@helloao/tools 0.0.17 → 0.0.19

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -36,7 +36,10 @@ const chapterHeadingSchema = import_zod.z.object({
36
36
  });
37
37
  const verseSchema = import_zod.z.object({
38
38
  type: import_zod.z.literal("text"),
39
- id: import_zod.z.string()
39
+ id: import_zod.z.string(),
40
+ bookCode: import_zod.z.string().optional(),
41
+ chapter: import_zod.z.number().optional(),
42
+ verse: import_zod.z.number().optional()
40
43
  });
41
44
  const paratextSchema = import_zod.z.object({
42
45
  type: import_zod.z.literal("paratext"),
@@ -95,11 +98,11 @@ class CodexParser {
95
98
  }
96
99
  chapter.content.push(...content);
97
100
  }
98
- function addReference(ref, content) {
99
- addChapterContent(ref.chapter, [
101
+ function addReference(ref2, content) {
102
+ addChapterContent(ref2.chapter, [
100
103
  {
101
104
  type: "verse",
102
- number: ref.verse,
105
+ number: ref2.verse,
103
106
  content
104
107
  }
105
108
  ]);
@@ -113,6 +116,24 @@ class CodexParser {
113
116
  content: [content]
114
117
  };
115
118
  }
119
+ function addLineBreaks(lines2) {
120
+ return lines2.reduce(
121
+ (prev, current) => {
122
+ if (prev.length === 0) return [current];
123
+ else
124
+ return [
125
+ ...prev,
126
+ {
127
+ lineBreak: true
128
+ },
129
+ current
130
+ ];
131
+ },
132
+ []
133
+ );
134
+ }
135
+ let previousReference = null;
136
+ let lines = [];
116
137
  for (let cell of data.cells) {
117
138
  if (cell.languageId === "html" && cell.value) {
118
139
  if (!cell.metadata) continue;
@@ -120,31 +141,57 @@ class CodexParser {
120
141
  if (!metadataResult.success) continue;
121
142
  const metadata = metadataResult.data;
122
143
  if (metadata.type === "text") {
123
- const reference = (0, import_utils.parseVerseReference)(metadata.id);
124
- if (!reference)
125
- throw new Error("Could not find verse reference.");
144
+ let reference = (0, import_utils.parseVerseReference)(metadata.id);
145
+ if (!reference) {
146
+ if (metadata.bookCode && metadata.chapter) {
147
+ const bookId = (0, import_utils.getBookId)(metadata.bookCode);
148
+ if (!bookId) {
149
+ throw new Error(
150
+ "Could not find book ID for code: " + metadata.bookCode
151
+ );
152
+ }
153
+ if (metadata.verse) {
154
+ reference = {
155
+ book: bookId,
156
+ chapter: metadata.chapter,
157
+ verse: metadata.verse
158
+ };
159
+ } else if (previousReference?.verse) {
160
+ if (previousReference.chapter !== metadata.chapter || previousReference.book !== bookId) {
161
+ throw new Error(
162
+ "Cannot infer verse number for non-consecutive verse reference."
163
+ );
164
+ }
165
+ reference = {
166
+ book: bookId,
167
+ chapter: metadata.chapter,
168
+ verse: previousReference.verse
169
+ };
170
+ } else {
171
+ throw new Error(
172
+ "Could not find verse reference in metadata."
173
+ );
174
+ }
175
+ } else {
176
+ throw new Error(
177
+ "Could not find verse reference in metadata."
178
+ );
179
+ }
180
+ }
181
+ if (previousReference && (reference.book !== previousReference.book || reference.chapter !== previousReference.chapter || reference.verse !== previousReference.verse)) {
182
+ addReference(previousReference, addLineBreaks(lines));
183
+ lines = [];
184
+ }
185
+ previousReference = reference;
126
186
  if (!root.id) root.id = reference.book;
127
187
  const content = stripHTML(cell.value);
128
- const lines = content.split("\n").reduce(
129
- (prev, current) => {
130
- if (prev.length === 0) return [current];
131
- else
132
- return [
133
- ...prev,
134
- {
135
- lineBreak: true
136
- },
137
- current
138
- ];
139
- },
140
- []
141
- );
142
- addReference(reference, lines);
188
+ const newLines = content.split("\n");
189
+ lines.push(...newLines);
143
190
  } else if (metadata.type === "paratext") {
144
191
  const reference = (0, import_utils.parseVerseReference)(`${metadata.id}:1`);
145
192
  if (reference) {
146
193
  const content = stripHTML(cell.value);
147
- const lines = content.split("\n").reduce((prev, current) => {
194
+ const lines2 = content.split("\n").reduce((prev, current) => {
148
195
  if (prev.length === 0)
149
196
  return [toHeading(current)];
150
197
  else
@@ -156,7 +203,7 @@ class CodexParser {
156
203
  toHeading(current)
157
204
  ];
158
205
  }, []);
159
- lines.forEach(
206
+ lines2.forEach(
160
207
  (line) => addChapterContent(reference.chapter, [line])
161
208
  );
162
209
  } else {
@@ -174,6 +221,9 @@ class CodexParser {
174
221
  }
175
222
  }
176
223
  }
224
+ if (previousReference) {
225
+ addReference(previousReference, addLineBreaks(lines));
226
+ }
177
227
  for (let chapter of chapters.values()) {
178
228
  root.content.push(chapter);
179
229
  }
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "version": 3,
3
3
  "sources": ["../../../parser/codex-parser.ts"],
4
- "sourcesContent": ["import { z } from 'zod';\nimport {\n Chapter,\n ChapterContent,\n ParseTree,\n Verse,\n VerseContent,\n} from './types.js';\nimport { parseVerseReference, VerseRef } from '../utils.js';\n\nexport const chapterHeadingSchema = z.object({\n type: z.literal('chapter-heading'),\n data: z.object({\n chapter: z.coerce.number().int(),\n }),\n});\n\nexport const verseSchema = z.object({\n type: z.literal('text'),\n id: z.string(),\n});\n\nexport const paratextSchema = z.object({\n type: z.literal('paratext'),\n id: z.string(),\n});\n\nexport const metadataSchema = z.discriminatedUnion('type', [\n chapterHeadingSchema,\n verseSchema,\n paratextSchema,\n]);\n\nexport const codexSchema = z.object({\n cells: z.array(\n z.object({\n kind: z.number(),\n languageId: z.string(),\n value: z.string(),\n metadata: z\n .object({\n type: z.string(),\n })\n .passthrough()\n .nullable()\n .optional(),\n })\n ),\n});\n\n/**\n * Defines a parser for codex files.\n */\nexport class CodexParser {\n // TODO:\n // /**\n // * Whether to preserve markdown in the parsed content.\n // */\n // preserveMarkdown: boolean = true;\n\n private _noteCounter: number = 0;\n\n /**\n * Parses the specified codex content.\n *\n * @param codex The codex content to parse.\n * @returns The parse tree that was generated.