bible-ref-parse 3.2.0 → 3.3.1

package/dist/ru-bcv-parser.js

@@ -548,7 +548,11 @@ class bcv_parser {
         const out = [];
         for (var osis of Array.from(this.parsed_entities())) {
             if (osis.osis.length > 0) {
-                out.push({ osis: osis.osis, translations: osis.translations, indices: osis.indices });
+                const entry = { osis: osis.osis, translations: osis.translations, indices: osis.indices };
+                if (osis.osises != null) {
+                    entry.osises = osis.osises;
+                }
+                out.push(entry);
             }
         }
         return out;
@@ -565,6 +569,28 @@ class bcv_parser {
             if (entity.passages == null) {
                 continue;
             }
+            if (this.options.consecutive_combination_strategy === "separate-chapters") {
+                const semicolon_segments = this.split_text_by_semicolon(entity.absolute_indices);
+                if (semicolon_segments.length > 1) {
+                    for (const segment of Array.from(semicolon_segments)) {
+                        const sub_parser = new bcv_parser;
+                        sub_parser.set_options({ ...this.options });
+                        sub_parser.parse(segment.text);
+                        for (const sub of Array.from(sub_parser.parsed_entities())) {
+                            if (sub.osis.length === 0) {
+                                continue;
+                            }
+                            const indices = [sub.indices[0] + segment.offset, sub.indices[1] + segment.offset];
+                            const entry = { osis: sub.osis, translations: sub.translations, indices, entity_id };
+                            if (sub.osises != null) {
+                                entry.osises = sub.osises;
+                            }
+                            out.push(entry);
+                        }
+                    }
+                    continue;
+                }
+            }
             if (((entity.type === "b") && (this.options.book_alone_strategy === "ignore")) || ((entity.type === "b_range") && (this.options.book_range_strategy === "ignore")) || (entity.type === "context")) {
                 continue;
             }
@@ -614,34 +640,18 @@ class bcv_parser {
                     passage.absolute_indices = entity.absolute_indices;
                 }
                 if ((this.options.consecutive_combination_strategy === 'separate-chapters') && (passage.start.c !== passage.end.c)) {
-                    var asc2, end2;
-                    for (i = passage.start.c, end2 = passage.end.c, asc2 = passage.start.c <= end2; asc2 ? i <= end2 : i >= end2; asc2 ? i++ : i--) {
-                        var passageStartVerse = i === passage.start.c ? passage.start.v : 1;
-                        var passageEndVerse = i === passage.end.c ? passage.end.v : this.translations[translation_alias]['chapters'][passage.start.b][i - 1];
-                        var passageStart = {
-                            b: passage.start.b,
-                            c: i,
-                            v: passage.start.type === 'bcv' ? passageStartVerse : null,
-                            type: passage.start.type
-                        };
-                        var passageEnd = {
-                            b: passage.end.b,
-                            c: i,
-                            v: passage.start.type === 'bcv' ? passageEndVerse : null,
-                            type: passage.start.type
-                        };
-                        nonCombinableOsises.push({
-                            osis: passage.valid.valid ? this.to_osis(passageStart, passageEnd, translation_alias) : "",
-                            type: passage.type,
-                            indices: passage.absolute_indices,
-                            translations,
-                            start: passageStart,
-                            end: passageEnd,
-                            enclosed_indices: passage.enclosed_absolute_indices,
-                            entity_id,
-                            entities: [passage]
-                        });
-                    }
+                    osises.push({
+                        osis: passage.valid.valid ? this.to_osis(passage.start, passage.end, translation_alias) : "",
+                        osises: this.get_osises_by_chapter({ osis: "", start: passage.start, end: passage.end }, translation_alias),
+                        type: passage.type,
+                        indices: passage.absolute_indices,
+                        translations,
+                        start: passage.start,
+                        end: passage.end,
+                        enclosed_indices: passage.enclosed_absolute_indices,
+                        entity_id,
+                        entities: [passage]
+                    });
                 }
                 else {
                     osises.push({
@@ -667,12 +677,54 @@ class bcv_parser {
                 }
                 osises = osises.concat(nonCombinableOsises).sort((osis1, osis2) => osis1.indices[0] - osis2.indices[0]);
             }
+            if (this.options.consecutive_combination_strategy === "separate-chapters") {
+                for (osis of Array.from(osises)) {
+                    if (osis.osis.length === 0) {
+                        continue;
+                    }
+                    const osises_by_chapter = this.get_osises_by_chapter(osis, translation_alias);
+                    if (osises_by_chapter.length > 1) {
+                        osis.osises = osises_by_chapter;
+                    }
+                }
+            }
             // Add the osises array to the existing array.
             if (this.options.sequence_combination_strategy === "separate") {
                 out = out.concat(osises);
                 // Add the OSIS string and some data to the array.
