+			exit 1
+		}
+
+		# Determine language flags
+		only_latin = (ENVIRON["ALLIOLI_ONLY_LATIN"] != "" && ENVIRON["ALLIOLI_ONLY_LATIN"] != "0")
+		only_german = (ENVIRON["ALLIOLI_ONLY_GERMAN"] != "" && ENVIRON["ALLIOLI_ONLY_GERMAN"] != "0")
+		no_footnotes = (ENVIRON["ALLIOLI_NOFOOTNOTES"] != "" && ENVIRON["ALLIOLI_NOFOOTNOTES"] != "0")
+
+		# Start JSON output
+		print "{"
+		printf("  \"book\": {\n")
+		printf("    \"name\": \"%s\",\n", json_book_name)
+		printf("    \"abbreviation\": \"%s\",\n", json_book_abbr)
+		printf("    \"number\": %d\n", json_book_num)
+		printf("  },\n")
+
+		# Output chapters
+		for (c_idx = 1; c_idx <= json_chapter_total; c_idx++) {
+			chapter = json_chapters[c_idx]
+
+			# Handle introduction (chapter 0)
+			if (chapter == 0) {
+				printf("  \"introduction\": \"%s\"", json_intro)
+				if (json_chapter_total > 1) {
+					printf(",\n")
+				} else {
+					printf("\n")
+				}
+				continue
+			}
+
+			# Regular chapter
+			printf("  \"chapter\": %d,\n", chapter)
+			printf("  \"verses\": [\n")
+
+			# Sort verses numerically before output
+			delete sorted_verses
+			for (v_idx = 1; v_idx <= json_verse_count[chapter]; v_idx++) {
+				sorted_verses[v_idx] = json_verses[chapter, v_idx]
+			}
+			# Simple bubble sort for numeric ordering
+			for (i = 1; i <= json_verse_count[chapter]; i++) {
+				for (j = i + 1; j <= json_verse_count[chapter]; j++) {
+					if (sorted_verses[i] + 0 > sorted_verses[j] + 0) {
+						temp = sorted_verses[i]
+						sorted_verses[i] = sorted_verses[j]
+						sorted_verses[j] = temp
+					}
+				}
+			}
+
+			# Output verses in sorted order
+			for (v_idx = 1; v_idx <= json_verse_count[chapter]; v_idx++) {
+				verse_num = sorted_verses[v_idx]
+
+				printf("    {\n")
+				printf("      \"verse\": %d,\n", verse_num)
+
+				# Text object
+				printf("      \"text\": {")
+
+				# Output text based on language flags
+				if (only_latin) {
+					printf("\n        \"latin\": \"%s\"\n", json_latin[chapter, verse_num])
+				} else if (only_german) {
+					# Remove superscript markers if footnotes disabled
+					german_text = json_german[chapter, verse_num]
+					if (no_footnotes) {
+						gsub(/[⁰¹²³⁴⁵⁶⁷⁸⁹]+/, "", german_text)
+					}
+					printf("\n        \"german\": \"%s\"\n", german_text)
+				} else {
+					# Both languages
+					german_text = json_german[chapter, verse_num]
+					if (no_footnotes) {
+						gsub(/[⁰¹²³⁴⁵⁶⁷⁸⁹]+/, "", german_text)
+					}
+					if (json_latin[chapter, verse_num] != "") {
+						printf("\n        \"latin\": \"%s\",\n", json_latin[chapter, verse_num])
+					}
+					if (german_text != "") {
+						printf("        \"german\": \"%s\"\n", german_text)
+					}
+				}
+
+				printf("      }")
+
+				# Footnotes array (if not disabled)
+				if (!no_footnotes && json_footnote_count[chapter, verse_num] > 0) {
+					printf(",\n      \"footnotes\": [\n")
+					for (f_idx = 1; f_idx <= json_footnote_count[chapter, verse_num]; f_idx++) {
+						fn_num = json_footnote_nums[chapter, verse_num, f_idx]
+						fn_text = json_footnotes[chapter, verse_num, fn_num]
+						printf("        {\n")
+						printf("          \"number\": %d,\n", fn_num)
+						printf("          \"text\": \"%s\"\n", fn_text)
+						if (f_idx < json_footnote_count[chapter, verse_num]) {
+							printf("        },\n")
+						} else {
+							printf("        }\n")
+						}
+					}
+					printf("      ]\n")
+				} else {
+					printf("\n")
+				}
+
+				# Close verse object
+				if (v_idx < json_verse_count[chapter]) {
+					printf("    },\n")
+				} else {
+					printf("    }\n")
+				}
+			}
+
+			printf("  ]\n")
+		}
+
+		print "}"
+		exit
+	}
+
+	# Normal text mode
 	if (cmd == "ref" && outputted_records == 0) {
 		print "Unknown reference: " ref
 	}