\n */\n public parse(codex: string): ParseTree {\n const json = JSON.parse(codex);\n const data = codexSchema.parse(json);\n\n let root: ParseTree = {\n type: 'root',\n content: [],\n };\n\n let chapters: Map<number, Chapter> = new Map();\n let lastChapter: Chapter | null = null;\n\n function addChapterContent(\n chapterNumber: number,\n content: ChapterContent[]\n ) {\n let chapter = chapters.get(chapterNumber);\n if (!chapter) {\n chapter = {\n type: 'chapter',\n number: chapterNumber,\n content: [],\n footnotes: [],\n };\n lastChapter = chapter;\n chapters.set(chapterNumber, chapter);\n }\n\n chapter.content.push(...content);\n }\n\n function addReference(ref: VerseRef, content: Verse['content']) {\n addChapterContent(ref.chapter, [\n {\n type: 'verse',\n number: ref.verse,\n content: content,\n },\n ]);\n }\n\n function stripHTML(value: string) {\n return value.replace(/<[^>]*>/g, '');\n }\n\n function toHeading(content: string) {\n return {\n type: 'heading',\n content: [content],\n } satisfies ChapterContent;\n }\n\n for (let cell of data.cells) {\n if (cell.languageId === 'html' && cell.value) {\n if (!cell.metadata) continue; // ignore html cells without metadata\n\n const metadataResult = metadataSchema.safeParse(cell.metadata);\n\n if (!metadataResult.success) continue; // ignore cells with invalid/unknown metadata\n const metadata = metadataResult.data;\n if (metadata.type === 'text') {\n const reference = parseVerseReference(metadata.id);\n if (!reference)\n throw new Error('Could not find verse reference.');\n\n if (!root.id) root.id = reference.book; // set book id if not already set\n\n const content = stripHTML(cell.value);\n\n // add line breaks in heading if there are multiple lines\n const lines = content.split('\\n').reduce(\n (prev, current) => {\n if (prev.length === 0) return [current];\n else\n return [\n ...prev,\n {\n lineBreak: true,\n },\n current,\n ] satisfies Verse['content'];\n },\n [] as Verse['content']\n );\n\n addReference(reference, lines);\n } else if (metadata.type === 'paratext') {\n const reference = parseVerseReference(`${metadata.id}:1`); // chapter headings do not have a verse reference always default to 1\n if (reference) {\n const content = stripHTML(cell.value);\n\n // add line breaks in heading if there are multiple lines\n const lines = content\n .split('\\n')\n .reduce<ChapterContent[]>((prev, current) => {\n if (prev.length === 0)\n return [toHeading(current)];\n else\n return [\n ...prev,\n {\n type: 'line_break',\n },\n toHeading(current),\n ] satisfies ChapterContent[];\n }, []);\n\n lines.forEach((line) =>\n addChapterContent(reference.chapter, [line])\n );\n } else {\n // parse paratext that isn't a chapter reference as footnote\n const content = stripHTML(cell.value);\n if (!cell.metadata) {\n if (lastChapter) {\n (lastChapter as Chapter).footnotes.push({\n noteId: this._noteCounter++,\n text: content,\n caller: null,\n });\n }\n }\n }\n }\n }\n }\n\n for (let chapter of chapters.values()) {\n root.content.push(chapter);\n }\n\n return root;\n }\n}\n"],
5
- "mappings": ";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,iBAAkB;AAQlB,mBAA8C;AAEvC,MAAM,uBAAuB,aAAE,OAAO;AAAA,EACzC,MAAM,aAAE,QAAQ,iBAAiB;AAAA,EACjC,MAAM,aAAE,OAAO;AAAA,IACX,SAAS,aAAE,OAAO,OAAO,EAAE,IAAI;AAAA,EACnC,CAAC;AACL,CAAC;AAEM,MAAM,cAAc,aAAE,OAAO;AAAA,EAChC,MAAM,aAAE,QAAQ,MAAM;AAAA,EACtB,IAAI,aAAE,OAAO;AACjB,CAAC;AAEM,MAAM,iBAAiB,aAAE,OAAO;AAAA,EACnC,MAAM,aAAE,QAAQ,UAAU;AAAA,EAC1B,IAAI,aAAE,OAAO;AACjB,CAAC;AAEM,MAAM,iBAAiB,aAAE,mBAAmB,QAAQ;AAAA,EACvD;AAAA,EACA;AAAA,EACA;AACJ,CAAC;AAEM,MAAM,cAAc,aAAE,OAAO;AAAA,EAChC,OAAO,aAAE;AAAA,IACL,aAAE,OAAO;AAAA,MACL,MAAM,aAAE,OAAO;AAAA,MACf,YAAY,aAAE,OAAO;AAAA,MACrB,OAAO,aAAE,OAAO;AAAA,MAChB,UAAU,aACL,OAAO;AAAA,QACJ,MAAM,aAAE,OAAO;AAAA,MACnB,CAAC,EACA,YAAY,EACZ,SAAS,EACT,SAAS;AAAA,IAClB,CAAC;AAAA,EACL;AACJ,CAAC;AAKM,MAAM,YAAY;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOb,eAAuB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQxB,MAAM,OAA0B;AACnC,UAAM,OAAO,KAAK,MAAM,KAAK;AAC7B,UAAM,OAAO,YAAY,MAAM,IAAI;AAEnC,QAAI,OAAkB;AAAA,MAClB,MAAM;AAAA,MACN,SAAS,CAAC;AAAA,IACd;AAEA,QAAI,WAAiC,oBAAI,IAAI;AAC7C,QAAI,cAA8B;AAElC,aAAS,kBACL,eACA,SACF;AACE,UAAI,UAAU,SAAS,IAAI,aAAa;AACxC,UAAI,CAAC,SAAS;AACV,kBAAU;AAAA,UACN,MAAM;AAAA,UACN,QAAQ;AAAA,UACR,SAAS,CAAC;AAAA,UACV,WAAW,CAAC;AAAA,QAChB;AACA,sBAAc;AACd,iBAAS,IAAI,eAAe,OAAO;AAAA,MACvC;AAEA,cAAQ,QAAQ,KAAK,GAAG,OAAO;AAAA,IACnC;AAEA,aAAS,aAAa,KAAe,SAA2B;AAC5D,wBAAkB,IAAI,SAAS;AAAA,QAC3B;AAAA,UACI,MAAM;AAAA,UACN,QAAQ,IAAI;AAAA,UACZ;AAAA,QACJ;AAAA,MACJ,CAAC;AAAA,IACL;AAEA,aAAS,UAAU,OAAe;AAC9B,aAAO,MAAM,QAAQ,YAAY,EAAE;AAAA,IACvC;AAEA,aAAS,UAAU,SAAiB;AAChC,aAAO;AAAA,QACH,MAAM;AAAA,QACN,SAAS,CAAC,OAAO;AAAA,MACrB;AAAA,IACJ;AAEA,aAAS,QAAQ,KAAK,OAAO;AACzB,UAAI,KAAK,eAAe,UAAU,KAAK,OAAO;AAC1C,YAAI,CAAC,KAAK,SAAU;AAEpB,cAAM,iBAAiB,eAAe,UAAU,KAAK,QAAQ;AAE7D,YAAI,CAAC,eAAe,QAAS;AAC7B,cAAM,WAAW,eAAe;AAChC,YAAI,SAAS,SAAS,QAAQ;AAC1B,gBAAM,gBAAY,kCAAoB,SAAS,EAAE;AACjD,cAAI,CAAC;AACD,kBAAM,IAAI,MAAM,iCAAiC;AAErD,cAAI,CAAC,KAAK,GAAI,MAAK,KAAK,UAAU;AAElC,gBAAM,UAAU,UAAU,KAAK,KAAK;AAGpC,gBAAM,QAAQ,QAAQ,MAAM,IAAI,EAAE;AAAA,YAC9B,CAAC,MAAM,YAAY;AACf,kBAAI,KAAK,WAAW,EAAG,QAAO,CAAC,OAAO;AAAA;AAElC,uBAAO;AAAA,kBACH,GAAG;AAAA,kBACH;AAAA,oBACI,WAAW;AAAA,kBACf;AAAA,kBACA;AAAA,gBACJ;AAAA,YACR;AAAA,YACA,CAAC;AAAA,UACL;AAEA,uBAAa,WAAW,KAAK;AAAA,QACjC,WAAW,SAAS,SAAS,YAAY;AACrC,gBAAM,gBAAY,kCAAoB,GAAG,SAAS,EAAE,IAAI;AACxD,cAAI,WAAW;AACX,kBAAM,UAAU,UAAU,KAAK,KAAK;AAGpC,kBAAM,QAAQ,QACT,MAAM,IAAI,EACV,OAAyB,CAAC,MAAM,YAAY;AACzC,kBAAI,KAAK,WAAW;AAChB,uBAAO,CAAC,UAAU,OAAO,CAAC;AAAA;AAE1B,uBAAO;AAAA,kBACH,GAAG;AAAA,kBACH;AAAA,oBACI,MAAM;AAAA,kBACV;AAAA,kBACA,UAAU,OAAO;AAAA,gBACrB;AAAA,YACR,GAAG,CAAC,CAAC;AAET,kBAAM;AAAA,cAAQ,CAAC,SACX,kBAAkB,UAAU,SAAS,CAAC,IAAI,CAAC;AAAA,YAC/C;AAAA,UACJ,OAAO;AAEH,kBAAM,UAAU,UAAU,KAAK,KAAK;AACpC,gBAAI,CAAC,KAAK,UAAU;AAChB,kBAAI,aAAa;AACb,gBAAC,YAAwB,UAAU,KAAK;AAAA,kBACpC,QAAQ,KAAK;AAAA,kBACb,MAAM;AAAA,kBACN,QAAQ;AAAA,gBACZ,CAAC;AAAA,cACL;AAAA,YACJ;AAAA,UACJ;AAAA,QACJ;AAAA,MACJ;AAAA,IACJ;AAEA,aAAS,WAAW,SAAS,OAAO,GAAG;AACnC,WAAK,QAAQ,KAAK,OAAO;AAAA,IAC7B;AAEA,WAAO;AAAA,EACX;AACJ;",
6
- "names": []
4
