USX implementations in node and web modules

node-usfm-parser/package.json

@@ -6,7 +6,7 @@
   "module": "./dist/es/index.mjs",
   "scripts": {
     "build": "parcel build ./src/index.js",
-    "test": "mocha --timeout 40000"
+    "test": "mocha --timeout 40000 --parallel"
   },
   "repository": {
     "type": "git",
@@ -27,13 +27,14 @@
   ],
   "dependencies": {
     "tree-sitter": "0.21.1",
-    "tree-sitter-usfm3": "file:../tree-sitter-usfm3"
+    "tree-sitter-usfm3": "file:../tree-sitter-usfm3",
+    "xmldom": "^0.6.0",
+    "xpath": "^0.0.34"
   },
   "devDependencies": {
     "ajv": "^8.17.1",
     "glob": "^11.0.0",
     "mocha": "^10.7.3",
-    "parcel": "^2.12.0",
-    "xml2js": "^0.6.2"
+    "parcel": "^2.12.0"
   }
 }

node-usfm-parser/src/usfmGenerator.js

@@ -1,4 +1,7 @@
 const { NO_USFM_USJ_TYPES, CLOSING_USJ_TYPES, NON_ATTRIB_USJ_KEYS, NO_NEWLINE_USJ_TYPES } = require("./utils/types");
+const { NON_ATTRIB_USX_KEYS, NO_NEWLINE_USX_TYPES } = require("./utils/types");
+const { DOMParser } = require('xmldom');
+
 class USFMGenerator {
   constructor() {
     this.usfmString = "";
@@ -73,6 +76,126 @@ class USFMGenerator {
     }
     return this.usfmString;
   }
+
+  usxToUsfm(xmlObj, nested=false) {
+        // Check if xmlObj is a string
+        // if (typeof xmlObj === 'string') {
+        //     // this.usfmString += xmlObj;
+        //     return;
+        // }
+
+        const objType = xmlObj.tagName;
+        let marker = null;
+        let usfmAttributes = [];
+
+        if (['verse', 'chapter'].includes(objType) && xmlObj.hasAttribute('eid')) {
+            return;
+        }
+
+        if (!NO_NEWLINE_USX_TYPES.includes(objType)) {
+            this.usfmString += '\n';
+        }
+
+        if (objType === 'optbreak') {
+            if (this.usfmString !== '' && !['\n', '\r', ' ', '\t'].includes(this.usfmString.slice(-1))) {
+                this.usfmString += ' ';
+            }
+            this.usfmString += '// ';
+        }
+
+        if (xmlObj.hasAttribute('style')) {
+            marker = xmlObj.getAttribute('style');
+            if (nested && objType === 'char' && !['xt', 'fv', 'ref'].includes(marker)) {
+                marker = `+${marker}`;
+            }
+            this.usfmString += `\\${marker} `;
+        } else if (objType === 'ref') {
+          marker = 'ref'
+          this.usfmString += `\\${marker} `;
+        }
+
+        if (xmlObj.hasAttribute('code')) {
+            this.usfmString += xmlObj.getAttribute('code');
+        }
+
+        if (xmlObj.hasAttribute('number')) {
+            this.usfmString += `${xmlObj.getAttribute('number')} `;
+        }
+
+        if (xmlObj.hasAttribute('caller')) {
+            this.usfmString += `${xmlObj.getAttribute('caller')} `;
+        }
+
+        if (xmlObj.hasAttribute('altnumber')) {
+            if (objType === 'verse') {
+                this.usfmString += `\\va ${xmlObj.getAttribute('altnumber')}\\va*`;
+            } else if (objType === 'chapter') {
+                this.usfmString += `\n\\ca ${xmlObj.getAttribute('altnumber')}\\ca*`;
+            }
+        }
+
+        if (xmlObj.hasAttribute('pubnumber')) {
+            if (objType === 'verse') {
+                this.usfmString += `\\vp ${xmlObj.getAttribute('pubnumber')}\\vp*`;
+            } else if (objType === 'chapter') {
+                this.usfmString += `\n\\cp ${xmlObj.getAttribute('pubnumber')}`;
+            }
+        }
+
+        if (xmlObj.hasAttribute('category')) {
+            this.usfmString += `\n\\cat ${xmlObj.getAttribute('category')} \\cat*`;
+        }
+
+        const children = Array.from(xmlObj.childNodes);
+        for (const child of children) {
+            if (child.nodeType === 1) { // Check if child is an element node
+                if (objType === 'char') {
+                    this.usxToUsfm(child, true);
+                } else {
+                    this.usxToUsfm(child, false);
+                }
+            }
+            if (child.nodeType === 3 && child.nodeValue.trim()) { // Check if child is a text node with content
+                if (this.usfmString !== '' && !['\n', '\r', ' ', '\t'].includes(this.usfmString.slice(-1))) {
+                    this.usfmString += ' ';
+                }
+                this.usfmString += child.nodeValue.trim();
+            }
+        }
+
+        const attributes = Array.from(xmlObj.attributes);
+        for (const attrNode of attributes) {
+            let key = attrNode.name;
+            let val = attrNode.value.replace(/"/g, '');
+            if (key === 'file' && objType === 'figure') {
+                usfmAttributes.push(`src="${val}"`);
+            } else if (!NON_ATTRIB_USX_KEYS.includes(key)) {
+                usfmAttributes.push(`${key}="${val}"`);
+            }
+            if (['sid', 'eid'].includes(key) && objType === 'ms') {
+                usfmAttributes.push(`${key}="${val}"`);
+            }
+        }
+
+        if (usfmAttributes.length > 0) {
+            this.usfmString += '|';
+            this.usfmString += usfmAttributes.join(' ');
+        }
+
+        if ((xmlObj.hasAttribute('closed') && xmlObj.getAttribute('closed') === 'true')
+            || CLOSING_USJ_TYPES.includes(objType)
+            || usfmAttributes.length > 0) {
+            if (objType === 'ms') {
+                this.usfmString += '\\*';
+            } else {
+                this.usfmString += `\\${marker}*`;
+            }
+        }
+
+        if (objType === 'sidebar') {
+            this.usfmString += '\n\\esbe\n';
+        }    
+  }
 }
 