             }
             else {
+                if ((this.options.consecutive_combination_strategy === "separate-chapters") && (entity.type === "sequence")) {
+                    const groups = this.group_osises_by_semicolon(osises);
+                    if (groups.length > 1) {
+                        for (const group of Array.from(groups)) {
+                            if (group.length === 0) {
+                                continue;
+                            }
+                            const group_last_i = group.length - 1;
+                            const indices = [group[0].indices[0], group[group_last_i].indices[1]];
+                            if ((group[group_last_i].enclosed_indices != null) && (group[group_last_i].enclosed_indices[1] >= 0)) {
+                                indices[1] = group[group_last_i].enclosed_indices[1];
+                            }
+                            const strings = [];
+                            let osises_by_chapter = [];
+                            for (osis of Array.from(group)) {
+                                if (osis.osis.length > 0) {
+                                    strings.push(osis.osis);
+                                }
+                                if (osis.osises != null) {
+                                    osises_by_chapter = osises_by_chapter.concat(osis.osises);
+                                }
+                            }
+                            const out_entry = { osis: strings.join(","), indices, translations, entity_id, entities: group };
+                            if (osises_by_chapter.length > 0) {
+                                out_entry.osises = osises_by_chapter;
+                            }
+                            out.push(out_entry);
+                        }
+                        continue;
+                    }
+                }
                 var osis;
                 var strings = [];
                 var last_i = osises.length - 1;
@@ -685,7 +737,19 @@ class bcv_parser {
                         strings.push(osis.osis);
                     }
                 }
-                out.push({ osis: strings.join(","), indices: entity.absolute_indices, translations, entity_id, entities: osises });
+                let osises_by_chapter = [];
+                if (this.options.consecutive_combination_strategy === "separate-chapters") {
+                    for (osis of Array.from(osises)) {
+                        if (osis.osises != null) {
+                            osises_by_chapter = osises_by_chapter.concat(osis.osises);
+                        }
+                    }
+                }
+                const out_entry = { osis: strings.join(","), indices: entity.absolute_indices, translations, entity_id, entities: osises };
+                if (osises_by_chapter.length > 0) {
+                    out_entry.osises = osises_by_chapter;
+                }
+                out.push(out_entry);
             }
         }
         return out;
@@ -765,6 +829,96 @@ class bcv_parser {
         }
         return out;
     }
+    get_osises_by_chapter(passage, translation) {
+        if ((passage == null) || (passage.start == null) || (passage.end == null)) {
+            return [];
+        }
+        if ((passage.start.b == null) || (passage.end.b == null) || (passage.start.c == null) || (passage.end.c == null)) {
+            return (passage.osis != null) && (passage.osis.length > 0) ? [passage.osis] : [];
+        }
+        if ((passage.start.b !== passage.end.b) || (passage.start.c === passage.end.c)) {
+            return (passage.osis != null) && (passage.osis.length > 0) ? [passage.osis] : [];
+        }
+        const out = [];
+        for (let c = passage.start.c, end = passage.end.c, asc = passage.start.c <= end; asc ? c <= end : c >= end; asc ? c++ : c--) {
+            const passageStartVerse = c === passage.start.c ? passage.start.v : 1;
+            const passageEndVerse = c === passage.end.c ? passage.end.v : this.translations[translation].chapters[passage.start.b][c - 1];
+            const passageStart = {
+                b: passage.start.b,
+                c,
+                v: passage.start.type === "bcv" ? passageStartVerse : null,
+                type: passage.start.type
+            };
+            const passageEnd = {