+ "sourcesContent": ["import { z } from 'zod';\nimport {\n Chapter,\n ChapterContent,\n ParseTree,\n Verse,\n VerseContent,\n} from './types.js';\nimport { getBookId, parseVerseReference, VerseRef } from '../utils.js';\nimport { ref } from 'process';\n\nexport const chapterHeadingSchema = z.object({\n type: z.literal('chapter-heading'),\n data: z.object({\n chapter: z.coerce.number().int(),\n }),\n});\n\nexport const verseSchema = z.object({\n type: z.literal('text'),\n id: z.string(),\n bookCode: z.string().optional(),\n chapter: z.number().optional(),\n verse: z.number().optional(),\n});\n\nexport const paratextSchema = z.object({\n type: z.literal('paratext'),\n id: z.string(),\n});\n\nexport const metadataSchema = z.discriminatedUnion('type', [\n chapterHeadingSchema,\n verseSchema,\n paratextSchema,\n]);\n\nexport const codexSchema = z.object({\n cells: z.array(\n z.object({\n kind: z.number(),\n languageId: z.string(),\n value: z.string(),\n metadata: z\n .object({\n type: z.string(),\n })\n .passthrough()\n .nullable()\n .optional(),\n })\n ),\n});\n\n/**\n * Defines a parser for codex files.\n */\nexport class CodexParser {\n // TODO:\n // /**\n // * Whether to preserve markdown in the parsed content.\n // */\n // preserveMarkdown: boolean = true;\n\n private _noteCounter: number = 0;\n\n /**\n * Parses the specified codex content.\n *\n * @param codex The codex content to parse.\n * @returns The parse tree that was generated.\n */\n public parse(codex: string): ParseTree {\n const json = JSON.parse(codex);\n const data = codexSchema.parse(json);\n\n let root: ParseTree = {\n type: 'root',\n content: [],\n };\n\n let chapters: Map<number, Chapter> = new Map();\n let lastChapter: Chapter | null = null;\n\n function addChapterContent(\n chapterNumber: number,\n content: ChapterContent[]\n ) {\n let chapter = chapters.get(chapterNumber);\n if (!chapter) {\n chapter = {\n type: 'chapter',\n number: chapterNumber,\n content: [],\n footnotes: [],\n };\n lastChapter = chapter;\n chapters.set(chapterNumber, chapter);\n }\n\n chapter.content.push(...content);\n }\n\n function addReference(ref: VerseRef, content: Verse['content']) {\n addChapterContent(ref.chapter, [\n {\n type: 'verse',\n number: ref.verse,\n content: content,\n },\n ]);\n }\n\n function stripHTML(value: string) {\n return value.replace(/<[^>]*>/g, '');\n }\n\n function toHeading(content: string) {\n return {\n type: 'heading',\n content: [content],\n } satisfies ChapterContent;\n }\n\n function addLineBreaks(lines: Verse['content']) {\n return lines.reduce(\n (prev, current) => {\n if (prev.length === 0) return [current];\n else\n return [\n ...prev,\n {\n lineBreak: true,\n },\n current,\n ] satisfies Verse['content'];\n },\n [] as Verse['content']\n );\n }\n\n let previousReference: VerseRef | null = null;\n let lines: Verse['content'] = [];\n\n for (let cell of data.cells) {\n if (cell.languageId === 'html' && cell.value) {\n if (!cell.metadata) continue; // ignore html cells without metadata\n\n const metadataResult = metadataSchema.safeParse(cell.metadata);\n\n if (!metadataResult.success) continue; // ignore cells with invalid/unknown metadata\n const metadata = metadataResult.data;\n if (metadata.type === 'text') {\n let reference = parseVerseReference(metadata.id);\n if (!reference) {\n if (metadata.bookCode && metadata.chapter) {\n const bookId = getBookId(metadata.bookCode);\n if (!bookId) {\n throw new Error(\n 'Could not find book ID for code: ' +\n metadata.bookCode\n );\n }\n\n if (metadata.verse) {\n reference = {\n book: bookId,\n chapter: metadata.chapter,\n verse: metadata.verse,\n };\n } else if (previousReference?.verse) {\n if (\n previousReference.chapter !==\n metadata.chapter ||\n previousReference.book !== bookId\n ) {\n throw new Error(\n 'Cannot infer verse number for non-consecutive verse reference.'\n );\n }\n\n // Sometimes codex doesn't include verse numbers for every cell, so we infer it from the previous reference\n reference = {\n book: bookId,\n chapter: metadata.chapter,\n verse: previousReference.verse,\n };\n } else {\n throw new Error(\n 'Could not find verse reference in metadata.'\n );\n }\n } else {\n throw new Error(\n 'Could not find verse reference in metadata.'\n );\n }\n }\n\n if (\n previousReference &&\n (reference.book !== previousReference.book ||\n reference.chapter !== previousReference.chapter ||\n reference.verse !== previousReference.verse)\n ) {\n addReference(previousReference, addLineBreaks(lines));\n lines = [];\n }\n\n previousReference = reference;\n\n if (!root.id) root.id = reference.book; // set book id if not already set\n\n const content = stripHTML(cell.value);\n\n // add line breaks in heading if there are multiple lines\n const newLines = content.split('\\n');\n\n lines.push(...newLines);\n } else if (metadata.type === 'paratext') {\n const reference = parseVerseReference(`${metadata.id}:1`); // chapter headings do not have a verse reference always default to 1\n if (reference) {\n const content = stripHTML(cell.value);\n\n // add line breaks in heading if there are multiple lines\n const lines = content\n .split('\\n')\n .reduce<ChapterContent[]>((prev, current) => {\n if (prev.length === 0)\n return [toHeading(current)];\n else\n return [\n ...prev,\n {\n type: 'line_break',\n },\n toHeading(current),\n ] satisfies ChapterContent[];\n }, []);\n\n lines.forEach((line) =>\n addChapterContent(reference.chapter, [line])\n );\n } else {\n // parse paratext that isn't a chapter reference as footnote\n const content = stripHTML(cell.value);\n if (!cell.metadata) {\n if (lastChapter) {\n (lastChapter as Chapter).footnotes.push({\n noteId: this._noteCounter++,\n text: content,\n caller: null,\n });\n }\n }\n }\n }\n }\n }\n\n if (previousReference) {\n addReference(previousReference, addLineBreaks(lines));\n }\n\n for (let chapter of chapters.values()) {\n root.content.push(chapter);\n }\n\n return root;\n }\n}\n"],
5