 exports.USFMGenerator = USFMGenerator;

node-usfm-parser/src/usfmParser.js

@@ -1,8 +1,10 @@
 const Parser = require('tree-sitter');
+const assert = require('assert');
 
 const {USFMGenerator} = require("./usfmGenerator");
 const {USJGenerator} = require("./usjGenerator"); 
 const {ListGenerator} = require("./listGenerator");
+const {USXGenerator} = require("./usxGenerator")
 const { includeMarkersInUsj, excludeMarkersInUsj, Filter } = require("./filters.js");
 const USFM3 = require('tree-sitter-usfm3');
 const { Query } = Parser;
@@ -39,7 +41,7 @@ Only one of USFM, USJ or USX is supported in one object.`)
         	this.usfm = this.convertUSJToUSFM()
         } else if (fromUsx !== null) {
         	this.usx = fromUsx;
-        	// this.usfm = this.convertUSXToUSFM()
+        	this.usfm = this.convertUSXToUSFM()
         }
 		this.parser = null;
 		this.initializeParser();
@@ -48,6 +50,7 @@ Only one of USFM, USJ or USX is supported in one object.`)
 		this.errors = [];
         this.warnings = [];
         this.parseUSFM();
+
 	}
 	initializeParser() {
 		this.parser = new Parser();
@@ -135,6 +138,33 @@ Only one of USFM, USJ or USX is supported in one object.`)
 		return outputUSFM;
 	}
 
+	convertUSXToUSFM() {
+		try {
+			assert(1 <= this.usx.nodeType && this.usx.nodeType <= 12 ,
+		        'Input must be an instance of xmldom Document or Element'
+		    );
+			if (this.usx.tagName !== "usx") {
+				assert(this.usx.getElementsByTagName('usx').length === 1,
+					'Expects a <usx> node. Refer docs: https://docs.usfm.bible/usfm/3.1/syntax.html#_usx_usfm_xml');
+
+				this.usx = this.usx.getElementsByTagName('usx')[0]
+			}
+			// assert(this.usx.childNodes[0].tagName === 'book', "<book> expected as first element in <usx>")
+
+		} catch(err) {
+			throw new Error("USX not in expected format. "+err.message)
+		}
+		try {
+			const usfmGen = new USFMGenerator()
+			usfmGen.usxToUsfm(this.usx);
+			// console.log(usfmGen.usfmString)
+			return usfmGen.usfmString;
+		} catch(err) {
+	        let message = "Unable to do the conversion from USX to USFM. ";
+	        throw new Error(message, { cause: err });
+		}
+	}
+
 	convertUSFMToUSJ(
 		excludeMarkers = null,
 		includeMarkers = null,
@@ -187,9 +217,9 @@ Only one of USFM, USJ or USX is supported in one object.`)
 	    /* Uses the toJSON function and converts JSON to CSV
 	       To be re-implemented to work with the flat JSON schema */
 