+                b: passage.end.b,
+                c,
+                v: passage.start.type === "bcv" ? passageEndVerse : null,
+                type: passage.start.type
+            };
+            out.push(this.to_osis(passageStart, passageEnd, translation));
+        }
+        return out;
+    }
+    get_separator_between(prev_indices, next_indices) {
+        if ((this.s == null) || (prev_indices == null) || (next_indices == null)) {
+            return "";
+        }
+        const start = prev_indices[1] + 1;
+        const end = next_indices[0];
+        if (end <= start) {
+            return "";
+        }
+        return this.s.slice(start, end);
+    }
+    has_semicolon_separator(prev_osis, next_osis) {
+        const between = this.get_separator_between(prev_osis.indices, next_osis.indices);
+        return /[;；]/.test(between);
+    }
+    group_osises_by_semicolon(osises) {
+        const out = [];
+        let current = [];
+        for (const osis of Array.from(osises)) {
+            if (current.length === 0) {
+                current.push(osis);
+                continue;
+            }
+            if (this.has_semicolon_separator(current[current.length - 1], osis)) {
+                out.push(current);
+                current = [osis];
+            }
+            else {
+                current.push(osis);
+            }
+        }
+        if (current.length > 0) {
+            out.push(current);
+        }
+        return out;
+    }
+    split_text_by_semicolon(absolute_indices) {
+        if ((this.s == null) || (absolute_indices == null) || (absolute_indices[0] == null) || (absolute_indices[1] == null)) {
+            return [];
+        }
+        const text = this.s.slice(absolute_indices[0], absolute_indices[1]);
+        if (!/[;；]/.test(text)) {
+            return [{ text, offset: absolute_indices[0] }];
+        }
+        const segments = [];
+        let start = 0;
+        for (let i = 0, end = text.length; i < end; i++) {
+            const ch = text[i];
+            if ((ch === ";") || (ch === "；")) {
+                if (i > start) {
+                    segments.push({ text: text.slice(start, i), offset: absolute_indices[0] + start });
+                }
+                start = i + 1;
+            }
+        }
+        if (start < text.length) {
+            segments.push({ text: text.slice(start), offset: absolute_indices[0] + start });
+        }
+        return segments;
+    }
     // If we want to treat Ps151 as a book rather than a chapter, we have to do some gymnastics to make sure it returns properly.
     fix_ps151(start, end, translation) {
         // Ps151 doesn't necessarily get promoted into the translation chapter list because during the string parsing, we treat it as `Ps` rather than `Ps151`.
@@ -809,6 +963,11 @@ class bcv_parser {
             if (osis.osis.length > 0) {
                 var prev_i = out.length - 1;
                 var is_enclosed_last = false;
+                if ((prev_i >= 0) && this.has_semicolon_separator(out[prev_i], osis)) {
+                    out.push(osis);
+                    prev = { b: osis.end.b, c: osis.end.c, v: osis.end.v };
+                    continue;
+                }
                 // Record the start index of the enclosed sequence for use in future iterations.
                 if (osis.enclosed_indices[0] !== enclosed_sequence_start) {
                     enclosed_sequence_start = osis.enclosed_indices[0];
@@ -2549,8 +2708,8 @@ bcv_parser.prototype.regexps.pre_book = "[^A-Za-zЀ-ҁ҃-҇Ҋ-ԧⷠ-ⷿꙀ-꙯
 bcv_parser.prototype.regexps.first = `(?:1-?я|1-?е|1)\\.?${bcv_parser.prototype.regexps.space}*`;
 bcv_parser.prototype.regexps.second = `(?:2-?я|2-?е|2)\\.?${bcv_parser.prototype.regexps.space}*`;
 bcv_parser.prototype.regexps.third = `(?:3-?я|3-?е|3)\\.?${bcv_parser.prototype.regexps.space}*`;
-bcv_parser.prototype.regexps.range_and = "(?:[&\\u2013\\u2014-]|и|—)";
-bcv_parser.prototype.regexps.range_only = "(?:[\\u2013\\u2014-]|—)";
+bcv_parser.prototype.regexps.range_and = `(?:[&\\u2013\\u2014-]|и|—)`;
+bcv_parser.prototype.regexps.range_only = `(?:[\\u2013\\u2014-]|—)`;
 // Each book regexp should return two parenthesized objects: an optional preliminary character and the book itself.
 bcv_parser.prototype.regexps.get_books = function (include_apocrypha, case_sensitive) {
     const books = [