+ "mappings": ";;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,iBAAkB;AAQlB,mBAAyD;AAGlD,MAAM,uBAAuB,aAAE,OAAO;AAAA,EACzC,MAAM,aAAE,QAAQ,iBAAiB;AAAA,EACjC,MAAM,aAAE,OAAO;AAAA,IACX,SAAS,aAAE,OAAO,OAAO,EAAE,IAAI;AAAA,EACnC,CAAC;AACL,CAAC;AAEM,MAAM,cAAc,aAAE,OAAO;AAAA,EAChC,MAAM,aAAE,QAAQ,MAAM;AAAA,EACtB,IAAI,aAAE,OAAO;AAAA,EACb,UAAU,aAAE,OAAO,EAAE,SAAS;AAAA,EAC9B,SAAS,aAAE,OAAO,EAAE,SAAS;AAAA,EAC7B,OAAO,aAAE,OAAO,EAAE,SAAS;AAC/B,CAAC;AAEM,MAAM,iBAAiB,aAAE,OAAO;AAAA,EACnC,MAAM,aAAE,QAAQ,UAAU;AAAA,EAC1B,IAAI,aAAE,OAAO;AACjB,CAAC;AAEM,MAAM,iBAAiB,aAAE,mBAAmB,QAAQ;AAAA,EACvD;AAAA,EACA;AAAA,EACA;AACJ,CAAC;AAEM,MAAM,cAAc,aAAE,OAAO;AAAA,EAChC,OAAO,aAAE;AAAA,IACL,aAAE,OAAO;AAAA,MACL,MAAM,aAAE,OAAO;AAAA,MACf,YAAY,aAAE,OAAO;AAAA,MACrB,OAAO,aAAE,OAAO;AAAA,MAChB,UAAU,aACL,OAAO;AAAA,QACJ,MAAM,aAAE,OAAO;AAAA,MACnB,CAAC,EACA,YAAY,EACZ,SAAS,EACT,SAAS;AAAA,IAClB,CAAC;AAAA,EACL;AACJ,CAAC;AAKM,MAAM,YAAY;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOb,eAAuB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQxB,MAAM,OAA0B;AACnC,UAAM,OAAO,KAAK,MAAM,KAAK;AAC7B,UAAM,OAAO,YAAY,MAAM,IAAI;AAEnC,QAAI,OAAkB;AAAA,MAClB,MAAM;AAAA,MACN,SAAS,CAAC;AAAA,IACd;AAEA,QAAI,WAAiC,oBAAI,IAAI;AAC7C,QAAI,cAA8B;AAElC,aAAS,kBACL,eACA,SACF;AACE,UAAI,UAAU,SAAS,IAAI,aAAa;AACxC,UAAI,CAAC,SAAS;AACV,kBAAU;AAAA,UACN,MAAM;AAAA,UACN,QAAQ;AAAA,UACR,SAAS,CAAC;AAAA,UACV,WAAW,CAAC;AAAA,QAChB;AACA,sBAAc;AACd,iBAAS,IAAI,eAAe,OAAO;AAAA,MACvC;AAEA,cAAQ,QAAQ,KAAK,GAAG,OAAO;AAAA,IACnC;AAEA,aAAS,aAAaA,MAAe,SAA2B;AAC5D,wBAAkBA,KAAI,SAAS;AAAA,QAC3B;AAAA,UACI,MAAM;AAAA,UACN,QAAQA,KAAI;AAAA,UACZ;AAAA,QACJ;AAAA,MACJ,CAAC;AAAA,IACL;AAEA,aAAS,UAAU,OAAe;AAC9B,aAAO,MAAM,QAAQ,YAAY,EAAE;AAAA,IACvC;AAEA,aAAS,UAAU,SAAiB;AAChC,aAAO;AAAA,QACH,MAAM;AAAA,QACN,SAAS,CAAC,OAAO;AAAA,MACrB;AAAA,IACJ;AAEA,aAAS,cAAcC,QAAyB;AAC5C,aAAOA,OAAM;AAAA,QACT,CAAC,MAAM,YAAY;AACf,cAAI,KAAK,WAAW,EAAG,QAAO,CAAC,OAAO;AAAA;AAElC,mBAAO;AAAA,cACH,GAAG;AAAA,cACH;AAAA,gBACI,WAAW;AAAA,cACf;AAAA,cACA;AAAA,YACJ;AAAA,QACR;AAAA,QACA,CAAC;AAAA,MACL;AAAA,IACJ;AAEA,QAAI,oBAAqC;AACzC,QAAI,QAA0B,CAAC;AAE/B,aAAS,QAAQ,KAAK,OAAO;AACzB,UAAI,KAAK,eAAe,UAAU,KAAK,OAAO;AAC1C,YAAI,CAAC,KAAK,SAAU;AAEpB,cAAM,iBAAiB,eAAe,UAAU,KAAK,QAAQ;AAE7D,YAAI,CAAC,eAAe,QAAS;AAC7B,cAAM,WAAW,eAAe;AAChC,YAAI,SAAS,SAAS,QAAQ;AAC1B,cAAI,gBAAY,kCAAoB,SAAS,EAAE;AAC/C,cAAI,CAAC,WAAW;AACZ,gBAAI,SAAS,YAAY,SAAS,SAAS;AACvC,oBAAM,aAAS,wBAAU,SAAS,QAAQ;AAC1C,kBAAI,CAAC,QAAQ;AACT,sBAAM,IAAI;AAAA,kBACN,sCACI,SAAS;AAAA,gBACjB;AAAA,cACJ;AAEA,kBAAI,SAAS,OAAO;AAChB,4BAAY;AAAA,kBACR,MAAM;AAAA,kBACN,SAAS,SAAS;AAAA,kBAClB,OAAO,SAAS;AAAA,gBACpB;AAAA,cACJ,WAAW,mBAAmB,OAAO;AACjC,oBACI,kBAAkB,YACd,SAAS,WACb,kBAAkB,SAAS,QAC7B;AACE,wBAAM,IAAI;AAAA,oBACN;AAAA,kBACJ;AAAA,gBACJ;AAGA,4BAAY;AAAA,kBACR,MAAM;AAAA,kBACN,SAAS,SAAS;AAAA,kBAClB,OAAO,kBAAkB;AAAA,gBAC7B;AAAA,cACJ,OAAO;AACH,sBAAM,IAAI;AAAA,kBACN;AAAA,gBACJ;AAAA,cACJ;AAAA,YACJ,OAAO;AACH,oBAAM,IAAI;AAAA,gBACN;AAAA,cACJ;AAAA,YACJ;AAAA,UACJ;AAEA,cACI,sBACC,UAAU,SAAS,kBAAkB,QAClC,UAAU,YAAY,kBAAkB,WACxC,UAAU,UAAU,kBAAkB,QAC5C;AACE,yBAAa,mBAAmB,cAAc,KAAK,CAAC;AACpD,oBAAQ,CAAC;AAAA,UACb;AAEA,8BAAoB;AAEpB,cAAI,CAAC,KAAK,GAAI,MAAK,KAAK,UAAU;AAElC,gBAAM,UAAU,UAAU,KAAK,KAAK;AAGpC,gBAAM,WAAW,QAAQ,MAAM,IAAI;AAEnC,gBAAM,KAAK,GAAG,QAAQ;AAAA,QAC1B,WAAW,SAAS,SAAS,YAAY;AACrC,gBAAM,gBAAY,kCAAoB,GAAG,SAAS,EAAE,IAAI;AACxD,cAAI,WAAW;AACX,kBAAM,UAAU,UAAU,KAAK,KAAK;AAGpC,kBAAMA,SAAQ,QACT,MAAM,IAAI,EACV,OAAyB,CAAC,MAAM,YAAY;AACzC,kBAAI,KAAK,WAAW;AAChB,uBAAO,CAAC,UAAU,OAAO,CAAC;AAAA;AAE1B,uBAAO;AAAA,kBACH,GAAG;AAAA,kBACH;AAAA,oBACI,MAAM;AAAA,kBACV;AAAA,kBACA,UAAU,OAAO;AAAA,gBACrB;AAAA,YACR,GAAG,CAAC,CAAC;AAET,YAAAA,OAAM;AAAA,cAAQ,CAAC,SACX,kBAAkB,UAAU,SAAS,CAAC,IAAI,CAAC;AAAA,YAC/C;AAAA,UACJ,OAAO;AAEH,kBAAM,UAAU,UAAU,KAAK,KAAK;AACpC,gBAAI,CAAC,KAAK,UAAU;AAChB,kBAAI,aAAa;AACb,gBAAC,YAAwB,UAAU,KAAK;AAAA,kBACpC,QAAQ,KAAK;AAAA,kBACb,MAAM;AAAA,kBACN,QAAQ;AAAA,gBACZ,CAAC;AAAA,cACL;AAAA,YACJ;AAAA,UACJ;AAAA,QACJ;AAAA,MACJ;AAAA,IACJ;AAEA,QAAI,mBAAmB;AACnB,mBAAa,mBAAmB,cAAc,KAAK,CAAC;AAAA,IACxD;AAEA,aAAS,WAAW,SAAS,OAAO,GAAG;AACnC,WAAK,QAAQ,KAAK,OAAO;AAAA,IAC7B;AAEA,WAAO;AAAA,EACX;AACJ;",
6
+ "names": ["ref", "lines"]
7
7
  }
@@ -1,6 +1,6 @@
1
1
  "use strict";
2
2
  import { z } from "zod";
3
- import { parseVerseReference } from "../utils.js";
3
+ import { getBookId, parseVerseReference } from "../utils.js";
4
4
  export const chapterHeadingSchema = z.object({
5
5
  type: z.literal("chapter-heading"),
6
6
  data: z.object({
@@ -9,7 +9,10 @@ export const chapterHeadingSchema = z.object({
9
9
  });
10
10
  export const verseSchema = z.object({
11
11
  type: z.literal("text"),
12
- id: z.string()
12
+ id: z.string(),
13
+ bookCode: z.string().optional(),
14
+ chapter: z.number().optional(),
15
+ verse: z.number().optional()
13
16
  });
14
17
  export const paratextSchema = z.object({
15
18
  type: z.literal("paratext"),
@@ -68,11 +71,11 @@ export class CodexParser {
68
71
  }
69
72
  chapter.content.push(...content);
70
73
  }
71
- function addReference(ref, content) {
72
- addChapterContent(ref.chapter, [
74
+ function addReference(ref2, content) {
75
+ addChapterContent(ref2.chapter, [
73
76
  {
74
77
  type: "verse",
75
- number: ref.verse,
78
+ number: ref2.verse,
76
79
  content
77
80
  }
78
81
  ]);
@@ -86,6 +89,24 @@ export class CodexParser {
86
89
  content: [content]
87
90
  };
88
91
  }
92
+ function addLineBreaks(lines2) {
93
+ return lines2.reduce(
94
+ (prev, current) => {
95
+ if (prev.length === 0) return [current];
96
+ else
97
+ return [
98
+ ...prev,
99
+ {
100
+ lineBreak: true
101
+ },
102
+ current
103
+ ];
104
+ },
105
+ []
106
+ );
107
+ }
108
+ let previousReference = null;
109
+ let lines = [];
89
110
  for (let cell of data.cells) {
90
111
  if (cell.languageId === "html" && cell.value) {
91
112
  if (!cell.metadata) continue;
@@ -93,31 +114,57 @@ export class CodexParser {
93
114
  if (!metadataResult.success) continue;
94
115
  const metadata = metadataResult.data;
95
116
  if (metadata.type === "text") {
96
- const reference = parseVerseReference(metadata.id);
97
- if (!reference)
98
- throw new Error("Could not find verse reference.");
117
+ let reference = parseVerseReference(metadata.id);
118
+ if (!reference) {
119
+ if (metadata.bookCode && metadata.chapter) {
120
+ const bookId = getBookId(metadata.bookCode);
121
+ if (!bookId) {
122
+ throw new Error(
123
+ "Could not find book ID for code: " + metadata.bookCode
124
+ );
125
+ }
126
+ if (metadata.verse) {
127
+ reference = {
128
+ book: bookId,
129
+ chapter: metadata.chapter,
130
+ verse: metadata.verse
131
+ };
132
+ } else if (previousReference?.verse) {
133
+ if (previousReference.chapter !== metadata.chapter || previousReference.book !== bookId) {
134
+ throw new Error(
135
+ "Cannot infer verse number for non-consecutive verse reference."