-	    if (!ignoreErrors && this.errors && this.errors.length > 0) {
-	        const errStr = this.errors.map(err => err.join(":")).join("\n\t");
-	        throw new Error(`Errors present:\n\t${errStr}\nUse ignoreErrors=true to generate output despite errors`);
+	    if (!ignoreErrors && this.errors.length > 0) {
+			let errorString = this.errors.join("\n\t");
+	        throw new Error(`Errors present:\n\t${errorString}\nUse ignoreErrors=true to generate output despite errors`);
 	    }
 
 	    try {
@@ -201,15 +231,48 @@ Only one of USFM, USJ or USX is supported in one object.`)
 
 	    } catch (exe) {
 	        let message = "Unable to do the conversion. ";
-	        if (this.errors && this.errors.length > 0) {
-	            const errStr = this.errors.map(err => err.join(":")).join("\n\t");
-	            message += `Could be due to an error in the USFM\n\t${errStr}`;
+	        if (this.errors.length > 0) {
+				let errorString = this.errors.join("\n\t");
+	            message += `Could be due to an error in the USFM\n\t${errorString}`;
+	        }
+	        throw new Error(message, { cause: exe });
+	    }
+
+	}
+
+	toUSX(ignoreErrors = false) {
+	    /* Convert the syntax_tree to the XML format (USX) */
+
+	    if (!ignoreErrors && this.errors.length > 0) {
+			let errorString = this.errors.join("\n\t");
+	        throw new Error(`Errors present:\n\t${errorString}\nUse ignoreErrors=true to generate output despite errors`);
+	    }
+	    let xmlContent = null;
+
+	    try {
+	        // Initialize the USX generator (assuming the constructor is already implemented in JS)
+	        const usxGenerator = new USXGenerator(USFM3,
+													this.usfm);
+
+	        // Process the syntax tree and convert to USX format
+	        usxGenerator.node2Usx(this.syntaxTree, usxGenerator.xmlRootNode);
+
+	        // xmlContent = usxSerializer.serializeToString(usxGenerator.xmlRootNode);
+	        xmlContent = usxGenerator.xmlRootNode;
+	    } catch (exe) {
+	        let message = "Unable to do the conversion. ";
+	        if (this.errors.length > 0) {
+				let errorString = this.errors.join("\n\t");
+	            message += `Could be due to an error in the USFM\n\t${errorString}`;
 	        }
 	        throw new Error(message, { cause: exe });
 	    }
 
+	    // Return the generated XML structure (in JSON format)
+	    return xmlContent;
 	}
 
+
 }
 
 

node-usfm-parser/src/usjGenerator.js

@@ -418,21 +418,21 @@ class USJGenerator {
     let style = this.usfm.substring(tagNode.startIndex, tagNode.endIndex);
     if (style.startsWith("\\")) {
       style = style.replace("\\", "").trim();
-    } else {
-      style = node.type;
+    // } else {
+    //   style = node.type;
     }
 
     // console.log(node.children.length, node.children[0].type, node.children[1].type)
     let childrenRangeStart = 1;
-    if (
-      node.children.length > 1 &&
-      node.children[1].type.startsWith("numbered")
-    ) {
-      const numNode = node.children[1];
-      const num = this.usfm.substring(numNode.startIndex, numNode.endIndex);
-      style += num;
-      childrenRangeStart = 2;
-    }
+    // if (
+    //   node.children.length > 1 &&
+    //   node.children[1].type.startsWith("numbered")
+    // ) {
+    //   const numNode = node.children[1];
+    //   const num = this.usfm.substring(numNode.startIndex, numNode.endIndex);
+    //   style += num;
+    //   childrenRangeStart = 2;
+    // }
     const paraJsonObj = { type: "para", marker: style, content: [] };
     parentJsonObj.content.push(paraJsonObj);