136
+ );
137
+ }
138
+ reference = {
139
+ book: bookId,
140
+ chapter: metadata.chapter,
141
+ verse: previousReference.verse
142
+ };
143
+ } else {
144
+ throw new Error(
145
+ "Could not find verse reference in metadata."
146
+ );
147
+ }
148
+ } else {
149
+ throw new Error(
150
+ "Could not find verse reference in metadata."
151
+ );
152
+ }
153
+ }
154
+ if (previousReference && (reference.book !== previousReference.book || reference.chapter !== previousReference.chapter || reference.verse !== previousReference.verse)) {
155
+ addReference(previousReference, addLineBreaks(lines));
156
+ lines = [];
157
+ }
158
+ previousReference = reference;
99
159
  if (!root.id) root.id = reference.book;
100
160
  const content = stripHTML(cell.value);
101
- const lines = content.split("\n").reduce(
102
- (prev, current) => {
103
- if (prev.length === 0) return [current];
104
- else
105
- return [
106
- ...prev,
107
- {
108
- lineBreak: true
109
- },
110
- current
111
- ];
112
- },
113
- []
114
- );
115
- addReference(reference, lines);
161
+ const newLines = content.split("\n");
162
+ lines.push(...newLines);
116
163
  } else if (metadata.type === "paratext") {
117
164
  const reference = parseVerseReference(`${metadata.id}:1`);
118
165
  if (reference) {
119
166
  const content = stripHTML(cell.value);
120
- const lines = content.split("\n").reduce((prev, current) => {
167
+ const lines2 = content.split("\n").reduce((prev, current) => {
121
168
  if (prev.length === 0)
122
169
  return [toHeading(current)];
123
170
  else
@@ -129,7 +176,7 @@ export class CodexParser {
129
176
  toHeading(current)
130
177
  ];
131
178
  }, []);
132
- lines.forEach(
179
+ lines2.forEach(
133
180
  (line) => addChapterContent(reference.chapter, [line])
134
181
  );
135
182
  } else {
@@ -147,6 +194,9 @@ export class CodexParser {
147
194
  }
148
195
  }
149
196
  }
197
+ if (previousReference) {
198
+ addReference(previousReference, addLineBreaks(lines));
199
+ }
150
200
  for (let chapter of chapters.values()) {
151
201
  root.content.push(chapter);
152
202
  }
@@ -1,7 +1,7 @@
1
1
  {
2
2
  "version": 3,
3
3
  "sources": ["../../../parser/codex-parser.ts"],
4
- "sourcesContent": ["import { z } from 'zod';\nimport {\n Chapter,\n ChapterContent,\n ParseTree,\n Verse,\n VerseContent,\n} from './types.js';\nimport { parseVerseReference, VerseRef } from '../utils.js';\n\nexport const chapterHeadingSchema = z.object({\n type: z.literal('chapter-heading'),\n data: z.object({\n chapter: z.coerce.number().int(),\n }),\n});\n\nexport const verseSchema = z.object({\n type: z.literal('text'),\n id: z.string(),\n});\n\nexport const paratextSchema = z.object({\n type: z.literal('paratext'),\n id: z.string(),\n});\n\nexport const metadataSchema = z.discriminatedUnion('type', [\n chapterHeadingSchema,\n verseSchema,\n paratextSchema,\n]);\n\nexport const codexSchema = z.object({\n cells: z.array(\n z.object({\n kind: z.number(),\n languageId: z.string(),\n value: z.string(),\n metadata: z\n .object({\n type: z.string(),\n })\n .passthrough()\n .nullable()\n .optional(),\n })\n ),\n});\n\n/**\n * Defines a parser for codex files.\n */\nexport class CodexParser {\n // TODO:\n // /**\n // * Whether to preserve markdown in the parsed content.\n // */\n // preserveMarkdown: boolean = true;\n\n private _noteCounter: number = 0;\n\n /**\n * Parses the specified codex content.\n *\n * @param codex The codex content to parse.\n * @returns The parse tree that was generated.\n */\n public parse(codex: string): ParseTree {\n const json = JSON.parse(codex);\n const data = codexSchema.parse(json);\n\n let root: ParseTree = {\n type: 'root',\n content: [],\n };\n\n let chapters: Map<number, Chapter> = new Map();\n let lastChapter: Chapter | null = null;\n\n function addChapterContent(\n chapterNumber: number,\n content: ChapterContent[]\n ) {\n let chapter = chapters.get(chapterNumber);\n if (!chapter) {\n chapter = {\n type: 'chapter',\n number: chapterNumber,\n content: [],\n footnotes: [],\n };\n lastChapter = chapter;\n chapters.set(chapterNumber, chapter);\n }\n\n chapter.content.push(...content);\n }\n\n function addReference(ref: VerseRef, content: Verse['content']) {\n addChapterContent(ref.chapter, [\n {\n type: 'verse',\n number: ref.verse,\n content: content,\n },\n ]);\n }\n\n function stripHTML(value: string) {\n return value.replace(/<[^>]*>/g, '');\n }\n\n function toHeading(content: string) {\n return {\n type: 'heading',\n content: [content],\n } satisfies ChapterContent;\n }\n\n for (let cell of data.cells) {\n if (cell.languageId === 'html' && cell.value) {\n if (!cell.metadata) continue; // ignore html cells without metadata\n\n const metadataResult = metadataSchema.safeParse(cell.metadata);\n\n if (!metadataResult.success) continue; // ignore cells with invalid/unknown metadata\n const metadata = metadataResult.data;\n if (metadata.type === 'text') {\n const reference = parseVerseReference(metadata.id);\n if (!reference)\n throw new Error('Could not find verse reference.');\n\n if (!root.id) root.id = reference.book; // set book id if not already set\n\n const content = stripHTML(cell.value);\n\n // add line breaks in heading if there are multiple lines\n const lines = content.split('\\n').reduce(\n (prev, current) => {\n if (prev.length === 0) return [current];\n else\n return [\n ...prev,\n {\n lineBreak: true,\n },\n current,\n ] satisfies Verse['content'];\n },\n [] as Verse['content']\n );\n\n addReference(reference, lines);\n } else if (metadata.type === 'paratext') {\n const reference = parseVerseReference(`${metadata.id}:1`); // chapter headings do not have a verse reference always default to 1\n if (reference) {\n const content = stripHTML(cell.value);\n\n // add line breaks in heading if there are multiple lines\n const lines = content\n .split('\\n')\n .reduce<ChapterContent[]>((prev, current) => {\n if (prev.length === 0)\n return [toHeading(current)];\n else\n return [\n ...prev,\n {\n type: 'line_break',\n },\n toHeading(current),\n ] satisfies ChapterContent[];\n }, []);\n\n lines.forEach((line) =>\n addChapterContent(reference.chapter, [line])\n );\n } else {\n // parse paratext that isn't a chapter reference as footnote\n const content = stripHTML(cell.value);\n if (!cell.metadata) {\n if (lastChapter) {\n (lastChapter as Chapter).footnotes.push({\n noteId: this._noteCounter++,\n text: content,\n caller: null,\n });\n }\n }\n }\n }\n }\n }\n\n for (let chapter of chapters.values()) {\n root.content.push(chapter);\n }\n\n return root;\n }\n}\n"],
5
- "mappings": ";AAAA,SAAS,SAAS;AAQlB,SAAS,2BAAqC;AAEvC,aAAM,uBAAuB,EAAE,OAAO;AAAA,EACzC,MAAM,EAAE,QAAQ,iBAAiB;AAAA,EACjC,MAAM,EAAE,OAAO;AAAA,IACX,SAAS,EAAE,OAAO,OAAO,EAAE,IAAI;AAAA,EACnC,CAAC;AACL,CAAC;AAEM,aAAM,cAAc,EAAE,OAAO;AAAA,EAChC,MAAM,EAAE,QAAQ,MAAM;AAAA,EACtB,IAAI,EAAE,OAAO;AACjB,CAAC;AAEM,aAAM,iBAAiB,EAAE,OAAO;AAAA,EACnC,MAAM,EAAE,QAAQ,UAAU;AAAA,EAC1B,IAAI,EAAE,OAAO;AACjB,CAAC;AAEM,aAAM,iBAAiB,EAAE,mBAAmB,QAAQ;AAAA,EACvD;AAAA,EACA;AAAA,EACA;AACJ,CAAC;AAEM,aAAM,cAAc,EAAE,OAAO;AAAA,EAChC,OAAO,EAAE;AAAA,IACL,EAAE,OAAO;AAAA,MACL,MAAM,EAAE,OAAO;AAAA,MACf,YAAY,EAAE,OAAO;AAAA,MACrB,OAAO,EAAE,OAAO;AAAA,MAChB,UAAU,EACL,OAAO;AAAA,QACJ,MAAM,EAAE,OAAO;AAAA,MACnB,CAAC,EACA,YAAY,EACZ,SAAS,EACT,SAAS;AAAA,IAClB,CAAC;AAAA,EACL;AACJ,CAAC;AAKM,aAAM,YAAY;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOb,eAAuB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQxB,MAAM,OAA0B;AACnC,UAAM,OAAO,KAAK,MAAM,KAAK;AAC7B,UAAM,OAAO,YAAY,MAAM,IAAI;AAEnC,QAAI,OAAkB;AAAA,MAClB,MAAM;AAAA,MACN,SAAS,CAAC;AAAA,IACd;AAEA,QAAI,WAAiC,oBAAI,IAAI;AAC7C,QAAI,cAA8B;AAElC,aAAS,kBACL,eACA,SACF;AACE,UAAI,UAAU,SAAS,IAAI,aAAa;AACxC,UAAI,CAAC,SAAS;AACV,kBAAU;AAAA,UACN,MAAM;AAAA,UACN,QAAQ;AAAA,UACR,SAAS,CAAC;AAAA,UACV,WAAW,CAAC;AAAA,QAChB;AACA,sBAAc;AACd,iBAAS,IAAI,eAAe,OAAO;AAAA,MACvC;AAEA,cAAQ,QAAQ,KAAK,GAAG,OAAO;AAAA,IACnC;AAEA,aAAS,aAAa,KAAe,SAA2B;AAC5D,wBAAkB,IAAI,SAAS;AAAA,QAC3B;AAAA,UACI,MAAM;AAAA,UACN,QAAQ,IAAI;AAAA,UACZ;AAAA,QACJ;AAAA,MACJ,CAAC;AAAA,IACL;AAEA,aAAS,UAAU,OAAe;AAC9B,aAAO,MAAM,QAAQ,YAAY,EAAE;AAAA,IACvC;AAEA,aAAS,UAAU,SAAiB;AAChC,aAAO;AAAA,QACH,MAAM;AAAA,QACN,SAAS,CAAC,OAAO;AAAA,MACrB;AAAA,IACJ;AAEA,aAAS,QAAQ,KAAK,OAAO;AACzB,UAAI,KAAK,eAAe,UAAU,KAAK,OAAO;AAC1C,YAAI,CAAC,KAAK,SAAU;AAEpB,cAAM,iBAAiB,eAAe,UAAU,KAAK,QAAQ;AAE7D,YAAI,CAAC,eAAe,QAAS;AAC7B,cAAM,WAAW,eAAe;AAChC,YAAI,SAAS,SAAS,QAAQ;AAC1B,gBAAM,YAAY,oBAAoB,SAAS,EAAE;AACjD,cAAI,CAAC;AACD,kBAAM,IAAI,MAAM,iCAAiC;AAErD,cAAI,CAAC,KAAK,GAAI,MAAK,KAAK,UAAU;AAElC,gBAAM,UAAU,UAAU,KAAK,KAAK;AAGpC,gBAAM,QAAQ,QAAQ,MAAM,IAAI,EAAE;AAAA,YAC9B,CAAC,MAAM,YAAY;AACf,kBAAI,KAAK,WAAW,EAAG,QAAO,CAAC,OAAO;AAAA;AAElC,uBAAO;AAAA,kBACH,GAAG;AAAA,kBACH;AAAA,oBACI,WAAW;AAAA,kBACf;AAAA,kBACA;AAAA,gBACJ;AAAA,YACR;AAAA,YACA,CAAC;AAAA,UACL;AAEA,uBAAa,WAAW,KAAK;AAAA,QACjC,WAAW,SAAS,SAAS,YAAY;AACrC,gBAAM,YAAY,oBAAoB,GAAG,SAAS,EAAE,IAAI;AACxD,cAAI,WAAW;AACX,kBAAM,UAAU,UAAU,KAAK,KAAK;AAGpC,kBAAM,QAAQ,QACT,MAAM,IAAI,EACV,OAAyB,CAAC,MAAM,YAAY;AACzC,kBAAI,KAAK,WAAW;AAChB,uBAAO,CAAC,UAAU,OAAO,CAAC;AAAA;AAE1B,uBAAO;AAAA,kBACH,GAAG;AAAA,kBACH;AAAA,oBACI,MAAM;AAAA,kBACV;AAAA,kBACA,UAAU,OAAO;AAAA,gBACrB;AAAA,YACR,GAAG,CAAC,CAAC;AAET,kBAAM;AAAA,cAAQ,CAAC,SACX,kBAAkB,UAAU,SAAS,CAAC,IAAI,CAAC;AAAA,YAC/C;AAAA,UACJ,OAAO;AAEH,kBAAM,UAAU,UAAU,KAAK,KAAK;AACpC,gBAAI,CAAC,KAAK,UAAU;AAChB,kBAAI,aAAa;AACb,gBAAC,YAAwB,UAAU,KAAK;AAAA,kBACpC,QAAQ,KAAK;AAAA,kBACb,MAAM;AAAA,kBACN,QAAQ;AAAA,gBACZ,CAAC;AAAA,cACL;AAAA,YACJ;AAAA,UACJ;AAAA,QACJ;AAAA,MACJ;AAAA,IACJ;AAEA,aAAS,WAAW,SAAS,OAAO,GAAG;AACnC,WAAK,QAAQ,KAAK,OAAO;AAAA,IAC7B;AAEA,WAAO;AAAA,EACX;AACJ;",
6
- "names": []
4
+ "sourcesContent": ["import { z } from 'zod';\nimport {\n Chapter,\n ChapterContent,\n ParseTree,\n Verse,\n VerseContent,\n} from './types.js';\nimport { getBookId, parseVerseReference, VerseRef } from '../utils.js';\nimport { ref } from 'process';\n\nexport const chapterHeadingSchema = z.object({\n type: z.literal('chapter-heading'),\n data: z.object({\n chapter: z.coerce.number().int(),\n }),\n});\n\nexport const verseSchema = z.object({\n type: z.literal('text'),\n id: z.string(),\n bookCode: z.string().optional(),\n chapter: z.number().optional(),\n verse: z.number().optional(),\n});\n\nexport const paratextSchema = z.object({\n type: z.literal('paratext'),\n id: z.string(),\n});\n\nexport const metadataSchema = z.discriminatedUnion('type', [\n chapterHeadingSchema,\n verseSchema,\n paratextSchema,\n]);\n\nexport const codexSchema = z.object({\n cells: z.array(\n z.object({\n kind: z.number(),\n languageId: z.string(),\n value: z.string(),\n metadata: z\n .object({\n type: z.string(),\n })\n .passthrough()\n .nullable()\n .optional(),\n })\n ),\n});\n\n/**\n * Defines a parser for codex files.\n */\nexport class CodexParser {\n // TODO:\n // /**\n // * Whether to preserve markdown in the parsed content.\n // */\n // preserveMarkdown: boolean = true;\n\n private _noteCounter: number = 0;\n\n /**\n * Parses the specified codex content.\n *\n * @param codex The codex content to parse.\n * @returns The parse tree that was generated.\n */\n public parse(codex: string): ParseTree {\n const json = JSON.parse(codex);\n const data = codexSchema.parse(json);\n\n let root: ParseTree = {\n type: 'root',\n content: [],\n };\n\n let chapters: Map<number, Chapter> = new Map();\n let lastChapter: Chapter | null = null;\n\n function addChapterContent(\n chapterNumber: number,\n content: ChapterContent[]\n ) {\n let chapter = chapters.get(chapterNumber);\n if (!chapter) {\n chapter = {\n type: 'chapter',\n number: chapterNumber,\n content: [],\n footnotes: [],\n };\n lastChapter = chapter;\n chapters.set(chapterNumber, chapter);\n }\n\n chapter.content.push(...content);\n }\n\n function addReference(ref: VerseRef, content: Verse['content']) {\n addChapterContent(ref.chapter, [\n {\n type: 'verse',\n number: ref.verse,\n content: content,\n },\n ]);\n }\n\n function stripHTML(value: string) {\n return value.replace(/<[^>]*>/g, '');\n }\n\n function toHeading(content: string) {\n return {\n type: 'heading',\n content: [content],\n } satisfies ChapterContent;\n }\n\n function addLineBreaks(lines: Verse['content']) {\n return lines.reduce(\n (prev, current) => {\n if (prev.length === 0) return [current];\n else\n return [\n ...prev,\n {\n lineBreak: true,\n },\n current,\n ] satisfies Verse['content'];\n },\n [] as Verse['content']\n );\n }\n\n let previousReference: VerseRef | null = null;\n let lines: Verse['content'] = [];\n\n for (let cell of data.cells) {\n if (cell.languageId === 'html' && cell.value) {\n if (!cell.metadata) continue; // ignore html cells without metadata\n\n const metadataResult = metadataSchema.safeParse(cell.metadata);\n\n if (!metadataResult.success) continue; // ignore cells with invalid/unknown metadata\n const metadata = metadataResult.data;\n if (metadata.type === 'text') {\n let reference = parseVerseReference(metadata.id);\n if (!reference) {\n if (metadata.bookCode && metadata.chapter) {\n const bookId = getBookId(metadata.bookCode);\n if (!bookId) {\n throw new Error(\n 'Could not find book ID for code: ' +\n metadata.bookCode\n );\n }\n\n if (metadata.verse) {\n reference = {\n book: bookId,\n chapter: metadata.chapter,\n verse: metadata.verse,\n };\n } else if (previousReference?.verse) {\n if (\n previousReference.chapter !==\n metadata.chapter ||\n previousReference.book !== bookId\n ) {\n throw new Error(\n 'Cannot infer verse number for non-consecutive verse reference.'\n );\n }\n\n // Sometimes codex doesn't include verse numbers for every cell, so we infer it from the previous reference\n reference = {\n book: bookId,\n chapter: metadata.chapter,\n verse: previousReference.verse,\n };\n } else {\n throw new Error(\n 'Could not find verse reference in metadata.'\n );\n }\n } else {\n throw new Error(\n 'Could not find verse reference in metadata.'\n );\n }\n }\n\n if (\n previousReference &&\n (reference.book !== previousReference.book ||\n reference.chapter !== previousReference.chapter ||\n reference.verse !== previousReference.verse)\n ) {\n addReference(previousReference, addLineBreaks(lines));\n lines = [];\n }\n\n previousReference = reference;\n\n if (!root.id) root.id = reference.book; // set book id if not already set\n\n const content = stripHTML(cell.value);\n\n // add line breaks in heading if there are multiple lines\n const newLines = content.split('\\n');\n\n lines.push(...newLines);\n } else if (metadata.type === 'paratext') {\n const reference = parseVerseReference(`${metadata.id}:1`); // chapter headings do not have a verse reference always default to 1\n if (reference) {\n const content = stripHTML(cell.value);\n\n // add line breaks in heading if there are multiple lines\n const lines = content\n .split('\\n')\n .reduce<ChapterContent[]>((prev, current) => {\n if (prev.length === 0)\n return [toHeading(current)];\n else\n return [\n ...prev,\n {\n type: 'line_break',\n },\n toHeading(current),\n ] satisfies ChapterContent[];\n }, []);\n\n lines.forEach((line) =>\n addChapterContent(reference.chapter, [line])\n );\n } else {\n // parse paratext that isn't a chapter reference as footnote\n const content = stripHTML(cell.value);\n if (!cell.metadata) {\n if (lastChapter) {\n (lastChapter as Chapter).footnotes.push({\n noteId: this._noteCounter++,\n text: content,\n caller: null,\n });\n }\n }\n }\n }\n }\n }\n\n if (previousReference) {\n addReference(previousReference, addLineBreaks(lines));\n }\n\n for (let chapter of chapters.values()) {\n root.content.push(chapter);\n }\n\n return root;\n }\n}\n"],
5
+ "mappings": ";AAAA,SAAS,SAAS;AAQlB,SAAS,WAAW,2BAAqC;AAGlD,aAAM,uBAAuB,EAAE,OAAO;AAAA,EACzC,MAAM,EAAE,QAAQ,iBAAiB;AAAA,EACjC,MAAM,EAAE,OAAO;AAAA,IACX,SAAS,EAAE,OAAO,OAAO,EAAE,IAAI;AAAA,EACnC,CAAC;AACL,CAAC;AAEM,aAAM,cAAc,EAAE,OAAO;AAAA,EAChC,MAAM,EAAE,QAAQ,MAAM;AAAA,EACtB,IAAI,EAAE,OAAO;AAAA,EACb,UAAU,EAAE,OAAO,EAAE,SAAS;AAAA,EAC9B,SAAS,EAAE,OAAO,EAAE,SAAS;AAAA,EAC7B,OAAO,EAAE,OAAO,EAAE,SAAS;AAC/B,CAAC;AAEM,aAAM,iBAAiB,EAAE,OAAO;AAAA,EACnC,MAAM,EAAE,QAAQ,UAAU;AAAA,EAC1B,IAAI,EAAE,OAAO;AACjB,CAAC;AAEM,aAAM,iBAAiB,EAAE,mBAAmB,QAAQ;AAAA,EACvD;AAAA,EACA;AAAA,EACA;AACJ,CAAC;AAEM,aAAM,cAAc,EAAE,OAAO;AAAA,EAChC,OAAO,EAAE;AAAA,IACL,EAAE,OAAO;AAAA,MACL,MAAM,EAAE,OAAO;AAAA,MACf,YAAY,EAAE,OAAO;AAAA,MACrB,OAAO,EAAE,OAAO;AAAA,MAChB,UAAU,EACL,OAAO;AAAA,QACJ,MAAM,EAAE,OAAO;AAAA,MACnB,CAAC,EACA,YAAY,EACZ,SAAS,EACT,SAAS;AAAA,IAClB,CAAC;AAAA,EACL;AACJ,CAAC;AAKM,aAAM,YAAY;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOb,eAAuB;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAQxB,MAAM,OAA0B;AACnC,UAAM,OAAO,KAAK,MAAM,KAAK;AAC7B,UAAM,OAAO,YAAY,MAAM,IAAI;AAEnC,QAAI,OAAkB;AAAA,MAClB,MAAM;AAAA,MACN,SAAS,CAAC;AAAA,IACd;AAEA,QAAI,WAAiC,oBAAI,IAAI;AAC7C,QAAI,cAA8B;AAElC,aAAS,kBACL,eACA,SACF;AACE,UAAI,UAAU,SAAS,IAAI,aAAa;AACxC,UAAI,CAAC,SAAS;AACV,kBAAU;AAAA,UACN,MAAM;AAAA,UACN,QAAQ;AAAA,UACR,SAAS,CAAC;AAAA,UACV,WAAW,CAAC;AAAA,QAChB;AACA,sBAAc;AACd,iBAAS,IAAI,eAAe,OAAO;AAAA,MACvC;AAEA,cAAQ,QAAQ,KAAK,GAAG,OAAO;AAAA,IACnC;AAEA,aAAS,aAAaA,MAAe,SAA2B;AAC5D,wBAAkBA,KAAI,SAAS;AAAA,QAC3B;AAAA,UACI,MAAM;AAAA,UACN,QAAQA,KAAI;AAAA,UACZ;AAAA,QACJ;AAAA,MACJ,CAAC;AAAA,IACL;AAEA,aAAS,UAAU,OAAe;AAC9B,aAAO,MAAM,QAAQ,YAAY,EAAE;AAAA,IACvC;AAEA,aAAS,UAAU,SAAiB;AAChC,aAAO;AAAA,QACH,MAAM;AAAA,QACN,SAAS,CAAC,OAAO;AAAA,MACrB;AAAA,IACJ;AAEA,aAAS,cAAcC,QAAyB;AAC5C,aAAOA,OAAM;AAAA,QACT,CAAC,MAAM,YAAY;AACf,cAAI,KAAK,WAAW,EAAG,QAAO,CAAC,OAAO;AAAA;AAElC,mBAAO;AAAA,cACH,GAAG;AAAA,cACH;AAAA,gBACI,WAAW;AAAA,cACf;AAAA,cACA;AAAA,YACJ;AAAA,QACR;AAAA,QACA,CAAC;AAAA,MACL;AAAA,IACJ;AAEA,QAAI,oBAAqC;AACzC,QAAI,QAA0B,CAAC;AAE/B,aAAS,QAAQ,KAAK,OAAO;AACzB,UAAI,KAAK,eAAe,UAAU,KAAK,OAAO;AAC1C,YAAI,CAAC,KAAK,SAAU;AAEpB,cAAM,iBAAiB,eAAe,UAAU,KAAK,QAAQ;AAE7D,YAAI,CAAC,eAAe,QAAS;AAC7B,cAAM,WAAW,eAAe;AAChC,YAAI,SAAS,SAAS,QAAQ;AAC1B,cAAI,YAAY,oBAAoB,SAAS,EAAE;AAC/C,cAAI,CAAC,WAAW;AACZ,gBAAI,SAAS,YAAY,SAAS,SAAS;AACvC,oBAAM,SAAS,UAAU,SAAS,QAAQ;AAC1C,kBAAI,CAAC,QAAQ;AACT,sBAAM,IAAI;AAAA,kBACN,sCACI,SAAS;AAAA,gBACjB;AAAA,cACJ;AAEA,kBAAI,SAAS,OAAO;AAChB,4BAAY;AAAA,kBACR,MAAM;AAAA,kBACN,SAAS,SAAS;AAAA,kBAClB,OAAO,SAAS;AAAA,gBACpB;AAAA,cACJ,WAAW,mBAAmB,OAAO;AACjC,oBACI,kBAAkB,YACd,SAAS,WACb,kBAAkB,SAAS,QAC7B;AACE,wBAAM,IAAI;AAAA,oBACN;AAAA,kBACJ;AAAA,gBACJ;AAGA,4BAAY;AAAA,kBACR,MAAM;AAAA,kBACN,SAAS,SAAS;AAAA,kBAClB,OAAO,kBAAkB;AAAA,gBAC7B;AAAA,cACJ,OAAO;AACH,sBAAM,IAAI;AAAA,kBACN;AAAA,gBACJ;AAAA,cACJ;AAAA,YACJ,OAAO;AACH,oBAAM,IAAI;AAAA,gBACN;AAAA,cACJ;AAAA,YACJ;AAAA,UACJ;AAEA,cACI,sBACC,UAAU,SAAS,kBAAkB,QAClC,UAAU,YAAY,kBAAkB,WACxC,UAAU,UAAU,kBAAkB,QAC5C;AACE,yBAAa,mBAAmB,cAAc,KAAK,CAAC;AACpD,oBAAQ,CAAC;AAAA,UACb;AAEA,8BAAoB;AAEpB,cAAI,CAAC,KAAK,GAAI,MAAK,KAAK,UAAU;AAElC,gBAAM,UAAU,UAAU,KAAK,KAAK;AAGpC,gBAAM,WAAW,QAAQ,MAAM,IAAI;AAEnC,gBAAM,KAAK,GAAG,QAAQ;AAAA,QAC1B,WAAW,SAAS,SAAS,YAAY;AACrC,gBAAM,YAAY,oBAAoB,GAAG,SAAS,EAAE,IAAI;AACxD,cAAI,WAAW;AACX,kBAAM,UAAU,UAAU,KAAK,KAAK;AAGpC,kBAAMA,SAAQ,QACT,MAAM,IAAI,EACV,OAAyB,CAAC,MAAM,YAAY;AACzC,kBAAI,KAAK,WAAW;AAChB,uBAAO,CAAC,UAAU,OAAO,CAAC;AAAA;AAE1B,uBAAO;AAAA,kBACH,GAAG;AAAA,kBACH;AAAA,oBACI,MAAM;AAAA,kBACV;AAAA,kBACA,UAAU,OAAO;AAAA,gBACrB;AAAA,YACR,GAAG,CAAC,CAAC;AAET,YAAAA,OAAM;AAAA,cAAQ,CAAC,SACX,kBAAkB,UAAU,SAAS,CAAC,IAAI,CAAC;AAAA,YAC/C;AAAA,UACJ,OAAO;AAEH,kBAAM,UAAU,UAAU,KAAK,KAAK;AACpC,gBAAI,CAAC,KAAK,UAAU;AAChB,kBAAI,aAAa;AACb,gBAAC,YAAwB,UAAU,KAAK;AAAA,kBACpC,QAAQ,KAAK;AAAA,kBACb,MAAM;AAAA,kBACN,QAAQ;AAAA,gBACZ,CAAC;AAAA,cACL;AAAA,YACJ;AAAA,UACJ;AAAA,QACJ;AAAA,MACJ;AAAA,IACJ;AAEA,QAAI,mBAAmB;AACnB,mBAAa,mBAAmB,cAAc,KAAK,CAAC;AAAA,IACxD;AAEA,aAAS,WAAW,SAAS,OAAO,GAAG;AACnC,WAAK,QAAQ,KAAK,OAAO;AAAA,IAC7B;AAEA,WAAO;AAAA,EACX;AACJ;",
6
+ "names": ["ref", "lines"]
7
7
  }
@@ -23,12 +23,21 @@ export declare const chapterHeadingSchema: z.ZodObject<{
23
23
  export declare const verseSchema: z.ZodObject<{
24
24
  type: z.ZodLiteral<"text">;
25
25
  id: z.ZodString;
26
+ bookCode: z.ZodOptional<z.ZodString>;
27
+ chapter: z.ZodOptional<z.ZodNumber>;
28
+ verse: z.ZodOptional<z.ZodNumber>;
26
29
  }, "strip", z.ZodTypeAny, {
27
30
  id: string;
28
31
  type: "text";
32
+ chapter?: number | undefined;
33
+ verse?: number | undefined;
34
+ bookCode?: string | undefined;
29
35
  }, {
30
36
  id: string;
31
37
  type: "text";
38
+ chapter?: number | undefined;
39
+ verse?: number | undefined;
40
+ bookCode?: string | undefined;
32
41
  }>;
33
42
  export declare const paratextSchema: z.ZodObject<{
34
43
  type: z.ZodLiteral<"paratext">;
@@ -62,12 +71,21 @@ export declare const metadataSchema: z.ZodDiscriminatedUnion<"type", [z.ZodObjec
62
71
  }>, z.ZodObject<{
63
72
  type: z.ZodLiteral<"text">;
64
73
  id: z.ZodString;
74
+ bookCode: z.ZodOptional<z.ZodString>;
75
+ chapter: z.ZodOptional<z.ZodNumber>;
76
+ verse: z.ZodOptional<z.ZodNumber>;
65
77
  }, "strip", z.ZodTypeAny, {
66
78
  id: string;
67
79
  type: "text";
80
+ chapter?: number | undefined;
81
+ verse?: number | undefined;
82
+ bookCode?: string | undefined;
68
83
  }, {
69
84
  id: string;
70
85
  type: "text";
86
+ chapter?: number | undefined;
87
+ verse?: number | undefined;
88
+ bookCode?: string | undefined;
71
89
  }>, z.ZodObject<{
72
90
  type: z.ZodLiteral<"paratext">;
73
91
  id: z.ZodString;
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@helloao/tools",
3
- "version": "0.0.17",
3
+ "version": "0.0.19",
4
4
  "description": "A set of tools for managing HelloAO's Free Bible API",
5
5
  "main": "./dist/cjs/index.cjs",
6
6
  "module": "./dist/esm/index.js",